Skip to contentSkip to conversation

ath10k: Check tx_stats before use it

PATCH In review Sasha Levin 8 participants 268 messages Jul 15, 2019 – Jul 24, 2019 Page 1 of 2

Subsystems (41)
lore
Share & tools
lore

Sasha Levin

author maintainer
#0 lore raw

PATCH 1/249 ath10k: Check tx_stats before use it

From: Yingying Tang <[email protected]>

[ Upstream commit 9e7251fa38978b85108c44743e1436d48e8d0d76 ]

tx_stats will be freed and set to NULL before debugfs_sta node is removed in station disconnetion process. So if read the debugfs_sta node there may be NULL pointer error. Add check for tx_stats before use it to resove this issue.

---
 drivers/net/wireless/ath/ath10k/debugfs_sta.c | 7 +++++++
 1 file changed, 7 insertions(+)
drivers/net/wireless/ath/ath10k/debugfs_sta.c +7 -0
index c704ae371c4d..42931a669b02 100644
@@ -663,6 +663,13 @@ static ssize_t ath10k_dbg_sta_dump_tx_stats(struct file *file,
 
 	mutex_lock(&ar->conf_mutex);
 
+	if (!arsta->tx_stats) {
+		ath10k_warn(ar, "failed to get tx stats");
+		mutex_unlock(&ar->conf_mutex);
+		kfree(buf);
+		return 0;
+	}
+
 	spin_lock_bh(&ar->data_lock);
 	for (k = 0; k < ATH10K_STATS_TYPE_MAX; k++) {
 		for (j = 0; j < ATH10K_COUNTER_TYPE_MAX; j++) {
Signature · 7 lines
2.20.1


_______________________________________________
ath10k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath10k
author maintainer
#1 lore raw

PATCH 2/249 ath10k: htt: don't use txdone_fifo with SDIO

From: Alagu Sankar <[email protected]>

[ Upstream commit e2a6b711282a371c5153239e0468a48254f17ca6 ]

HTT High Latency (ATH10K_DEV_TYPE_HL) does not use txdone_fifo at all, we don't even initialise it by skipping ath10k_htt_tx_alloc_buf() in ath10k_htt_tx_start(). Because of this using QCA6174 SDIO ath10k_htt_rx_tx_compl_ind() will crash when it accesses unitialised txdone_fifo. So skip txdone_fifo when using High Latency mode.

Tested with QCA6174 SDIO with firmware WLAN.RMH.4.4.1-00007-QCARMSWP-1.

---
 drivers/net/wireless/ath/ath10k/htt_rx.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/net/wireless/ath/ath10k/htt_rx.c +3 -1
index 1acc622d2183..f22840bbc389 100644
@@ -2277,7 +2277,9 @@ static void ath10k_htt_rx_tx_compl_ind(struct ath10k *ar,
 		 *  Note that with only one concurrent reader and one concurrent
 		 *  writer, you don't need extra locking to use these macro.
 		 */
-		if (!kfifo_put(&htt->txdone_fifo, tx_done)) {
+		if (ar->bus_param.dev_type == ATH10K_DEV_TYPE_HL) {
+			ath10k_txrx_tx_unref(htt, &tx_done);
+		} else if (!kfifo_put(&htt->txdone_fifo, tx_done)) {
 			ath10k_warn(ar, "txdone fifo overrun, msdu_id %d status %d\n",
 				    tx_done.msdu_id, tx_done.status);
 			ath10k_txrx_tx_unref(htt, &tx_done);
Signature · 7 lines
2.20.1


_______________________________________________
ath10k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath10k
author maintainer
#2 lore raw

PATCH 3/249 ath10k: fix incorrect multicast/broadcast rate setting

In reply to Sasha Levin

From: Pradeep kumar Chitrapu <[email protected]>

[ Upstream commit 93ee3d108fc77e19efeac3ec5aa7d5886711bfef ]

Invalid rate code is sent to firmware when multicast rate value of 0 is sent to driver indicating disabled case, causing broken mesh path. so fix that.

Tested on QCA9984 with firmware 10.4-3.6.1-00827

Sven tested on IPQ4019 with 10.4-3.5.3-00057 and QCA9888 with 10.4-3.5.3-00053 (ath10k-firmware) and 10.4-3.6-00140 (linux-firmware 2018-12-16-211de167).

Fixes cd93b83ad92 ("ath10k: support for multicast rate control")
---
 drivers/net/wireless/ath/ath10k/mac.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)
drivers/net/wireless/ath/ath10k/mac.c +7 -3
index 9c703d287333..e8997e22ceec 100644
@@ -5588,8 +5588,8 @@ static void ath10k_bss_info_changed(struct ieee80211_hw *hw,
 	struct cfg80211_chan_def def;
 	u32 vdev_param, pdev_param, slottime, preamble;
 	u16 bitrate, hw_value;
-	u8 rate, basic_rate_idx;
-	int rateidx, ret = 0, hw_rate_code;
+	u8 rate, basic_rate_idx, rateidx;
+	int ret = 0, hw_rate_code, mcast_rate;
 	enum nl80211_band band;
 	const struct ieee80211_supported_band *sband;
 
@@ -5776,7 +5776,11 @@ static void ath10k_bss_info_changed(struct ieee80211_hw *hw,
 	if (changed & BSS_CHANGED_MCAST_RATE &&
 	    !ath10k_mac_vif_chan(arvif->vif, &def)) {
 		band = def.chan->band;
-		rateidx = vif->bss_conf.mcast_rate[band] - 1;
+		mcast_rate = vif->bss_conf.mcast_rate[band];
+		if (mcast_rate > 0)
+			rateidx = mcast_rate - 1;
+		else
+			rateidx = ffs(vif->bss_conf.basic_rates) - 1;
 
 		if (ar->phy_capability & WHAL_WLAN_11A_CAPABILITY)
 			rateidx += ATH10K_MAC_FIRST_OFDM_RATE_IDX;
Signature · 7 lines
2.20.1


_______________________________________________
ath10k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath10k
author maintainer
#3 lore raw

PATCH 4/249 ath9k: Don't trust TX status TID number when reporting airtime

In reply to Sasha Levin

From: Toke Høiland-Jørgensen <[email protected]>

[ Upstream commit 389b72e58259336c2d56d58b660b79cf4b9e0dcb ]

As already noted a comment in ath_tx_complete_aggr(), the hardware will occasionally send a TX status with the wrong tid number. If we trust the value, airtime usage will be reported to the wrong AC, which can cause the deficit on that AC to become very low, blocking subsequent attempts to transmit.

To fix this, account airtime usage to the TID number from the original skb, instead of the one in the hardware TX status report.

---
 drivers/net/wireless/ath/ath9k/xmit.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
drivers/net/wireless/ath/ath9k/xmit.c +4 -3
index b17e1ca40995..3be0aeedb9b5 100644
@@ -668,7 +668,8 @@ static bool bf_is_ampdu_not_probing(struct ath_buf *bf)
 static void ath_tx_count_airtime(struct ath_softc *sc,
 				 struct ieee80211_sta *sta,
 				 struct ath_buf *bf,
-				 struct ath_tx_status *ts)
+				 struct ath_tx_status *ts,
+				 u8 tid)
 {
 	u32 airtime = 0;
 	int i;
@@ -679,7 +680,7 @@ static void ath_tx_count_airtime(struct ath_softc *sc,
 		airtime += rate_dur * bf->rates[i].count;
 	}
 
-	ieee80211_sta_register_airtime(sta, ts->tid, airtime, 0);
+	ieee80211_sta_register_airtime(sta, tid, airtime, 0);
 }
 
 static void ath_tx_process_buffer(struct ath_softc *sc, struct ath_txq *txq,
@@ -709,7 +710,7 @@ static void ath_tx_process_buffer(struct ath_softc *sc, struct ath_txq *txq,
 	if (sta) {
 		struct ath_node *an = (struct ath_node *)sta->drv_priv;
 		tid = ath_get_skb_tid(sc, an, bf->bf_mpdu);
-		ath_tx_count_airtime(sc, sta, bf, ts);
+		ath_tx_count_airtime(sc, sta, bf, ts, tid->tidno);
 		if (ts->ts_status & (ATH9K_TXERR_FILT | ATH9K_TXERR_XRETRY))
 			tid->clear_ps_filter = true;
 	}
Signature · 1 line
2.20.1
author maintainer
#4 lore raw

PATCH 5/249 wil6210: fix potential out-of-bounds read

In reply to Sasha Levin

From: "Gustavo A. R. Silva" <[email protected]>

[ Upstream commit bfabdd6997323adbedccb13a3fed1967fb8cf8f5 ]

Notice that *rc* can evaluate to up to 5, include/linux/netdevice.h:

enum gro_result {
GRO_MERGED,
GRO_MERGED_FREE,
GRO_HELD,
GRO_NORMAL,
GRO_DROP,
GRO_CONSUMED,
};
typedef enum gro_result gro_result_t;

In case *rc* evaluates to 5, we end up having an out-of-bounds read at drivers/net/wireless/ath/wil6210/txrx.c:821:

	wil_dbg_txrx(wil, "Rx complete %d bytes => %s\n",
		     len, gro_res_str[rc]);

Fix this by adding element "GRO_CONSUMED" to array gro_res_str.

Addresses-Coverity-ID: 1444666 ("Out-of-bounds read")

Fixes 194b482b5055 ("wil6210: Debug print GRO Rx result")
---
 drivers/net/wireless/ath/wil6210/txrx.c | 1 +
 1 file changed, 1 insertion(+)
drivers/net/wireless/ath/wil6210/txrx.c +1 -0
index 4ccfd1404458..d74837cce67f 100644
@@ -750,6 +750,7 @@ void wil_netif_rx_any(struct sk_buff *skb, struct net_device *ndev)
 		[GRO_HELD]		= "GRO_HELD",
 		[GRO_NORMAL]		= "GRO_NORMAL",
 		[GRO_DROP]		= "GRO_DROP",
+		[GRO_CONSUMED]		= "GRO_CONSUMED",
 	};
 
 	wil->txrx_ops.get_netif_rx_params(skb, &cid, &security);
Signature · 1 line
2.20.1
author maintainer
#5 lore raw

PATCH 6/249 ath10k: Do not send probe response template for mesh

In reply to Sasha Levin

From: Surabhi Vishnoi <[email protected]>

[ Upstream commit 97354f2c432788e3163134df6bb144f4b6289d87 ]

Currently mac80211 do not support probe response template for mesh point. When WMI_SERVICE_BEACON_OFFLOAD is enabled, host driver tries to configure probe response template for mesh, but it fails because the interface type is not NL80211_IFTYPE_AP but NL80211_IFTYPE_MESH_POINT.

To avoid this failure, skip sending probe response template to firmware for mesh point.

Tested HW: WCN3990/QCA6174/QCA9984

---
 drivers/net/wireless/ath/ath10k/mac.c | 4 ++++
 1 file changed, 4 insertions(+)
drivers/net/wireless/ath/ath10k/mac.c +4 -0
index e8997e22ceec..b500fd427595 100644
@@ -1630,6 +1630,10 @@ static int ath10k_mac_setup_prb_tmpl(struct ath10k_vif *arvif)
 	if (arvif->vdev_type != WMI_VDEV_TYPE_AP)
 		return 0;
 
+	 /* For mesh, probe response and beacon share the same template */
+	if (ieee80211_vif_is_mesh(vif))
+		return 0;
+
 	prb = ieee80211_proberesp_get(hw, vif);
 	if (!prb) {
 		ath10k_warn(ar, "failed to get probe resp template from mac80211\n");
Signature · 1 line
2.20.1
author maintainer
#6 lore raw

PATCH 7/249 spi: rockchip: turn down tx dma bursts

In reply to Sasha Levin

From: Emil Renner Berthing <[email protected]>

[ Upstream commit 47300728fb213486a830565d2af49da967c9d16a ]

This fixes tx and bi-directional dma transfers on rk3399-gru-kevin.

It seems the SPI fifo must have room for 2 bursts when the dma_tx_req signal is generated or it might skip some words. This in turn makes the rx dma channel never complete for bi-directional transfers.

Fix it by setting tx burst length to fifo_len / 4 and the dma watermark to fifo_len / 2.

However the rk3399 TRM says (sic):
"DMAC support incrementing-address burst and fixed-address burst. But in the case of access SPI and UART at byte or halfword size, DMAC only support fixed-address burst and the address must be aligned to word."

So this relies on fifo_len being a multiple of 16 such that the burst length (= fifo_len / 4) is a multiple of 4 and the addresses will be word-aligned.

Fixes dcfc861d24ec ("spi: rockchip: adjust dma watermark and burstlen")
---
 drivers/spi/spi-rockchip.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/spi/spi-rockchip.c +2 -2
index 9b91188a85f9..2cc6d9951b52 100644
@@ -417,7 +417,7 @@ static int rockchip_spi_prepare_dma(struct rockchip_spi *rs,
 			.direction = DMA_MEM_TO_DEV,
 			.dst_addr = rs->dma_addr_tx,
 			.dst_addr_width = rs->n_bytes,
-			.dst_maxburst = rs->fifo_len / 2,
+			.dst_maxburst = rs->fifo_len / 4,
 		};
 
 		dmaengine_slave_config(master->dma_tx, &txconf);
@@ -518,7 +518,7 @@ static void rockchip_spi_config(struct rockchip_spi *rs,
 	else
 		writel_relaxed(rs->fifo_len / 2 - 1, rs->regs + ROCKCHIP_SPI_RXFTLR);
 
-	writel_relaxed(rs->fifo_len / 2 - 1, rs->regs + ROCKCHIP_SPI_DMATDLR);
+	writel_relaxed(rs->fifo_len / 2, rs->regs + ROCKCHIP_SPI_DMATDLR);
 	writel_relaxed(0, rs->regs + ROCKCHIP_SPI_DMARDLR);
 	writel_relaxed(dmacr, rs->regs + ROCKCHIP_SPI_DMACR);
 
Signature · 1 line
2.20.1
author maintainer
#7 lore raw

PATCH 8/249 ath9k: Check for errors when reading SREV register

In reply to Sasha Levin

From: Tim Schumacher <[email protected]>

[ Upstream commit 2f90c7e5d09437a4d8d5546feaae9f1cf48cfbe1 ]

Right now, if an error is encountered during the SREV register read (i.e. an EIO in ath9k_regread()), that error code gets passed all the way to __ath9k_hw_init(), where it is visible during the "Chip rev not supported" message.

ath9k_htc 1-1.4:1.0: ath9k_htc: HTC initialized with 33 credits
ath: phy2: Mac Chip Rev 0x0f.3 is not supported by this driver
ath: phy2: Unable to initialize hardware; initialization status: -95
ath: phy2: Unable to initialize hardware; initialization status: -95
ath9k_htc: Failed to initialize the device

Check for -EIO explicitly in ath9k_hw_read_revisions() and return a boolean based on the success of the operation. Check for that in __ath9k_hw_init() and abort with a more debugging-friendly message if reading the revisions wasn't successful.

ath9k_htc 1-1.4:1.0: ath9k_htc: HTC initialized with 33 credits
ath: phy2: Failed to read SREV register
ath: phy2: Could not read hardware revision
ath: phy2: Unable to initialize hardware; initialization status: -95
ath: phy2: Unable to initialize hardware; initialization status: -95
ath9k_htc: Failed to initialize the device

This helps when debugging by directly showing the first point of failure and it could prevent possible errors if a 0x0f.3 revision is ever supported.

---
 drivers/net/wireless/ath/ath9k/hw.c | 32 +++++++++++++++++++++--------
 1 file changed, 23 insertions(+), 9 deletions(-)
drivers/net/wireless/ath/ath9k/hw.c +23 -9
index 8581d917635a..b6773d613f0c 100644
@@ -252,8 +252,9 @@ void ath9k_hw_get_channel_centers(struct ath_hw *ah,
 /* Chip Revisions */
 /******************/
 
-static void ath9k_hw_read_revisions(struct ath_hw *ah)
+static bool ath9k_hw_read_revisions(struct ath_hw *ah)
 {
+	u32 srev;
 	u32 val;
 
 	if (ah->get_mac_revision)
@@ -269,25 +270,33 @@ static void ath9k_hw_read_revisions(struct ath_hw *ah)
 			val = REG_READ(ah, AR_SREV);
 			ah->hw_version.macRev = MS(val, AR_SREV_REVISION2);
 		}
-		return;
+		return true;
 	case AR9300_DEVID_AR9340:
 		ah->hw_version.macVersion = AR_SREV_VERSION_9340;
-		return;
+		return true;
 	case AR9300_DEVID_QCA955X:
 		ah->hw_version.macVersion = AR_SREV_VERSION_9550;
-		return;
+		return true;
 	case AR9300_DEVID_AR953X:
 		ah->hw_version.macVersion = AR_SREV_VERSION_9531;
-		return;
+		return true;
 	case AR9300_DEVID_QCA956X:
 		ah->hw_version.macVersion = AR_SREV_VERSION_9561;
-		return;
+		return true;
 	}
 
-	val = REG_READ(ah, AR_SREV) & AR_SREV_ID;
+	srev = REG_READ(ah, AR_SREV);
+
+	if (srev == -EIO) {
+		ath_err(ath9k_hw_common(ah),
+			"Failed to read SREV register");
+		return false;
+	}
+
+	val = srev & AR_SREV_ID;
 
 	if (val == 0xFF) {
-		val = REG_READ(ah, AR_SREV);
+		val = srev;
 		ah->hw_version.macVersion =
 			(val & AR_SREV_VERSION2) >> AR_SREV_TYPE2_S;
 		ah->hw_version.macRev = MS(val, AR_SREV_REVISION2);
@@ -306,6 +315,8 @@ static void ath9k_hw_read_revisions(struct ath_hw *ah)
 		if (ah->hw_version.macVersion == AR_SREV_VERSION_5416_PCIE)
 			ah->is_pciexpress = true;
 	}
+
+	return true;
 }
 
 /************************************/
@@ -559,7 +570,10 @@ static int __ath9k_hw_init(struct ath_hw *ah)
 	struct ath_common *common = ath9k_hw_common(ah);
 	int r = 0;
 
-	ath9k_hw_read_revisions(ah);
+	if (!ath9k_hw_read_revisions(ah)) {
+		ath_err(common, "Could not read hardware revisions");
+		return -EOPNOTSUPP;
+	}
 
 	switch (ah->hw_version.macVersion) {
 	case AR_SREV_VERSION_5416_PCI:
Signature · 1 line
2.20.1
author maintainer
#8 lore raw

PATCH 9/249 ath10k: Fix the wrong value of enums for wmi tlv stats id

In reply to Sasha Levin

From: Surabhi Vishnoi <[email protected]>

[ Upstream commit 9280f4fc06f44d0b4dc9e831f72d97b3d7cd35d3 ]

The enum value for WMI_TLV_STAT_PDEV, WMI_TLV_STAT_VDEV and WMI_TLV_STAT_PEER is wrong, due to which the vdev stats are not received from firmware in wmi_update_stats event.

Fix the enum values for above stats to receive all stats from firmware in WMI_TLV_UPDATE_STATS_EVENTID.

Tested HW: WCN3990
Tested FW: WLAN.HL.3.1-00784-QCAHLSWMTPLZ-1

Fixes f40a307eb92c ("ath10k: Fill rx duration for each peer in fw_stats for WCN3990)
---
 drivers/net/wireless/ath/ath10k/wmi.h | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
drivers/net/wireless/ath/ath10k/wmi.h +4 -3
index e1c40bb69932..12f57f9adbba 100644
@@ -4535,9 +4535,10 @@ enum wmi_10_4_stats_id {
 };
 
 enum wmi_tlv_stats_id {
-	WMI_TLV_STAT_PDEV	= BIT(0),
-	WMI_TLV_STAT_VDEV	= BIT(1),
-	WMI_TLV_STAT_PEER	= BIT(2),
+	WMI_TLV_STAT_PEER	= BIT(0),
+	WMI_TLV_STAT_AP		= BIT(1),
+	WMI_TLV_STAT_PDEV	= BIT(2),
+	WMI_TLV_STAT_VDEV	= BIT(3),
 	WMI_TLV_STAT_PEER_EXTD  = BIT(10),
 };
 
Signature · 7 lines
2.20.1


_______________________________________________
ath10k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath10k
author maintainer
#9 lore raw

PATCH 10/249 wil6210: fix missed MISC mbox interrupt

In reply to Sasha Levin

From: Maya Erez <[email protected]>

[ Upstream commit 7441be71ba7e07791fd4fa2b07c932dff14ff4d9 ]

When MISC interrupt is triggered due to HALP bit, in parallel to mbox events handling by the MISC threaded IRQ, new mbox interrupt can be missed in the following scenario:
1. MISC ICR is read in the IRQ handler
2. Threaded IRQ is completed and all MISC interrupts are unmasked
3. mbox interrupt is set by FW
4. HALP is masked
The mbox interrupt in step 3 can be missed due to constant high level of ICM.
Masking all MISC IRQs instead of masking only HALP bit in step 4 will guarantee that ICM will drop to 0 and interrupt will be triggered once MISC interrupts will be unmasked.

---
 drivers/net/wireless/ath/wil6210/interrupt.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/wireless/ath/wil6210/interrupt.c +1 -1
index 3f5bd177d55f..e41ba24011d8 100644
@@ -580,7 +580,7 @@ static irqreturn_t wil6210_irq_misc(int irq, void *cookie)
 			/* no need to handle HALP ICRs until next vote */
 			wil->halp.handle_icr = false;
 			wil_dbg_irq(wil, "irq_misc: HALP IRQ invoked\n");
-			wil6210_mask_halp(wil);
+			wil6210_mask_irq_misc(wil, true);
 			complete(&wil->halp.comp);
 		}
 	}
Signature · 1 line
2.20.1
author maintainer
#10 lore raw

PATCH 11/249 ath6kl: add some bounds checking

In reply to Sasha Levin

From: Dan Carpenter <[email protected]>

[ Upstream commit 5d6751eaff672ea77642e74e92e6c0ac7f9709ab ]

The "ev->traffic_class" and "reply->ac" variables come from the network and they're used as an offset into the wmi->stream_exist_for_ac[] array. Those variables are u8 so they can be 0-255 but the stream_exist_for_ac[] array only has WMM_NUM_AC (4) elements. We need to add a couple bounds checks to prevent array overflows.

I also modified one existing check from "if (traffic_class > 3) {" to "if (traffic_class >= WMM_NUM_AC) {" just to make them all consistent.

Fixes bdcd81707973 (" Add ath6kl cleaned up driver")
---
 drivers/net/wireless/ath/ath6kl/wmi.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)
drivers/net/wireless/ath/ath6kl/wmi.c +9 -1
index 68854c45d0a4..9ab6aa9ded5c 100644
@@ -1176,6 +1176,10 @@ static int ath6kl_wmi_pstream_timeout_event_rx(struct wmi *wmi, u8 *datap,
 		return -EINVAL;
 
 	ev = (struct wmi_pstream_timeout_event *) datap;
+	if (ev->traffic_class >= WMM_NUM_AC) {
+		ath6kl_err("invalid traffic class: %d\n", ev->traffic_class);
+		return -EINVAL;
+	}
 
 	/*
 	 * When the pstream (fat pipe == AC) timesout, it means there were
@@ -1517,6 +1521,10 @@ static int ath6kl_wmi_cac_event_rx(struct wmi *wmi, u8 *datap, int len,
 		return -EINVAL;
 
 	reply = (struct wmi_cac_event *) datap;
+	if (reply->ac >= WMM_NUM_AC) {
+		ath6kl_err("invalid AC: %d\n", reply->ac);
+		return -EINVAL;
+	}
 
 	if ((reply->cac_indication == CAC_INDICATION_ADMISSION_RESP) &&
 	    (reply->status_code != IEEE80211_TSPEC_STATUS_ADMISS_ACCEPTED)) {
@@ -2633,7 +2641,7 @@ int ath6kl_wmi_delete_pstream_cmd(struct wmi *wmi, u8 if_idx, u8 traffic_class,
 	u16 active_tsids = 0;
 	int ret;
 
-	if (traffic_class > 3) {
+	if (traffic_class >= WMM_NUM_AC) {
 		ath6kl_err("invalid traffic class: %d\n", traffic_class);
 		return -EINVAL;
 	}
Signature · 1 line
2.20.1
author maintainer
#11 lore raw

PATCH 12/249 ath10k: add peer id check in ath10k_peer_find_by_id

In reply to Sasha Levin

From: Wen Gong <[email protected]>

[ Upstream commit 49ed34b835e231aa941257394716bc689bc98d9f ]

For some SDIO chip, the peer id is 65535 for MPDU with error status, then test_bit will trigger buffer overflow for peer's memory, if kasan enabled, it will report error.

Reason is when station is in disconnecting status, firmware do not delete the peer info since it not disconnected completely, meanwhile some AP will still send data packet to station, then hardware will receive the packet and send to firmware, firmware's logic will report peer id of 65535 for MPDU with error status.

Add check for overflow the size of peer's peer_ids will avoid the buffer overflow access.

Call trace of kasan:

dump_backtrace+0x0/0x2ec
show_stack+0x20/0x2c
__dump_stack+0x20/0x28
dump_stack+0xc8/0xec
print_address_description+0x74/0x240
kasan_report+0x250/0x26c
__asan_report_load8_noabort+0x20/0x2c
ath10k_peer_find_by_id+0x180/0x1e4 [ath10k_core]
ath10k_htt_t2h_msg_handler+0x100c/0x2fd4 [ath10k_core]
ath10k_htt_htc_t2h_msg_handler+0x20/0x34 [ath10k_core]
ath10k_sdio_irq_handler+0xcc8/0x1678 [ath10k_sdio]
process_sdio_pending_irqs+0xec/0x370
sdio_run_irqs+0x68/0xe4
sdio_irq_work+0x1c/0x28
process_one_work+0x3d8/0x8b0
worker_thread+0x508/0x7cc
kthread+0x24c/0x264
ret_from_fork+0x10/0x18

Tested with QCA6174 SDIO with firmware
WLAN.RMH.4.4.1-00007-QCARMSWP-1.

---
 drivers/net/wireless/ath/ath10k/txrx.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/net/wireless/ath/ath10k/txrx.c +3 -0
index c5818d28f55a..4102df016931 100644
@@ -150,6 +150,9 @@ struct ath10k_peer *ath10k_peer_find_by_id(struct ath10k *ar, int peer_id)
 {
 	struct ath10k_peer *peer;
 
+	if (peer_id >= BITS_PER_TYPE(peer->peer_ids))
+		return NULL;
+
 	lockdep_assert_held(&ar->data_lock);
 
 	list_for_each_entry(peer, &ar->peers, list)
Signature · 7 lines
2.20.1


_______________________________________________
ath10k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath10k
author maintainer
#12 lore raw

PATCH 13/249 x86/tsc: Use CPUID.0x16 to calculate missing crystal frequency

In reply to Sasha Levin

From: Daniel Drake <[email protected]>

[ Upstream commit 604dc9170f2435d27da5039a3efd757dceadc684 ]

native_calibrate_tsc() had a data mapping Intel CPU families and crystal clock speed, but hardcoded tables are not ideal, and this approach was already problematic at least in the Skylake X case, as seen in commit:

b51120309348 ("x86/tsc: Fix erroneous TSC rate on Skylake Xeon")

By examining CPUID data from http://instlatx64.atw.hu/ and units in the lab, we have found that 3 different scenarios need to be dealt with, and we can eliminate most of the hardcoded data using an approach a little more advanced than before:

1. ApolloLake, GeminiLake, CannonLake (and presumably all new chipsets from this point) report the crystal frequency directly via CPUID.0x15. That's definitive data that we can rely upon.

2. Skylake, Kabylake and all variants of those two chipsets report a crystal frequency of zero, however we can calculate the crystal clock speed by condidering data from CPUID.0x16.

This method correctly distinguishes between the two crystal clock frequencies present on different Skylake X variants that caused headaches before.

As the calculations do not quite match the previously-hardcoded values in some cases (e.g. 23913043Hz instead of 24MHz), TSC refinement is enabled on all platforms where we had to calculate the crystal frequency in this way.

3. Denverton (GOLDMONT_X) reports a crystal frequency of zero and does

    not support CPUID.0x16, so we leave this entry hardcoded.
---
 arch/x86/kernel/tsc.c | 47 +++++++++++++++++++++++++------------------
 1 file changed, 27 insertions(+), 20 deletions(-)
arch/x86/kernel/tsc.c +27 -20
index 0b29e58f288e..da029fd21460 100644
@@ -632,31 +632,38 @@ unsigned long native_calibrate_tsc(void)
 
 	crystal_khz = ecx_hz / 1000;
 
-	if (crystal_khz == 0) {
-		switch (boot_cpu_data.x86_model) {
-		case INTEL_FAM6_SKYLAKE_MOBILE:
-		case INTEL_FAM6_SKYLAKE_DESKTOP:
-		case INTEL_FAM6_KABYLAKE_MOBILE:
-		case INTEL_FAM6_KABYLAKE_DESKTOP:
-			crystal_khz = 24000;	/* 24.0 MHz */
-			break;
-		case INTEL_FAM6_ATOM_GOLDMONT_X:
-			crystal_khz = 25000;	/* 25.0 MHz */
-			break;
-		case INTEL_FAM6_ATOM_GOLDMONT:
-			crystal_khz = 19200;	/* 19.2 MHz */
-			break;
-		}
-	}
+	/*
+	 * Denverton SoCs don't report crystal clock, and also don't support
+	 * CPUID.0x16 for the calculation below, so hardcode the 25MHz crystal
+	 * clock.
+	 */
+	if (crystal_khz == 0 &&
+			boot_cpu_data.x86_model == INTEL_FAM6_ATOM_GOLDMONT_X)
+		crystal_khz = 25000;
 
-	if (crystal_khz == 0)
-		return 0;
 	/*
-	 * TSC frequency determined by CPUID is a "hardware reported"
+	 * TSC frequency reported directly by CPUID is a "hardware reported"
 	 * frequency and is the most accurate one so far we have. This
 	 * is considered a known frequency.
 	 */
-	setup_force_cpu_cap(X86_FEATURE_TSC_KNOWN_FREQ);
+	if (crystal_khz != 0)
+		setup_force_cpu_cap(X86_FEATURE_TSC_KNOWN_FREQ);
+
+	/*
+	 * Some Intel SoCs like Skylake and Kabylake don't report the crystal
+	 * clock, but we can easily calculate it to a high degree of accuracy
+	 * by considering the crystal ratio and the CPU speed.
+	 */
+	if (crystal_khz == 0 && boot_cpu_data.cpuid_level >= 0x16) {
+		unsigned int eax_base_mhz, ebx, ecx, edx;
+
+		cpuid(0x16, &eax_base_mhz, &ebx, &ecx, &edx);
+		crystal_khz = eax_base_mhz * 1000 *
+			eax_denominator / ebx_numerator;
+	}
+
+	if (crystal_khz == 0)
+		return 0;
 
 	/*
 	 * For Atom SoCs TSC is the only reliable clocksource.
Signature · 1 line
2.20.1
author maintainer
#13 lore raw

PATCH 14/249 wil6210: fix spurious interrupts in 3-msi

In reply to Sasha Levin

From: Maya Erez <[email protected]>

[ Upstream commit e10b0eddd5235aa5aef4e40b970e34e735611a80 ]

Interrupt is set in ICM (ICR & ~IMV) rising trigger. As the driver masks the IRQ after clearing it, there can be a race where an additional spurious interrupt is triggered when the driver unmask the IRQ.
This can happen in case HW triggers an interrupt after the clear and before the mask.

To prevent the second spurious interrupt the driver needs to mask the IRQ before reading and clearing it.

---
 drivers/net/wireless/ath/wil6210/interrupt.c | 65 ++++++++++++--------
 1 file changed, 40 insertions(+), 25 deletions(-)
drivers/net/wireless/ath/wil6210/interrupt.c +40 -25
index e41ba24011d8..b00a13d6d530 100644
@@ -296,21 +296,24 @@ void wil_configure_interrupt_moderation(struct wil6210_priv *wil)
 static irqreturn_t wil6210_irq_rx(int irq, void *cookie)
 {
 	struct wil6210_priv *wil = cookie;
-	u32 isr = wil_ioread32_and_clear(wil->csr +
-					 HOSTADDR(RGF_DMA_EP_RX_ICR) +
-					 offsetof(struct RGF_ICR, ICR));
+	u32 isr;
 	bool need_unmask = true;
 
+	wil6210_mask_irq_rx(wil);
+
+	isr = wil_ioread32_and_clear(wil->csr +
+				     HOSTADDR(RGF_DMA_EP_RX_ICR) +
+				     offsetof(struct RGF_ICR, ICR));
+
 	trace_wil6210_irq_rx(isr);
 	wil_dbg_irq(wil, "ISR RX 0x%08x\n", isr);
 
 	if (unlikely(!isr)) {
 		wil_err_ratelimited(wil, "spurious IRQ: RX\n");
+		wil6210_unmask_irq_rx(wil);
 		return IRQ_NONE;
 	}
 
-	wil6210_mask_irq_rx(wil);
-
 	/* RX_DONE and RX_HTRSH interrupts are the same if interrupt
 	 * moderation is not used. Interrupt moderation may cause RX
 	 * buffer overflow while RX_DONE is delayed. The required
@@ -355,21 +358,24 @@ static irqreturn_t wil6210_irq_rx(int irq, void *cookie)
 static irqreturn_t wil6210_irq_rx_edma(int irq, void *cookie)
 {
 	struct wil6210_priv *wil = cookie;
-	u32 isr = wil_ioread32_and_clear(wil->csr +
-					 HOSTADDR(RGF_INT_GEN_RX_ICR) +
-					 offsetof(struct RGF_ICR, ICR));
+	u32 isr;
 	bool need_unmask = true;
 
+	wil6210_mask_irq_rx_edma(wil);
+
+	isr = wil_ioread32_and_clear(wil->csr +
+				     HOSTADDR(RGF_INT_GEN_RX_ICR) +
+				     offsetof(struct RGF_ICR, ICR));
+
 	trace_wil6210_irq_rx(isr);
 	wil_dbg_irq(wil, "ISR RX 0x%08x\n", isr);
 
 	if (unlikely(!isr)) {
 		wil_err(wil, "spurious IRQ: RX\n");
+		wil6210_unmask_irq_rx_edma(wil);
 		return IRQ_NONE;
 	}
 
-	wil6210_mask_irq_rx_edma(wil);
-
 	if (likely(isr & BIT_RX_STATUS_IRQ)) {
 		wil_dbg_irq(wil, "RX status ring\n");
 		isr &= ~BIT_RX_STATUS_IRQ;
@@ -403,21 +409,24 @@ static irqreturn_t wil6210_irq_rx_edma(int irq, void *cookie)
 static irqreturn_t wil6210_irq_tx_edma(int irq, void *cookie)
 {
 	struct wil6210_priv *wil = cookie;
-	u32 isr = wil_ioread32_and_clear(wil->csr +
-					 HOSTADDR(RGF_INT_GEN_TX_ICR) +
-					 offsetof(struct RGF_ICR, ICR));
+	u32 isr;
 	bool need_unmask = true;
 
+	wil6210_mask_irq_tx_edma(wil);
+
+	isr = wil_ioread32_and_clear(wil->csr +
+				     HOSTADDR(RGF_INT_GEN_TX_ICR) +
+				     offsetof(struct RGF_ICR, ICR));
+
 	trace_wil6210_irq_tx(isr);
 	wil_dbg_irq(wil, "ISR TX 0x%08x\n", isr);
 
 	if (unlikely(!isr)) {
 		wil_err(wil, "spurious IRQ: TX\n");
+		wil6210_unmask_irq_tx_edma(wil);
 		return IRQ_NONE;
 	}
 
-	wil6210_mask_irq_tx_edma(wil);
-
 	if (likely(isr & BIT_TX_STATUS_IRQ)) {
 		wil_dbg_irq(wil, "TX status ring\n");
 		isr &= ~BIT_TX_STATUS_IRQ;
@@ -446,21 +455,24 @@ static irqreturn_t wil6210_irq_tx_edma(int irq, void *cookie)
 static irqreturn_t wil6210_irq_tx(int irq, void *cookie)
 {
 	struct wil6210_priv *wil = cookie;
-	u32 isr = wil_ioread32_and_clear(wil->csr +
-					 HOSTADDR(RGF_DMA_EP_TX_ICR) +
-					 offsetof(struct RGF_ICR, ICR));
+	u32 isr;
 	bool need_unmask = true;
 
+	wil6210_mask_irq_tx(wil);
+
+	isr = wil_ioread32_and_clear(wil->csr +
+				     HOSTADDR(RGF_DMA_EP_TX_ICR) +
+				     offsetof(struct RGF_ICR, ICR));
+
 	trace_wil6210_irq_tx(isr);
 	wil_dbg_irq(wil, "ISR TX 0x%08x\n", isr);
 
 	if (unlikely(!isr)) {
 		wil_err_ratelimited(wil, "spurious IRQ: TX\n");
+		wil6210_unmask_irq_tx(wil);
 		return IRQ_NONE;
 	}
 
-	wil6210_mask_irq_tx(wil);
-
 	if (likely(isr & BIT_DMA_EP_TX_ICR_TX_DONE)) {
 		wil_dbg_irq(wil, "TX done\n");
 		isr &= ~BIT_DMA_EP_TX_ICR_TX_DONE;
@@ -532,20 +544,23 @@ static bool wil_validate_mbox_regs(struct wil6210_priv *wil)
 static irqreturn_t wil6210_irq_misc(int irq, void *cookie)
 {
 	struct wil6210_priv *wil = cookie;
-	u32 isr = wil_ioread32_and_clear(wil->csr +
-					 HOSTADDR(RGF_DMA_EP_MISC_ICR) +
-					 offsetof(struct RGF_ICR, ICR));
+	u32 isr;
+
+	wil6210_mask_irq_misc(wil, false);
+
+	isr = wil_ioread32_and_clear(wil->csr +
+				     HOSTADDR(RGF_DMA_EP_MISC_ICR) +
+				     offsetof(struct RGF_ICR, ICR));
 
 	trace_wil6210_irq_misc(isr);
 	wil_dbg_irq(wil, "ISR MISC 0x%08x\n", isr);
 
 	if (!isr) {
 		wil_err(wil, "spurious IRQ: MISC\n");
+		wil6210_unmask_irq_misc(wil, false);
 		return IRQ_NONE;
 	}
 
-	wil6210_mask_irq_misc(wil, false);
-
 	if (isr & ISR_MISC_FW_ERROR) {
 		u32 fw_assert_code = wil_r(wil, wil->rgf_fw_assert_code_addr);
 		u32 ucode_assert_code =
Signature · 1 line
2.20.1
author maintainer
#14 lore raw

PATCH 15/249 ath: DFS JP domain W56 fixed pulse type 3 RADAR detection

In reply to Sasha Levin

From: Anilkumar Kolli <[email protected]>

[ Upstream commit d8792393a783158cbb2c39939cb897dc5e5299b6 ]

Increase pulse width range from 1-2usec to 0-4usec.
During data traffic HW occasionally fails detecting radar pulses, so that SW cannot get enough radar reports to achieve the success rate.

Tested ath10k hw and fw:
* QCA9888(10.4-3.5.1-00052)
* QCA4019(10.4-3.2.1.1-00017)
* QCA9984(10.4-3.6-00104)
* QCA988X(10.2.4-1.0-00041)

Tested ath9k hw: AR9300

---
 drivers/net/wireless/ath/dfs_pattern_detector.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/wireless/ath/dfs_pattern_detector.c +1 -1
index d52b31b45df7..a274eb0d1968 100644
@@ -111,7 +111,7 @@ static const struct radar_detector_specs jp_radar_ref_types[] = {
 	JP_PATTERN(0, 0, 1, 1428, 1428, 1, 18, 29, false),
 	JP_PATTERN(1, 2, 3, 3846, 3846, 1, 18, 29, false),
 	JP_PATTERN(2, 0, 1, 1388, 1388, 1, 18, 50, false),
-	JP_PATTERN(3, 1, 2, 4000, 4000, 1, 18, 50, false),
+	JP_PATTERN(3, 0, 4, 4000, 4000, 1, 18, 50, false),
 	JP_PATTERN(4, 0, 5, 150, 230, 1, 23, 50, false),
 	JP_PATTERN(5, 6, 10, 200, 500, 1, 16, 50, false),
 	JP_PATTERN(6, 11, 20, 200, 500, 1, 12, 50, false),
Signature · 1 line
2.20.1
author maintainer
#15 lore raw

PATCH 16/249 ath10k: Fix encoding for protected management frames

In reply to Sasha Levin

From: Rakesh Pillai <[email protected]>

[ Upstream commit 42f1bc43e6a97b9ddbe976eba9bd05306c990c75 ]

Currently the protected management frames are
not appended with the MIC_LEN which results in
the protected management frames being encoded
incorrectly.

Add the extra space at the end of the protected
management frames to fix this encoding error for
the protected management frames.

Tested HW: WCN3990
Tested FW: WLAN.HL.3.1-00784-QCAHLSWMTPLZ-1

Fixes 1807da49733e ("ath10k: wmi: add management tx by reference support over wmi")
---
 drivers/net/wireless/ath/ath10k/wmi-tlv.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/net/wireless/ath/ath10k/wmi-tlv.c +3 -1
index 582fb11f648a..02709fc99034 100644
@@ -2840,8 +2840,10 @@ ath10k_wmi_tlv_op_gen_mgmt_tx_send(struct ath10k *ar, struct sk_buff *msdu,
 	if ((ieee80211_is_action(hdr->frame_control) ||
 	     ieee80211_is_deauth(hdr->frame_control) ||
 	     ieee80211_is_disassoc(hdr->frame_control)) &&
-	     ieee80211_has_protected(hdr->frame_control))
+	     ieee80211_has_protected(hdr->frame_control)) {
+		skb_put(msdu, IEEE80211_CCMP_MIC_LEN);
 		buf_len += IEEE80211_CCMP_MIC_LEN;
+	}
 
 	buf_len = min_t(u32, buf_len, WMI_TLV_MGMT_TX_FRAME_MAX_LEN);
 	buf_len = round_up(buf_len, 4);
Signature · 7 lines
2.20.1


_______________________________________________
ath10k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath10k
author maintainer
#16 lore raw

PATCH 17/249 regmap: debugfs: Fix memory leak in regmap_debugfs_init

In reply to Sasha Levin

From: Daniel Baluta <[email protected]>

[ Upstream commit 2899872b627e99b7586fe3b6c9f861da1b4d5072 ]

As detected by kmemleak running on i.MX6ULL board:

nreferenced object 0xd8366600 (size 64):
comm "swapper/0", pid 1, jiffies 4294937370 (age 933.220s) hex dump (first 32 bytes):

    64 75 6d 6d 79 2d 69 6f 6d 75 78 63 2d 67 70 72  dummy-iomuxc-gpr
    40 32 30 65 34 30 30 30 00 e3 f3 ab fe d1 1b dd  @20e4000........
  backtrace:
    [<b0402aec>] kasprintf+0x2c/0x54
    [<a6fbad2c>] regmap_debugfs_init+0x7c/0x31c
    [<9c8d91fa>] __regmap_init+0xb5c/0xcf4
    [<5b1c3d2a>] of_syscon_register+0x164/0x2c4
    [<596a5d80>] syscon_node_to_regmap+0x64/0x90
    [<49bd597b>] imx6ul_init_machine+0x34/0xa0
    [<250a4dac>] customize_machine+0x1c/0x30
    [<2d19fdaf>] do_one_initcall+0x7c/0x398
    [<e6084469>] kernel_init_freeable+0x328/0x448
    [<168c9101>] kernel_init+0x8/0x114
    [<913268aa>] ret_from_fork+0x14/0x20
    [<ce7b131a>] 0x0

Root cause is that map->debugfs_name is allocated using kasprintf and then the pointer is lost by assigning it other memory address.

---
 drivers/base/regmap/regmap-debugfs.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/base/regmap/regmap-debugfs.c +2 -0
index 263f82516ff4..e5e1b3a01b1a 100644
@@ -579,6 +579,8 @@ void regmap_debugfs_init(struct regmap *map, const char *name)
 	}
 
 	if (!strcmp(name, "dummy")) {
+		kfree(map->debugfs_name);
+
 		map->debugfs_name = kasprintf(GFP_KERNEL, "dummy%d",
 						dummy_index);
 		name = map->debugfs_name;
Signature · 1 line
2.20.1
author maintainer
#17 lore raw

PATCH 18/249 batman-adv: fix for leaked TVLV handler.

In reply to Sasha Levin

From: Jeremy Sowden <[email protected]>

[ Upstream commit 17f78dd1bd624a4dd78ed5db3284a63ee807fcc3 ]

A handler for BATADV_TVLV_ROAM was being registered when the translation-table was initialized, but not unregistered when the translation-table was freed. Unregister it.

Fixes 122edaa05940 ("batman-adv: tvlv - convert roaming adv packet to use tvlv unicast packets")
Reported-by [email protected]
---
 net/batman-adv/translation-table.c | 2 ++
 1 file changed, 2 insertions(+)
net/batman-adv/translation-table.c +2 -0
index 1ddfd5e011ee..8a482c5ec67b 100644
@@ -3813,6 +3813,8 @@ static void batadv_tt_purge(struct work_struct *work)
  */
 void batadv_tt_free(struct batadv_priv *bat_priv)
 {
+	batadv_tvlv_handler_unregister(bat_priv, BATADV_TVLV_ROAM, 1);
+
 	batadv_tvlv_container_unregister(bat_priv, BATADV_TVLV_TT, 1);
 	batadv_tvlv_handler_unregister(bat_priv, BATADV_TVLV_TT, 1);
 
Signature · 1 line
2.20.1
author maintainer
#18 lore raw

PATCH 19/249 media: dvb: usb: fix use after free in dvb_usb_device_exit

In reply to Sasha Levin

From: Oliver Neukum <[email protected]>

[ Upstream commit 6cf97230cd5f36b7665099083272595c55d72be7 ]

dvb_usb_device_exit() frees and uses the device name in that order. Fix by storing the name in a buffer before freeing it.

---
 drivers/media/usb/dvb-usb/dvb-usb-init.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
drivers/media/usb/dvb-usb/dvb-usb-init.c +5 -2
index e97f6edc98de..65f2b1a20ca1 100644
@@ -284,12 +284,15 @@ EXPORT_SYMBOL(dvb_usb_device_init);
 void dvb_usb_device_exit(struct usb_interface *intf)
 {
 	struct dvb_usb_device *d = usb_get_intfdata(intf);
-	const char *name = "generic DVB-USB module";
+	const char *default_name = "generic DVB-USB module";
+	char name[40];
 
 	usb_set_intfdata(intf, NULL);
 	if (d != NULL && d->desc != NULL) {
-		name = d->desc->name;
+		strscpy(name, d->desc->name, sizeof(name));
 		dvb_usb_exit(d);
+	} else {
+		strscpy(name, default_name, sizeof(name));
 	}
 	info("%s successfully deinitialized and disconnected.", name);
 
Signature · 1 line
2.20.1
author maintainer
#19 lore raw

PATCH 20/249 media: spi: IR LED: add missing of table registration

In reply to Sasha Levin

From: Daniel Gomez <[email protected]>

[ Upstream commit 24e4cf770371df6ad49ed873f21618d9878f64c8 ]

MODULE_DEVICE_TABLE(of, <of_match_table> should be called to complete DT OF mathing mechanism and register it.

Before this patch:
modinfo drivers/media/rc/ir-spi.ko | grep alias

After this patch:
modinfo drivers/media/rc/ir-spi.ko | grep alias

alias:          of:N*T*Cir-spi-ledC*
alias:          of:N*T*Cir-spi-led
---
 drivers/media/rc/ir-spi.c | 1 +
 1 file changed, 1 insertion(+)
drivers/media/rc/ir-spi.c +1 -0
index 66334e8d63ba..c58f2d38a458 100644
@@ -161,6 +161,7 @@ static const struct of_device_id ir_spi_of_match[] = {
 	{ .compatible = "ir-spi-led" },
 	{},
 };
+MODULE_DEVICE_TABLE(of, ir_spi_of_match);
 
 static struct spi_driver ir_spi_driver = {
 	.probe = ir_spi_probe,
Signature · 1 line
2.20.1
author maintainer
#20 lore raw

PATCH 21/249 crypto: talitos - fix skcipher failure due to wrong output IV

In reply to Sasha Levin

From: Christophe Leroy <[email protected]>

[ Upstream commit 3e03e792865ae48b8cfc69a0b4d65f02f467389f ]

Selftests report the following:

[    2.984845] alg: skcipher: cbc-aes-talitos encryption test failed (wrong output IV) on test vector 0, cfg="in-place"
[    2.995377] 00000000: 3d af ba 42 9d 9e b4 30 b4 22 da 80 2c 9f ac 41
[    3.032673] alg: skcipher: cbc-des-talitos encryption test failed (wrong output IV) on test vector 0, cfg="in-place"
[    3.043185] 00000000: fe dc ba 98 76 54 32 10
[    3.063238] alg: skcipher: cbc-3des-talitos encryption test failed (wrong output IV) on test vector 0, cfg="in-place"
[    3.073818] 00000000: 7d 33 88 93 0f 93 b2 42

This above dumps show that the actual output IV is indeed the input IV. This is due to the IV not being copied back into the request.

This patch fixes that.

---
 drivers/crypto/talitos.c | 4 ++++
 1 file changed, 4 insertions(+)
drivers/crypto/talitos.c +4 -0
index 427c78d4d948..2d9a0971a7fc 100644
@@ -1624,11 +1624,15 @@ static void ablkcipher_done(struct device *dev,
 			    int err)
 {
 	struct ablkcipher_request *areq = context;
+	struct crypto_ablkcipher *cipher = crypto_ablkcipher_reqtfm(areq);
+	struct talitos_ctx *ctx = crypto_ablkcipher_ctx(cipher);
+	unsigned int ivsize = crypto_ablkcipher_ivsize(cipher);
 	struct talitos_edesc *edesc;
 
 	edesc = container_of(desc, struct talitos_edesc, desc);
 
 	common_nonsnoop_unmap(dev, edesc, areq);
+	memcpy(areq->info, ctx->iv, ivsize);
 
 	kfree(edesc);
 
Signature · 1 line
2.20.1
author maintainer
#21 lore raw

PATCH 22/249 media: ov7740: avoid invalid framesize setting

In reply to Sasha Levin

From: Akinobu Mita <[email protected]>

[ Upstream commit 6e4ab830ac6d6a0d7cd7f87dc5d6536369bf24a8 ]

If the requested framesize by VIDIOC_SUBDEV_S_FMT is larger than supported framesizes, it causes an out of bounds array access and the resulting framesize is unexpected.

Avoid out of bounds array access and select the default framesize.

---
 drivers/media/i2c/ov7740.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
drivers/media/i2c/ov7740.c +5 -1
index 54e80a60aa57..63011d4b4738 100644
@@ -785,7 +785,11 @@ static int ov7740_try_fmt_internal(struct v4l2_subdev *sd,
 
 		fsize++;
 	}
-
+	if (i >= ARRAY_SIZE(ov7740_framesizes)) {
+		fsize = &ov7740_framesizes[0];
+		fmt->width = fsize->width;
+		fmt->height = fsize->height;
+	}
 	if (ret_frmsize != NULL)
 		*ret_frmsize = fsize;
 
Signature · 1 line
2.20.1
author maintainer
#22 lore raw

PATCH 23/249 media: marvell-ccic: fix DMA s/g desc number calculation

In reply to Sasha Levin

From: Lubomir Rintel <[email protected]>

[ Upstream commit 0c7aa32966dab0b8a7424e1b34c7f206817953ec ]

The commit d790b7eda953 ("[media] vb2-dma-sg: move dma_(un)map_sg here") left dma_desc_nent unset. It previously contained the number of DMA descriptors as returned from dma_map_sg().

We can now (since the commit referred to above) obtain the same value from the sg_table and drop dma_desc_nent altogether.

Tested on OLPC XO-1.75 machine. Doesn't affect the OLPC XO-1's Cafe driver, since that one doesn't do DMA.

[[email protected]: fix a checkpatch warning]

Fixes d790b7eda953 ("[media] vb2-dma-sg: move dma_(un)map_sg here")
---
 drivers/media/platform/marvell-ccic/mcam-core.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
drivers/media/platform/marvell-ccic/mcam-core.c +3 -2
index f1b301810260..0a6411b877e9 100644
@@ -200,7 +200,6 @@ struct mcam_vb_buffer {
 	struct list_head queue;
 	struct mcam_dma_desc *dma_desc;	/* Descriptor virtual address */
 	dma_addr_t dma_desc_pa;		/* Descriptor physical address */
-	int dma_desc_nent;		/* Number of mapped descriptors */
 };
 
 static inline struct mcam_vb_buffer *vb_to_mvb(struct vb2_v4l2_buffer *vb)
@@ -608,9 +607,11 @@ static void mcam_dma_contig_done(struct mcam_camera *cam, int frame)
 static void mcam_sg_next_buffer(struct mcam_camera *cam)
 {
 	struct mcam_vb_buffer *buf;
+	struct sg_table *sg_table;
 
 	buf = list_first_entry(&cam->buffers, struct mcam_vb_buffer, queue);
 	list_del_init(&buf->queue);
+	sg_table = vb2_dma_sg_plane_desc(&buf->vb_buf.vb2_buf, 0);
 	/*
 	 * Very Bad Not Good Things happen if you don't clear
 	 * C1_DESC_ENA before making any descriptor changes.
@@ -618,7 +619,7 @@ static void mcam_sg_next_buffer(struct mcam_camera *cam)
 	mcam_reg_clear_bit(cam, REG_CTRL1, C1_DESC_ENA);
 	mcam_reg_write(cam, REG_DMA_DESC_Y, buf->dma_desc_pa);
 	mcam_reg_write(cam, REG_DESC_LEN_Y,
-			buf->dma_desc_nent*sizeof(struct mcam_dma_desc));
+			sg_table->nents * sizeof(struct mcam_dma_desc));
 	mcam_reg_write(cam, REG_DESC_LEN_U, 0);
 	mcam_reg_write(cam, REG_DESC_LEN_V, 0);
 	mcam_reg_set_bit(cam, REG_CTRL1, C1_DESC_ENA);
Signature · 1 line
2.20.1
author maintainer
#23 lore raw

PATCH 24/249 selftests/bpf: adjust verifier scale test

In reply to Sasha Levin

From: Alexei Starovoitov <[email protected]>

[ Upstream commit 7c0c6095d48dcd0e67c917aa73cdbb2715aafc36 ]

Adjust scale tests to check for new jmp sequence limit.

BPF_JGT had to be changed to BPF_JEQ because the verifier was too smart. It tracked the known safe range of R0 values and pruned the search earlier before hitting exact 8192 limit. bpf_semi_rand_get() was too (un)?lucky.

k = 0; was missing in bpf_fill_scale2.
It was testing a bit shorter sequence of jumps than intended.

---
 tools/testing/selftests/bpf/test_verifier.c | 31 +++++++++++----------
 1 file changed, 17 insertions(+), 14 deletions(-)
tools/testing/selftests/bpf/test_verifier.c +17 -14
index 288cb740e005..6438d4dc8ae1 100644
@@ -207,33 +207,35 @@ static void bpf_fill_rand_ld_dw(struct bpf_test *self)
 	self->retval = (uint32_t)res;
 }
 
-/* test the sequence of 1k jumps */
+#define MAX_JMP_SEQ 8192
+
+/* test the sequence of 8k jumps */
 static void bpf_fill_scale1(struct bpf_test *self)
 {
 	struct bpf_insn *insn = self->fill_insns;
 	int i = 0, k = 0;
 
 	insn[i++] = BPF_MOV64_REG(BPF_REG_6, BPF_REG_1);
-	/* test to check that the sequence of 1024 jumps is acceptable */
-	while (k++ < 1024) {
+	/* test to check that the long sequence of jumps is acceptable */
+	while (k++ < MAX_JMP_SEQ) {
 		insn[i++] = BPF_RAW_INSN(BPF_JMP | BPF_CALL, 0, 0, 0,
 					 BPF_FUNC_get_prandom_u32);
-		insn[i++] = BPF_JMP_IMM(BPF_JGT, BPF_REG_0, bpf_semi_rand_get(), 2);
+		insn[i++] = BPF_JMP_IMM(BPF_JEQ, BPF_REG_0, bpf_semi_rand_get(), 2);
 		insn[i++] = BPF_MOV64_REG(BPF_REG_1, BPF_REG_10);
 		insn[i++] = BPF_STX_MEM(BPF_DW, BPF_REG_1, BPF_REG_6,
 					-8 * (k % 64 + 1));
 	}
-	/* every jump adds 1024 steps to insn_processed, so to stay exactly
-	 * within 1m limit add MAX_TEST_INSNS - 1025 MOVs and 1 EXIT
+	/* every jump adds 1 step to insn_processed, so to stay exactly
+	 * within 1m limit add MAX_TEST_INSNS - MAX_JMP_SEQ - 1 MOVs and 1 EXIT
 	 */
-	while (i < MAX_TEST_INSNS - 1025)
+	while (i < MAX_TEST_INSNS - MAX_JMP_SEQ - 1)
 		insn[i++] = BPF_ALU32_IMM(BPF_MOV, BPF_REG_0, 42);
 	insn[i] = BPF_EXIT_INSN();
 	self->prog_len = i + 1;
 	self->retval = 42;
 }
 
-/* test the sequence of 1k jumps in inner most function (function depth 8)*/
+/* test the sequence of 8k jumps in inner most function (function depth 8)*/
 static void bpf_fill_scale2(struct bpf_test *self)
 {
 	struct bpf_insn *insn = self->fill_insns;
@@ -245,19 +247,20 @@ static void bpf_fill_scale2(struct bpf_test *self)
 		insn[i++] = BPF_EXIT_INSN();
 	}
 	insn[i++] = BPF_MOV64_REG(BPF_REG_6, BPF_REG_1);
-	/* test to check that the sequence of 1024 jumps is acceptable */
-	while (k++ < 1024) {
+	/* test to check that the long sequence of jumps is acceptable */
+	k = 0;
+	while (k++ < MAX_JMP_SEQ) {
 		insn[i++] = BPF_RAW_INSN(BPF_JMP | BPF_CALL, 0, 0, 0,
 					 BPF_FUNC_get_prandom_u32);
-		insn[i++] = BPF_JMP_IMM(BPF_JGT, BPF_REG_0, bpf_semi_rand_get(), 2);
+		insn[i++] = BPF_JMP_IMM(BPF_JEQ, BPF_REG_0, bpf_semi_rand_get(), 2);
 		insn[i++] = BPF_MOV64_REG(BPF_REG_1, BPF_REG_10);
 		insn[i++] = BPF_STX_MEM(BPF_DW, BPF_REG_1, BPF_REG_6,
 					-8 * (k % (64 - 4 * FUNC_NEST) + 1));
 	}
-	/* every jump adds 1024 steps to insn_processed, so to stay exactly
-	 * within 1m limit add MAX_TEST_INSNS - 1025 MOVs and 1 EXIT
+	/* every jump adds 1 step to insn_processed, so to stay exactly
+	 * within 1m limit add MAX_TEST_INSNS - MAX_JMP_SEQ - 1 MOVs and 1 EXIT
 	 */
-	while (i < MAX_TEST_INSNS - 1025)
+	while (i < MAX_TEST_INSNS - MAX_JMP_SEQ - 1)
 		insn[i++] = BPF_ALU32_IMM(BPF_MOV, BPF_REG_0, 42);
 	insn[i] = BPF_EXIT_INSN();
 	self->prog_len = i + 1;
Signature · 1 line
2.20.1
author maintainer
#24 lore raw

PATCH 25/249 media: vpss: fix a potential NULL pointer dereference

In reply to Sasha Levin

From: Kangjie Lu <[email protected]>

[ Upstream commit e08f0761234def47961d3252eac09ccedfe4c6a0 ]

In case ioremap fails, the fix returns -ENOMEM to avoid NULL pointer dereference.

---
 drivers/media/platform/davinci/vpss.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/media/platform/davinci/vpss.c +5 -0
index 3f079ac1b080..be91b0c7d20b 100644
@@ -509,6 +509,11 @@ static int __init vpss_init(void)
 		return -EBUSY;
 
 	oper_cfg.vpss_regs_base2 = ioremap(VPSS_CLK_CTRL, 4);
+	if (unlikely(!oper_cfg.vpss_regs_base2)) {
+		release_mem_region(VPSS_CLK_CTRL, 4);
+		return -ENOMEM;
+	}
+
 	writel(VPSS_CLK_CTRL_VENCCLKEN |
 		     VPSS_CLK_CTRL_DACCLKEN, oper_cfg.vpss_regs_base2);
 
Signature · 1 line
2.20.1
author maintainer
#25 lore raw

PATCH 26/249 media: media_device_enum_links32: clean a reserved field

In reply to Sasha Levin

From: Jungo Lin <[email protected]>

[ Upstream commit f49308878d7202e07d8761238e01bd0e5fce2750 ]

In v4l2-compliance utility, test MEDIA_IOC_ENUM_ENTITIES will check whether reserved field of media_links_enum filled with zero.

However, for 32 bit program, the reserved field is missing copy from kernel space to user space in media_device_enum_links32 function.

This patch adds the cleaning a reserved field logic in media_device_enum_links32 function.

---
 drivers/media/media-device.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)
drivers/media/media-device.c +8 -1
index 9ae481ddd975..b9bb4904bba1 100644
@@ -494,6 +494,7 @@ static long media_device_enum_links32(struct media_device *mdev,
 {
 	struct media_links_enum links;
 	compat_uptr_t pads_ptr, links_ptr;
+	int ret;
 
 	memset(&links, 0, sizeof(links));
 
@@ -505,7 +506,13 @@ static long media_device_enum_links32(struct media_device *mdev,
 	links.pads = compat_ptr(pads_ptr);
 	links.links = compat_ptr(links_ptr);
 
-	return media_device_enum_links(mdev, &links);
+	ret = media_device_enum_links(mdev, &links);
+	if (ret)
+		return ret;
+
+	memset(ulinks->reserved, 0, sizeof(ulinks->reserved));
+
+	return 0;
 }
 
 #define MEDIA_IOC_ENUM_LINKS32		_IOWR('|', 0x02, struct media_links_enum32)
Signature · 1 line
2.20.1
author maintainer
#26 lore raw

PATCH 27/249 ice: Gracefully handle reset failure in ice_alloc_vfs()

In reply to Sasha Levin

From: Brett Creeley <[email protected]>

[ Upstream commit 72f9c2039859e6303550f202d6cc6b8d8af0178c ]

Currently if ice_reset_all_vfs() fails in ice_alloc_vfs() we fail to free some resources, reset variables, and return an error value. Fix this by adding another unroll case to free the pf->vf array, set the pf->num_alloc_vfs to 0, and return an error code.

Without this, if ice_reset_all_vfs() fails in ice_alloc_vfs() we will not be able to do SRIOV without hard rebooting the system because rmmod'ing the driver does not work.

---
 drivers/net/ethernet/intel/ice/ice_virtchnl_pf.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)
drivers/net/ethernet/intel/ice/ice_virtchnl_pf.c +9 -2
index a805cbdd69be..81ea77978355 100644
@@ -1134,7 +1134,7 @@ static int ice_alloc_vfs(struct ice_pf *pf, u16 num_alloc_vfs)
 			   GFP_KERNEL);
 	if (!vfs) {
 		ret = -ENOMEM;
-		goto err_unroll_sriov;
+		goto err_pci_disable_sriov;
 	}
 	pf->vf = vfs;
 
@@ -1154,12 +1154,19 @@ static int ice_alloc_vfs(struct ice_pf *pf, u16 num_alloc_vfs)
 	pf->num_alloc_vfs = num_alloc_vfs;
 
 	/* VF resources get allocated during reset */
-	if (!ice_reset_all_vfs(pf, true))
+	if (!ice_reset_all_vfs(pf, true)) {
+		ret = -EIO;
 		goto err_unroll_sriov;
+	}
 
 	goto err_unroll_intr;
 
 err_unroll_sriov:
+	pf->vf = NULL;
+	devm_kfree(&pf->pdev->dev, vfs);
+	vfs = NULL;
+	pf->num_alloc_vfs = 0;
+err_pci_disable_sriov:
 	pci_disable_sriov(pf->pdev);
 err_unroll_intr:
 	/* rearm interrupts here */
Signature · 1 line
2.20.1
author maintainer
#27 lore raw

PATCH 28/249 media: venus: firmware: fix leaked of_node references

In reply to Sasha Levin

From: Wen Yang <[email protected]>

[ Upstream commit 2c41cc0be07b5ee2f1167f41cd8a86fc5b53d82c ]

The call to of_parse_phandle returns a node pointer with refcount incremented thus it must be explicitly decremented after the last usage.

Detected by coccinelle with the following warnings:
drivers/media/platform/qcom/venus/firmware.c:90:2-8: ERROR: missing of_node_put; acquired a node pointer with refcount incremented on line 82, but without a corresponding object release within this function. drivers/media/platform/qcom/venus/firmware.c:94:2-8: ERROR: missing of_node_put; acquired a node pointer with refcount incremented on line 82, but without a corresponding object release within this function. drivers/media/platform/qcom/venus/firmware.c:128:1-7: ERROR: missing of_node_put; acquired a node pointer with refcount incremented on line 82, but without a corresponding object release within this function.

---
 drivers/media/platform/qcom/venus/firmware.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/media/platform/qcom/venus/firmware.c +4 -2
index 1eba23409ff3..d3d1748a7ef6 100644
@@ -78,11 +78,11 @@ static int venus_load_fw(struct venus_core *core, const char *fwname,
 
 	ret = of_address_to_resource(node, 0, &r);
 	if (ret)
-		return ret;
+		goto err_put_node;
 
 	ret = request_firmware(&mdt, fwname, dev);
 	if (ret < 0)
-		return ret;
+		goto err_put_node;
 
 	fw_size = qcom_mdt_get_size(mdt);
 	if (fw_size < 0) {
@@ -116,6 +116,8 @@ static int venus_load_fw(struct venus_core *core, const char *fwname,
 	memunmap(mem_va);
 err_release_fw:
 	release_firmware(mdt);
+err_put_node:
+	of_node_put(node);
 	return ret;
 }
 
Signature · 1 line
2.20.1
author maintainer
#28 lore raw

PATCH 29/249 crypto: caam - avoid S/G table fetching for AEAD zero-length output

In reply to Sasha Levin

From: Horia Geantă <[email protected]>

[ Upstream commit dcd9c76e5a183af4f793beb5141efcd260b8d09f ]

When enabling IOMMU support, the following issue becomes visible in the AEAD zero-length case.

Even though the output sequence length is set to zero, the crypto engine tries to prefetch 4 S/G table entries (since SGF bit is set in SEQ OUT PTR command - which is either generated in SW in case of caam/jr or in HW in case of caam/qi, caam/qi2).
The DMA read operation will trigger an IOMMU fault since the address in the SEQ OUT PTR is "dummy" (set to zero / not obtained via DMA API mapping).

1. In case of caam/jr, avoid the IOMMU fault by clearing the SGF bit in SEQ OUT PTR command.

2. In case of caam/qi - setting address, bpid, length to zero for output entry in the compound frame has a special meaning (cf. CAAM RM):
"Output frame = Unspecified, Input address = Y. A unspecified frame is indicated by an unused SGT entry (an entry in which the Address, Length, and BPID fields are all zero). SEC obtains output buffers from BMan as prescribed by the preheader."

Since no output buffers are needed, modify the preheader by setting (ABS = 1, ADDBUF = 0):
-"ABS = 1 means obtain the number of buffers in ADDBUF (0 or 1) from the pool POOL ID"
-ADDBUF: "If ABS is set, ADD BUF specifies whether to allocate a buffer or not"

3. In case of caam/qi2, since engine:
-does not support FLE[FMT]=2'b11 ("unused" entry) mentioned in DPAA2 RM -requires output entry to be present, even if not used
the solution chosen is to leave output frame list entry zeroized.

Fixes 763069ba49d3 ("crypto: caam - handle zero-length AEAD output")
---
 drivers/crypto/caam/caamalg.c     | 1 +
 drivers/crypto/caam/caamalg_qi.c  | 2 +-
 drivers/crypto/caam/caamalg_qi2.c | 9 +++++++++
 drivers/crypto/caam/qi.c          | 3 +++
 4 files changed, 14 insertions(+), 1 deletion(-)
drivers/crypto/caam/caamalg.c +1 -0
index c0ece44f303b..df416e6c1468 100644
@@ -1106,6 +1106,7 @@ static void init_aead_job(struct aead_request *req,
 	if (unlikely(req->src != req->dst)) {
 		if (!edesc->mapped_dst_nents) {
 			dst_dma = 0;
+			out_options = 0;
 		} else if (edesc->mapped_dst_nents == 1) {
 			dst_dma = sg_dma_address(req->dst);
 			out_options = 0;
drivers/crypto/caam/caamalg_qi.c +1 -1
index d290d6b41825..116cbc81fa8d 100644
@@ -1109,7 +1109,7 @@ static struct aead_edesc *aead_edesc_alloc(struct aead_request *req,
 			dma_to_qm_sg_one_ext(&fd_sgt[0], qm_sg_dma +
 					     (1 + !!ivsize) * sizeof(*sg_table),
 					     out_len, 0);
-	} else if (mapped_dst_nents == 1) {
+	} else if (mapped_dst_nents <= 1) {
 		dma_to_qm_sg_one(&fd_sgt[0], sg_dma_address(req->dst), out_len,
 				 0);
 	} else {
drivers/crypto/caam/caamalg_qi2.c +9 -0
index 2b2980a8a9b9..b949944c8e55 100644
@@ -559,6 +559,14 @@ static struct aead_edesc *aead_edesc_alloc(struct aead_request *req,
 			dpaa2_fl_set_addr(out_fle, qm_sg_dma +
 					  (1 + !!ivsize) * sizeof(*sg_table));
 		}
+	} else if (!mapped_dst_nents) {
+		/*
+		 * crypto engine requires the output entry to be present when
+		 * "frame list" FD is used.
+		 * Since engine does not support FMT=2'b11 (unused entry type),
+		 * leaving out_fle zeroized is the best option.
+		 */
+		goto skip_out_fle;
 	} else if (mapped_dst_nents == 1) {
 		dpaa2_fl_set_format(out_fle, dpaa2_fl_single);
 		dpaa2_fl_set_addr(out_fle, sg_dma_address(req->dst));
@@ -570,6 +578,7 @@ static struct aead_edesc *aead_edesc_alloc(struct aead_request *req,
 
 	dpaa2_fl_set_len(out_fle, out_len);
 
+skip_out_fle:
 	return edesc;
 }
 
drivers/crypto/caam/qi.c +3 -0
index 9f08f84cca59..2d9b0485141f 100644
@@ -18,6 +18,7 @@
 #include "desc_constr.h"
 
 #define PREHDR_RSLS_SHIFT	31
+#define PREHDR_ABS		BIT(25)
 
 /*
  * Use a reasonable backlog of frames (per CPU) as congestion threshold,
@@ -346,6 +347,7 @@ int caam_drv_ctx_update(struct caam_drv_ctx *drv_ctx, u32 *sh_desc)
 	 */
 	drv_ctx->prehdr[0] = cpu_to_caam32((1 << PREHDR_RSLS_SHIFT) |
 					   num_words);
+	drv_ctx->prehdr[1] = cpu_to_caam32(PREHDR_ABS);
 	memcpy(drv_ctx->sh_desc, sh_desc, desc_bytes(sh_desc));
 	dma_sync_single_for_device(qidev, drv_ctx->context_a,
 				   sizeof(drv_ctx->sh_desc) +
@@ -401,6 +403,7 @@ struct caam_drv_ctx *caam_drv_ctx_init(struct device *qidev,
 	 */
 	drv_ctx->prehdr[0] = cpu_to_caam32((1 << PREHDR_RSLS_SHIFT) |
 					   num_words);
+	drv_ctx->prehdr[1] = cpu_to_caam32(PREHDR_ABS);
 	memcpy(drv_ctx->sh_desc, sh_desc, desc_bytes(sh_desc));
 	size = sizeof(drv_ctx->prehdr) + sizeof(drv_ctx->sh_desc);
 	hwdesc = dma_map_single(qidev, drv_ctx->prehdr, size,
Signature · 1 line
2.20.1
author maintainer
#29 lore raw

PATCH 30/249 net: stmmac: dwmac1000: Clear unused address entries

In reply to Sasha Levin

From: Jose Abreu <[email protected]>

[ Upstream commit 9463c445590091202659cdfdd44b236acadfbd84 ]

In case we don't use a given address entry we need to clear it because it could contain previous values that are no longer valid.

Found out while running stmmac selftests.

Cc Joao Pinto <[email protected]>
Cc David S. Miller <[email protected]>
Cc Giuseppe Cavallaro <[email protected]>
Cc Alexandre Torgue <[email protected]>
---
 drivers/net/ethernet/stmicro/stmmac/dwmac1000_core.c | 6 ++++++
 1 file changed, 6 insertions(+)
drivers/net/ethernet/stmicro/stmmac/dwmac1000_core.c +6 -0
index 9fff81170163..54f4ffb36d60 100644
@@ -206,6 +206,12 @@ static void dwmac1000_set_filter(struct mac_device_info *hw,
 					    GMAC_ADDR_LOW(reg));
 			reg++;
 		}
+
+		while (reg <= perfect_addr_number) {
+			writel(0, ioaddr + GMAC_ADDR_HIGH(reg));
+			writel(0, ioaddr + GMAC_ADDR_LOW(reg));
+			reg++;
+		}
 	}
 
 #ifdef FRAME_FILTER_DEBUG
Signature · 1 line
2.20.1
author maintainer
#30 lore raw

PATCH 31/249 net: stmmac: dwmac4/5: Clear unused address entries

In reply to Sasha Levin

From: Jose Abreu <[email protected]>

[ Upstream commit 0620ec6c62a5a07625b65f699adc5d1b90394ee6 ]

In case we don't use a given address entry we need to clear it because it could contain previous values that are no longer valid.

Found out while running stmmac selftests.

Cc Joao Pinto <[email protected]>
Cc David S. Miller <[email protected]>
Cc Giuseppe Cavallaro <[email protected]>
Cc Alexandre Torgue <[email protected]>
---
 drivers/net/ethernet/stmicro/stmmac/dwmac4_core.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)
drivers/net/ethernet/stmicro/stmmac/dwmac4_core.c +8 -2
index 99d772517242..206170d0bf81 100644
@@ -443,14 +443,20 @@ static void dwmac4_set_filter(struct mac_device_info *hw,
 		 * are required
 		 */
 		value |= GMAC_PACKET_FILTER_PR;
-	} else if (!netdev_uc_empty(dev)) {
-		int reg = 1;
+	} else {
 		struct netdev_hw_addr *ha;
+		int reg = 1;
 
 		netdev_for_each_uc_addr(ha, dev) {
 			dwmac4_set_umac_addr(hw, ha->addr, reg);
 			reg++;
 		}
+
+		while (reg <= GMAC_MAX_PERFECT_ADDRESSES) {
+			writel(0, ioaddr + GMAC_ADDR_HIGH(reg));
+			writel(0, ioaddr + GMAC_ADDR_LOW(reg));
+			reg++;
+		}
 	}
 
 	writel(value, ioaddr + GMAC_PACKET_FILTER);
Signature · 1 line
2.20.1
author maintainer
#31 lore raw

PATCH 32/249 net: stmmac: Prevent missing interrupts when running NAPI

In reply to Sasha Levin

From: Jose Abreu <[email protected]>

[ Upstream commit a976ca79e23f13bff79c14e7266cea4a0ea51e67 ]

When we trigger NAPI we are disabling interrupts but in case we receive or send a packet in the meantime, as interrupts are disabled, we will miss this event.

Trigger both NAPI instances (RX and TX) when at least one event happens so that we don't miss any interrupts.

Cc Joao Pinto <[email protected]>
Cc David S. Miller <[email protected]>
Cc Giuseppe Cavallaro <[email protected]>
Cc Alexandre Torgue <[email protected]>
---
 drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/net/ethernet/stmicro/stmmac/stmmac_main.c +3 -0
index 06358fe5b245..dbee9b0113e3 100644
@@ -2048,6 +2048,9 @@ static int stmmac_napi_check(struct stmmac_priv *priv, u32 chan)
 						 &priv->xstats, chan);
 	struct stmmac_channel *ch = &priv->channel[chan];
 
+	if (status)
+		status |= handle_rx | handle_tx;
+
 	if ((status & handle_rx) && (chan < priv->plat->rx_queues_to_use)) {
 		stmmac_disable_dma_irq(priv, priv->ioaddr, chan);
 		napi_schedule_irqoff(&ch->rx_napi);
Signature · 1 line
2.20.1
author maintainer
#32 lore raw

PATCH 33/249 ice: Fix couple of issues in ice_vsi_release

In reply to Sasha Levin

From: Brett Creeley <[email protected]>

[ Upstream commit aa6ccf3f2d7042f94c4e91538956ce7051e7856e ]

Currently the driver is calling ice_napi_del() and then unregister_netdev(). The call to unregister_netdev() will result in a call to ice_stop() and then ice_vsi_close(). This is where we call napi_disable() for all the MSI-X vectors. This flow is reversed so make the changes to ensure napi_disable() happens prior to napi_del().

Before calling napi_del() and free_netdev() make sure unregister_netdev() was called. This is done by making sure the __ICE_DOWN bit is set in the vsi->state for the interested VSI.

---
 drivers/net/ethernet/intel/ice/ice.h      |  1 -
 drivers/net/ethernet/intel/ice/ice_lib.c  | 24 ++++++++++++-----------
 drivers/net/ethernet/intel/ice/ice_main.c |  2 +-
 3 files changed, 14 insertions(+), 13 deletions(-)
drivers/net/ethernet/intel/ice/ice.h +0 -1
index 792e6e42030e..754c7080c3fc 100644
@@ -451,7 +451,6 @@ int ice_set_rss(struct ice_vsi *vsi, u8 *seed, u8 *lut, u16 lut_size);
 int ice_get_rss(struct ice_vsi *vsi, u8 *seed, u8 *lut, u16 lut_size);
 void ice_fill_rss_lut(u8 *lut, u16 rss_table_size, u16 rss_size);
 void ice_print_link_msg(struct ice_vsi *vsi, bool isup);
-void ice_napi_del(struct ice_vsi *vsi);
 #ifdef CONFIG_DCB
 int ice_pf_ena_all_vsi(struct ice_pf *pf, bool locked);
 void ice_pf_dis_all_vsi(struct ice_pf *pf, bool locked);
drivers/net/ethernet/intel/ice/ice_lib.c +13 -11
index fbf1eba0cc2a..f14fa51cc704 100644
@@ -2754,19 +2754,14 @@ int ice_vsi_release(struct ice_vsi *vsi)
 
 	if (vsi->type == ICE_VSI_VF)
 		vf = &pf->vf[vsi->vf_id];
-	/* do not unregister and free netdevs while driver is in the reset
-	 * recovery pending state. Since reset/rebuild happens through PF
-	 * service task workqueue, its not a good idea to unregister netdev
-	 * that is associated to the PF that is running the work queue items
-	 * currently. This is done to avoid check_flush_dependency() warning
-	 * on this wq
+	/* do not unregister while driver is in the reset recovery pending
+	 * state. Since reset/rebuild happens through PF service task workqueue,
+	 * it's not a good idea to unregister netdev that is associated to the
+	 * PF that is running the work queue items currently. This is done to
+	 * avoid check_flush_dependency() warning on this wq
 	 */
-	if (vsi->netdev && !ice_is_reset_in_progress(pf->state)) {
-		ice_napi_del(vsi);
+	if (vsi->netdev && !ice_is_reset_in_progress(pf->state))
 		unregister_netdev(vsi->netdev);
-		free_netdev(vsi->netdev);
-		vsi->netdev = NULL;
-	}
 
 	if (test_bit(ICE_FLAG_RSS_ENA, pf->flags))
 		ice_rss_clean(vsi);
@@ -2799,6 +2794,13 @@ int ice_vsi_release(struct ice_vsi *vsi)
 	ice_rm_vsi_lan_cfg(vsi->port_info, vsi->idx);
 	ice_vsi_delete(vsi);
 	ice_vsi_free_q_vectors(vsi);
+
+	/* make sure unregister_netdev() was called by checking __ICE_DOWN */
+	if (vsi->netdev && test_bit(__ICE_DOWN, vsi->state)) {
+		free_netdev(vsi->netdev);
+		vsi->netdev = NULL;
+	}
+
 	ice_vsi_clear_rings(vsi);
 
 	ice_vsi_put_qs(vsi);
drivers/net/ethernet/intel/ice/ice_main.c +1 -1
index 7843abf4d44d..dbf3d39ad8b1 100644
@@ -1667,7 +1667,7 @@ static int ice_req_irq_msix_misc(struct ice_pf *pf)
  * ice_napi_del - Remove NAPI handler for the VSI
  * @vsi: VSI for which NAPI handler is to be removed
  */
-void ice_napi_del(struct ice_vsi *vsi)
+static void ice_napi_del(struct ice_vsi *vsi)
 {
 	int v_idx;
 
Signature · 1 line
2.20.1
author maintainer
#33 lore raw

PATCH 34/249 net: mvpp2: cls: Extract the RSS context when parsing the ethtool rule

In reply to Sasha Levin

From: Maxime Chevallier <[email protected]>

[ Upstream commit c561da68038a738f30eca21456534c2d1872d13d ]

ethtool_rx_flow_rule_create takes into parameter the ethtool flow spec, which doesn't contain the rss context id. We therefore need to extract it ourself before parsing the ethtool rule.

The FLOW_RSS flag is only set in info->fs.flow_type, and not info->flow_type.

---
 drivers/net/ethernet/marvell/mvpp2/mvpp2_cls.c | 6 ++++++
 1 file changed, 6 insertions(+)
drivers/net/ethernet/marvell/mvpp2/mvpp2_cls.c +6 -0
index a57d17ab91f0..fb06c0aa620a 100644
@@ -1242,6 +1242,12 @@ int mvpp2_ethtool_cls_rule_ins(struct mvpp2_port *port,
 
 	input.fs = &info->fs;
 
+	/* We need to manually set the rss_ctx, since this info isn't present
+	 * in info->fs
+	 */
+	if (info->fs.flow_type & FLOW_RSS)
+		input.rss_ctx = info->rss_context;
+
 	ethtool_rule = ethtool_rx_flow_rule_create(&input);
 	if (IS_ERR(ethtool_rule)) {
 		ret = PTR_ERR(ethtool_rule);
Signature · 1 line
2.20.1
author maintainer
#34 lore raw

PATCH 35/249 net: hns3: initialize CPU reverse mapping

In reply to Sasha Levin

From: Jian Shen <[email protected]>

[ Upstream commit ffab9691bcb2fe2594f4c38bfceb4d9685b93b87 ]

Allocate CPU rmap and add entry for each irq. CPU rmap is used in aRFS to get the queue number of the rx completion interrupts.

In additional, remove the calling of irq_set_affinity_notifier() in hns3_nic_init_irq(), because we have registered notifier in irq_cpu_rmap_add() for each vector, otherwise it may cause use-after-free issue.

---
 .../net/ethernet/hisilicon/hns3/hns3_enet.c   | 77 ++++++++++++-------
 1 file changed, 48 insertions(+), 29 deletions(-)
drivers/net/ethernet/hisilicon/hns3/hns3_enet.c +48 -29
index f326805543a4..cd59c0cc636a 100644
@@ -4,6 +4,9 @@
 #include <linux/dma-mapping.h>
 #include <linux/etherdevice.h>
 #include <linux/interrupt.h>
+#ifdef CONFIG_RFS_ACCEL
+#include <linux/cpu_rmap.h>
+#endif
 #include <linux/if_vlan.h>
 #include <linux/ip.h>
 #include <linux/ipv6.h>
@@ -79,23 +82,6 @@ static irqreturn_t hns3_irq_handle(int irq, void *vector)
 	return IRQ_HANDLED;
 }
 
-/* This callback function is used to set affinity changes to the irq affinity
- * masks when the irq_set_affinity_notifier function is used.
- */
-static void hns3_nic_irq_affinity_notify(struct irq_affinity_notify *notify,
-					 const cpumask_t *mask)
-{
-	struct hns3_enet_tqp_vector *tqp_vectors =
-		container_of(notify, struct hns3_enet_tqp_vector,
-			     affinity_notify);
-
-	tqp_vectors->affinity_mask = *mask;
-}
-
-static void hns3_nic_irq_affinity_release(struct kref *ref)
-{
-}
-
 static void hns3_nic_uninit_irq(struct hns3_nic_priv *priv)
 {
 	struct hns3_enet_tqp_vector *tqp_vectors;
@@ -107,8 +93,7 @@ static void hns3_nic_uninit_irq(struct hns3_nic_priv *priv)
 		if (tqp_vectors->irq_init_flag != HNS3_VECTOR_INITED)
 			continue;
 
-		/* clear the affinity notifier and affinity mask */
-		irq_set_affinity_notifier(tqp_vectors->vector_irq, NULL);
+		/* clear the affinity mask */
 		irq_set_affinity_hint(tqp_vectors->vector_irq, NULL);
 
 		/* release the irq resource */
@@ -161,12 +146,6 @@ static int hns3_nic_init_irq(struct hns3_nic_priv *priv)
 			return ret;
 		}
 
-		tqp_vectors->affinity_notify.notify =
-					hns3_nic_irq_affinity_notify;
-		tqp_vectors->affinity_notify.release =
-					hns3_nic_irq_affinity_release;
-		irq_set_affinity_notifier(tqp_vectors->vector_irq,
-					  &tqp_vectors->affinity_notify);
 		irq_set_affinity_hint(tqp_vectors->vector_irq,
 				      &tqp_vectors->affinity_mask);
 
@@ -340,6 +319,40 @@ static void hns3_tqp_disable(struct hnae3_queue *tqp)
 	hns3_write_dev(tqp, HNS3_RING_EN_REG, rcb_reg);
 }
 
+static void hns3_free_rx_cpu_rmap(struct net_device *netdev)
+{
+#ifdef CONFIG_RFS_ACCEL
+	free_irq_cpu_rmap(netdev->rx_cpu_rmap);
+	netdev->rx_cpu_rmap = NULL;
+#endif
+}
+
+static int hns3_set_rx_cpu_rmap(struct net_device *netdev)
+{
+#ifdef CONFIG_RFS_ACCEL
+	struct hns3_nic_priv *priv = netdev_priv(netdev);
+	struct hns3_enet_tqp_vector *tqp_vector;
+	int i, ret;
+
+	if (!netdev->rx_cpu_rmap) {
+		netdev->rx_cpu_rmap = alloc_irq_cpu_rmap(priv->vector_num);
+		if (!netdev->rx_cpu_rmap)
+			return -ENOMEM;
+	}
+
+	for (i = 0; i < priv->vector_num; i++) {
+		tqp_vector = &priv->tqp_vector[i];
+		ret = irq_cpu_rmap_add(netdev->rx_cpu_rmap,
+				       tqp_vector->vector_irq);
+		if (ret) {
+			hns3_free_rx_cpu_rmap(netdev);
+			return ret;
+		}
+	}
+#endif
+	return 0;
+}
+
 static int hns3_nic_net_up(struct net_device *netdev)
 {
 	struct hns3_nic_priv *priv = netdev_priv(netdev);
@@ -351,11 +364,16 @@ static int hns3_nic_net_up(struct net_device *netdev)
 	if (ret)
 		return ret;
 
+	/* the device can work without cpu rmap, only aRFS needs it */
+	ret = hns3_set_rx_cpu_rmap(netdev);
+	if (ret)
+		netdev_warn(netdev, "set rx cpu rmap fail, ret=%d!\n", ret);
+
 	/* get irq resource for all vectors */
 	ret = hns3_nic_init_irq(priv);
 	if (ret) {
 		netdev_err(netdev, "hns init irq failed! ret=%d\n", ret);
-		return ret;
+		goto free_rmap;
 	}
 
 	clear_bit(HNS3_NIC_STATE_DOWN, &priv->state);
@@ -384,7 +402,8 @@ static int hns3_nic_net_up(struct net_device *netdev)
 		hns3_vector_disable(&priv->tqp_vector[j]);
 
 	hns3_nic_uninit_irq(priv);
-
+free_rmap:
+	hns3_free_rx_cpu_rmap(netdev);
 	return ret;
 }
 
@@ -467,6 +486,8 @@ static void hns3_nic_net_down(struct net_device *netdev)
 	if (ops->stop)
 		ops->stop(priv->ae_handle);
 
+	hns3_free_rx_cpu_rmap(netdev);
+
 	/* free irq resources */
 	hns3_nic_uninit_irq(priv);
 
@@ -3331,8 +3352,6 @@ static void hns3_nic_uninit_vector_data(struct hns3_nic_priv *priv)
 		hns3_free_vector_ring_chain(tqp_vector, &vector_ring_chain);
 
 		if (tqp_vector->irq_init_flag == HNS3_VECTOR_INITED) {
-			irq_set_affinity_notifier(tqp_vector->vector_irq,
-						  NULL);
 			irq_set_affinity_hint(tqp_vector->vector_irq, NULL);
 			free_irq(tqp_vector->vector_irq, tqp_vector);
 			tqp_vector->irq_init_flag = HNS3_VECTOR_NOT_INITED;
Signature · 1 line
2.20.1
author maintainer
#35 lore raw

PATCH 36/249 net: hns3: fix for FEC configuration

In reply to Sasha Levin

From: Jian Shen <[email protected]>

[ Upstream commit f438bfe9d4fe2e491505abfbf04d7c506e00d146 ]

The FEC capbility may be changed with port speed changes. Driver needs to read the active FEC mode, and update FEC capability when port speed changes.

Fixes 7e6ec9148a1d ("net: hns3: add support for FEC encoding control")
---
 drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c | 7 +++++++
 1 file changed, 7 insertions(+)
drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c +7 -0
index d3b1f8cb1155..4d9bcad26f06 100644
@@ -2508,6 +2508,9 @@ static void hclge_update_link_status(struct hclge_dev *hdev)
 
 static void hclge_update_port_capability(struct hclge_mac *mac)
 {
+	/* update fec ability by speed */
+	hclge_convert_setting_fec(mac);
+
 	/* firmware can not identify back plane type, the media type
 	 * read from configuration can help deal it
 	 */
@@ -2580,6 +2583,10 @@ static int hclge_get_sfp_info(struct hclge_dev *hdev, struct hclge_mac *mac)
 		mac->speed_ability = le32_to_cpu(resp->speed_ability);
 		mac->autoneg = resp->autoneg;
 		mac->support_autoneg = resp->autoneg_ability;
+		if (!resp->active_fec)
+			mac->fec_mode = 0;
+		else
+			mac->fec_mode = BIT(resp->active_fec);
 	} else {
 		mac->speed_type = QUERY_SFP_SPEED;
 	}
Signature · 1 line
2.20.1
author maintainer
#36 lore raw

PATCH 37/249 qed: Set the doorbell address correctly

In reply to Sasha Levin

From: Michal Kalderon <[email protected]>

[ Upstream commit 8366d520019f366fabd6c7a13032bdcd837e18d4 ]

In 100g mode the doorbell bar is united for both engines. Set the correct offset in the hwfn so that the doorbell returned for RoCE is in the affined hwfn.

---
 drivers/net/ethernet/qlogic/qed/qed_dev.c  | 29 ++++++++++++++--------
 drivers/net/ethernet/qlogic/qed/qed_rdma.c |  2 +-
 2 files changed, 19 insertions(+), 12 deletions(-)
drivers/net/ethernet/qlogic/qed/qed_dev.c +18 -11
index fccdb06fc5c5..8c40739e0d1b 100644
@@ -3443,6 +3443,7 @@ static void qed_nvm_info_free(struct qed_hwfn *p_hwfn)
 static int qed_hw_prepare_single(struct qed_hwfn *p_hwfn,
 				 void __iomem *p_regview,
 				 void __iomem *p_doorbells,
+				 u64 db_phys_addr,
 				 enum qed_pci_personality personality)
 {
 	struct qed_dev *cdev = p_hwfn->cdev;
@@ -3451,6 +3452,7 @@ static int qed_hw_prepare_single(struct qed_hwfn *p_hwfn,
 	/* Split PCI bars evenly between hwfns */
 	p_hwfn->regview = p_regview;
 	p_hwfn->doorbells = p_doorbells;
+	p_hwfn->db_phys_addr = db_phys_addr;
 
 	if (IS_VF(p_hwfn->cdev))
 		return qed_vf_hw_prepare(p_hwfn);
@@ -3546,7 +3548,9 @@ int qed_hw_prepare(struct qed_dev *cdev,
 	/* Initialize the first hwfn - will learn number of hwfns */
 	rc = qed_hw_prepare_single(p_hwfn,
 				   cdev->regview,
-				   cdev->doorbells, personality);
+				   cdev->doorbells,
+				   cdev->db_phys_addr,
+				   personality);
 	if (rc)
 		return rc;
 
@@ -3555,22 +3559,25 @@ int qed_hw_prepare(struct qed_dev *cdev,
 	/* Initialize the rest of the hwfns */
 	if (cdev->num_hwfns > 1) {
 		void __iomem *p_regview, *p_doorbell;
-		u8 __iomem *addr;
+		u64 db_phys_addr;
+		u32 offset;
 
 		/* adjust bar offset for second engine */
-		addr = cdev->regview +
-		       qed_hw_bar_size(p_hwfn, p_hwfn->p_main_ptt,
-				       BAR_ID_0) / 2;
-		p_regview = addr;
+		offset = qed_hw_bar_size(p_hwfn, p_hwfn->p_main_ptt,
+					 BAR_ID_0) / 2;
+		p_regview = cdev->regview + offset;
 
-		addr = cdev->doorbells +
-		       qed_hw_bar_size(p_hwfn, p_hwfn->p_main_ptt,
-				       BAR_ID_1) / 2;
-		p_doorbell = addr;
+		offset = qed_hw_bar_size(p_hwfn, p_hwfn->p_main_ptt,
+					 BAR_ID_1) / 2;
+
+		p_doorbell = cdev->doorbells + offset;
+
+		db_phys_addr = cdev->db_phys_addr + offset;
 
 		/* prepare second hw function */
 		rc = qed_hw_prepare_single(&cdev->hwfns[1], p_regview,
-					   p_doorbell, personality);
+					   p_doorbell, db_phys_addr,
+					   personality);
 
 		/* in case of error, need to free the previously
 		 * initiliazed hwfn 0.
drivers/net/ethernet/qlogic/qed/qed_rdma.c +1 -1
index 7873d6dfd91f..13802b825d65 100644
@@ -803,7 +803,7 @@ static int qed_rdma_add_user(void *rdma_cxt,
 				     dpi_start_offset +
 				     ((out_params->dpi) * p_hwfn->dpi_size));
 
-	out_params->dpi_phys_addr = p_hwfn->cdev->db_phys_addr +
+	out_params->dpi_phys_addr = p_hwfn->db_phys_addr +
 				    dpi_start_offset +
 				    ((out_params->dpi) * p_hwfn->dpi_size);
 
Signature · 1 line
2.20.1
author maintainer
#37 lore raw

PATCH 38/249 signal/pid_namespace: Fix reboot_pid_ns to use send_sig not force_sig

In reply to Sasha Levin

From: "Eric W. Biederman" <[email protected]>

[ Upstream commit f9070dc94542093fd516ae4ccea17ef46a4362c5 ]

The locking in force_sig_info is not prepared to deal with a task that exits or execs (as sighand may change). The is not a locking problem in force_sig as force_sig is only built to handle synchronous exceptions.

Further the function force_sig_info changes the signal state if the signal is ignored, or blocked or if SIGNAL_UNKILLABLE will prevent the delivery of the signal. The signal SIGKILL can not be ignored and can not be blocked and SIGNAL_UNKILLABLE won't prevent it from being delivered.

So using force_sig rather than send_sig for SIGKILL is confusing and pointless.

Because it won't impact the sending of the signal and and because using force_sig is wrong, replace force_sig with send_sig.

Cc Daniel Lezcano <[email protected]>
Cc Serge Hallyn <[email protected]>
Cc Oleg Nesterov <[email protected]>
Fixes cf3f89214ef6 ("pidns: add reboot_pid_ns() to handle the reboot syscall")
---
 kernel/pid_namespace.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/pid_namespace.c +1 -1
index f54bc7cb6c2d..6d726cef241c 100644
@@ -326,7 +326,7 @@ int reboot_pid_ns(struct pid_namespace *pid_ns, int cmd)
 	}
 
 	read_lock(&tasklist_lock);
-	force_sig(SIGKILL, pid_ns->child_reaper);
+	send_sig(SIGKILL, pid_ns->child_reaper, 1);
 	read_unlock(&tasklist_lock);
 
 	do_exit(0);
Signature · 1 line
2.20.1
author maintainer
#38 lore raw

PATCH 39/249 signal/cifs: Fix cifs_put_tcp_session to call send_sig instead of force_sig

In reply to Sasha Levin

From: "Eric W. Biederman" <[email protected]>

[ Upstream commit 72abe3bcf0911d69b46c1e8bdb5612675e0ac42c ]

The locking in force_sig_info is not prepared to deal with a task that exits or execs (as sighand may change). The is not a locking problem in force_sig as force_sig is only built to handle synchronous exceptions.

Further the function force_sig_info changes the signal state if the signal is ignored, or blocked or if SIGNAL_UNKILLABLE will prevent the delivery of the signal. The signal SIGKILL can not be ignored and can not be blocked and SIGNAL_UNKILLABLE won't prevent it from being delivered.

So using force_sig rather than send_sig for SIGKILL is confusing and pointless.

Because it won't impact the sending of the signal and and because using force_sig is wrong, replace force_sig with send_sig.

Cc Namjae Jeon <[email protected]>
Cc Jeff Layton <[email protected]>
Cc Steve French <[email protected]>
Fixes a5c3e1c725af ("Revert "cifs: No need to send SIGKILL to demux_thread during umount"")
Fixes e7ddee9037e7 ("cifs: disable sharing session and tcon and add new TCP sharing code")
---
 fs/cifs/connect.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/cifs/connect.c +1 -1
index 8dd6637a3cbb..714a359c7c8d 100644
@@ -2631,7 +2631,7 @@ cifs_put_tcp_session(struct TCP_Server_Info *server, int from_reconnect)
 
 	task = xchg(&server->tsk, NULL);
 	if (task)
-		force_sig(SIGKILL, task);
+		send_sig(SIGKILL, task, 1);
 }
 
 static struct TCP_Server_Info *
Signature · 1 line
2.20.1
#39 lore raw

Re: [PATCH AUTOSEL 5.2 039/249] signal/cifs: Fix cifs_put_tcp_session to call send_sig instead of force_sig

Quoted 16 lines · Sasha Levin

From: "Eric W. Biederman" <[email protected]>

[ Upstream commit 72abe3bcf0911d69b46c1e8bdb5612675e0ac42c ]

The locking in force_sig_info is not prepared to deal with a task that exits or execs (as sighand may change). The is not a locking problem in force_sig as force_sig is only built to handle synchronous exceptions.

Further the function force_sig_info changes the signal state if the signal is ignored, or blocked or if SIGNAL_UNKILLABLE will prevent the delivery of the signal. The signal SIGKILL can not be ignored and can not be blocked and SIGNAL_UNKILLABLE won't prevent it from being delivered.

Quoted 5 lines

So using force_sig rather than send_sig for SIGKILL is confusing and pointless.

Because it won't impact the sending of the signal and and because using force_sig is wrong, replace force_sig with send_sig.

I think this patch broke the cifs module.
The issue is that the use count is now not updated properly and thus it is no longer possible to
rmmod the module.

Quoted 28 lines
Cc Namjae Jeon <[email protected]>
Cc Jeff Layton <[email protected]>
Cc Steve French <[email protected]>
Fixes a5c3e1c725af ("Revert "cifs: No need to send SIGKILL to demux_thread during umount"")
Fixes e7ddee9037e7 ("cifs: disable sharing session and tcon and add new TCP sharing code")
---
 fs/cifs/connect.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/cifs/connect.c +1 -2
index 8dd6637a3cbb..714a359c7c8d 100644
@@ -2631,7 +2631,7 @@ cifs_put_tcp_session(struct TCP_Server_Info *server, int from_reconnect)
 
        task = xchg(&server->tsk, NULL);
        if (task)
-               force_sig(SIGKILL, task);
+               send_sig(SIGKILL, task, 1);
 }
 
 static struct TCP_Server_Info *
--

2.20.1

#40 lore raw

Re: [PATCH AUTOSEL 5.2 039/249] signal/cifs: Fix cifs_put_tcp_session to call send_sig instead of force_sig

Very easy to see what caused the regression with this global change:

mount (which launches "cifsd" thread to read the socket) umount (which kills the "cifsd" thread)
rmmod (rmmod now fails since "cifsd" thread is still active)

mount launches a thread to read from the socket ("cifsd") umount is supposed to kill that thread (but with the patch "signal/cifs: Fix cifs_put_tcp_session to call send_sig instead of force_sig" that no longer works). So the regression is that after unmount you still see the "cifsd" thread, and the reason that cifsd thread is still around is that that patch no longer force kills the process (see line 2652 of fs/cifs/connect.c) which regresses module removal.

- force_sig(SIGKILL, task);
+ send_sig(SIGKILL, task, 1);

The comment in the changeset indicates "The signal SIGKILL can not be ignored" but obviously it can be ignored - at least on 5.3-rc1 it is being ignored.

If send_sig(SIGKILL ...) doesn't work and if force_sig(SIGKILL, task) is removed and no longer possible - how do we kill a helper process
...

Quoted 58 lines · ronnie sahlberg
Quoted 16 lines · Sasha Levin

From: "Eric W. Biederman" <[email protected]>

[ Upstream commit 72abe3bcf0911d69b46c1e8bdb5612675e0ac42c ]

The locking in force_sig_info is not prepared to deal with a task that exits or execs (as sighand may change). The is not a locking problem in force_sig as force_sig is only built to handle synchronous exceptions.

Further the function force_sig_info changes the signal state if the signal is ignored, or blocked or if SIGNAL_UNKILLABLE will prevent the delivery of the signal. The signal SIGKILL can not be ignored and can not be blocked and SIGNAL_UNKILLABLE won't prevent it from being delivered.

Quoted 5 lines

So using force_sig rather than send_sig for SIGKILL is confusing and pointless.

Because it won't impact the sending of the signal and and because using force_sig is wrong, replace force_sig with send_sig.

I think this patch broke the cifs module.
The issue is that the use count is now not updated properly and thus it is no longer possible to
rmmod the module.

Quoted 28 lines
Cc Namjae Jeon <[email protected]>
Cc Jeff Layton <[email protected]>
Cc Steve French <[email protected]>
Fixes a5c3e1c725af ("Revert "cifs: No need to send SIGKILL to demux_thread during umount"")
Fixes e7ddee9037e7 ("cifs: disable sharing session and tcon and add new TCP sharing code")
---
 fs/cifs/connect.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/cifs/connect.c +1 -2
index 8dd6637a3cbb..714a359c7c8d 100644
@@ -2631,7 +2631,7 @@ cifs_put_tcp_session(struct TCP_Server_Info *server, int from_reconnect)
 
        task = xchg(&server->tsk, NULL);
        if (task)
-               force_sig(SIGKILL, task);
+               send_sig(SIGKILL, task, 1);
 }
 
 static struct TCP_Server_Info *
--

2.20.1

Signature · 3 lines
Thanks,

Steve
#41 lore raw

Re: [PATCH AUTOSEL 5.2 039/249] signal/cifs: Fix cifs_put_tcp_session to call send_sig instead of force_sig

Quoted 18 lines · Steve French

Very easy to see what caused the regression with this global change:

mount (which launches "cifsd" thread to read the socket) umount (which kills the "cifsd" thread)
rmmod (rmmod now fails since "cifsd" thread is still active)

mount launches a thread to read from the socket ("cifsd") umount is supposed to kill that thread (but with the patch "signal/cifs: Fix cifs_put_tcp_session to call send_sig instead of force_sig" that no longer works). So the regression is that after unmount you still see the "cifsd" thread, and the reason that cifsd thread is still around is that that patch no longer force kills the process (see line 2652 of fs/cifs/connect.c) which regresses module removal.

(quoted diff) +1 -1
-               force_sig(SIGKILL, task);
+               send_sig(SIGKILL, task, 1);
Quoted 7 lines

The comment in the changeset indicates "The signal SIGKILL can not be ignored" but obviously it can be ignored - at least on 5.3-rc1 it is being ignored.

If send_sig(SIGKILL ...) doesn't work and if force_sig(SIGKILL, task) is removed and no longer possible - how do we kill a helper process
...

I think I see what is happening. It looks like as well as misuinsg force_sig, cifs is also violating the invariant that keeps SIGKILL out of the blocked signal set.

For that force_sig will act differently. I did not consider it because that is never supposed to happen.

Can someone test this code below and confirm the issue goes away?

fs/cifs/transport.c +1 -0
index 5d6d44bfe10a..2a782ebc7b65 100644
@@ -347,6 +347,7 @@ __smb_send_rqst(struct TCP_Server_Info *server, int num_rqst,
 	 */
 
 	sigfillset(&mask);
+	sigdelset(&mask, SIGKILL);
 	sigprocmask(SIG_BLOCK, &mask, &oldmask);
 
 	/* Generate a rfc1002 marker for SMB2+ */

Eric

#42 lore raw

Re: [PATCH AUTOSEL 5.2 039/249] signal/cifs: Fix cifs_put_tcp_session to call send_sig instead of force_sig

Quoted 52 lines · Eric W. Biederman
Quoted 18 lines · Steve French

Very easy to see what caused the regression with this global change:

mount (which launches "cifsd" thread to read the socket) umount (which kills the "cifsd" thread)
rmmod (rmmod now fails since "cifsd" thread is still active)

mount launches a thread to read from the socket ("cifsd") umount is supposed to kill that thread (but with the patch "signal/cifs: Fix cifs_put_tcp_session to call send_sig instead of force_sig" that no longer works). So the regression is that after unmount you still see the "cifsd" thread, and the reason that cifsd thread is still around is that that patch no longer force kills the process (see line 2652 of fs/cifs/connect.c) which regresses module removal.

(quoted diff) +1 -1
-               force_sig(SIGKILL, task);
+               send_sig(SIGKILL, task, 1);
Quoted 7 lines

The comment in the changeset indicates "The signal SIGKILL can not be ignored" but obviously it can be ignored - at least on 5.3-rc1 it is being ignored.

If send_sig(SIGKILL ...) doesn't work and if force_sig(SIGKILL, task) is removed and no longer possible - how do we kill a helper process
...

I think I see what is happening. It looks like as well as misuinsg force_sig, cifs is also violating the invariant that keeps SIGKILL out of the blocked signal set.

For that force_sig will act differently. I did not consider it because that is never supposed to happen.

Can someone test this code below and confirm the issue goes away?

fs/cifs/transport.c +1 -0
index 5d6d44bfe10a..2a782ebc7b65 100644
@@ -347,6 +347,7 @@ __smb_send_rqst(struct TCP_Server_Info *server, int num_rqst,
         */
 
        sigfillset(&mask);
+       sigdelset(&mask, SIGKILL);
        sigprocmask(SIG_BLOCK, &mask, &oldmask);
 
        /* Generate a rfc1002 marker for SMB2+ */
Quoted 1 line

Eric

I just tried your suggestion and it didn't work. I also tried doing
a similar thing on the thread we are trying to kills ("cifsd" - ie which is blocked in the function cifs_demultiplex_thread waiting to read from the socket)

# git diff -a
fs/cifs/connect.c +3 -0
index a4830ced0f98..b73062520a17 100644
@@ -1104,6 +1104,7 @@ cifs_demultiplex_thread(void *p)
        struct task_struct *task_to_wake = NULL;
        struct mid_q_entry *mids[MAX_COMPOUND];
        char *bufs[MAX_COMPOUND];
+       sigset_t mask;
 
        current->flags |= PF_MEMALLOC;
        cifs_dbg(FYI, "Demultiplex PID: %d\n", task_pid_nr(current));
@@ -1113,6 +1114,8 @@ cifs_demultiplex_thread(void *p)
                mempool_resize(cifs_req_poolp, length + cifs_min_rcv);
 
        set_freezable();
+       sigfillset(&mask);
+       sigdelset(&mask, SIGKILL);
        while (server->tcpStatus != CifsExiting) {
                if (try_to_freeze())
                        continue;

That also didn't work. The only thing I have been able to find
which worked was:

fs/cifs/connect.c +1 -0
index a4830ced0f98..e74f04163fc9 100644
@@ -1113,6 +1113,7 @@ cifs_demultiplex_thread(void *p)
                mempool_resize(cifs_req_poolp, length + cifs_min_rcv);
 
        set_freezable();
+      allow_signal(SIGKILL);
        while (server->tcpStatus != CifsExiting) {
                if (try_to_freeze())
                        continue;

That fixes the problem ... but ... as Ronnie and others have noted it would allow a userspace process to make the mount unusable (all you would have to do would be to do a kill -9 of the "cifsd" process from some userspace process like bash and the mount would be unusable - so this sounds dangerous.

Is there an alternative that, in the process doing the unmount in kernel, would allow us to do the equivalent of:
"allow_signal(SIGKILL, <the id of the cifsd process>"
In otherwords, to minimize the risk of some userspace process killing cifsd, could we delay enabling allow_signal(SIGKILL) till the unmount begins by doing it for a different process (have the unmount process
enable signals for the cifsd process). Otherwise is there a way to
force kill a process from the kernel as we used to do - without running the risk of a user space process killing cifsd (which is bad).

Signature · 3 lines
Thanks,

Steve
#43 lore raw

Re: [PATCH AUTOSEL 5.2 039/249] signal/cifs: Fix cifs_put_tcp_session to call send_sig instead of force_sig

Pavel noticed I missed a line from the attempt to do a similar patch to Eric's suggestion
(it still didn't work though - although "allow_signal" does albeit is possibly dangerous as user space can kill cifsd)

# git diff -a
fs/cifs/connect.c +4 -0
index a4830ced0f98..8758dff18c15 100644
@@ -1104,6 +1104,7 @@ cifs_demultiplex_thread(void *p)
        struct task_struct *task_to_wake = NULL;
        struct mid_q_entry *mids[MAX_COMPOUND];
        char *bufs[MAX_COMPOUND];
+       sigset_t mask, oldmask;
 
        current->flags |= PF_MEMALLOC;
        cifs_dbg(FYI, "Demultiplex PID: %d\n", task_pid_nr(current));
@@ -1113,6 +1114,9 @@ cifs_demultiplex_thread(void *p)
                mempool_resize(cifs_req_poolp, length + cifs_min_rcv);
 
        set_freezable();
+       sigfillset(&mask);
+       sigdelset(&mask, SIGKILL);
+       sigprocmask(SIG_BLOCK, &mask, &oldmask);
        while (server->tcpStatus != CifsExiting) {
                if (try_to_freeze())
                        continue;
Quoted 121 lines · Steve French
Quoted 52 lines · Eric W. Biederman
Quoted 18 lines · Steve French

Very easy to see what caused the regression with this global change:

mount (which launches "cifsd" thread to read the socket) umount (which kills the "cifsd" thread)
rmmod (rmmod now fails since "cifsd" thread is still active)

mount launches a thread to read from the socket ("cifsd") umount is supposed to kill that thread (but with the patch "signal/cifs: Fix cifs_put_tcp_session to call send_sig instead of force_sig" that no longer works). So the regression is that after unmount you still see the "cifsd" thread, and the reason that cifsd thread is still around is that that patch no longer force kills the process (see line 2652 of fs/cifs/connect.c) which regresses module removal.

(quoted diff) +1 -1
-               force_sig(SIGKILL, task);
+               send_sig(SIGKILL, task, 1);
Quoted 7 lines

The comment in the changeset indicates "The signal SIGKILL can not be ignored" but obviously it can be ignored - at least on 5.3-rc1 it is being ignored.

If send_sig(SIGKILL ...) doesn't work and if force_sig(SIGKILL, task) is removed and no longer possible - how do we kill a helper process
...

I think I see what is happening. It looks like as well as misuinsg force_sig, cifs is also violating the invariant that keeps SIGKILL out of the blocked signal set.

For that force_sig will act differently. I did not consider it because that is never supposed to happen.

Can someone test this code below and confirm the issue goes away?

fs/cifs/transport.c +1 -0
index 5d6d44bfe10a..2a782ebc7b65 100644
@@ -347,6 +347,7 @@ __smb_send_rqst(struct TCP_Server_Info *server, int num_rqst,
         */
 
        sigfillset(&mask);
+       sigdelset(&mask, SIGKILL);
        sigprocmask(SIG_BLOCK, &mask, &oldmask);
 
        /* Generate a rfc1002 marker for SMB2+ */
Quoted 1 line

Eric

I just tried your suggestion and it didn't work. I also tried doing
a similar thing on the thread we are trying to kills ("cifsd" - ie which is blocked in the function cifs_demultiplex_thread waiting to read from the socket)

# git diff -a
fs/cifs/connect.c +3 -0
index a4830ced0f98..b73062520a17 100644
@@ -1104,6 +1104,7 @@ cifs_demultiplex_thread(void *p)
        struct task_struct *task_to_wake = NULL;
        struct mid_q_entry *mids[MAX_COMPOUND];
        char *bufs[MAX_COMPOUND];
+       sigset_t mask;
 
        current->flags |= PF_MEMALLOC;
        cifs_dbg(FYI, "Demultiplex PID: %d\n", task_pid_nr(current));
@@ -1113,6 +1114,8 @@ cifs_demultiplex_thread(void *p)
                mempool_resize(cifs_req_poolp, length + cifs_min_rcv);
 
        set_freezable();
+       sigfillset(&mask);
+       sigdelset(&mask, SIGKILL);
        while (server->tcpStatus != CifsExiting) {
                if (try_to_freeze())
                        continue;

That also didn't work. The only thing I have been able to find
which worked was:

fs/cifs/connect.c +1 -0
index a4830ced0f98..e74f04163fc9 100644
@@ -1113,6 +1113,7 @@ cifs_demultiplex_thread(void *p)
                mempool_resize(cifs_req_poolp, length + cifs_min_rcv);
 
        set_freezable();
+      allow_signal(SIGKILL);
        while (server->tcpStatus != CifsExiting) {
                if (try_to_freeze())
                        continue;

That fixes the problem ... but ... as Ronnie and others have noted it would allow a userspace process to make the mount unusable (all you would have to do would be to do a kill -9 of the "cifsd" process from some userspace process like bash and the mount would be unusable - so this sounds dangerous.

Is there an alternative that, in the process doing the unmount in kernel, would allow us to do the equivalent of:
"allow_signal(SIGKILL, <the id of the cifsd process>"
In otherwords, to minimize the risk of some userspace process killing cifsd, could we delay enabling allow_signal(SIGKILL) till the unmount begins by doing it for a different process (have the unmount process
enable signals for the cifsd process). Otherwise is there a way to
force kill a process from the kernel as we used to do - without running the risk of a user space process killing cifsd (which is bad).

--
Thanks,

Steve

Signature · 3 lines
Thanks,

Steve
#44 lore raw

Re: [PATCH AUTOSEL 5.2 039/249] signal/cifs: Fix cifs_put_tcp_session to call send_sig instead of force_sig

I did some additional testing and it looks like the "allow_signal" change may be safe enough

# git diff -a
fs/cifs/connect.c +1 -0
index a4830ced0f98..a15a6e738eb5 100644
@@ -1113,6 +1113,7 @@ cifs_demultiplex_thread(void *p)
                mempool_resize(cifs_req_poolp, length + cifs_min_rcv);
 
        set_freezable();
+       allow_signal(SIGKILL);
        while (server->tcpStatus != CifsExiting) {
                if (try_to_freeze())
                        continue;

See below:

root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# insmod ./cifs.ko
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# mount -t cifs
//localhost/scratch /mnt -o username=sfrench
Password for sfrench@//localhost/scratch:  ************
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# ps -A | grep cifsd
 5176 ?        00:00:00 cifsd
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# kill -9 5176
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# ls /mnt
0444  dir0750  dir0754  newfile
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# umount /mnt
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# ps -A | grep cifsd
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# rmmod cifs
Quoted 159 lines · Steve French

Pavel noticed I missed a line from the attempt to do a similar patch to Eric's suggestion
(it still didn't work though - although "allow_signal" does albeit is possibly dangerous as user space can kill cifsd)

# git diff -a
fs/cifs/connect.c +4 -0
index a4830ced0f98..8758dff18c15 100644
@@ -1104,6 +1104,7 @@ cifs_demultiplex_thread(void *p)
        struct task_struct *task_to_wake = NULL;
        struct mid_q_entry *mids[MAX_COMPOUND];
        char *bufs[MAX_COMPOUND];
+       sigset_t mask, oldmask;
 
        current->flags |= PF_MEMALLOC;
        cifs_dbg(FYI, "Demultiplex PID: %d\n", task_pid_nr(current));
@@ -1113,6 +1114,9 @@ cifs_demultiplex_thread(void *p)
                mempool_resize(cifs_req_poolp, length + cifs_min_rcv);
 
        set_freezable();
+       sigfillset(&mask);
+       sigdelset(&mask, SIGKILL);
+       sigprocmask(SIG_BLOCK, &mask, &oldmask);
        while (server->tcpStatus != CifsExiting) {
                if (try_to_freeze())
                        continue;
Quoted 117 lines · Steve French
Quoted 52 lines · Eric W. Biederman
Quoted 18 lines · Steve French

Very easy to see what caused the regression with this global change:

mount (which launches "cifsd" thread to read the socket) umount (which kills the "cifsd" thread)
rmmod (rmmod now fails since "cifsd" thread is still active)

mount launches a thread to read from the socket ("cifsd") umount is supposed to kill that thread (but with the patch "signal/cifs: Fix cifs_put_tcp_session to call send_sig instead of force_sig" that no longer works). So the regression is that after unmount you still see the "cifsd" thread, and the reason that cifsd thread is still around is that that patch no longer force kills the process (see line 2652 of fs/cifs/connect.c) which regresses module removal.

(quoted diff) +1 -1
-               force_sig(SIGKILL, task);
+               send_sig(SIGKILL, task, 1);
Quoted 7 lines

The comment in the changeset indicates "The signal SIGKILL can not be ignored" but obviously it can be ignored - at least on 5.3-rc1 it is being ignored.

If send_sig(SIGKILL ...) doesn't work and if force_sig(SIGKILL, task) is removed and no longer possible - how do we kill a helper process
...

I think I see what is happening. It looks like as well as misuinsg force_sig, cifs is also violating the invariant that keeps SIGKILL out of the blocked signal set.

For that force_sig will act differently. I did not consider it because that is never supposed to happen.

Can someone test this code below and confirm the issue goes away?

fs/cifs/transport.c +1 -0
index 5d6d44bfe10a..2a782ebc7b65 100644
@@ -347,6 +347,7 @@ __smb_send_rqst(struct TCP_Server_Info *server, int num_rqst,
         */
 
        sigfillset(&mask);
+       sigdelset(&mask, SIGKILL);
        sigprocmask(SIG_BLOCK, &mask, &oldmask);
 
        /* Generate a rfc1002 marker for SMB2+ */
Quoted 1 line

Eric

I just tried your suggestion and it didn't work. I also tried doing
a similar thing on the thread we are trying to kills ("cifsd" - ie which is blocked in the function cifs_demultiplex_thread waiting to read from the socket)

# git diff -a
fs/cifs/connect.c +3 -0
index a4830ced0f98..b73062520a17 100644
@@ -1104,6 +1104,7 @@ cifs_demultiplex_thread(void *p)
        struct task_struct *task_to_wake = NULL;
        struct mid_q_entry *mids[MAX_COMPOUND];
        char *bufs[MAX_COMPOUND];
+       sigset_t mask;
 
        current->flags |= PF_MEMALLOC;
        cifs_dbg(FYI, "Demultiplex PID: %d\n", task_pid_nr(current));
@@ -1113,6 +1114,8 @@ cifs_demultiplex_thread(void *p)
                mempool_resize(cifs_req_poolp, length + cifs_min_rcv);
 
        set_freezable();
+       sigfillset(&mask);
+       sigdelset(&mask, SIGKILL);
        while (server->tcpStatus != CifsExiting) {
                if (try_to_freeze())
                        continue;

That also didn't work. The only thing I have been able to find
which worked was:

fs/cifs/connect.c +1 -0
index a4830ced0f98..e74f04163fc9 100644
@@ -1113,6 +1113,7 @@ cifs_demultiplex_thread(void *p)
                mempool_resize(cifs_req_poolp, length + cifs_min_rcv);
 
        set_freezable();
+      allow_signal(SIGKILL);
        while (server->tcpStatus != CifsExiting) {
                if (try_to_freeze())
                        continue;

That fixes the problem ... but ... as Ronnie and others have noted it would allow a userspace process to make the mount unusable (all you would have to do would be to do a kill -9 of the "cifsd" process from some userspace process like bash and the mount would be unusable - so this sounds dangerous.

Is there an alternative that, in the process doing the unmount in kernel, would allow us to do the equivalent of:
"allow_signal(SIGKILL, <the id of the cifsd process>"
In otherwords, to minimize the risk of some userspace process killing cifsd, could we delay enabling allow_signal(SIGKILL) till the unmount begins by doing it for a different process (have the unmount process
enable signals for the cifsd process). Otherwise is there a way to
force kill a process from the kernel as we used to do - without running the risk of a user space process killing cifsd (which is bad).

Quoted 4 lines

--
Thanks,

Steve

--
Thanks,

Steve

Signature · 3 lines
Thanks,

Steve
Reviewed-by Ronnie Sahlberg
#45 lore raw

Re: [PATCH AUTOSEL 5.2 039/249] signal/cifs: Fix cifs_put_tcp_session to call send_sig instead of force_sig

Patch attached - tests out ok.

Quoted 199 lines · Steve French

I did some additional testing and it looks like the "allow_signal" change may be safe enough

# git diff -a
fs/cifs/connect.c +1 -0
index a4830ced0f98..a15a6e738eb5 100644
@@ -1113,6 +1113,7 @@ cifs_demultiplex_thread(void *p)
                mempool_resize(cifs_req_poolp, length + cifs_min_rcv);
 
        set_freezable();
+       allow_signal(SIGKILL);
        while (server->tcpStatus != CifsExiting) {
                if (try_to_freeze())
                        continue;

See below:

root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# insmod ./cifs.ko
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# mount -t cifs
//localhost/scratch /mnt -o username=sfrench
Password for sfrench@//localhost/scratch:  ************
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# ps -A | grep cifsd
 5176 ?        00:00:00 cifsd
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# kill -9 5176
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# ls /mnt
0444  dir0750  dir0754  newfile
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# umount /mnt
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# ps -A | grep cifsd
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# rmmod cifs
Quoted 151 lines · Steve French

Pavel noticed I missed a line from the attempt to do a similar patch to Eric's suggestion
(it still didn't work though - although "allow_signal" does albeit is possibly dangerous as user space can kill cifsd)

# git diff -a
fs/cifs/connect.c +4 -0
index a4830ced0f98..8758dff18c15 100644
@@ -1104,6 +1104,7 @@ cifs_demultiplex_thread(void *p)
        struct task_struct *task_to_wake = NULL;
        struct mid_q_entry *mids[MAX_COMPOUND];
        char *bufs[MAX_COMPOUND];
+       sigset_t mask, oldmask;
 
        current->flags |= PF_MEMALLOC;
        cifs_dbg(FYI, "Demultiplex PID: %d\n", task_pid_nr(current));
@@ -1113,6 +1114,9 @@ cifs_demultiplex_thread(void *p)
                mempool_resize(cifs_req_poolp, length + cifs_min_rcv);
 
        set_freezable();
+       sigfillset(&mask);
+       sigdelset(&mask, SIGKILL);
+       sigprocmask(SIG_BLOCK, &mask, &oldmask);
        while (server->tcpStatus != CifsExiting) {
                if (try_to_freeze())
                        continue;
Quoted 120 lines · Steve French
Quoted 52 lines · Eric W. Biederman
Quoted 18 lines · Steve French

Very easy to see what caused the regression with this global change:

mount (which launches "cifsd" thread to read the socket) umount (which kills the "cifsd" thread)
rmmod (rmmod now fails since "cifsd" thread is still active)

mount launches a thread to read from the socket ("cifsd") umount is supposed to kill that thread (but with the patch "signal/cifs: Fix cifs_put_tcp_session to call send_sig instead of force_sig" that no longer works). So the regression is that after unmount you still see the "cifsd" thread, and the reason that cifsd thread is still around is that that patch no longer force kills the process (see line 2652 of fs/cifs/connect.c) which regresses module removal.

(quoted diff) +1 -1
-               force_sig(SIGKILL, task);
+               send_sig(SIGKILL, task, 1);
Quoted 7 lines

The comment in the changeset indicates "The signal SIGKILL can not be ignored" but obviously it can be ignored - at least on 5.3-rc1 it is being ignored.

If send_sig(SIGKILL ...) doesn't work and if force_sig(SIGKILL, task) is removed and no longer possible - how do we kill a helper process
...

I think I see what is happening. It looks like as well as misuinsg force_sig, cifs is also violating the invariant that keeps SIGKILL out of the blocked signal set.

For that force_sig will act differently. I did not consider it because that is never supposed to happen.

Can someone test this code below and confirm the issue goes away?

fs/cifs/transport.c +1 -0
index 5d6d44bfe10a..2a782ebc7b65 100644
@@ -347,6 +347,7 @@ __smb_send_rqst(struct TCP_Server_Info *server, int num_rqst,
         */
 
        sigfillset(&mask);
+       sigdelset(&mask, SIGKILL);
        sigprocmask(SIG_BLOCK, &mask, &oldmask);
 
        /* Generate a rfc1002 marker for SMB2+ */
Quoted 1 line

Eric

I just tried your suggestion and it didn't work. I also tried doing
a similar thing on the thread we are trying to kills ("cifsd" - ie which is blocked in the function cifs_demultiplex_thread waiting to read from the socket)

# git diff -a
fs/cifs/connect.c +3 -0
index a4830ced0f98..b73062520a17 100644
@@ -1104,6 +1104,7 @@ cifs_demultiplex_thread(void *p)
        struct task_struct *task_to_wake = NULL;
        struct mid_q_entry *mids[MAX_COMPOUND];
        char *bufs[MAX_COMPOUND];
+       sigset_t mask;
 
        current->flags |= PF_MEMALLOC;
        cifs_dbg(FYI, "Demultiplex PID: %d\n", task_pid_nr(current));
@@ -1113,6 +1114,8 @@ cifs_demultiplex_thread(void *p)
                mempool_resize(cifs_req_poolp, length + cifs_min_rcv);
 
        set_freezable();
+       sigfillset(&mask);
+       sigdelset(&mask, SIGKILL);
        while (server->tcpStatus != CifsExiting) {
                if (try_to_freeze())
                        continue;

That also didn't work. The only thing I have been able to find
which worked was:

fs/cifs/connect.c +1 -0
index a4830ced0f98..e74f04163fc9 100644
@@ -1113,6 +1113,7 @@ cifs_demultiplex_thread(void *p)
                mempool_resize(cifs_req_poolp, length + cifs_min_rcv);
 
        set_freezable();
+      allow_signal(SIGKILL);
        while (server->tcpStatus != CifsExiting) {
                if (try_to_freeze())
                        continue;

That fixes the problem ... but ... as Ronnie and others have noted it would allow a userspace process to make the mount unusable (all you would have to do would be to do a kill -9 of the "cifsd" process from some userspace process like bash and the mount would be unusable - so this sounds dangerous.

Is there an alternative that, in the process doing the unmount in kernel, would allow us to do the equivalent of:
"allow_signal(SIGKILL, <the id of the cifsd process>"
In otherwords, to minimize the risk of some userspace process killing cifsd, could we delay enabling allow_signal(SIGKILL) till the unmount begins by doing it for a different process (have the unmount process
enable signals for the cifsd process). Otherwise is there a way to
force kill a process from the kernel as we used to do - without running the risk of a user space process killing cifsd (which is bad).

--
Thanks,

Quoted 8 lines
Quoted 1 line

Steve

--
Thanks,

Steve

--
Thanks,

Steve

Signature · 3 lines
Thanks,

Steve

From 6966af5a177be7fdab29cd6e871c2850c078385c Mon Sep 17 00:00:00 2001
From: Steve French <[email protected]>
Date: Tue, 23 Jul 2019 22:14:29 -0500
Subject: [PATCH] cifs: fix rmmod regression in cifs.ko caused by force_sig
changes

Fixes 72abe3bcf091 ("signal/cifs: Fix cifs_put_tcp_session to call send_sig instead of force_sig")

The global change from force_sig caused module unloading of cifs.ko to fail (since the cifsd process could not be killed, "rmmod cifs" now would always fail)

---
 fs/cifs/connect.c | 1 +
 1 file changed, 1 insertion(+)
fs/cifs/connect.c +1 -0
index a4830ced0f98..a15a6e738eb5 100644
@@ -1113,6 +1113,7 @@ cifs_demultiplex_thread(void *p)
 		mempool_resize(cifs_req_poolp, length + cifs_min_rcv);
 
 	set_freezable();
+	allow_signal(SIGKILL);
 	while (server->tcpStatus != CifsExiting) {
 		if (try_to_freeze())
 			continue;
Signature · 1 line
2.20.1
#46 lore raw

Re: [PATCH AUTOSEL 5.2 039/249] signal/cifs: Fix cifs_put_tcp_session to call send_sig instead of force_sig

In reply to Steve French
Quoted 18 lines · Steve French

I did some additional testing and it looks like the "allow_signal" change may be safe enough

# git diff -a
fs/cifs/connect.c +1 -0
index a4830ced0f98..a15a6e738eb5 100644
@@ -1113,6 +1113,7 @@ cifs_demultiplex_thread(void *p)
                mempool_resize(cifs_req_poolp, length + cifs_min_rcv);
 
        set_freezable();
+       allow_signal(SIGKILL);
        while (server->tcpStatus != CifsExiting) {
                if (try_to_freeze())
                        continue;

See below:

Quoted 12 lines
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# insmod ./cifs.ko
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# mount -t cifs
//localhost/scratch /mnt -o username=sfrench
Password for sfrench@//localhost/scratch:  ************
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# ps -A | grep cifsd
 5176 ?        00:00:00 cifsd
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# kill -9 5176
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# ls /mnt
0444  dir0750  dir0754  newfile
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# umount /mnt
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# ps -A | grep cifsd
root@smf-Thinkpad-P51:~/cifs-2.6/fs/cifs# rmmod cifs

Yes. I just discovered that kthreadd calls a function named ignore_signals that set all signals to SIG_IGN. Which becomes the default for all kernel threads. So something like allow_signal to change the signal handler is necessary. The blocking of SIGKILL is also concerning but apparently that is not the issue here.

Ideally I think cifs should use kthread_stop, instead of signals for this purpose. The logic is convoluted enough that reading through the cifs code quickly I don't see how sending SIGKILL to the daemon causes it to stop.

Eric

Quoted 160 lines · Steve French
Quoted 159 lines

Pavel noticed I missed a line from the attempt to do a similar patch to Eric's suggestion
(it still didn't work though - although "allow_signal" does albeit is possibly dangerous as user space can kill cifsd)

# git diff -a
fs/cifs/connect.c +4 -0
index a4830ced0f98..8758dff18c15 100644
@@ -1104,6 +1104,7 @@ cifs_demultiplex_thread(void *p)
        struct task_struct *task_to_wake = NULL;
        struct mid_q_entry *mids[MAX_COMPOUND];
        char *bufs[MAX_COMPOUND];
+       sigset_t mask, oldmask;
 
        current->flags |= PF_MEMALLOC;
        cifs_dbg(FYI, "Demultiplex PID: %d\n", task_pid_nr(current));
@@ -1113,6 +1114,9 @@ cifs_demultiplex_thread(void *p)
                mempool_resize(cifs_req_poolp, length + cifs_min_rcv);
 
        set_freezable();
+       sigfillset(&mask);
+       sigdelset(&mask, SIGKILL);
+       sigprocmask(SIG_BLOCK, &mask, &oldmask);
        while (server->tcpStatus != CifsExiting) {
                if (try_to_freeze())
                        continue;
Quoted 117 lines · Steve French
Quoted 52 lines · Eric W. Biederman
Quoted 18 lines · Steve French

Very easy to see what caused the regression with this global change:

mount (which launches "cifsd" thread to read the socket) umount (which kills the "cifsd" thread)
rmmod (rmmod now fails since "cifsd" thread is still active)

mount launches a thread to read from the socket ("cifsd") umount is supposed to kill that thread (but with the patch "signal/cifs: Fix cifs_put_tcp_session to call send_sig instead of force_sig" that no longer works). So the regression is that after unmount you still see the "cifsd" thread, and the reason that cifsd thread is still around is that that patch no longer force kills the process (see line 2652 of fs/cifs/connect.c) which regresses module removal.

(quoted diff) +1 -1
-               force_sig(SIGKILL, task);
+               send_sig(SIGKILL, task, 1);
Quoted 7 lines

The comment in the changeset indicates "The signal SIGKILL can not be ignored" but obviously it can be ignored - at least on 5.3-rc1 it is being ignored.

If send_sig(SIGKILL ...) doesn't work and if force_sig(SIGKILL, task) is removed and no longer possible - how do we kill a helper process
...

I think I see what is happening. It looks like as well as misuinsg force_sig, cifs is also violating the invariant that keeps SIGKILL out of the blocked signal set.

For that force_sig will act differently. I did not consider it because that is never supposed to happen.

Can someone test this code below and confirm the issue goes away?

fs/cifs/transport.c +1 -0
index 5d6d44bfe10a..2a782ebc7b65 100644
@@ -347,6 +347,7 @@ __smb_send_rqst(struct TCP_Server_Info *server, int num_rqst,
         */
 
        sigfillset(&mask);
+       sigdelset(&mask, SIGKILL);
        sigprocmask(SIG_BLOCK, &mask, &oldmask);
 
        /* Generate a rfc1002 marker for SMB2+ */
Quoted 1 line

Eric

I just tried your suggestion and it didn't work. I also tried doing
a similar thing on the thread we are trying to kills ("cifsd" - ie which is blocked in the function cifs_demultiplex_thread waiting to read from the socket)

# git diff -a
fs/cifs/connect.c +3 -0
index a4830ced0f98..b73062520a17 100644
@@ -1104,6 +1104,7 @@ cifs_demultiplex_thread(void *p)
        struct task_struct *task_to_wake = NULL;
        struct mid_q_entry *mids[MAX_COMPOUND];
        char *bufs[MAX_COMPOUND];
+       sigset_t mask;
 
        current->flags |= PF_MEMALLOC;
        cifs_dbg(FYI, "Demultiplex PID: %d\n", task_pid_nr(current));
@@ -1113,6 +1114,8 @@ cifs_demultiplex_thread(void *p)
                mempool_resize(cifs_req_poolp, length + cifs_min_rcv);
 
        set_freezable();
+       sigfillset(&mask);
+       sigdelset(&mask, SIGKILL);
        while (server->tcpStatus != CifsExiting) {
                if (try_to_freeze())
                        continue;

That also didn't work. The only thing I have been able to find
which worked was:

fs/cifs/connect.c +1 -0
index a4830ced0f98..e74f04163fc9 100644
@@ -1113,6 +1113,7 @@ cifs_demultiplex_thread(void *p)
                mempool_resize(cifs_req_poolp, length + cifs_min_rcv);
 
        set_freezable();
+      allow_signal(SIGKILL);
        while (server->tcpStatus != CifsExiting) {
                if (try_to_freeze())
                        continue;

That fixes the problem ... but ... as Ronnie and others have noted it would allow a userspace process to make the mount unusable (all you would have to do would be to do a kill -9 of the "cifsd" process from some userspace process like bash and the mount would be unusable - so this sounds dangerous.

Is there an alternative that, in the process doing the unmount in kernel, would allow us to do the equivalent of:
"allow_signal(SIGKILL, <the id of the cifsd process>"
In otherwords, to minimize the risk of some userspace process killing cifsd, could we delay enabling allow_signal(SIGKILL) till the unmount begins by doing it for a different process (have the unmount process
enable signals for the cifsd process). Otherwise is there a way to
force kill a process from the kernel as we used to do - without running the risk of a user space process killing cifsd (which is bad).

Quoted 4 lines

--
Thanks,

Steve

--
Thanks,

Steve

author maintainer
#47 lore raw

PATCH 40/249 af_key: fix leaks in key_pol_get_resp and dump_sp.

In reply to Sasha Levin

From: Jeremy Sowden <[email protected]>

[ Upstream commit 7c80eb1c7e2b8420477fbc998971d62a648035d9 ]

In both functions, if pfkey_xfrm_policy2msg failed we leaked the newly allocated sk_buff. Free it on error.

Fixes 55569ce256ce ("Fix conversion between IPSEC_MODE_xxx and XFRM_MODE_xxx.")
Reported-by [email protected]
---
 net/key/af_key.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
net/key/af_key.c +6 -2
index a50dd6f34b91..fe5fc4bab7ee 100644
@@ -2438,8 +2438,10 @@ static int key_pol_get_resp(struct sock *sk, struct xfrm_policy *xp, const struc
 		goto out;
 	}
 	err = pfkey_xfrm_policy2msg(out_skb, xp, dir);
-	if (err < 0)
+	if (err < 0) {
+		kfree_skb(out_skb);
 		goto out;
+	}
 
 	out_hdr = (struct sadb_msg *) out_skb->data;
 	out_hdr->sadb_msg_version = hdr->sadb_msg_version;
@@ -2690,8 +2692,10 @@ static int dump_sp(struct xfrm_policy *xp, int dir, int count, void *ptr)
 		return PTR_ERR(out_skb);
 
 	err = pfkey_xfrm_policy2msg(out_skb, xp, dir);
-	if (err < 0)
+	if (err < 0) {
+		kfree_skb(out_skb);
 		return err;
+	}
 
 	out_hdr = (struct sadb_msg *) out_skb->data;
 	out_hdr->sadb_msg_version = pfk->dump.msg_version;
Signature · 1 line
2.20.1
author maintainer
#48 lore raw

PATCH 41/249 xfrm: Fix xfrm sel prefix length validation

In reply to Sasha Levin

From: Anirudh Gupta <[email protected]>

[ Upstream commit b38ff4075a80b4da5cb2202d7965332ca0efb213 ]

Family of src/dst can be different from family of selector src/dst. Use xfrm selector family to validate address prefix length, while verifying new sa from userspace.

Validated patch with this command:
ip xfrm state add src 1.1.6.1 dst 1.1.6.2 proto esp spi 4260196 \
reqid 20004 mode tunnel aead "rfc4106(gcm(aes))" \
0x1111016400000000000000000000000044440001 128 \
sel src 1011:1:4::2/128 sel dst 1021:1:4::2/128 dev Port5

Fixes 07bf7908950a ("xfrm: Validate address prefix lengths in the xfrm selector.")
---
 net/xfrm/xfrm_user.c | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)
net/xfrm/xfrm_user.c +16 -0
index 173477211e40..76ad7e201626 100644
@@ -151,6 +151,22 @@ static int verify_newsa_info(struct xfrm_usersa_info *p,
 
 	err = -EINVAL;
 	switch (p->family) {
+	case AF_INET:
+		break;
+
+	case AF_INET6:
+#if IS_ENABLED(CONFIG_IPV6)
+		break;
+#else
+		err = -EAFNOSUPPORT;
+		goto out;
+#endif
+
+	default:
+		goto out;
+	}
+
+	switch (p->sel.family) {
 	case AF_INET:
 		if (p->sel.prefixlen_d > 32 || p->sel.prefixlen_s > 32)
 			goto out;
Signature · 1 line
2.20.1
author maintainer
#49 lore raw

PATCH 42/249 media: vim2m: fix two double-free issues

In reply to Sasha Levin

From: Kefeng Wang <[email protected]>

[ Upstream commit 20059cbbf981ca954be56f7963ae494d18e2dda1 ]

vim2m_device_release() will be called by video_unregister_device() to release various objects.

There are two double-free issue,
1. dev->m2m_dev will be freed twice in error_m2m path/vim2m_device_release
2. the error_v4l2 and error_free path in vim2m_probe() will release same objects, since vim2m_device_release has done.

Fixes ea6c7e34f3b2 ("media: vim2m: replace devm_kzalloc by kzalloc")
---
 drivers/media/platform/vim2m.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/media/platform/vim2m.c +3 -3
index 243c82b5d537..acd3bd48c7e2 100644
@@ -1359,7 +1359,7 @@ static int vim2m_probe(struct platform_device *pdev)
 						 MEDIA_ENT_F_PROC_VIDEO_SCALER);
 	if (ret) {
 		v4l2_err(&dev->v4l2_dev, "Failed to init mem2mem media controller\n");
-		goto error_m2m;
+		goto error_dev;
 	}
 
 	ret = media_device_register(&dev->mdev);
@@ -1373,11 +1373,11 @@ static int vim2m_probe(struct platform_device *pdev)
 #ifdef CONFIG_MEDIA_CONTROLLER
 error_m2m_mc:
 	v4l2_m2m_unregister_media_controller(dev->m2m_dev);
-error_m2m:
-	v4l2_m2m_release(dev->m2m_dev);
 #endif
 error_dev:
 	video_unregister_device(&dev->vfd);
+	/* vim2m_device_release called by video_unregister_device to release various objects */
+	return ret;
 error_v4l2:
 	v4l2_device_unregister(&dev->v4l2_dev);
 error_free:
Signature · 1 line
2.20.1
author maintainer
#50 lore raw

PATCH 43/249 media: v4l2-core: fix use-after-free error

In reply to Sasha Levin

From: sumitg <[email protected]>

[ Upstream commit 3e0f724346e96daae7792262c6767449795ac3b5 ]

Fixing use-after-free within __v4l2_ctrl_handler_setup(). Memory is being freed with kfree(new_ref) for duplicate control reference entry but ctrl->cluster pointer is still referring to freed duplicate entry resulting in error on access. Change done to update cluster pointer only when new control reference is added.

==================================================================

 BUG: KASAN: use-after-free in __v4l2_ctrl_handler_setup+0x388/0x428
 Read of size 8 at addr ffffffc324e78618 by task systemd-udevd/312

 Allocated by task 312:

 Freed by task 312:

 The buggy address belongs to the object at ffffffc324e78600
  which belongs to the cache kmalloc-64 of size 64
 The buggy address is located 24 bytes inside of
  64-byte region [ffffffc324e78600, ffffffc324e78640)
 The buggy address belongs to the page:
 page:ffffffbf0c939e00 count:1 mapcount:0 mapping:
					(null) index:0xffffffc324e78f80
 flags: 0x4000000000000100(slab)
 raw: 4000000000000100 0000000000000000 ffffffc324e78f80 000000018020001a
 raw: 0000000000000000 0000000100000001 ffffffc37040fb80 0000000000000000
 page dumped because: kasan: bad access detected

 Memory state around the buggy address:
  ffffffc324e78500: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
  ffffffc324e78580: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
 >ffffffc324e78600: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
                             ^
  ffffffc324e78680: 00 00 00 00 00 00 00 00 fc fc fc fc fc fc fc fc
  ffffffc324e78700: 00 00 00 00 00 fc fc fc fc fc fc fc fc fc fc fc
 ==================================================================
---
 drivers/media/v4l2-core/v4l2-ctrls.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)
drivers/media/v4l2-core/v4l2-ctrls.c +9 -9
index 7d3a33258748..3c720f54efa8 100644
@@ -2149,15 +2149,6 @@ static int handler_new_ref(struct v4l2_ctrl_handler *hdl,
 	if (size_extra_req)
 		new_ref->p_req.p = &new_ref[1];
 
-	if (ctrl->handler == hdl) {
-		/* By default each control starts in a cluster of its own.
-		   new_ref->ctrl is basically a cluster array with one
-		   element, so that's perfect to use as the cluster pointer.
-		   But only do this for the handler that owns the control. */
-		ctrl->cluster = &new_ref->ctrl;
-		ctrl->ncontrols = 1;
-	}
-
 	INIT_LIST_HEAD(&new_ref->node);
 
 	mutex_lock(hdl->lock);
@@ -2190,6 +2181,15 @@ static int handler_new_ref(struct v4l2_ctrl_handler *hdl,
 	hdl->buckets[bucket] = new_ref;
 	if (ctrl_ref)
 		*ctrl_ref = new_ref;
+	if (ctrl->handler == hdl) {
+		/* By default each control starts in a cluster of its own.
+		 * new_ref->ctrl is basically a cluster array with one
+		 * element, so that's perfect to use as the cluster pointer.
+		 * But only do this for the handler that owns the control.
+		 */
+		ctrl->cluster = &new_ref->ctrl;
+		ctrl->ncontrols = 1;
+	}
 
 unlock:
 	mutex_unlock(hdl->lock);
Signature · 1 line
2.20.1
author maintainer
#51 lore raw

PATCH 44/249 fscrypt: clean up some BUG_ON()s in block encryption/decryption

In reply to Sasha Levin

From: Eric Biggers <[email protected]>

[ Upstream commit eeacfdc68a104967162dfcba60f53f6f5b62a334 ]

Replace some BUG_ON()s with WARN_ON_ONCE() and returning an error code, and move the check for len divisible by FS_CRYPTO_BLOCK_SIZE into fscrypt_crypt_block() so that it's done for both encryption and decryption, not just encryption.

---
 fs/crypto/crypto.c | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)
fs/crypto/crypto.c +9 -6
index 335a362ee446..6f753198eeef 100644
@@ -154,7 +154,10 @@ int fscrypt_do_page_crypto(const struct inode *inode, fscrypt_direction_t rw,
 	struct crypto_skcipher *tfm = ci->ci_ctfm;
 	int res = 0;
 
-	BUG_ON(len == 0);
+	if (WARN_ON_ONCE(len <= 0))
+		return -EINVAL;
+	if (WARN_ON_ONCE(len % FS_CRYPTO_BLOCK_SIZE != 0))
+		return -EINVAL;
 
 	fscrypt_generate_iv(&iv, lblk_num, ci);
 
@@ -238,8 +241,6 @@ struct page *fscrypt_encrypt_page(const struct inode *inode,
 	struct page *ciphertext_page = page;
 	int err;
 
-	BUG_ON(len % FS_CRYPTO_BLOCK_SIZE != 0);
-
 	if (inode->i_sb->s_cop->flags & FS_CFLG_OWN_PAGES) {
 		/* with inplace-encryption we just encrypt the page */
 		err = fscrypt_do_page_crypto(inode, FS_ENCRYPT, lblk_num, page,
@@ -251,7 +252,8 @@ struct page *fscrypt_encrypt_page(const struct inode *inode,
 		return ciphertext_page;
 	}
 
-	BUG_ON(!PageLocked(page));
+	if (WARN_ON_ONCE(!PageLocked(page)))
+		return ERR_PTR(-EINVAL);
 
 	ctx = fscrypt_get_ctx(gfp_flags);
 	if (IS_ERR(ctx))
@@ -299,8 +301,9 @@ EXPORT_SYMBOL(fscrypt_encrypt_page);
 int fscrypt_decrypt_page(const struct inode *inode, struct page *page,
 			unsigned int len, unsigned int offs, u64 lblk_num)
 {
-	if (!(inode->i_sb->s_cop->flags & FS_CFLG_OWN_PAGES))
-		BUG_ON(!PageLocked(page));
+	if (WARN_ON_ONCE(!PageLocked(page) &&
+			 !(inode->i_sb->s_cop->flags & FS_CFLG_OWN_PAGES)))
+		return -EINVAL;
 
 	return fscrypt_do_page_crypto(inode, FS_DECRYPT, lblk_num, page, page,
 				      len, offs, GFP_NOFS);
Signature · 1 line
2.20.1
author maintainer
#52 lore raw

PATCH 45/249 media: usb:zr364xx:Fix KASAN:null-ptr-deref Read in zr364xx_vidioc_querycap

In reply to Sasha Levin

From: Vandana BN <[email protected]>

[ Upstream commit 5d2e73a5f80a5b5aff3caf1ec6d39b5b3f54b26e ]

SyzKaller hit the null pointer deref while reading from uninitialized udev->product in zr364xx_vidioc_querycap().

==================================================================

BUG: KASAN: null-ptr-deref in read_word_at_a_time+0xe/0x20
include/linux/compiler.h:274
Read of size 1 at addr 0000000000000000 by task v4l_id/5287

CPU: 1 PID: 5287 Comm: v4l_id Not tainted 5.1.0-rc3-319004-g43151d6 #6
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS
Google 01/01/2011
Call Trace:
  __dump_stack lib/dump_stack.c:77 [inline]
  dump_stack+0xe8/0x16e lib/dump_stack.c:113
  kasan_report.cold+0x5/0x3c mm/kasan/report.c:321
  read_word_at_a_time+0xe/0x20 include/linux/compiler.h:274
  strscpy+0x8a/0x280 lib/string.c:207
  zr364xx_vidioc_querycap+0xb5/0x210 drivers/media/usb/zr364xx/zr364xx.c:706
  v4l_querycap+0x12b/0x340 drivers/media/v4l2-core/v4l2-ioctl.c:1062
  __video_do_ioctl+0x5bb/0xb40 drivers/media/v4l2-core/v4l2-ioctl.c:2874
  video_usercopy+0x44e/0xf00 drivers/media/v4l2-core/v4l2-ioctl.c:3056
  v4l2_ioctl+0x14e/0x1a0 drivers/media/v4l2-core/v4l2-dev.c:364
  vfs_ioctl fs/ioctl.c:46 [inline]
  file_ioctl fs/ioctl.c:509 [inline]
  do_vfs_ioctl+0xced/0x12f0 fs/ioctl.c:696
  ksys_ioctl+0xa0/0xc0 fs/ioctl.c:713
  __do_sys_ioctl fs/ioctl.c:720 [inline]
  __se_sys_ioctl fs/ioctl.c:718 [inline]
  __x64_sys_ioctl+0x74/0xb0 fs/ioctl.c:718
  do_syscall_64+0xcf/0x4f0 arch/x86/entry/common.c:290
  entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x7f3b56d8b347
Code: 90 90 90 48 8b 05 f1 fa 2a 00 64 c7 00 26 00 00 00 48 c7 c0 ff ff ff

ff c3 90 90 90 90 90 90 90 90 90 90 b8 10 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d c1 fa 2a 00 31 d2 48 29 c2 64
RSP: 002b:00007ffe005d5d68 EFLAGS: 00000202 ORIG_RAX: 0000000000000010
RAX: ffffffffffffffda RBX: 0000000000000003 RCX: 00007f3b56d8b347
RDX: 00007ffe005d5d70 RSI: 0000000080685600 RDI: 0000000000000003
RBP: 0000000000000000 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000202 R12: 0000000000400884
R13: 00007ffe005d5ec0 R14: 0000000000000000 R15: 0000000000000000
==================================================================

For this device udev->product is not initialized and accessing it causes a NULL pointer deref.

The fix is to check for NULL before strscpy() and copy empty string, if product is NULL

---
 drivers/media/usb/zr364xx/zr364xx.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/media/usb/zr364xx/zr364xx.c +2 -1
index 37a7992585df..48803eb773ed 100644
@@ -694,7 +694,8 @@ static int zr364xx_vidioc_querycap(struct file *file, void *priv,
 	struct zr364xx_camera *cam = video_drvdata(file);
 
 	strscpy(cap->driver, DRIVER_DESC, sizeof(cap->driver));
-	strscpy(cap->card, cam->udev->product, sizeof(cap->card));
+	if (cam->udev->product)
+		strscpy(cap->card, cam->udev->product, sizeof(cap->card));
 	strscpy(cap->bus_info, dev_name(&cam->udev->dev),
 		sizeof(cap->bus_info));
 	cap->device_caps = V4L2_CAP_VIDEO_CAPTURE |
Signature · 1 line
2.20.1
author maintainer
#53 lore raw

PATCH 46/249 perf annotate TUI browser: Do not use member from variable within its own initialization

In reply to Sasha Levin

From: Arnaldo Carvalho de Melo <[email protected]>

[ Upstream commit da2019633f0b5c105ce658aada333422d8cb28fe ]

Some compilers will complain when using a member of a struct to initialize another member, in the same struct initialization.

For instance:

debian:8 Debian clang version 3.5.0-10 (tags/RELEASE_350/final) (based on LLVM 3.5.0)
oraclelinux:7 clang version 3.4.2 (tags/RELEASE_34/dot2-final)

Produce:

ui/browsers/annotate.c:104:12: error: variable 'ops' is uninitialized when used within its own initialization [-Werror,-Wuninitialized]

                                              (!ops.current_entry ||
                                                ^~~

1 error generated.

So use an extra variable, initialized just before that struct, to have the value used in the expressions used to init two of the struct members.

Cc Adrian Hunter <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Fixes c298304bd747 ("perf annotate: Use a ops table for annotation_line__write()")
---
 tools/perf/ui/browsers/annotate.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
tools/perf/ui/browsers/annotate.c +3 -2
index 98d934a36d86..b0d089a95dac 100644
@@ -97,11 +97,12 @@ static void annotate_browser__write(struct ui_browser *browser, void *entry, int
 	struct annotate_browser *ab = container_of(browser, struct annotate_browser, b);
 	struct annotation *notes = browser__annotation(browser);
 	struct annotation_line *al = list_entry(entry, struct annotation_line, node);
+	const bool is_current_entry = ui_browser__is_current_entry(browser, row);
 	struct annotation_write_ops ops = {
 		.first_line		 = row == 0,
-		.current_entry		 = ui_browser__is_current_entry(browser, row),
+		.current_entry		 = is_current_entry,
 		.change_color		 = (!notes->options->hide_src_code &&
-					    (!ops.current_entry ||
+					    (!is_current_entry ||
 					     (browser->use_navkeypressed &&
 					      !browser->navkeypressed))),
 		.width			 = browser->width,
Signature · 1 line
2.20.1
author maintainer
#54 lore raw

PATCH 47/249 Revert "e1000e: fix cyclic resets at link up with active tx"

In reply to Sasha Levin

From: Konstantin Khlebnikov <[email protected]>

[ Upstream commit caff422ea81e144842bc44bab408d85ac449377b ]

This reverts commit 0f9e980bf5ee1a97e2e401c846b2af989eb21c61.

That change cased false-positive warning about hardware hang:

e1000e: eth0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: Rx/Tx
IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready e1000e 0000:00:1f.6 eth0: Detected Hardware Unit Hang:

   TDH                  <0>
   TDT                  <1>
   next_to_use          <1>
   next_to_clean        <0>
buffer_info[next_to_clean]:
   time_stamp           <fffba7a7>
   next_to_watch        <0>
   jiffies              <fffbb140>
   next_to_watch.status <0>
MAC Status             <40080080>
PHY Status             <7949>
PHY 1000BASE-T Status  <0>
PHY Extended Status    <3000>
PCI Status             <10>

e1000e: eth0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: Rx/Tx

Besides warning everything works fine.
Original issue will be fixed property in following patch.

---
 drivers/net/ethernet/intel/e1000e/netdev.c | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)
drivers/net/ethernet/intel/e1000e/netdev.c +9 -6
index 0e09bede42a2..e21b2ffd1e92 100644
@@ -5308,13 +5308,8 @@ static void e1000_watchdog_task(struct work_struct *work)
 			/* 8000ES2LAN requires a Rx packet buffer work-around
 			 * on link down event; reset the controller to flush
 			 * the Rx packet buffer.
-			 *
-			 * If the link is lost the controller stops DMA, but
-			 * if there is queued Tx work it cannot be done.  So
-			 * reset the controller to flush the Tx packet buffers.
 			 */
-			if ((adapter->flags & FLAG_RX_NEEDS_RESTART) ||
-			    e1000_desc_unused(tx_ring) + 1 < tx_ring->count)
+			if (adapter->flags & FLAG_RX_NEEDS_RESTART)
 				adapter->flags |= FLAG_RESTART_NOW;
 			else
 				pm_schedule_suspend(netdev->dev.parent,
@@ -5337,6 +5332,14 @@ static void e1000_watchdog_task(struct work_struct *work)
 	adapter->gotc_old = adapter->stats.gotc;
 	spin_unlock(&adapter->stats64_lock);
 
+	/* If the link is lost the controller stops DMA, but
+	 * if there is queued Tx work it cannot be done.  So
+	 * reset the controller to flush the Tx packet buffers.
+	 */
+	if (!netif_carrier_ok(netdev) &&
+	    (e1000_desc_unused(tx_ring) + 1 < tx_ring->count))
+		adapter->flags |= FLAG_RESTART_NOW;
+
 	/* If reset is necessary, do it outside of interrupt context. */
 	if (adapter->flags & FLAG_RESTART_NOW) {
 		schedule_work(&adapter->reset_task);
Signature · 1 line
2.20.1
author maintainer
#55 lore raw

PATCH 48/249 e1000e: start network tx queue only when link is up

In reply to Sasha Levin

From: Konstantin Khlebnikov <[email protected]>

[ Upstream commit d17ba0f616a08f597d9348c372d89b8c0405ccf3 ]

Driver does not want to keep packets in Tx queue when link is lost. But present code only reset NIC to flush them, but does not prevent queuing new packets. Moreover reset sequence itself could generate new packets via netconsole and NIC falls into endless reset loop.

This patch wakes Tx queue only when NIC is ready to send packets.

This is proper fix for problem addressed by commit 0f9e980bf5ee ("e1000e: fix cyclic resets at link up with active tx").

---
 drivers/net/ethernet/intel/e1000e/netdev.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/net/ethernet/intel/e1000e/netdev.c +4 -2
index e21b2ffd1e92..b081a1ef6859 100644
@@ -4208,7 +4208,7 @@ void e1000e_up(struct e1000_adapter *adapter)
 		e1000_configure_msix(adapter);
 	e1000_irq_enable(adapter);
 
-	netif_start_queue(adapter->netdev);
+	/* Tx queue started by watchdog timer when link is up */
 
 	e1000e_trigger_lsc(adapter);
 }
@@ -4606,6 +4606,7 @@ int e1000e_open(struct net_device *netdev)
 	pm_runtime_get_sync(&pdev->dev);
 
 	netif_carrier_off(netdev);
+	netif_stop_queue(netdev);
 
 	/* allocate transmit descriptors */
 	err = e1000e_setup_tx_resources(adapter->tx_ring);
@@ -4666,7 +4667,6 @@ int e1000e_open(struct net_device *netdev)
 	e1000_irq_enable(adapter);
 
 	adapter->tx_hang_recheck = false;
-	netif_start_queue(netdev);
 
 	hw->mac.get_link_status = true;
 	pm_runtime_put(&pdev->dev);
@@ -5288,6 +5288,7 @@ static void e1000_watchdog_task(struct work_struct *work)
 			if (phy->ops.cfg_on_link_up)
 				phy->ops.cfg_on_link_up(hw);
 
+			netif_wake_queue(netdev);
 			netif_carrier_on(netdev);
 
 			if (!test_bit(__E1000_DOWN, &adapter->state))
@@ -5301,6 +5302,7 @@ static void e1000_watchdog_task(struct work_struct *work)
 			/* Link status message must follow this format */
 			pr_info("%s NIC Link is Down\n", adapter->netdev->name);
 			netif_carrier_off(netdev);
+			netif_stop_queue(netdev);
 			if (!test_bit(__E1000_DOWN, &adapter->state))
 				mod_timer(&adapter->phy_info_timer,
 					  round_jiffies(jiffies + 2 * HZ));
Signature · 1 line
2.20.1
author maintainer
#56 lore raw

PATCH 49/249 ice: Check all VFs for MDD activity, don't disable

In reply to Sasha Levin

From: Mitch Williams <[email protected]>

[ Upstream commit 23c0112246b454e408fb0579b3f9089353d4d327 ]

Don't use the mdd_detected variable as an exit condition for this loop; the first VF to NOT have an MDD event will cause the loop to terminate.

Instead just look at all of the VFs, but don't disable them. This prevents proper release of resources if the VFs are rebooted or the VF driver reloaded. Instead, just log a message and call out repeat offenders.

To make it clear what we are doing, use a differently-named variable in the loop.

---
 drivers/net/ethernet/intel/ice/ice_main.c | 23 +++++++++++------------
 1 file changed, 11 insertions(+), 12 deletions(-)
drivers/net/ethernet/intel/ice/ice_main.c +11 -12
index dbf3d39ad8b1..1c803106e301 100644
@@ -1161,16 +1161,16 @@ static void ice_handle_mdd_event(struct ice_pf *pf)
 		}
 	}
 
-	/* see if one of the VFs needs to be reset */
-	for (i = 0; i < pf->num_alloc_vfs && mdd_detected; i++) {
+	/* check to see if one of the VFs caused the MDD */
+	for (i = 0; i < pf->num_alloc_vfs; i++) {
 		struct ice_vf *vf = &pf->vf[i];
 
-		mdd_detected = false;
+		bool vf_mdd_detected = false;
 
 		reg = rd32(hw, VP_MDET_TX_PQM(i));
 		if (reg & VP_MDET_TX_PQM_VALID_M) {
 			wr32(hw, VP_MDET_TX_PQM(i), 0xFFFF);
-			mdd_detected = true;
+			vf_mdd_detected = true;
 			dev_info(&pf->pdev->dev, "TX driver issue detected on VF %d\n",
 				 i);
 		}
@@ -1178,7 +1178,7 @@ static void ice_handle_mdd_event(struct ice_pf *pf)
 		reg = rd32(hw, VP_MDET_TX_TCLAN(i));
 		if (reg & VP_MDET_TX_TCLAN_VALID_M) {
 			wr32(hw, VP_MDET_TX_TCLAN(i), 0xFFFF);
-			mdd_detected = true;
+			vf_mdd_detected = true;
 			dev_info(&pf->pdev->dev, "TX driver issue detected on VF %d\n",
 				 i);
 		}
@@ -1186,7 +1186,7 @@ static void ice_handle_mdd_event(struct ice_pf *pf)
 		reg = rd32(hw, VP_MDET_TX_TDPU(i));
 		if (reg & VP_MDET_TX_TDPU_VALID_M) {
 			wr32(hw, VP_MDET_TX_TDPU(i), 0xFFFF);
-			mdd_detected = true;
+			vf_mdd_detected = true;
 			dev_info(&pf->pdev->dev, "TX driver issue detected on VF %d\n",
 				 i);
 		}
@@ -1194,19 +1194,18 @@ static void ice_handle_mdd_event(struct ice_pf *pf)
 		reg = rd32(hw, VP_MDET_RX(i));
 		if (reg & VP_MDET_RX_VALID_M) {
 			wr32(hw, VP_MDET_RX(i), 0xFFFF);
-			mdd_detected = true;
+			vf_mdd_detected = true;
 			dev_info(&pf->pdev->dev, "RX driver issue detected on VF %d\n",
 				 i);
 		}
 
-		if (mdd_detected) {
+		if (vf_mdd_detected) {
 			vf->num_mdd_events++;
-			dev_info(&pf->pdev->dev,
-				 "Use PF Control I/F to re-enable the VF\n");
-			set_bit(ICE_VF_STATE_DIS, vf->vf_states);
+			if (vf->num_mdd_events > 1)
+				dev_info(&pf->pdev->dev, "VF %d has had %llu MDD events since last boot\n",
+					 i, vf->num_mdd_events);
 		}
 	}
-
 }
 
 /**
Signature · 1 line
2.20.1
author maintainer
#57 lore raw

PATCH 50/249 media: mc-device.c: don't memset __user pointer contents

In reply to Sasha Levin

From: Hans Verkuil <[email protected]>

[ Upstream commit 518fa4e0e0da97ea2e17c95ab57647ce748a96e2 ]

You can't memset the contents of a __user pointer. Instead, call copy_to_user to copy links.reserved (which is zeroed) to the user memory.

This fixes this sparse warning:

SPARSE:drivers/media/mc/mc-device.c drivers/media/mc/mc-device.c:521:16: warning: incorrect type in argument 1 (different address spaces)

Fixes f49308878d720 ("media: media_device_enum_links32: clean a reserved field")
---
 drivers/media/media-device.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
drivers/media/media-device.c +3 -2
index b9bb4904bba1..e19df5165e78 100644
@@ -510,8 +510,9 @@ static long media_device_enum_links32(struct media_device *mdev,
 	if (ret)
 		return ret;
 
-	memset(ulinks->reserved, 0, sizeof(ulinks->reserved));
-
+	if (copy_to_user(ulinks->reserved, links.reserved,
+			 sizeof(ulinks->reserved)))
+		return -EFAULT;
 	return 0;
 }
 
Signature · 1 line
2.20.1
author maintainer
#58 lore raw

PATCH 51/249 media: saa7164: fix remove_proc_entry warning

In reply to Sasha Levin

From: Kefeng Wang <[email protected]>

[ Upstream commit 50710eeefbc1ed25375942aad0c4d1eb4af0f330 ]

if saa7164_proc_create() fails, saa7164_fini() will trigger a warning,

name 'saa7164'

WARNING: CPU: 1 PID: 6311 at fs/proc/generic.c:672 remove_proc_entry+0x1e8/0x3a0
  ? remove_proc_entry+0x1e8/0x3a0
  ? try_stop_module+0x7b/0x240
  ? proc_readdir+0x70/0x70
  ? rcu_read_lock_sched_held+0xd7/0x100
  saa7164_fini+0x13/0x1f [saa7164]
  __x64_sys_delete_module+0x30c/0x480
  ? __ia32_sys_delete_module+0x480/0x480
  ? __x64_sys_clock_gettime+0x11e/0x1c0
  ? __x64_sys_timer_create+0x1a0/0x1a0
  ? trace_hardirqs_off_caller+0x40/0x180
  ? do_syscall_64+0x18/0x450
  do_syscall_64+0x9f/0x450
  entry_SYSCALL_64_after_hwframe+0x49/0xbe

Fix it by checking the return of proc_create_single() before calling remove_proc_entry().

[[email protected]: use 0444 instead of S_IRUGO]
[[email protected]: use pr_info instead of KERN_INFO]
---
 drivers/media/pci/saa7164/saa7164-core.c | 33 ++++++++++++++++--------
 1 file changed, 22 insertions(+), 11 deletions(-)
drivers/media/pci/saa7164/saa7164-core.c +22 -11
index c594aff92e70..9ae04e18e6c6 100644
@@ -1112,16 +1112,25 @@ static int saa7164_proc_show(struct seq_file *m, void *v)
 	return 0;
 }
 
+static struct proc_dir_entry *saa7164_pe;
+
 static int saa7164_proc_create(void)
 {
-	struct proc_dir_entry *pe;
-
-	pe = proc_create_single("saa7164", S_IRUGO, NULL, saa7164_proc_show);
-	if (!pe)
+	saa7164_pe = proc_create_single("saa7164", 0444, NULL, saa7164_proc_show);
+	if (!saa7164_pe)
 		return -ENOMEM;
 
 	return 0;
 }
+
+static void saa7164_proc_destroy(void)
+{
+	if (saa7164_pe)
+		remove_proc_entry("saa7164", NULL);
+}
+#else
+static int saa7164_proc_create(void) { return 0; }
+static void saa7164_proc_destroy(void) {}
 #endif
 
 static int saa7164_thread_function(void *data)
@@ -1493,19 +1502,21 @@ static struct pci_driver saa7164_pci_driver = {
 
 static int __init saa7164_init(void)
 {
-	printk(KERN_INFO "saa7164 driver loaded\n");
+	int ret = pci_register_driver(&saa7164_pci_driver);
+
+	if (ret)
+		return ret;
 
-#ifdef CONFIG_PROC_FS
 	saa7164_proc_create();
-#endif
-	return pci_register_driver(&saa7164_pci_driver);
+
+	pr_info("saa7164 driver loaded\n");
+
+	return 0;
 }
 
 static void __exit saa7164_fini(void)
 {
-#ifdef CONFIG_PROC_FS
-	remove_proc_entry("saa7164", NULL);
-#endif
+	saa7164_proc_destroy();
 	pci_unregister_driver(&saa7164_pci_driver);
 }
 
Signature · 1 line
2.20.1
author maintainer
#59 lore raw

PATCH 52/249 ASoC: Intel: sof-rt5682: fix undefined references with Baytrail-only support

In reply to Sasha Levin

From: Pierre-Louis Bossart <[email protected]>

[ Upstream commit 17fc24875da1bef4650cf007edae3b2e26d2fa4e ]

The sof-rt5682 machine driver supports both legacy Baytrail devices and more recent ApolloLake/CometLake platforms. When only Baytrail is selected, the compilation fails with the following errors:

ERROR: "hdac_hdmi_jack_port_init" [sound/soc/intel/boards/snd-soc-sof_rt5682.ko] undefined!

ERROR: "hdac_hdmi_jack_init" [sound/soc/intel/boards/snd-soc-sof_rt5682.ko] undefined!

Fix by selecting SND_SOC_HDAC_HDMI unconditionally. The code for HDMI support is not reachable on Baytrail so this change has no functional impact.

Fixes f70abd75b7c6 ("ASoC: Intel: add sof-rt5682 machine driver")
---
 sound/soc/intel/boards/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
sound/soc/intel/boards/Kconfig +1 -1
index 5407d217228e..c0aef45d335a 100644
@@ -392,7 +392,7 @@ config SND_SOC_INTEL_SOF_RT5682_MACH
 		   (SND_SOC_SOF_BAYTRAIL && X86_INTEL_LPSS)
 	select SND_SOC_RT5682
 	select SND_SOC_DMIC
-	select SND_SOC_HDAC_HDMI if SND_SOC_SOF_HDA_COMMON
+	select SND_SOC_HDAC_HDMI
 	help
 	   This adds support for ASoC machine driver for SOF platforms
 	   with rt5682 codec.
Signature · 1 line
2.20.1
author maintainer
#60 lore raw

PATCH 53/249 media: staging: media: davinci_vpfe: - Fix for memory leak if decoder initialization fails.

In reply to Sasha Levin

From: Shailendra Verma <[email protected]>

[ Upstream commit 6995a659101bd4effa41cebb067f9dc18d77520d ]

Fix to avoid possible memory leak if the decoder initialization got failed.Free the allocated memory for file handle object before return in case decoder initialization fails.

---
 drivers/staging/media/davinci_vpfe/vpfe_video.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/staging/media/davinci_vpfe/vpfe_video.c +3 -0
index 510202a3b091..84cca18e3e9d 100644
@@ -419,6 +419,9 @@ static int vpfe_open(struct file *file)
 	/* If decoder is not initialized. initialize it */
 	if (!video->initialized && vpfe_update_pipe_state(video)) {
 		mutex_unlock(&video->lock);
+		v4l2_fh_del(&handle->vfh);
+		v4l2_fh_exit(&handle->vfh);
+		kfree(handle);
 		return -ENODEV;
 	}
 	/* Increment device users counter */
Signature · 1 line
2.20.1
author maintainer
#61 lore raw

PATCH 54/249 net: phy: Check against net_device being NULL

In reply to Sasha Levin

From: Ioana Ciornei <[email protected]>

[ Upstream commit 82c76aca81187b3d28a6fb3062f6916450ce955e ]

In general, we don't want MAC drivers calling phy_attach_direct with the net_device being NULL. Add checks against this in all the functions calling it: phy_attach() and phy_connect_direct().

---
 drivers/net/phy/phy_device.c | 6 ++++++
 1 file changed, 6 insertions(+)
drivers/net/phy/phy_device.c +6 -0
index dcc93a873174..a3f8740c6163 100644
@@ -948,6 +948,9 @@ int phy_connect_direct(struct net_device *dev, struct phy_device *phydev,
 {
 	int rc;
 
+	if (!dev)
+		return -EINVAL;
+
 	rc = phy_attach_direct(dev, phydev, phydev->dev_flags, interface);
 	if (rc)
 		return rc;
@@ -1290,6 +1293,9 @@ struct phy_device *phy_attach(struct net_device *dev, const char *bus_id,
 	struct device *d;
 	int rc;
 
+	if (!dev)
+		return ERR_PTR(-EINVAL);
+
 	/* Search the list of PHY devices on the mdio bus for the
 	 * PHY with the requested name
 	 */
Signature · 1 line
2.20.1
author maintainer
#62 lore raw

PATCH 55/249 net: dsa: sja1105: Fix broken fixed-link interfaces on user ports

In reply to Sasha Levin

From: Vladimir Oltean <[email protected]>

[ Upstream commit af7cd0366ee994e8b35985d407261dc0ed9dfb4d ]

PHYLIB and PHYLINK handle fixed-link interfaces differently. PHYLIB wraps them in a software PHY ("pseudo fixed link") phydev construct such that .adjust_link driver callbacks see an unified API. Whereas PHYLINK simply creates a phylink_link_state structure and passes it to .mac_config.

At the time the driver was introduced, DSA was using PHYLIB for the CPU/cascade ports (the ones with no net devices) and PHYLINK for everything else.

As explained below:

commit aab9c4067d2389d0adfc9c53806437df7b0fe3d5

Author: Florian Fainelli <[email protected]>
Date:   Thu May 10 13:17:36 2018 -0700

net: dsa: Plug in PHYLINK support

Drivers that utilize fixed links for user-facing ports (e.g: bcm_sf2) will need to implement phylink_mac_ops from now on to preserve functionality, since PHYLINK *does not* create a phy_device instance for fixed links.

In the above patch, DSA guards the .phylink_mac_config callback against a NULL phydev pointer. Therefore, .adjust_link is not called in case of a fixed-link user port.

This patch fixes the situation by converting the driver from using .adjust_link to .phylink_mac_config. This can be done now in a unified fashion for both slave and CPU/cascade ports because DSA now uses PHYLINK for all ports.

---
 drivers/net/dsa/sja1105/sja1105_main.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)
drivers/net/dsa/sja1105/sja1105_main.c +6 -5
index 1c3959efebc4..844e038f3dc6 100644
@@ -734,15 +734,16 @@ static int sja1105_adjust_port_config(struct sja1105_private *priv, int port,
 	return sja1105_clocking_setup_port(priv, port);
 }
 
-static void sja1105_adjust_link(struct dsa_switch *ds, int port,
-				struct phy_device *phydev)
+static void sja1105_mac_config(struct dsa_switch *ds, int port,
+			       unsigned int link_an_mode,
+			       const struct phylink_link_state *state)
 {
 	struct sja1105_private *priv = ds->priv;
 
-	if (!phydev->link)
+	if (!state->link)
 		sja1105_adjust_port_config(priv, port, 0, false);
 	else
-		sja1105_adjust_port_config(priv, port, phydev->speed, true);
+		sja1105_adjust_port_config(priv, port, state->speed, true);
 }
 
 static void sja1105_phylink_validate(struct dsa_switch *ds, int port,
@@ -1515,9 +1516,9 @@ static int sja1105_set_ageing_time(struct dsa_switch *ds,
 static const struct dsa_switch_ops sja1105_switch_ops = {
 	.get_tag_protocol	= sja1105_get_tag_protocol,
 	.setup			= sja1105_setup,
-	.adjust_link		= sja1105_adjust_link,
 	.set_ageing_time	= sja1105_set_ageing_time,
 	.phylink_validate	= sja1105_phylink_validate,
+	.phylink_mac_config	= sja1105_mac_config,
 	.get_strings		= sja1105_get_strings,
 	.get_ethtool_stats	= sja1105_get_ethtool_stats,
 	.get_sset_count		= sja1105_get_sset_count,
Signature · 1 line
2.20.1
author maintainer
#63 lore raw

PATCH 56/249 crypto: talitos - properly handle split ICV.

In reply to Sasha Levin

From: Christophe Leroy <[email protected]>

[ Upstream commit eae55a586c3c8b50982bad3c3426e9c9dd7a0075 ]

The driver assumes that the ICV is as a single piece in the last element of the scatterlist. This assumption is wrong.

This patch ensures that the ICV is properly handled regardless of the scatterlist layout.

Fixes 9c4a79653b35 ("crypto: talitos - Freescale integrated security engine (SEC) driver")
---
 drivers/crypto/talitos.c | 26 +++++++++++++++-----------
 1 file changed, 15 insertions(+), 11 deletions(-)
drivers/crypto/talitos.c +15 -11
index 2d9a0971a7fc..eec880909fdf 100644
@@ -1050,7 +1050,6 @@ static void ipsec_esp_encrypt_done(struct device *dev,
 	unsigned int authsize = crypto_aead_authsize(authenc);
 	unsigned int ivsize = crypto_aead_ivsize(authenc);
 	struct talitos_edesc *edesc;
-	struct scatterlist *sg;
 	void *icvdata;
 
 	edesc = container_of(desc, struct talitos_edesc, desc);
@@ -1064,9 +1063,8 @@ static void ipsec_esp_encrypt_done(struct device *dev,
 		else
 			icvdata = &edesc->link_tbl[edesc->src_nents +
 						   edesc->dst_nents + 2];
-		sg = sg_last(areq->dst, edesc->dst_nents);
-		memcpy((char *)sg_virt(sg) + sg->length - authsize,
-		       icvdata, authsize);
+		sg_pcopy_from_buffer(areq->dst, edesc->dst_nents ? : 1, icvdata,
+				     authsize, areq->assoclen + areq->cryptlen);
 	}
 
 	dma_unmap_single(dev, edesc->iv_dma, ivsize, DMA_TO_DEVICE);
@@ -1084,7 +1082,6 @@ static void ipsec_esp_decrypt_swauth_done(struct device *dev,
 	struct crypto_aead *authenc = crypto_aead_reqtfm(req);
 	unsigned int authsize = crypto_aead_authsize(authenc);
 	struct talitos_edesc *edesc;
-	struct scatterlist *sg;
 	char *oicv, *icv;
 	struct talitos_private *priv = dev_get_drvdata(dev);
 	bool is_sec1 = has_ftr_sec1(priv);
@@ -1094,9 +1091,18 @@ static void ipsec_esp_decrypt_swauth_done(struct device *dev,
 	ipsec_esp_unmap(dev, edesc, req);
 
 	if (!err) {
+		char icvdata[SHA512_DIGEST_SIZE];
+		int nents = edesc->dst_nents ? : 1;
+		unsigned int len = req->assoclen + req->cryptlen;
+
 		/* auth check */
-		sg = sg_last(req->dst, edesc->dst_nents ? : 1);
-		icv = (char *)sg_virt(sg) + sg->length - authsize;
+		if (nents > 1) {
+			sg_pcopy_to_buffer(req->dst, nents, icvdata, authsize,
+					   len - authsize);
+			icv = icvdata;
+		} else {
+			icv = (char *)sg_virt(req->dst) + len - authsize;
+		}
 
 		if (edesc->dma_len) {
 			if (is_sec1)
@@ -1516,7 +1522,6 @@ static int aead_decrypt(struct aead_request *req)
 	struct talitos_ctx *ctx = crypto_aead_ctx(authenc);
 	struct talitos_private *priv = dev_get_drvdata(ctx->dev);
 	struct talitos_edesc *edesc;
-	struct scatterlist *sg;
 	void *icvdata;
 
 	req->cryptlen -= authsize;
@@ -1550,9 +1555,8 @@ static int aead_decrypt(struct aead_request *req)
 	else
 		icvdata = &edesc->link_tbl[0];
 
-	sg = sg_last(req->src, edesc->src_nents ? : 1);
-
-	memcpy(icvdata, (char *)sg_virt(sg) + sg->length - authsize, authsize);
+	sg_pcopy_to_buffer(req->src, edesc->src_nents ? : 1, icvdata, authsize,
+			   req->assoclen + req->cryptlen - authsize);
 
 	return ipsec_esp(edesc, req, ipsec_esp_decrypt_swauth_done);
 }
Signature · 1 line
2.20.1
author maintainer
#64 lore raw

PATCH 57/249 crypto: talitos - Align SEC1 accesses to 32 bits boundaries.

In reply to Sasha Levin

From: Christophe Leroy <[email protected]>

[ Upstream commit c9cca7034b34a2d82e9a03b757de2485c294851c ]

The MPC885 reference manual states:

SEC Lite-initiated 8xx writes can occur only on 32-bit-word boundaries, but reads can occur on any byte boundary. Writing back a header read from a non-32-bit-word boundary will yield unpredictable results.

In order to ensure that, cra_alignmask is set to 3 for SEC1.

Fixes 9c4a79653b35 ("crypto: talitos - Freescale integrated security engine (SEC) driver")
---
 drivers/crypto/talitos.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
drivers/crypto/talitos.c +4 -1
index eec880909fdf..b6433d58df41 100644
@@ -3278,7 +3278,10 @@ static struct talitos_crypto_alg *talitos_alg_alloc(struct device *dev,
 		alg->cra_priority = t_alg->algt.priority;
 	else
 		alg->cra_priority = TALITOS_CRA_PRIORITY;
-	alg->cra_alignmask = 0;
+	if (has_ftr_sec1(priv))
+		alg->cra_alignmask = 3;
+	else
+		alg->cra_alignmask = 0;
 	alg->cra_ctxsize = sizeof(struct talitos_ctx);
 	alg->cra_flags |= CRYPTO_ALG_KERN_DRIVER_ONLY;
 
Signature · 1 line
2.20.1
author maintainer
#65 lore raw

PATCH 58/249 tua6100: Avoid build warnings.

In reply to Sasha Levin

From: "David S. Miller" <[email protected]>

[ Upstream commit 621ccc6cc5f8d6730b740d31d4818227866c93c9 ]

Rename _P to _P_VAL and _R to _R_VAL to avoid global namespace conflicts:

drivers/media/dvb-frontends/tua6100.c: In function ‘tua6100_set_params’:
drivers/media/dvb-frontends/tua6100.c:79: warning: "_P" redefined
#define _P 32

In file included from ./include/acpi/platform/aclinux.h:54,

                 from ./include/acpi/platform/acenv.h:152,
                 from ./include/acpi/acpi.h:22,
                 from ./include/linux/acpi.h:34,
                 from ./include/linux/i2c.h:17,
                 from drivers/media/dvb-frontends/tua6100.h:30,
                 from drivers/media/dvb-frontends/tua6100.c:32:

./include/linux/ctype.h:14: note: this is the location of the previous definition
#define _P 0x10 /* punct */

---
 drivers/media/dvb-frontends/tua6100.c | 22 +++++++++++-----------
 1 file changed, 11 insertions(+), 11 deletions(-)
drivers/media/dvb-frontends/tua6100.c +11 -11
index f7c3e6be8e4d..2483f614d0e7 100644
@@ -67,8 +67,8 @@ static int tua6100_set_params(struct dvb_frontend *fe)
 	struct i2c_msg msg1 = { .addr = priv->i2c_address, .flags = 0, .buf = reg1, .len = 4 };
 	struct i2c_msg msg2 = { .addr = priv->i2c_address, .flags = 0, .buf = reg2, .len = 3 };
 
-#define _R 4
-#define _P 32
+#define _R_VAL 4
+#define _P_VAL 32
 #define _ri 4000000
 
 	// setup register 0
@@ -83,14 +83,14 @@ static int tua6100_set_params(struct dvb_frontend *fe)
 	else
 		reg1[1] = 0x0c;
 
-	if (_P == 64)
+	if (_P_VAL == 64)
 		reg1[1] |= 0x40;
 	if (c->frequency >= 1525000)
 		reg1[1] |= 0x80;
 
 	// register 2
-	reg2[1] = (_R >> 8) & 0x03;
-	reg2[2] = _R;
+	reg2[1] = (_R_VAL >> 8) & 0x03;
+	reg2[2] = _R_VAL;
 	if (c->frequency < 1455000)
 		reg2[1] |= 0x1c;
 	else if (c->frequency < 1630000)
@@ -102,18 +102,18 @@ static int tua6100_set_params(struct dvb_frontend *fe)
 	 * The N divisor ratio (note: c->frequency is in kHz, but we
 	 * need it in Hz)
 	 */
-	prediv = (c->frequency * _R) / (_ri / 1000);
-	div = prediv / _P;
+	prediv = (c->frequency * _R_VAL) / (_ri / 1000);
+	div = prediv / _P_VAL;
 	reg1[1] |= (div >> 9) & 0x03;
 	reg1[2] = div >> 1;
 	reg1[3] = (div << 7);
-	priv->frequency = ((div * _P) * (_ri / 1000)) / _R;
+	priv->frequency = ((div * _P_VAL) * (_ri / 1000)) / _R_VAL;
 
 	// Finally, calculate and store the value for A
-	reg1[3] |= (prediv - (div*_P)) & 0x7f;
+	reg1[3] |= (prediv - (div*_P_VAL)) & 0x7f;
 
-#undef _R
-#undef _P
+#undef _R_VAL
+#undef _P_VAL
 #undef _ri
 
 	if (fe->ops.i2c_gate_ctrl)
Signature · 1 line
2.20.1
author maintainer
#66 lore raw

PATCH 59/249 batman-adv: Fix duplicated OGMs on NETDEV_UP

In reply to Sasha Levin

From: Sven Eckelmann <[email protected]>

[ Upstream commit 9e6b5648bbc4cd48fab62cecbb81e9cc3c6e7e88 ]

The state of slave interfaces are handled differently depending on whether the interface is up or not. All active interfaces (IFF_UP) will transmit OGMs. But for B.A.T.M.A.N. IV, also non-active interfaces are scheduling (low TTL) OGMs on active interfaces. The code which setups and schedules the OGMs must therefore already be called when the interfaces gets added as slave interface and the transmit function must then check whether it has to send out the OGM or not on the specific slave interface.

But the commit f0d97253fb5f ("batman-adv: remove ogm_emit and ogm_schedule API calls") moved the setup code from the enable function to the activate function. The latter is called either when the added slave was already up when batadv_hardif_enable_interface processed the new interface or when a NETDEV_UP event was received for this slave interfac. As result, each NETDEV_UP would schedule a new OGM worker for the interface and thus OGMs would be send a lot more than expected.

Fixes f0d97253fb5f ("batman-adv: remove ogm_emit and ogm_schedule API calls")
---
 net/batman-adv/bat_iv_ogm.c     | 4 ++--
 net/batman-adv/hard-interface.c | 3 +++
 net/batman-adv/types.h          | 3 +++
 3 files changed, 8 insertions(+), 2 deletions(-)
net/batman-adv/bat_iv_ogm.c +2 -2
index bd4138ddf7e0..240ed70912d6 100644
@@ -2337,7 +2337,7 @@ batadv_iv_ogm_neigh_is_sob(struct batadv_neigh_node *neigh1,
 	return ret;
 }
 
-static void batadv_iv_iface_activate(struct batadv_hard_iface *hard_iface)
+static void batadv_iv_iface_enabled(struct batadv_hard_iface *hard_iface)
 {
 	/* begin scheduling originator messages on that interface */
 	batadv_iv_ogm_schedule(hard_iface);
@@ -2683,8 +2683,8 @@ static void batadv_iv_gw_dump(struct sk_buff *msg, struct netlink_callback *cb,
 static struct batadv_algo_ops batadv_batman_iv __read_mostly = {
 	.name = "BATMAN_IV",
 	.iface = {
-		.activate = batadv_iv_iface_activate,
 		.enable = batadv_iv_ogm_iface_enable,
+		.enabled = batadv_iv_iface_enabled,
 		.disable = batadv_iv_ogm_iface_disable,
 		.update_mac = batadv_iv_ogm_iface_update_mac,
 		.primary_set = batadv_iv_ogm_primary_iface_set,
net/batman-adv/hard-interface.c +3 -0
index 79d1731b8306..3719cfd026f0 100644
@@ -795,6 +795,9 @@ int batadv_hardif_enable_interface(struct batadv_hard_iface *hard_iface,
 
 	batadv_hardif_recalc_extra_skbroom(soft_iface);
 
+	if (bat_priv->algo_ops->iface.enabled)
+		bat_priv->algo_ops->iface.enabled(hard_iface);
+
 out:
 	return 0;
 
net/batman-adv/types.h +3 -0
index 74b644738a36..e0b25104cbfa 100644
@@ -2129,6 +2129,9 @@ struct batadv_algo_iface_ops {
 	/** @enable: init routing info when hard-interface is enabled */
 	int (*enable)(struct batadv_hard_iface *hard_iface);
 
+	/** @enabled: notification when hard-interface was enabled (optional) */
+	void (*enabled)(struct batadv_hard_iface *hard_iface);
+
 	/** @disable: de-init routing info when hard-interface is disabled */
 	void (*disable)(struct batadv_hard_iface *hard_iface);
 
Signature · 1 line
2.20.1
author maintainer
#67 lore raw

PATCH 60/249 locking/lockdep: Fix OOO unlock when hlocks need merging

In reply to Sasha Levin

From: Imre Deak <[email protected]>

[ Upstream commit 8c8889d8eaf4501ae4aaf870b6f8f55db5d5109a ]

The sequence

	static DEFINE_WW_CLASS(test_ww_class);

	struct ww_acquire_ctx ww_ctx;
	struct ww_mutex ww_lock_a;
	struct ww_mutex ww_lock_b;
	struct mutex lock_c;
	struct mutex lock_d;

	ww_acquire_init(&ww_ctx, &test_ww_class);

	ww_mutex_init(&ww_lock_a, &test_ww_class);
	ww_mutex_init(&ww_lock_b, &test_ww_class);

	mutex_init(&lock_c);

	ww_mutex_lock(&ww_lock_a, &ww_ctx);

	mutex_lock(&lock_c);

	ww_mutex_lock(&ww_lock_b, &ww_ctx);

	mutex_unlock(&lock_c);		(*)

	ww_mutex_unlock(&ww_lock_b);
	ww_mutex_unlock(&ww_lock_a);

	ww_acquire_fini(&ww_ctx);

triggers the following WARN in __lock_release() when doing the unlock at *:

	DEBUG_LOCKS_WARN_ON(curr->lockdep_depth != depth - 1);

The problem is that the WARN check doesn't take into account the merging of ww_lock_a and ww_lock_b which results in decreasing curr->lockdep_depth by 2 not only 1.

Note that the following sequence doesn't trigger the WARN, since there won't be any hlock merging.

	ww_acquire_init(&ww_ctx, &test_ww_class);

	ww_mutex_init(&ww_lock_a, &test_ww_class);
	ww_mutex_init(&ww_lock_b, &test_ww_class);

	mutex_init(&lock_c);
	mutex_init(&lock_d);

	ww_mutex_lock(&ww_lock_a, &ww_ctx);

	mutex_lock(&lock_c);
	mutex_lock(&lock_d);

	ww_mutex_lock(&ww_lock_b, &ww_ctx);

	mutex_unlock(&lock_d);

	ww_mutex_unlock(&ww_lock_b);
	ww_mutex_unlock(&ww_lock_a);

	mutex_unlock(&lock_c);

	ww_acquire_fini(&ww_ctx);

In general both of the above two sequences are valid and shouldn't trigger any lockdep warning.

Fix this by taking the decrement due to the hlock merging into account during lock release and hlock class re-setting. Merging can't happen during lock downgrading since there won't be a new possibility to merge hlocks in that case, so add a WARN if merging still happens then.

---
 kernel/locking/lockdep.c | 41 ++++++++++++++++++++++++++++------------
 1 file changed, 29 insertions(+), 12 deletions(-)
kernel/locking/lockdep.c +29 -12
index c47788fa85f9..82361e1bce0f 100644
@@ -3715,7 +3715,7 @@ static int __lock_acquire(struct lockdep_map *lock, unsigned int subclass,
 				hlock->references = 2;
 			}
 
-			return 1;
+			return 2;
 		}
 	}
 
@@ -3921,22 +3921,33 @@ static struct held_lock *find_held_lock(struct task_struct *curr,
 }
 
 static int reacquire_held_locks(struct task_struct *curr, unsigned int depth,
-			      int idx)
+				int idx, unsigned int *merged)
 {
 	struct held_lock *hlock;
+	int first_idx = idx;
 
 	if (DEBUG_LOCKS_WARN_ON(!irqs_disabled()))
 		return 0;
 
 	for (hlock = curr->held_locks + idx; idx < depth; idx++, hlock++) {
-		if (!__lock_acquire(hlock->instance,
+		switch (__lock_acquire(hlock->instance,
 				    hlock_class(hlock)->subclass,
 				    hlock->trylock,
 				    hlock->read, hlock->check,
 				    hlock->hardirqs_off,
 				    hlock->nest_lock, hlock->acquire_ip,
-				    hlock->references, hlock->pin_count))
+				    hlock->references, hlock->pin_count)) {
+		case 0:
 			return 1;
+		case 1:
+			break;
+		case 2:
+			*merged += (idx == first_idx);
+			break;
+		default:
+			WARN_ON(1);
+			return 0;
+		}
 	}
 	return 0;
 }
@@ -3947,9 +3958,9 @@ __lock_set_class(struct lockdep_map *lock, const char *name,
 		 unsigned long ip)
 {
 	struct task_struct *curr = current;
+	unsigned int depth, merged = 0;
 	struct held_lock *hlock;
 	struct lock_class *class;
-	unsigned int depth;
 	int i;
 
 	if (unlikely(!debug_locks))
@@ -3974,14 +3985,14 @@ __lock_set_class(struct lockdep_map *lock, const char *name,
 	curr->lockdep_depth = i;
 	curr->curr_chain_key = hlock->prev_chain_key;
 
-	if (reacquire_held_locks(curr, depth, i))
+	if (reacquire_held_locks(curr, depth, i, &merged))
 		return 0;
 
 	/*
 	 * I took it apart and put it back together again, except now I have
 	 * these 'spare' parts.. where shall I put them.
 	 */
-	if (DEBUG_LOCKS_WARN_ON(curr->lockdep_depth != depth))
+	if (DEBUG_LOCKS_WARN_ON(curr->lockdep_depth != depth - merged))
 		return 0;
 	return 1;
 }
@@ -3989,8 +4000,8 @@ __lock_set_class(struct lockdep_map *lock, const char *name,
 static int __lock_downgrade(struct lockdep_map *lock, unsigned long ip)
 {
 	struct task_struct *curr = current;
+	unsigned int depth, merged = 0;
 	struct held_lock *hlock;
-	unsigned int depth;
 	int i;
 
 	if (unlikely(!debug_locks))
@@ -4015,7 +4026,11 @@ static int __lock_downgrade(struct lockdep_map *lock, unsigned long ip)
 	hlock->read = 1;
 	hlock->acquire_ip = ip;
 
-	if (reacquire_held_locks(curr, depth, i))
+	if (reacquire_held_locks(curr, depth, i, &merged))
+		return 0;
+
+	/* Merging can't happen with unchanged classes.. */
+	if (DEBUG_LOCKS_WARN_ON(merged))
 		return 0;
 
 	/*
@@ -4024,6 +4039,7 @@ static int __lock_downgrade(struct lockdep_map *lock, unsigned long ip)
 	 */
 	if (DEBUG_LOCKS_WARN_ON(curr->lockdep_depth != depth))
 		return 0;
+
 	return 1;
 }
 
@@ -4038,8 +4054,8 @@ static int
 __lock_release(struct lockdep_map *lock, int nested, unsigned long ip)
 {
 	struct task_struct *curr = current;
+	unsigned int depth, merged = 1;
 	struct held_lock *hlock;
-	unsigned int depth;
 	int i;
 
 	if (unlikely(!debug_locks))
@@ -4094,14 +4110,15 @@ __lock_release(struct lockdep_map *lock, int nested, unsigned long ip)
 	if (i == depth-1)
 		return 1;
 
-	if (reacquire_held_locks(curr, depth, i + 1))
+	if (reacquire_held_locks(curr, depth, i + 1, &merged))
 		return 0;
 
 	/*
 	 * We had N bottles of beer on the wall, we drank one, but now
 	 * there's not N-1 bottles of beer left on the wall...
+	 * Pouring two of the bottles together is acceptable.
 	 */
-	DEBUG_LOCKS_WARN_ON(curr->lockdep_depth != depth-1);
+	DEBUG_LOCKS_WARN_ON(curr->lockdep_depth != depth - merged);
 
 	/*
 	 * Since reacquire_held_locks() would have called check_chain_key()
Signature · 1 line
2.20.1
author maintainer
#68 lore raw

PATCH 61/249 locking/lockdep: Fix merging of hlocks with non-zero references

In reply to Sasha Levin

From: Imre Deak <[email protected]>

[ Upstream commit d9349850e188b8b59e5322fda17ff389a1c0cd7d ]

The sequence

	static DEFINE_WW_CLASS(test_ww_class);

	struct ww_acquire_ctx ww_ctx;
	struct ww_mutex ww_lock_a;
	struct ww_mutex ww_lock_b;
	struct ww_mutex ww_lock_c;
	struct mutex lock_c;

	ww_acquire_init(&ww_ctx, &test_ww_class);

	ww_mutex_init(&ww_lock_a, &test_ww_class);
	ww_mutex_init(&ww_lock_b, &test_ww_class);
	ww_mutex_init(&ww_lock_c, &test_ww_class);

	mutex_init(&lock_c);

	ww_mutex_lock(&ww_lock_a, &ww_ctx);

	mutex_lock(&lock_c);

	ww_mutex_lock(&ww_lock_b, &ww_ctx);
	ww_mutex_lock(&ww_lock_c, &ww_ctx);

	mutex_unlock(&lock_c);	(*)

	ww_mutex_unlock(&ww_lock_c);
	ww_mutex_unlock(&ww_lock_b);
	ww_mutex_unlock(&ww_lock_a);

	ww_acquire_fini(&ww_ctx); (**)

will trigger the following error in __lock_release() when calling mutex_release() at **:

	DEBUG_LOCKS_WARN_ON(depth <= 0)

The problem is that the hlock merging happening at * updates the references for test_ww_class incorrectly to 3 whereas it should've updated it to 4 (representing all the instances for ww_ctx and ww_lock_[abc]).

Fix this by updating the references during merging correctly taking into account that we can have non-zero references (both for the hlock that we merge into another hlock or for the hlock we are merging into).

---
 kernel/locking/lockdep.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)
kernel/locking/lockdep.c +9 -9
index 82361e1bce0f..dbc936ccf149 100644
@@ -3703,17 +3703,17 @@ static int __lock_acquire(struct lockdep_map *lock, unsigned int subclass,
 	if (depth) {
 		hlock = curr->held_locks + depth - 1;
 		if (hlock->class_idx == class_idx && nest_lock) {
-			if (hlock->references) {
-				/*
-				 * Check: unsigned int references:12, overflow.
-				 */
-				if (DEBUG_LOCKS_WARN_ON(hlock->references == (1 << 12)-1))
-					return 0;
+			if (!references)
+				references++;
 
+			if (!hlock->references)
 				hlock->references++;
-			} else {
-				hlock->references = 2;
-			}
+
+			hlock->references += references;
+
+			/* Overflow */
+			if (DEBUG_LOCKS_WARN_ON(hlock->references < references))
+				return 0;
 
 			return 2;
 		}
Signature · 1 line
2.20.1
author maintainer
#69 lore raw

PATCH 62/249 media: platform: ao-cec-g12a: disable regmap fast_io for cec bus regmap

In reply to Sasha Levin

From: Neil Armstrong <[email protected]>

[ Upstream commit 9f7406d6b56b4b71a12480b68221755ea7b3e0ee ]

With fast_io enabled, spinlock_irq is used for read/write operations, thus leading to :

BUG: sleeping function called from invalid context at [snip]/ao-cec-g12a.c:379
 in_atomic(): 1, irqs_disabled(): 128, pid: 1451, name: irq/14-ff800280
[snip]
Call trace:
 dump_backtrace+0x0/0x180
 show_stack+0x14/0x1c
 dump_stack+0xa8/0xe0
 ___might_sleep+0xf4/0x104
 __might_sleep+0x4c/0x80
 meson_ao_cec_g12a_read+0x7c/0x164
 regmap_read+0x16c/0x1b0
 meson_ao_cec_g12a_irq_thread+0xcc/0x200
 irq_thread_fn+0x2c/0x60
 irq_thread+0x14c/0x1fc
 kthread+0x11c/0x12c
 ret_from_fork+0x10/0x18

Simply remove fast_io to use mutexes instead.

Fixes b7778c46683c ("media: platform: meson: Add Amlogic Meson G12A AO CEC Controller driver")
---
 drivers/media/platform/meson/ao-cec-g12a.c | 1 -
 1 file changed, 1 deletion(-)
drivers/media/platform/meson/ao-cec-g12a.c +0 -1
index 3620a1e310f5..ddfd060625da 100644
@@ -415,7 +415,6 @@ static const struct regmap_config meson_ao_cec_g12a_cec_regmap_conf = {
 	.reg_read = meson_ao_cec_g12a_read,
 	.reg_write = meson_ao_cec_g12a_write,
 	.max_register = 0xffff,
-	.fast_io = true,
 };
 
 static inline void
Signature · 1 line
2.20.1
author maintainer
#70 lore raw

PATCH 63/249 media: wl128x: Fix some error handling in fm_v4l2_init_video_device()

In reply to Sasha Levin

From: Kefeng Wang <[email protected]>

[ Upstream commit 69fbb3f47327d959830c94bf31893972b8c8f700 ]

X-Originating-IP: [10.175.113.25]
X-CFilter-Loop: Reflected
The fm_v4l2_init_video_device() forget to unregister v4l2/video device in the error path, it could lead to UAF issue, eg,

  BUG: KASAN: use-after-free in atomic64_read include/asm-generic/atomic-instrumented.h:836 [inline]
  BUG: KASAN: use-after-free in atomic_long_read include/asm-generic/atomic-long.h:28 [inline]
  BUG: KASAN: use-after-free in __mutex_unlock_slowpath+0x92/0x690 kernel/locking/mutex.c:1206
  Read of size 8 at addr ffff8881e84a7c70 by task v4l_id/3659

  CPU: 1 PID: 3659 Comm: v4l_id Not tainted 5.1.0 #8
  Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014
  Call Trace:
   __dump_stack lib/dump_stack.c:77 [inline]
   dump_stack+0xa9/0x10e lib/dump_stack.c:113
   print_address_description+0x65/0x270 mm/kasan/report.c:187
   kasan_report+0x149/0x18d mm/kasan/report.c:317
   atomic64_read include/asm-generic/atomic-instrumented.h:836 [inline]
   atomic_long_read include/asm-generic/atomic-long.h:28 [inline]
   __mutex_unlock_slowpath+0x92/0x690 kernel/locking/mutex.c:1206
   fm_v4l2_fops_open+0xac/0x120 [fm_drv]
   v4l2_open+0x191/0x390 [videodev]
   chrdev_open+0x20d/0x570 fs/char_dev.c:417
   do_dentry_open+0x700/0xf30 fs/open.c:777
   do_last fs/namei.c:3416 [inline]
   path_openat+0x7c4/0x2a90 fs/namei.c:3532
   do_filp_open+0x1a5/0x2b0 fs/namei.c:3563
   do_sys_open+0x302/0x490 fs/open.c:1069
   do_syscall_64+0x9f/0x450 arch/x86/entry/common.c:290
   entry_SYSCALL_64_after_hwframe+0x49/0xbe
  RIP: 0033:0x7f8180c17c8e
  ...
  Allocated by task 3642:
   set_track mm/kasan/common.c:87 [inline]
   __kasan_kmalloc.constprop.3+0xa0/0xd0 mm/kasan/common.c:497
   fm_drv_init+0x13/0x1000 [fm_drv]
   do_one_initcall+0xbc/0x47d init/main.c:901
   do_init_module+0x1b5/0x547 kernel/module.c:3456
   load_module+0x6405/0x8c10 kernel/module.c:3804
   __do_sys_finit_module+0x162/0x190 kernel/module.c:3898
   do_syscall_64+0x9f/0x450 arch/x86/entry/common.c:290
   entry_SYSCALL_64_after_hwframe+0x49/0xbe

  Freed by task 3642:
   set_track mm/kasan/common.c:87 [inline]
   __kasan_slab_free+0x130/0x180 mm/kasan/common.c:459
   slab_free_hook mm/slub.c:1429 [inline]
   slab_free_freelist_hook mm/slub.c:1456 [inline]
   slab_free mm/slub.c:3003 [inline]
   kfree+0xe1/0x270 mm/slub.c:3958
   fm_drv_init+0x1e6/0x1000 [fm_drv]
   do_one_initcall+0xbc/0x47d init/main.c:901
   do_init_module+0x1b5/0x547 kernel/module.c:3456
   load_module+0x6405/0x8c10 kernel/module.c:3804
   __do_sys_finit_module+0x162/0x190 kernel/module.c:3898
   do_syscall_64+0x9f/0x450 arch/x86/entry/common.c:290
   entry_SYSCALL_64_after_hwframe+0x49/0xbe

Add relevant unregister functions to fix it.

---
 drivers/media/radio/wl128x/fmdrv_v4l2.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/media/radio/wl128x/fmdrv_v4l2.c +3 -0
index c80a6df47f5e..469366dae1d5 100644
@@ -541,6 +541,7 @@ int fm_v4l2_init_video_device(struct fmdev *fmdev, int radio_nr)
 
 	/* Register with V4L2 subsystem as RADIO device */
 	if (video_register_device(&gradio_dev, VFL_TYPE_RADIO, radio_nr)) {
+		v4l2_device_unregister(&fmdev->v4l2_dev);
 		fmerr("Could not register video device\n");
 		return -ENOMEM;
 	}
@@ -554,6 +555,8 @@ int fm_v4l2_init_video_device(struct fmdev *fmdev, int radio_nr)
 	if (ret < 0) {
 		fmerr("(fmdev): Can't init ctrl handler\n");
 		v4l2_ctrl_handler_free(&fmdev->ctrl_handler);
+		video_unregister_device(fmdev->radio_dev);
+		v4l2_device_unregister(&fmdev->v4l2_dev);
 		return -EBUSY;
 	}
 
Signature · 1 line
2.20.1
author maintainer
#71 lore raw

PATCH 64/249 net: hns3: add a check to pointer in error_detected and slot_reset

In reply to Sasha Levin

From: Weihang Li <[email protected]>

[ Upstream commit 661262bc3e0ecc9a1aed39c6b2a99766da2c22e2 ]

If we add a VF without loading hclgevf.ko and then there is a RAS error occurs, PCIe AER will call error_detected and slot_reset of all functions, and will get a NULL pointer when we check ad_dev->ops->handle_hw_ras_error. This will cause a call trace and failures on handling of follow-up RAS errors.

This patch check ae_dev and ad_dev->ops at first to solve above issues.

---
 drivers/net/ethernet/hisilicon/hns3/hns3_enet.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
drivers/net/ethernet/hisilicon/hns3/hns3_enet.c +5 -2
index cd59c0cc636a..5611b990ac34 100644
@@ -1916,9 +1916,9 @@ static pci_ers_result_t hns3_error_detected(struct pci_dev *pdev,
 	if (state == pci_channel_io_perm_failure)
 		return PCI_ERS_RESULT_DISCONNECT;
 
-	if (!ae_dev) {
+	if (!ae_dev || !ae_dev->ops) {
 		dev_err(&pdev->dev,
-			"Can't recover - error happened during device init\n");
+			"Can't recover - error happened before device initialized\n");
 		return PCI_ERS_RESULT_NONE;
 	}
 
@@ -1937,6 +1937,9 @@ static pci_ers_result_t hns3_slot_reset(struct pci_dev *pdev)
 
 	dev_info(dev, "requesting reset due to PCI error\n");
 
+	if (!ae_dev || !ae_dev->ops)
+		return PCI_ERS_RESULT_NONE;
+
 	/* request the reset */
 	if (ae_dev->ops->reset_event) {
 		if (!ae_dev->override_pci_need_reset)
Signature · 1 line
2.20.1
author maintainer
#72 lore raw

PATCH 65/249 net: hns3: set ops to null when unregister ad_dev

In reply to Sasha Levin

From: Weihang Li <[email protected]>

[ Upstream commit 594a81b39525f0a17e92c2e0b167ae1400650380 ]

The hclge/hclgevf and hns3 module can be unloaded independently, when hclge/hclgevf unloaded firstly, the ops of ae_dev should be set to NULL, otherwise it will cause an use-after-free problem.

Fixes 38caee9d3ee8 ("net: hns3: Add support of the HNAE3 framework")
---
 drivers/net/ethernet/hisilicon/hns3/hnae3.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/net/ethernet/hisilicon/hns3/hnae3.c +2 -0
index fa8b8506b120..738e01393b68 100644
@@ -251,6 +251,7 @@ void hnae3_unregister_ae_algo(struct hnae3_ae_algo *ae_algo)
 
 		ae_algo->ops->uninit_ae_dev(ae_dev);
 		hnae3_set_bit(ae_dev->flag, HNAE3_DEV_INITED_B, 0);
+		ae_dev->ops = NULL;
 	}
 
 	list_del(&ae_algo->node);
@@ -351,6 +352,7 @@ void hnae3_unregister_ae_dev(struct hnae3_ae_dev *ae_dev)
 
 		ae_algo->ops->uninit_ae_dev(ae_dev);
 		hnae3_set_bit(ae_dev->flag, HNAE3_DEV_INITED_B, 0);
+		ae_dev->ops = NULL;
 	}
 
 	list_del(&ae_dev->node);
Signature · 1 line
2.20.1
author maintainer
#73 lore raw

PATCH 66/249 cpupower : frequency-set -r option misses the last cpu in related cpu list

In reply to Sasha Levin

From: Abhishek Goel <[email protected]>

[ Upstream commit 04507c0a9385cc8280f794a36bfff567c8cc1042 ]

To set frequency on specific cpus using cpupower, following syntax can be used :
cpupower -c #i frequency-set -f #f -r

While setting frequency using cpupower frequency-set command, if we use '-r' option, it is expected to set frequency for all cpus related to cpu #i. But it is observed to be missing the last cpu in related cpu list. This patch fixes the problem.

---
 tools/power/cpupower/utils/cpufreq-set.c | 2 ++
 1 file changed, 2 insertions(+)
tools/power/cpupower/utils/cpufreq-set.c +2 -0
index f49bc4aa2a08..6ed82fba5aaa 100644
@@ -305,6 +305,8 @@ int cmd_freq_set(int argc, char **argv)
 				bitmask_setbit(cpus_chosen, cpus->cpu);
 				cpus = cpus->next;
 			}
+			/* Set the last cpu in related cpus list */
+			bitmask_setbit(cpus_chosen, cpus->cpu);
 			cpufreq_put_related_cpus(cpus);
 		}
 	}
Signature · 1 line
2.20.1
author maintainer
#74 lore raw

PATCH 67/249 arm64: mm: make CONFIG_ZONE_DMA32 configurable

In reply to Sasha Levin

From: Miles Chen <[email protected]>

[ Upstream commit 0c1f14ed12262f45a3af1d588e4d7bd12438b8f5 ]

This change makes CONFIG_ZONE_DMA32 defuly y and allows users to overwrite it only when CONFIG_EXPERT=y.

For the SoCs that do not need CONFIG_ZONE_DMA32, this is the first step to manage all available memory by a single zone(normal zone) to reduce the overhead of multiple zones.

The change also fixes a build error when CONFIG_NUMA=y and CONFIG_ZONE_DMA32=n.

arch/arm64/mm/init.c:195:17: error: use of undeclared identifier 'ZONE_DMA32'

                max_zone_pfns[ZONE_DMA32] = PFN_DOWN(max_zone_dma_phys());

Change since v1:
1. only expose CONFIG_ZONE_DMA32 when CONFIG_EXPERT=y
2. remove redundant IS_ENABLED(CONFIG_ZONE_DMA32)

---
 arch/arm64/Kconfig   | 3 ++-
 arch/arm64/mm/init.c | 5 +++--
 2 files changed, 5 insertions(+), 3 deletions(-)
arch/arm64/Kconfig +2 -1
index 697ea0510729..cf5f1dafcf74 100644
@@ -260,7 +260,8 @@ config GENERIC_CALIBRATE_DELAY
 	def_bool y
 
 config ZONE_DMA32
-	def_bool y
+	bool "Support DMA32 zone" if EXPERT
+	default y
 
 config HAVE_GENERIC_GUP
 	def_bool y
arch/arm64/mm/init.c +3 -2
index 749c9b269f08..f3c795278def 100644
@@ -180,8 +180,9 @@ static void __init zone_sizes_init(unsigned long min, unsigned long max)
 {
 	unsigned long max_zone_pfns[MAX_NR_ZONES]  = {0};
 
-	if (IS_ENABLED(CONFIG_ZONE_DMA32))
-		max_zone_pfns[ZONE_DMA32] = PFN_DOWN(max_zone_dma_phys());
+#ifdef CONFIG_ZONE_DMA32
+	max_zone_pfns[ZONE_DMA32] = PFN_DOWN(max_zone_dma_phys());
+#endif
 	max_zone_pfns[ZONE_NORMAL] = max;
 
 	free_area_init_nodes(max_zone_pfns);
Signature · 1 line
2.20.1
author maintainer
#75 lore raw

PATCH 68/249 media: imx7-mipi-csis: Propagate the error if clock enabling fails

In reply to Sasha Levin

From: Fabio Estevam <[email protected]>

[ Upstream commit 2b393f91c651c16d5c09f5c7aa689e58a79df34e ]

Currently the return value from clk_bulk_prepare_enable() is checked, but it is not propagate it in the case of failure.

Fix it and also move the error message to the caller of mipi_csis_clk_enable().

---
 drivers/staging/media/imx/imx7-mipi-csis.c | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)
drivers/staging/media/imx/imx7-mipi-csis.c +7 -7
index 19455f425416..7d7bdfdd852a 100644
@@ -456,13 +456,9 @@ static void mipi_csis_set_params(struct csi_state *state)
 			MIPI_CSIS_CMN_CTRL_UPDATE_SHADOW_CTRL);
 }
 
-static void mipi_csis_clk_enable(struct csi_state *state)
+static int mipi_csis_clk_enable(struct csi_state *state)
 {
-	int ret;
-
-	ret = clk_bulk_prepare_enable(state->num_clks, state->clks);
-	if (ret < 0)
-		dev_err(state->dev, "failed to enable clocks\n");
+	return clk_bulk_prepare_enable(state->num_clks, state->clks);
 }
 
 static void mipi_csis_clk_disable(struct csi_state *state)
@@ -973,7 +969,11 @@ static int mipi_csis_probe(struct platform_device *pdev)
 	if (ret < 0)
 		return ret;
 
-	mipi_csis_clk_enable(state);
+	ret = mipi_csis_clk_enable(state);
+	if (ret < 0) {
+		dev_err(state->dev, "failed to enable clocks: %d\n", ret);
+		return ret;
+	}
 
 	ret = devm_request_irq(dev, state->irq, mipi_csis_irq_handler,
 			       0, dev_name(dev), state);
Signature · 1 line
2.20.1
author maintainer
#76 lore raw

PATCH 69/249 perf jvmti: Address gcc string overflow warning for strncpy()

In reply to Sasha Levin

From: Jiri Olsa <[email protected]>

[ Upstream commit 279ab04dbea1370d2eac0f854270369ccaef8a44 ]

We are getting false positive gcc warning when we compile with gcc9 (9.1.1):

CC jvmti/libjvmti.o
In file included from /usr/include/string.h:494,
from jvmti/libjvmti.c:5:
In function ‘strncpy’,
inlined from ‘copy_class_filename.constprop’ at jvmti/libjvmti.c:166:3:
/usr/include/bits/string_fortified.h:106:10: error: ‘__builtin_strncpy’ specified bound depends on the length of the source argument [-Werror=stringop-overflow=]

     106 |   return __builtin___strncpy_chk (__dest, __src, __len, __bos (__dest));
         |          ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
   jvmti/libjvmti.c: In function ‘copy_class_filename.constprop’:
   jvmti/libjvmti.c:165:26: note: length computed here
     165 |   size_t file_name_len = strlen(file_name);
         |                          ^~~~~~~~~~~~~~~~~

cc1: all warnings being treated as errors

As per Arnaldo's suggestion use strlcpy(), which does the same thing and keeps gcc silent.

---
 tools/perf/jvmti/libjvmti.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
tools/perf/jvmti/libjvmti.c +2 -2
index aea7b1fe85aa..c441a34cb1c0 100644
@@ -1,5 +1,6 @@
 // SPDX-License-Identifier: GPL-2.0
 #include <linux/compiler.h>
+#include <linux/string.h>
 #include <sys/types.h>
 #include <stdio.h>
 #include <string.h>
@@ -162,8 +163,7 @@ copy_class_filename(const char * class_sign, const char * file_name, char * resu
 		result[i] = '\0';
 	} else {
 		/* fallback case */
-		size_t file_name_len = strlen(file_name);
-		strncpy(result, file_name, file_name_len < max_length ? file_name_len : max_length);
+		strlcpy(result, file_name, max_length);
 	}
 }
 
Signature · 1 line
2.20.1
author maintainer
#77 lore raw

PATCH 70/249 media: aspeed: change irq to threaded irq

In reply to Sasha Levin

From: Jae Hyun Yoo <[email protected]>

[ Upstream commit 12ae1c1bf5db2f33fcd9092a96f630291c4b181a ]

Differently from other Aspeed drivers, this driver calls clock control APIs in interrupt context. Since ECLK is coupled with a reset bit in clk-aspeed module, aspeed_clk_enable will make 10ms of busy waiting delay for triggering the reset and it will eventually disturb other drivers' interrupt handling. To fix this issue, this commit changes this driver's irq to threaded irq so that the delay can be happened in a thread context.

---
 drivers/media/platform/aspeed-video.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
drivers/media/platform/aspeed-video.c +3 -2
index 8144fe36ad48..76d7512c82a3 100644
@@ -1589,8 +1589,9 @@ static int aspeed_video_init(struct aspeed_video *video)
 		return -ENODEV;
 	}
 
-	rc = devm_request_irq(dev, irq, aspeed_video_irq, IRQF_SHARED,
-			      DEVICE_NAME, video);
+	rc = devm_request_threaded_irq(dev, irq, NULL, aspeed_video_irq,
+				       IRQF_ONESHOT | IRQF_SHARED, DEVICE_NAME,
+				       video);
 	if (rc < 0) {
 		dev_err(dev, "Unable to request IRQ %d\n", irq);
 		return rc;
Signature · 1 line
2.20.1
author maintainer
#78 lore raw

PATCH 71/249 media: aspeed: fix a kernel warning on clk control

In reply to Sasha Levin

From: Jae Hyun Yoo <[email protected]>

[ Upstream commit 9698ed4d4a2993ce54b9f7d71a2891e972caa117 ]

Video engine clock control can be double disabled and eventually it causes a kernel warning with stack dump printing out like below:

[  515.540498] ------------[ cut here ]------------
[  515.545174] WARNING: CPU: 0 PID: 1310 at drivers/clk/clk.c:684 clk_core_unprepare+0x13c/0x170
[  515.553806] vclk-gate already unprepared
[  515.557841] CPU: 0 PID: 1310 Comm: obmc-ikvm Tainted: G        W         5.0.6-df66fbc97853fbba90a0bfa44de32f3d5f7602b4 #1
[  515.568973] Hardware name: Generic DT based system
[  515.573777] Backtrace:
[  515.576272] [<80107cdc>] (dump_backtrace) from [<80107f10>] (show_stack+0x20/0x24)
[  515.583930]  r7:803a5614 r6:00000009 r5:00000000 r4:9d88fe1c
[  515.589712] [<80107ef0>] (show_stack) from [<80690184>] (dump_stack+0x20/0x28)
[  515.597053] [<80690164>] (dump_stack) from [<80116044>] (__warn.part.3+0xb4/0xdc)
[  515.604557] [<80115f90>] (__warn.part.3) from [<801160d8>] (warn_slowpath_fmt+0x6c/0x90)
[  515.612734]  r6:000002ac r5:8080befc r4:80a07008
[  515.617463] [<80116070>] (warn_slowpath_fmt) from [<803a5614>] (clk_core_unprepare+0x13c/0x170)
[  515.626167]  r3:8080cdf4 r2:8080bfc0
[  515.629834]  r7:98d682a8 r6:9d8a9200 r5:9e5151a0 r4:97abd620
[  515.635530] [<803a54d8>] (clk_core_unprepare) from [<803a76a4>] (clk_unprepare+0x34/0x3c)
[  515.643812]  r5:9e5151a0 r4:97abd620
[  515.647529] [<803a7670>] (clk_unprepare) from [<804f36ec>] (aspeed_video_off+0x38/0x50)
[  515.655539]  r5:9e5151a0 r4:9e504000
[  515.659242] [<804f36b4>] (aspeed_video_off) from [<804f4358>] (aspeed_video_release+0x90/0x114)
[  515.668036]  r5:9e5044b0 r4:9e504000
[  515.671643] [<804f42c8>] (aspeed_video_release) from [<804d302c>] (v4l2_release+0xd4/0xe8)
[  515.679999]  r7:98d682a8 r6:9d087810 r5:9d8a9200 r4:9e504318
[  515.685695] [<804d2f58>] (v4l2_release) from [<80236454>] (__fput+0x98/0x1c4)
[  515.692914]  r5:9e51b608 r4:9d8a9200
[  515.696597] [<802363bc>] (__fput) from [<802365e8>] (____fput+0x18/0x1c)
[  515.703315]  r9:80a0700c r8:801011e4 r7:00000000 r6:80a64b9c r5:9d8e35a0 r4:9d8e38dc
[  515.711167] [<802365d0>] (____fput) from [<80131ca4>] (task_work_run+0x7c/0xa0)
[  515.718596] [<80131c28>] (task_work_run) from [<80106884>] (do_work_pending+0x4a8/0x578)
[  515.726777]  r7:801011e4 r6:80a07008 r5:9d88ffb0 r4:ffffe000
[  515.732466] [<801063dc>] (do_work_pending) from [<8010106c>] (slow_work_pending+0xc/0x20)
[  515.740727] Exception stack(0x9d88ffb0 to 0x9d88fff8)
[  515.745840] ffa0:                                     00000000 76f18094 00000000 00000000
[  515.754122] ffc0: 00000007 00176778 7eda4c20 00000006 00000000 00000000 48e20fa4 00000000
[  515.762386] ffe0: 00000002 7eda4b08 00000000 48f91efc 80000010 00000007
[  515.769097]  r10:00000000 r9:9d88e000 r8:801011e4 r7:00000006 r6:7eda4c20 r5:00176778
[  515.777006]  r4:00000007
[  515.779558] ---[ end trace 12c04aadef8afbbb ]---
[  515.784176] ------------[ cut here ]------------
[  515.788817] WARNING: CPU: 0 PID: 1310 at drivers/clk/clk.c:825 clk_core_disable+0x18c/0x204
[  515.797161] eclk-gate already disabled
[  515.800916] CPU: 0 PID: 1310 Comm: obmc-ikvm Tainted: G        W         5.0.6-df66fbc97853fbba90a0bfa44de32f3d5f7602b4 #1
[  515.811945] Hardware name: Generic DT based system
[  515.816730] Backtrace:
[  515.819210] [<80107cdc>] (dump_backtrace) from [<80107f10>] (show_stack+0x20/0x24)
[  515.826782]  r7:803a5900 r6:00000009 r5:00000000 r4:9d88fe04
[  515.832454] [<80107ef0>] (show_stack) from [<80690184>] (dump_stack+0x20/0x28)
[  515.839687] [<80690164>] (dump_stack) from [<80116044>] (__warn.part.3+0xb4/0xdc)
[  515.847170] [<80115f90>] (__warn.part.3) from [<801160d8>] (warn_slowpath_fmt+0x6c/0x90)
[  515.855247]  r6:00000339 r5:8080befc r4:80a07008
[  515.859868] [<80116070>] (warn_slowpath_fmt) from [<803a5900>] (clk_core_disable+0x18c/0x204)
[  515.868385]  r3:8080cdd0 r2:8080c00c
[  515.871957]  r7:98d682a8 r6:9d8a9200 r5:97abd560 r4:97abd560
[  515.877615] [<803a5774>] (clk_core_disable) from [<803a59a0>] (clk_core_disable_lock+0x28/0x34)
[  515.886301]  r7:98d682a8 r6:9d8a9200 r5:97abd560 r4:a0000013
[  515.891960] [<803a5978>] (clk_core_disable_lock) from [<803a7714>] (clk_disable+0x2c/0x30)
[  515.900216]  r5:9e5151a0 r4:9e515f60
[  515.903816] [<803a76e8>] (clk_disable) from [<804f36f8>] (aspeed_video_off+0x44/0x50)
[  515.911656] [<804f36b4>] (aspeed_video_off) from [<804f4358>] (aspeed_video_release+0x90/0x114)
[  515.920341]  r5:9e5044b0 r4:9e504000
[  515.923921] [<804f42c8>] (aspeed_video_release) from [<804d302c>] (v4l2_release+0xd4/0xe8)
[  515.932184]  r7:98d682a8 r6:9d087810 r5:9d8a9200 r4:9e504318
[  515.937851] [<804d2f58>] (v4l2_release) from [<80236454>] (__fput+0x98/0x1c4)
[  515.944980]  r5:9e51b608 r4:9d8a9200
[  515.948559] [<802363bc>] (__fput) from [<802365e8>] (____fput+0x18/0x1c)
[  515.955257]  r9:80a0700c r8:801011e4 r7:00000000 r6:80a64b9c r5:9d8e35a0 r4:9d8e38dc
[  515.963008] [<802365d0>] (____fput) from [<80131ca4>] (task_work_run+0x7c/0xa0)
[  515.970333] [<80131c28>] (task_work_run) from [<80106884>] (do_work_pending+0x4a8/0x578)
[  515.978421]  r7:801011e4 r6:80a07008 r5:9d88ffb0 r4:ffffe000
[  515.984086] [<801063dc>] (do_work_pending) from [<8010106c>] (slow_work_pending+0xc/0x20)
[  515.992247] Exception stack(0x9d88ffb0 to 0x9d88fff8)
[  515.997296] ffa0:                                     00000000 76f18094 00000000 00000000
[  516.005473] ffc0: 00000007 00176778 7eda4c20 00000006 00000000 00000000 48e20fa4 00000000
[  516.013642] ffe0: 00000002 7eda4b08 00000000 48f91efc 80000010 00000007
[  516.020257]  r10:00000000 r9:9d88e000 r8:801011e4 r7:00000006 r6:7eda4c20 r5:00176778
[  516.028072]  r4:00000007
[  516.030606] ---[ end trace 12c04aadef8afbbc ]---

To prevent this issue, this commit adds clock status checking logic into the Aspeed video engine driver.

---
 drivers/media/platform/aspeed-video.c | 11 +++++++++++
 1 file changed, 11 insertions(+)
drivers/media/platform/aspeed-video.c +11 -0
index 76d7512c82a3..de0f192afa8b 100644
@@ -187,6 +187,7 @@ enum {
 	VIDEO_STREAMING,
 	VIDEO_FRAME_INPRG,
 	VIDEO_STOPPED,
+	VIDEO_CLOCKS_ON,
 };
 
 struct aspeed_video_addr {
@@ -483,19 +484,29 @@ static void aspeed_video_enable_mode_detect(struct aspeed_video *video)
 
 static void aspeed_video_off(struct aspeed_video *video)
 {
+	if (!test_bit(VIDEO_CLOCKS_ON, &video->flags))
+		return;
+
 	/* Disable interrupts */
 	aspeed_video_write(video, VE_INTERRUPT_CTRL, 0);
 
 	/* Turn off the relevant clocks */
 	clk_disable_unprepare(video->vclk);
 	clk_disable_unprepare(video->eclk);
+
+	clear_bit(VIDEO_CLOCKS_ON, &video->flags);
 }
 
 static void aspeed_video_on(struct aspeed_video *video)
 {
+	if (test_bit(VIDEO_CLOCKS_ON, &video->flags))
+		return;
+
 	/* Turn on the relevant clocks */
 	clk_prepare_enable(video->eclk);
 	clk_prepare_enable(video->vclk);
+
+	set_bit(VIDEO_CLOCKS_ON, &video->flags);
 }
 
 static void aspeed_video_bufs_done(struct aspeed_video *video,
Signature · 1 line
2.20.1
author maintainer
#79 lore raw

PATCH 72/249 net: stmmac: dwmac4: fix flow control issue

In reply to Sasha Levin

From: Biao Huang <[email protected]>

[ Upstream commit ee326fd01e79dfa42014d55931260b68b9fa3273 ]

Current dwmac4_flow_ctrl will not clear GMAC_RX_FLOW_CTRL_RFE/GMAC_RX_FLOW_CTRL_RFE bits, so MAC hw will keep flow control on although expecting flow control off by ethtool. Add codes to fix it.

Fixes 477286b53f55 ("stmmac: add GMAC4 core support")
---
 drivers/net/ethernet/stmicro/stmmac/dwmac4_core.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
drivers/net/ethernet/stmicro/stmmac/dwmac4_core.c +6 -2
index 206170d0bf81..e3850938cf2f 100644
@@ -474,8 +474,9 @@ static void dwmac4_flow_ctrl(struct mac_device_info *hw, unsigned int duplex,
 	if (fc & FLOW_RX) {
 		pr_debug("\tReceive Flow-Control ON\n");
 		flow |= GMAC_RX_FLOW_CTRL_RFE;
-		writel(flow, ioaddr + GMAC_RX_FLOW_CTRL);
 	}
+	writel(flow, ioaddr + GMAC_RX_FLOW_CTRL);
+
 	if (fc & FLOW_TX) {
 		pr_debug("\tTransmit Flow-Control ON\n");
 
@@ -483,7 +484,7 @@ static void dwmac4_flow_ctrl(struct mac_device_info *hw, unsigned int duplex,
 			pr_debug("\tduplex mode: PAUSE %d\n", pause_time);
 
 		for (queue = 0; queue < tx_cnt; queue++) {
-			flow |= GMAC_TX_FLOW_CTRL_TFE;
+			flow = GMAC_TX_FLOW_CTRL_TFE;
 
 			if (duplex)
 				flow |=
@@ -491,6 +492,9 @@ static void dwmac4_flow_ctrl(struct mac_device_info *hw, unsigned int duplex,
 
 			writel(flow, ioaddr + GMAC_QX_TX_FLOW_CTRL(queue));
 		}
+	} else {
+		for (queue = 0; queue < tx_cnt; queue++)
+			writel(0, ioaddr + GMAC_QX_TX_FLOW_CTRL(queue));
 	}
 }
 
Signature · 1 line
2.20.1
author maintainer
#80 lore raw

PATCH 73/249 net: stmmac: modify default value of tx-frames

In reply to Sasha Levin

From: Biao Huang <[email protected]>

[ Upstream commit d2facb4b3983425f6776c24dd678a82dbe673773 ]

the default value of tx-frames is 25, it's too late when passing tstamp to stack, then the ptp4l will fail:

ptp4l -i eth0 -f gPTP.cfg -m
ptp4l: selected /dev/ptp0 as PTP clock
ptp4l: port 1: INITIALIZING to LISTENING on INITIALIZE
ptp4l: port 0: INITIALIZING to LISTENING on INITIALIZE
ptp4l: port 1: link up
ptp4l: timed out while polling for tx timestamp
ptp4l: increasing tx_timestamp_timeout may correct this issue,
but it is likely caused by a driver bug
ptp4l: port 1: send peer delay response failed
ptp4l: port 1: LISTENING to FAULTY on FAULT_DETECTED (FT_UNSPECIFIED)

ptp4l tests pass when changing the tx-frames from 25 to 1 with ethtool -C option.
It should be fine to set tx-frames default value to 1, so ptp4l will pass by default.

---
 drivers/net/ethernet/stmicro/stmmac/common.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/stmicro/stmmac/common.h +1 -1
index ceb0d23f5041..c265cc5770e8 100644
@@ -251,7 +251,7 @@ struct stmmac_safety_stats {
 #define STMMAC_COAL_TX_TIMER	1000
 #define STMMAC_MAX_COAL_TX_TICK	100000
 #define STMMAC_TX_MAX_FRAMES	256
-#define STMMAC_TX_FRAMES	25
+#define STMMAC_TX_FRAMES	1
 
 /* Packets types */
 enum packets_types {
Signature · 1 line
2.20.1
author maintainer
#81 lore raw

PATCH 74/249 crypto: inside-secure - do not rely on the hardware last bit for result descriptors

In reply to Sasha Levin

From: Antoine Tenart <[email protected]>

[ Upstream commit 89332590427235680236b9470e851afc49b3caa1 ]

When performing a transformation the hardware is given result descriptors to save the result data. Those result descriptors are batched using a 'first' and a 'last' bit. There are cases were more descriptors than needed are given to the engine, leading to the engine only using some of them, and not setting the last bit on the last descriptor we gave. This causes issues were the driver and the hardware aren't in sync anymore about the number of result descriptors given (as the driver do not give a pool of descriptor to use for any transformation, but a pool of descriptors to use *per* transformation).

This patch fixes it by attaching the number of given result descriptors to the requests, and by using this number instead of the 'last' bit found on the descriptors to process them.

---
 .../crypto/inside-secure/safexcel_cipher.c    | 24 ++++++++++++++-----
 1 file changed, 18 insertions(+), 6 deletions(-)
drivers/crypto/inside-secure/safexcel_cipher.c +18 -6
index de4be10b172f..ccacdcf07ffc 100644
@@ -51,6 +51,8 @@ struct safexcel_cipher_ctx {
 
 struct safexcel_cipher_req {
 	enum safexcel_cipher_direction direction;
+	/* Number of result descriptors associated to the request */
+	unsigned int rdescs;
 	bool needs_inv;
 };
 
@@ -333,7 +335,10 @@ static int safexcel_handle_req_result(struct safexcel_crypto_priv *priv, int rin
 
 	*ret = 0;
 
-	do {
+	if (unlikely(!sreq->rdescs))
+		return 0;
+
+	while (sreq->rdescs--) {
 		rdesc = safexcel_ring_next_rptr(priv, &priv->ring[ring].rdr);
 		if (IS_ERR(rdesc)) {
 			dev_err(priv->dev,
@@ -346,7 +351,7 @@ static int safexcel_handle_req_result(struct safexcel_crypto_priv *priv, int rin
 			*ret = safexcel_rdesc_check_errors(priv, rdesc);
 
 		ndesc++;
-	} while (!rdesc->last_seg);
+	}
 
 	safexcel_complete(priv, ring);
 
@@ -501,6 +506,7 @@ static int safexcel_send_req(struct crypto_async_request *base, int ring,
 static int safexcel_handle_inv_result(struct safexcel_crypto_priv *priv,
 				      int ring,
 				      struct crypto_async_request *base,
+				      struct safexcel_cipher_req *sreq,
 				      bool *should_complete, int *ret)
 {
 	struct safexcel_cipher_ctx *ctx = crypto_tfm_ctx(base->tfm);
@@ -509,7 +515,10 @@ static int safexcel_handle_inv_result(struct safexcel_crypto_priv *priv,
 
 	*ret = 0;
 
-	do {
+	if (unlikely(!sreq->rdescs))
+		return 0;
+
+	while (sreq->rdescs--) {
 		rdesc = safexcel_ring_next_rptr(priv, &priv->ring[ring].rdr);
 		if (IS_ERR(rdesc)) {
 			dev_err(priv->dev,
@@ -522,7 +531,7 @@ static int safexcel_handle_inv_result(struct safexcel_crypto_priv *priv,
 			*ret = safexcel_rdesc_check_errors(priv, rdesc);
 
 		ndesc++;
-	} while (!rdesc->last_seg);
+	}
 
 	safexcel_complete(priv, ring);
 
@@ -564,7 +573,7 @@ static int safexcel_skcipher_handle_result(struct safexcel_crypto_priv *priv,
 
 	if (sreq->needs_inv) {
 		sreq->needs_inv = false;
-		err = safexcel_handle_inv_result(priv, ring, async,
+		err = safexcel_handle_inv_result(priv, ring, async, sreq,
 						 should_complete, ret);
 	} else {
 		err = safexcel_handle_req_result(priv, ring, async, req->src,
@@ -587,7 +596,7 @@ static int safexcel_aead_handle_result(struct safexcel_crypto_priv *priv,
 
 	if (sreq->needs_inv) {
 		sreq->needs_inv = false;
-		err = safexcel_handle_inv_result(priv, ring, async,
+		err = safexcel_handle_inv_result(priv, ring, async, sreq,
 						 should_complete, ret);
 	} else {
 		err = safexcel_handle_req_result(priv, ring, async, req->src,
@@ -633,6 +642,8 @@ static int safexcel_skcipher_send(struct crypto_async_request *async, int ring,
 		ret = safexcel_send_req(async, ring, sreq, req->src,
 					req->dst, req->cryptlen, 0, 0, req->iv,
 					commands, results);
+
+	sreq->rdescs = *results;
 	return ret;
 }
 
@@ -655,6 +666,7 @@ static int safexcel_aead_send(struct crypto_async_request *async, int ring,
 					req->cryptlen, req->assoclen,
 					crypto_aead_authsize(tfm), req->iv,
 					commands, results);
+	sreq->rdescs = *results;
 	return ret;
 }
 
Signature · 1 line
2.20.1
author maintainer
#82 lore raw

PATCH 75/249 net: fec: Do not use netdev messages too early

In reply to Sasha Levin

From: Fabio Estevam <[email protected]>

[ Upstream commit a19a0582363b9a5f8ba812f34f1b8df394898780 ]

When a valid MAC address is not found the current messages are shown:

fec 2188000.ethernet (unnamed net_device) (uninitialized): Invalid MAC address: 00:00:00:00:00:00 fec 2188000.ethernet (unnamed net_device) (uninitialized): Using random MAC address: aa:9f:25:eb:7e:aa

Since the network device has not been registered at this point, it is better to use dev_err()/dev_info() instead, which will provide cleaner log messages like these:

fec 2188000.ethernet: Invalid MAC address: 00:00:00:00:00:00 fec 2188000.ethernet: Using random MAC address: aa:9f:25:eb:7e:aa

Tested on a imx6dl-pico-pi board.

---
 drivers/net/ethernet/freescale/fec_main.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/net/ethernet/freescale/fec_main.c +3 -3
index 38f10f7dcbc3..831bb709e783 100644
@@ -1689,10 +1689,10 @@ static void fec_get_mac(struct net_device *ndev)
 	 */
 	if (!is_valid_ether_addr(iap)) {
 		/* Report it and use a random ethernet address instead */
-		netdev_err(ndev, "Invalid MAC address: %pM\n", iap);
+		dev_err(&fep->pdev->dev, "Invalid MAC address: %pM\n", iap);
 		eth_hw_addr_random(ndev);
-		netdev_info(ndev, "Using random MAC address: %pM\n",
-			    ndev->dev_addr);
+		dev_info(&fep->pdev->dev, "Using random MAC address: %pM\n",
+			 ndev->dev_addr);
 		return;
 	}
 
Signature · 1 line
2.20.1
author maintainer
#83 lore raw

PATCH 76/249 net: axienet: Fix race condition causing TX hang

In reply to Sasha Levin

From: Robert Hancock <[email protected]>

[ Upstream commit 7de44285c1f69ccfbe8be1d6a16fcd956681fee6 ]

It is possible that the interrupt handler fires and frees up space in the TX ring in between checking for sufficient TX ring space and stopping the TX queue in axienet_start_xmit. If this happens, the queue wake from the interrupt handler will occur before the queue is stopped, causing a lost wakeup and the adapter's transmit hanging.

To avoid this, after stopping the queue, check again whether there is sufficient space in the TX ring. If so, wake up the queue again.

---
 .../net/ethernet/xilinx/xilinx_axienet_main.c | 20 ++++++++++++++++---
 1 file changed, 17 insertions(+), 3 deletions(-)
drivers/net/ethernet/xilinx/xilinx_axienet_main.c +17 -3
index 831967f6eff8..65c16772e589 100644
@@ -615,6 +615,10 @@ static void axienet_start_xmit_done(struct net_device *ndev)
 
 	ndev->stats.tx_packets += packets;
 	ndev->stats.tx_bytes += size;
+
+	/* Matches barrier in axienet_start_xmit */
+	smp_mb();
+
 	netif_wake_queue(ndev);
 }
 
@@ -670,9 +674,19 @@ axienet_start_xmit(struct sk_buff *skb, struct net_device *ndev)
 	cur_p = &lp->tx_bd_v[lp->tx_bd_tail];
 
 	if (axienet_check_tx_bd_space(lp, num_frag)) {
-		if (!netif_queue_stopped(ndev))
-			netif_stop_queue(ndev);
-		return NETDEV_TX_BUSY;
+		if (netif_queue_stopped(ndev))
+			return NETDEV_TX_BUSY;
+
+		netif_stop_queue(ndev);
+
+		/* Matches barrier in axienet_start_xmit_done */
+		smp_mb();
+
+		/* Space might have just been freed - check again */
+		if (axienet_check_tx_bd_space(lp, num_frag))
+			return NETDEV_TX_BUSY;
+
+		netif_wake_queue(ndev);
 	}
 
 	if (skb->ip_summed == CHECKSUM_PARTIAL) {
Signature · 1 line
2.20.1
author maintainer
#84 lore raw

PATCH 77/249 s390/qdio: handle PENDING state for QEBSM devices

In reply to Sasha Levin

From: Julian Wiedmann <[email protected]>

[ Upstream commit 04310324c6f482921c071444833e70fe861b73d9 ]

When a CQ-enabled device uses QEBSM for SBAL state inspection, get_buf_states() can return the PENDING state for an Output Queue. get_outbound_buffer_frontier() isn't prepared for this, and any PENDING buffer will permanently stall all further completion processing on this Queue.

This isn't a concern for non-QEBSM devices, as get_buf_states() for such devices will manually turn PENDING buffers into EMPTY ones.

Fixes 104ea556ee7f ("qdio: support asynchronous delivery of storage blocks")
---
 drivers/s390/cio/qdio_main.c | 1 +
 1 file changed, 1 insertion(+)
drivers/s390/cio/qdio_main.c +1 -0
index 7b7620de2acd..730c4e68094b 100644
@@ -736,6 +736,7 @@ static int get_outbound_buffer_frontier(struct qdio_q *q, unsigned int start)
 
 	switch (state) {
 	case SLSB_P_OUTPUT_EMPTY:
+	case SLSB_P_OUTPUT_PENDING:
 		/* the adapter got it */
 		DBF_DEV_EVENT(DBF_INFO, q->irq_ptr,
 			"out empty:%1d %02x", q->nr, count);
Signature · 1 line
2.20.1
author maintainer
#85 lore raw

PATCH 78/249 RAS/CEC: Fix pfn insertion

In reply to Sasha Levin

From: Borislav Petkov <[email protected]>

[ Upstream commit 6d8e294bf5f0e85c34e8b14b064e2965f53f38b0 ]

When inserting random PFNs for debugging the CEC through (debugfs)/ras/cec/pfn, depending on the return value of pfn_set(), multiple values get inserted per a single write.

That is because simple_attr_write() interprets a retval of 0 as success and claims the whole input. However, pfn_set() returns the cec_add_elem() value, which, if > 0 and smaller than the whole input length, makes glibc continue issuing the write syscall until there's input left:

pfn_set
simple_attr_write
debugfs_attr_write
full_proxy_write
vfs_write
ksys_write
do_syscall_64
entry_SYSCALL_64_after_hwframe

leading to those repeated calls.

Return 0 to fix that.

---
 drivers/ras/cec.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/ras/cec.c +3 -1
index 673f8a128397..f5795adc5a6e 100644
@@ -369,7 +369,9 @@ static int pfn_set(void *data, u64 val)
 {
 	*(u64 *)data = val;
 
-	return cec_add_elem(val);
+	cec_add_elem(val);
+
+	return 0;
 }
 
 DEFINE_DEBUGFS_ATTRIBUTE(pfn_ops, u64_get, pfn_set, "0x%llx\n");
Signature · 1 line
2.20.1
author maintainer
#86 lore raw

PATCH 79/249 net: sfp: add mutex to prevent concurrent state checks

In reply to Sasha Levin

From: Robert Hancock <[email protected]>

[ Upstream commit 2158e856f56bb762ef90f3ec244d41a519826f75 ]

sfp_check_state can potentially be called by both a threaded IRQ handler and delayed work. If it is concurrently called, it could result in incorrect state management. Add a st_mutex to protect the state - this lock gets taken outside of code that checks and handle state changes, and the existing sm_mutex nests inside of it.

---
 drivers/net/phy/sfp.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
drivers/net/phy/sfp.c +5 -1
index 71812be0ac64..b6efd2d41dce 100644
@@ -186,10 +186,11 @@ struct sfp {
 	struct gpio_desc *gpio[GPIO_MAX];
 
 	bool attached;
+	struct mutex st_mutex;			/* Protects state */
 	unsigned int state;
 	struct delayed_work poll;
 	struct delayed_work timeout;
-	struct mutex sm_mutex;
+	struct mutex sm_mutex;			/* Protects state machine */
 	unsigned char sm_mod_state;
 	unsigned char sm_dev_state;
 	unsigned short sm_state;
@@ -1719,6 +1720,7 @@ static void sfp_check_state(struct sfp *sfp)
 {
 	unsigned int state, i, changed;
 
+	mutex_lock(&sfp->st_mutex);
 	state = sfp_get_state(sfp);
 	changed = state ^ sfp->state;
 	changed &= SFP_F_PRESENT | SFP_F_LOS | SFP_F_TX_FAULT;
@@ -1744,6 +1746,7 @@ static void sfp_check_state(struct sfp *sfp)
 		sfp_sm_event(sfp, state & SFP_F_LOS ?
 				SFP_E_LOS_HIGH : SFP_E_LOS_LOW);
 	rtnl_unlock();
+	mutex_unlock(&sfp->st_mutex);
 }
 
 static irqreturn_t sfp_irq(int irq, void *data)
@@ -1774,6 +1777,7 @@ static struct sfp *sfp_alloc(struct device *dev)
 	sfp->dev = dev;
 
 	mutex_init(&sfp->sm_mutex);
+	mutex_init(&sfp->st_mutex);
 	INIT_DELAYED_WORK(&sfp->poll, sfp_poll);
 	INIT_DELAYED_WORK(&sfp->timeout, sfp_timeout);
 
Signature · 1 line
2.20.1
author maintainer
#87 lore raw

PATCH 80/249 netfilter: ipset: fix a missing check of nla_parse

In reply to Sasha Levin

From: Aditya Pakki <[email protected]>

[ Upstream commit f4f5748bfec94cf418e49bf05f0c81a1b9ebc950 ]

When nla_parse fails, we should not use the results (the first argument). The fix checks if it fails, and if so, returns its error code upstream.

---
 net/netfilter/ipset/ip_set_core.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)
net/netfilter/ipset/ip_set_core.c +7 -3
index 3cdf171cd468..16afa0df4004 100644
@@ -1541,10 +1541,14 @@ call_ad(struct sock *ctnl, struct sk_buff *skb, struct ip_set *set,
 		memcpy(&errmsg->msg, nlh, nlh->nlmsg_len);
 		cmdattr = (void *)&errmsg->msg + min_len;
 
-		nla_parse_deprecated(cda, IPSET_ATTR_CMD_MAX, cmdattr,
-				     nlh->nlmsg_len - min_len,
-				     ip_set_adt_policy, NULL);
+		ret = nla_parse_deprecated(cda, IPSET_ATTR_CMD_MAX, cmdattr,
+					   nlh->nlmsg_len - min_len,
+					   ip_set_adt_policy, NULL);
 
+		if (ret) {
+			nlmsg_free(skb2);
+			return ret;
+		}
 		errline = nla_data(cda[IPSET_ATTR_LINENO]);
 
 		*errline = lineno;
Signature · 1 line
2.20.1
author maintainer
#88 lore raw

PATCH 81/249 ipset: Fix memory accounting for hash types on resize

In reply to Sasha Levin

From: Stefano Brivio <[email protected]>

[ Upstream commit 11921796f4799ca9c61c4b22cc54d84aa69f8a35 ]

If a fresh array block is allocated during resize, the current in-memory set size should be increased by the size of the block, not replaced by it.

Before the fix, adding entries to a hash set type, leading to a table resize, caused an inconsistent memory size to be reported. This becomes more obvious when swapping sets with similar sizes:

  # cat hash_ip_size.sh
  #!/bin/sh
  FAIL_RETRIES=10
  tries=0
  while [ ${tries} -lt ${FAIL_RETRIES} ]; do
  	ipset create t1 hash:ip
  	for i in `seq 1 4345`; do
  		ipset add t1 1.2.$((i / 255)).$((i % 255))
  	done
  	t1_init="$(ipset list t1|sed -n 's/Size in memory: \(.*\)/\1/p')"

  	ipset create t2 hash:ip
  	for i in `seq 1 4360`; do
  		ipset add t2 1.2.$((i / 255)).$((i % 255))
  	done
  	t2_init="$(ipset list t2|sed -n 's/Size in memory: \(.*\)/\1/p')"

  	ipset swap t1 t2
  	t1_swap="$(ipset list t1|sed -n 's/Size in memory: \(.*\)/\1/p')"
  	t2_swap="$(ipset list t2|sed -n 's/Size in memory: \(.*\)/\1/p')"

  	ipset destroy t1
  	ipset destroy t2
  	tries=$((tries + 1))

  	if [ ${t1_init} -lt 10000 ] || [ ${t2_init} -lt 10000 ]; then
  		echo "FAIL after ${tries} tries:"
  		echo "T1 size ${t1_init}, after swap ${t1_swap}"
  		echo "T2 size ${t2_init}, after swap ${t2_swap}"
  		exit 1
  	fi
  done
  echo "PASS"
  # echo -n 'func hash_ip4_resize +p' > /sys/kernel/debug/dynamic_debug/control
  # ./hash_ip_size.sh
  [ 2035.018673] attempt to resize set t1 from 10 to 11, t 00000000fe6551fa
  [ 2035.078583] set t1 resized from 10 (00000000fe6551fa) to 11 (00000000172a0163)
  [ 2035.080353] Table destroy by resize 00000000fe6551fa
  FAIL after 4 tries:
  T1 size 9064, after swap 71128
  T2 size 71128, after swap 9064
Fixes 9e41f26a505c ("netfilter: ipset: Count non-static extension memory for userspace")
---
 net/netfilter/ipset/ip_set_hash_gen.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/netfilter/ipset/ip_set_hash_gen.h +1 -1
index 10f619625abd..175f8fedcfaf 100644
@@ -622,7 +622,7 @@ mtype_resize(struct ip_set *set, bool retried)
 					goto cleanup;
 				}
 				m->size = AHASH_INIT_SIZE;
-				extsize = ext_size(AHASH_INIT_SIZE, dsize);
+				extsize += ext_size(AHASH_INIT_SIZE, dsize);
 				RCU_INIT_POINTER(hbucket(t, key), m);
 			} else if (m->pos >= m->size) {
 				struct hbucket *ht;
Signature · 1 line
2.20.1
author maintainer
#89 lore raw

PATCH 82/249 perf cs-etm: Properly set the value of 'old' and 'head' in snapshot mode

In reply to Sasha Levin

From: Mathieu Poirier <[email protected]>

[ Upstream commit e45c48a9a4d20ebc7b639a62c3ef8f4b08007027 ]

This patch adds the necessary intelligence to properly compute the value of 'old' and 'head' when operating in snapshot mode. That way we can get the latest information in the AUX buffer and be compatible with the generic AUX ring buffer mechanic.

Tester notes:

Quoted 1 line

Leo, have you had the chance to test/review this one? Suzuki?

Sure. I applied this patch on the perf/core branch (with latest commit 3e4fbf36c1e3 'perf augmented_raw_syscalls: Move reading filename to the loop') and passed testing with below steps:

  # perf record -e cs_etm/@tmc_etr0/ -S -m,64 --per-thread ./sort &
  [1] 19097
  Bubble sorting array of 30000 elements
  # kill -USR2 19097
  # kill -USR2 19097
  # kill -USR2 19097
  [ perf record: Woken up 4 times to write data ]
  [ perf record: Captured and wrote 0.753 MB perf.data ]
---
 tools/perf/arch/arm/util/cs-etm.c | 127 +++++++++++++++++++++++++++++-
 1 file changed, 123 insertions(+), 4 deletions(-)
tools/perf/arch/arm/util/cs-etm.c +123 -4
index 911426721170..0a278bbcaba6 100644
@@ -31,6 +31,8 @@ struct cs_etm_recording {
 	struct auxtrace_record	itr;
 	struct perf_pmu		*cs_etm_pmu;
 	struct perf_evlist	*evlist;
+	int			wrapped_cnt;
+	bool			*wrapped;
 	bool			snapshot_mode;
 	size_t			snapshot_size;
 };
@@ -536,16 +538,131 @@ static int cs_etm_info_fill(struct auxtrace_record *itr,
 	return 0;
 }
 
-static int cs_etm_find_snapshot(struct auxtrace_record *itr __maybe_unused,
+static int cs_etm_alloc_wrapped_array(struct cs_etm_recording *ptr, int idx)
+{
+	bool *wrapped;
+	int cnt = ptr->wrapped_cnt;
+
+	/* Make @ptr->wrapped as big as @idx */
+	while (cnt <= idx)
+		cnt++;
+
+	/*
+	 * Free'ed in cs_etm_recording_free().  Using realloc() to avoid
+	 * cross compilation problems where the host's system supports
+	 * reallocarray() but not the target.
+	 */
+	wrapped = realloc(ptr->wrapped, cnt * sizeof(bool));
+	if (!wrapped)
+		return -ENOMEM;
+
+	wrapped[cnt - 1] = false;
+	ptr->wrapped_cnt = cnt;
+	ptr->wrapped = wrapped;
+
+	return 0;
+}
+
+static bool cs_etm_buffer_has_wrapped(unsigned char *buffer,
+				      size_t buffer_size, u64 head)
+{
+	u64 i, watermark;
+	u64 *buf = (u64 *)buffer;
+	size_t buf_size = buffer_size;
+
+	/*
+	 * We want to look the very last 512 byte (chosen arbitrarily) in
+	 * the ring buffer.
+	 */
+	watermark = buf_size - 512;
+
+	/*
+	 * @head is continuously increasing - if its value is equal or greater
+	 * than the size of the ring buffer, it has wrapped around.
+	 */
+	if (head >= buffer_size)
+		return true;
+
+	/*
+	 * The value of @head is somewhere within the size of the ring buffer.
+	 * This can be that there hasn't been enough data to fill the ring
+	 * buffer yet or the trace time was so long that @head has numerically
+	 * wrapped around.  To find we need to check if we have data at the very
+	 * end of the ring buffer.  We can reliably do this because mmap'ed
+	 * pages are zeroed out and there is a fresh mapping with every new
+	 * session.
+	 */
+
+	/* @head is less than 512 byte from the end of the ring buffer */
+	if (head > watermark)
+		watermark = head;
+
+	/*
+	 * Speed things up by using 64 bit transactions (see "u64 *buf" above)
+	 */
+	watermark >>= 3;
+	buf_size >>= 3;
+
+	/*
+	 * If we find trace data at the end of the ring buffer, @head has
+	 * been there and has numerically wrapped around at least once.
+	 */
+	for (i = watermark; i < buf_size; i++)
+		if (buf[i])
+			return true;
+
+	return false;
+}
+
+static int cs_etm_find_snapshot(struct auxtrace_record *itr,
 				int idx, struct auxtrace_mmap *mm,
-				unsigned char *data __maybe_unused,
+				unsigned char *data,
 				u64 *head, u64 *old)
 {
+	int err;
+	bool wrapped;
+	struct cs_etm_recording *ptr =
+			container_of(itr, struct cs_etm_recording, itr);
+
+	/*
+	 * Allocate memory to keep track of wrapping if this is the first
+	 * time we deal with this *mm.
+	 */
+	if (idx >= ptr->wrapped_cnt) {
+		err = cs_etm_alloc_wrapped_array(ptr, idx);
+		if (err)
+			return err;
+	}
+
+	/*
+	 * Check to see if *head has wrapped around.  If it hasn't only the
+	 * amount of data between *head and *old is snapshot'ed to avoid
+	 * bloating the perf.data file with zeros.  But as soon as *head has
+	 * wrapped around the entire size of the AUX ring buffer it taken.
+	 */
+	wrapped = ptr->wrapped[idx];
+	if (!wrapped && cs_etm_buffer_has_wrapped(data, mm->len, *head)) {
+		wrapped = true;
+		ptr->wrapped[idx] = true;
+	}
+
 	pr_debug3("%s: mmap index %d old head %zu new head %zu size %zu\n",
 		  __func__, idx, (size_t)*old, (size_t)*head, mm->len);
 
-	*old = *head;
-	*head += mm->len;
+	/* No wrap has occurred, we can just use *head and *old. */
+	if (!wrapped)
+		return 0;
+
+	/*
+	 * *head has wrapped around - adjust *head and *old to pickup the
+	 * entire content of the AUX buffer.
+	 */
+	if (*head >= mm->len) {
+		*old = *head - mm->len;
+	} else {
+		*head += mm->len;
+		*old = *head - mm->len;
+	}
 
 	return 0;
 }
@@ -586,6 +703,8 @@ static void cs_etm_recording_free(struct auxtrace_record *itr)
 {
 	struct cs_etm_recording *ptr =
 			container_of(itr, struct cs_etm_recording, itr);
+
+	zfree(&ptr->wrapped);
 	free(ptr);
 }
 
Signature · 1 line
2.20.1
author maintainer
#90 lore raw

PATCH 83/249 perf test 6: Fix missing kvm module load for s390

In reply to Sasha Levin

From: Thomas Richter <[email protected]>

[ Upstream commit 53fe307dfd309e425b171f6272d64296a54f4dff ]

Command

   # perf test -Fv 6

fails with error

running test 100 'kvm-s390:kvm_s390_create_vm' failed to parse

    event 'kvm-s390:kvm_s390_create_vm', err -1, str 'unknown tracepoint'
    event syntax error: 'kvm-s390:kvm_s390_create_vm'
                         \___ unknown tracepoint

when the kvm module is not loaded or not built in.

Fix this by adding a valid function which tests if the module is loaded. Loaded modules (or builtin KVM support) have a directory named
/sys/kernel/debug/tracing/events/kvm-s390
for this tracepoint.

Check for existence of this directory.

---
 tools/perf/tests/parse-events.c | 27 +++++++++++++++++++++++++++
 1 file changed, 27 insertions(+)
tools/perf/tests/parse-events.c +27 -0
index 4a69c07f4101..8f3c80e13584 100644
@@ -18,6 +18,32 @@
 #define PERF_TP_SAMPLE_TYPE (PERF_SAMPLE_RAW | PERF_SAMPLE_TIME | \
 			     PERF_SAMPLE_CPU | PERF_SAMPLE_PERIOD)
 
+#if defined(__s390x__)
+/* Return true if kvm module is available and loaded. Test this
+ * and retun success when trace point kvm_s390_create_vm
+ * exists. Otherwise this test always fails.
+ */
+static bool kvm_s390_create_vm_valid(void)
+{
+	char *eventfile;
+	bool rc = false;
+
+	eventfile = get_events_file("kvm-s390");
+
+	if (eventfile) {
+		DIR *mydir = opendir(eventfile);
+
+		if (mydir) {
+			rc = true;
+			closedir(mydir);
+		}
+		put_events_file(eventfile);
+	}
+
+	return rc;
+}
+#endif
+
 static int test__checkevent_tracepoint(struct perf_evlist *evlist)
 {
 	struct perf_evsel *evsel = perf_evlist__first(evlist);
@@ -1642,6 +1668,7 @@ static struct evlist_test test__events[] = {
 	{
 		.name  = "kvm-s390:kvm_s390_create_vm",
 		.check = test__checkevent_tracepoint,
+		.valid = kvm_s390_create_vm_valid,
 		.id    = 100,
 	},
 #endif
Signature · 1 line
2.20.1
author maintainer
#91 lore raw

PATCH 84/249 perf report: Fix OOM error in TUI mode on s390

In reply to Sasha Levin

From: Thomas Richter <[email protected]>

[ Upstream commit 8a07aa4e9b7b0222129c07afff81634a884b2866 ]

Debugging a OOM error using the TUI interface revealed this issue on s390:

[tmricht@m83lp54 perf]$ cat /proc/kallsyms |sort
....
00000001119b7158 B radix_tree_node_cachep 00000001119b8000 B __bss_stop
00000001119b8000 B _end
000003ff80002850 t autofs_mount [autofs4]
000003ff80002868 t autofs_show_options [autofs4]
000003ff80002a98 t autofs_evict_inode [autofs4]
....

There is a huge gap between the last kernel symbol __bss_stop/_end and the first kernel module symbol autofs_mount (from autofs4 module).

After reading the kernel symbol table via functions:

dso__load()

 +--> dso__load_kernel_sym()
      +--> dso__load_kallsyms()
	   +--> __dso_load_kallsyms()
	        +--> symbols__fixup_end()

the symbol __bss_stop has a start address of 1119b8000 and an end address of 3ff80002850, as can be seen by this debug statement:

symbols__fixup_end __bss_stop start:0x1119b8000 end:0x3ff80002850

The size of symbol __bss_stop is 0x3fe6e64a850 bytes! It is the last kernel symbol and fills up the space until the first kernel module symbol.

This size kills the TUI interface when executing the following code:

process_sample_event()

    hist_entry_iter__add()
      hist_iter__report_callback()
        hist_entry__inc_addr_samples()
          symbol__inc_addr_samples(symbol = __bss_stop)
            symbol__cycles_hist()
               annotated_source__alloc_histograms(...,
				                symbol__size(sym),
		                                ...)

This function allocates memory to save sample histograms. The symbol_size() marco is defined as sym->end - sym->start, which results in above value of 0x3fe6e64a850 bytes and
the call to calloc() in annotated_source__alloc_histograms() fails.

The histgram memory allocation might fail, make this failure no-fatal and continue processing.

Output before:

[tmricht@m83lp54 perf]$ ./perf --debug stderr=1 report -vvvvv \
					      -i ~/slow.data 2>/tmp/2
[tmricht@m83lp54 perf]$ tail -5 /tmp/2
  __symbol__inc_addr_samples(875): ENOMEM! sym->name=__bss_stop,
		start=0x1119b8000, addr=0x2aa0005eb08, end=0x3ff80002850,
		func: 0

problem adding hist entry, skipping event
0x938b8 [0x8]: failed to process type: 68 [Cannot allocate memory] [tmricht@m83lp54 perf]$

Output after:

[tmricht@m83lp54 perf]$ ./perf --debug stderr=1 report -vvvvv \
					      -i ~/slow.data 2>/tmp/2

[tmricht@m83lp54 perf]$ tail -5 /tmp/2
symbol__inc_addr_samples map:0x1597830 start:0x110730000 end:0x3ff80002850 symbol__hists notes->src:0x2aa2a70 nr_hists:1 symbol__inc_addr_samples sym:unlink_anon_vmas src:0x2aa2a70 __symbol__inc_addr_samples: addr=0x11094c69e
0x11094c670 unlink_anon_vmas: period++ [addr: 0x11094c69e, 0x2e, evidx=0]
=> nr_samples: 1, period: 526008
[tmricht@m83lp54 perf]$

There is no error about failed memory allocation and the TUI interface shows all entries.

---
 tools/perf/util/annotate.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)
tools/perf/util/annotate.c +2 -3
index 79db038b56f2..c8ce13419d9b 100644
@@ -931,9 +931,8 @@ static int symbol__inc_addr_samples(struct symbol *sym, struct map *map,
 	if (sym == NULL)
 		return 0;
 	src = symbol__hists(sym, evsel->evlist->nr_entries);
-	if (src == NULL)
-		return -ENOMEM;
-	return __symbol__inc_addr_samples(sym, map, src, evsel->idx, addr, sample);
+	return (src) ?  __symbol__inc_addr_samples(sym, map, src, evsel->idx,
+						   addr, sample) : 0;
 }
 
 static int symbol__account_cycles(u64 addr, u64 start,
Signature · 1 line
2.20.1
author maintainer
#92 lore raw

PATCH 85/249 selftests/bpf : clean up feature/ when make clean

In reply to Sasha Levin

From: Hechao Li <[email protected]>

[ Upstream commit 89cceaa939171fafa153d4bf637b39e396bbd785 ]

An error "implicit declaration of function 'reallocarray'" can be thrown with the following steps:

$ cd tools/testing/selftests/bpf
$ make clean && make CC=<Path to GCC 4.8.5>
$ make clean && make CC=<Path to GCC 7.x>

The cause is that the feature folder generated by GCC 4.8.5 is not removed, leaving feature-reallocarray being 1, which causes reallocarray not defined when re-compliing with GCC 7.x. This diff adds feature folder to EXTRA_CLEAN to avoid this problem.

v2: Rephrase the commit message.

---
 tools/testing/selftests/bpf/Makefile | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
tools/testing/selftests/bpf/Makefile +2 -1
index e36356e2377e..1c9511262947 100644
@@ -275,4 +275,5 @@ $(OUTPUT)/verifier/tests.h: $(VERIFIER_TESTS_DIR) $(VERIFIER_TEST_FILES)
 		 ) > $(VERIFIER_TESTS_H))
 
 EXTRA_CLEAN := $(TEST_CUSTOM_PROGS) $(ALU32_BUILD_DIR) \
-	$(VERIFIER_TESTS_H) $(PROG_TESTS_H) $(MAP_TESTS_H)
+	$(VERIFIER_TESTS_H) $(PROG_TESTS_H) $(MAP_TESTS_H) \
+	feature
Signature · 1 line
2.20.1
author maintainer
#93 lore raw

PATCH 86/249 irqchip/meson-gpio: Add support for Meson-G12A SoC

In reply to Sasha Levin

From: Xingyu Chen <[email protected]>

[ Upstream commit c64a9e804ccf86eb202bfd1c6a8c5233c75a0431 ]

The Meson-G12A SoC uses the same GPIO interrupt controller IP block as the other Meson SoCs, A totle of 100 pins can be spied on, which is the sum of:

- 223:100 undefined (no interrupt)

- 99:97   3 pins on bank GPIOE
- 96:77   20 pins on bank GPIOX
- 76:61   16 pins on bank GPIOA
- 60:53   8 pins on bank GPIOC
- 52:37   16 pins on bank BOOT
- 36:28   9 pins on bank GPIOH
- 27:12   16 pins on bank GPIOZ
- 11:0    12 pins in the AO domain
---
 drivers/irqchip/irq-meson-gpio.c | 1 +
 1 file changed, 1 insertion(+)
drivers/irqchip/irq-meson-gpio.c +1 -0
index 8eb92eb98f54..dcdc23b9dce6 100644
@@ -60,6 +60,7 @@ static const struct of_device_id meson_irq_gpio_matches[] = {
 	{ .compatible = "amlogic,meson-gxbb-gpio-intc", .data = &gxbb_params },
 	{ .compatible = "amlogic,meson-gxl-gpio-intc", .data = &gxl_params },
 	{ .compatible = "amlogic,meson-axg-gpio-intc", .data = &axg_params },
+	{ .compatible = "amlogic,meson-g12a-gpio-intc", .data = &axg_params },
 	{ }
 };
 
Signature · 7 lines
2.20.1


_______________________________________________
linux-amlogic mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-amlogic
author maintainer
#94 lore raw

PATCH 87/249 media: uvcvideo: Fix access to uninitialized fields on probe error

In reply to Sasha Levin

From: Oliver Neukum <[email protected]>

[ Upstream commit 11a087f484bf15ff65f0a9f277aa5a61fd07ed2a ]

We need to check whether this work we are canceling actually is initialized.

---
 drivers/media/usb/uvc/uvc_ctrl.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/media/usb/uvc/uvc_ctrl.c +3 -1
index 26163a5bde7d..e399b9fad757 100644
@@ -2345,7 +2345,9 @@ void uvc_ctrl_cleanup_device(struct uvc_device *dev)
 	struct uvc_entity *entity;
 	unsigned int i;
 
-	cancel_work_sync(&dev->async_ctrl.work);
+	/* Can be uninitialized if we are aborting on probe error. */
+	if (dev->async_ctrl.work.func)
+		cancel_work_sync(&dev->async_ctrl.work);
 
 	/* Free controls and control mappings for all entities. */
 	list_for_each_entry(entity, &dev->entities, list) {
Signature · 1 line
2.20.1
author maintainer
#95 lore raw

PATCH 88/249 media: fdp1: Support M3N and E3 platforms

In reply to Sasha Levin

From: Kieran Bingham <[email protected]>

[ Upstream commit 4e8c120de9268fc26f583268b9d22e7d37c4595f ]

New Gen3 R-Car platforms incorporate the FDP1 with an updated version register. No code change is required to support these targets, but they will currently report an error stating that the device can not be identified.

Update the driver to match against the new device types.

---
 drivers/media/platform/rcar_fdp1.c | 8 ++++++++
 1 file changed, 8 insertions(+)
drivers/media/platform/rcar_fdp1.c +8 -0
index 6a90bc4c476e..b8615a288e2b 100644
@@ -257,6 +257,8 @@ MODULE_PARM_DESC(debug, "activate debug info");
 #define FD1_IP_H3_ES1			0x02010101
 #define FD1_IP_M3W			0x02010202
 #define FD1_IP_H3			0x02010203
+#define FD1_IP_M3N			0x02010204
+#define FD1_IP_E3			0x02010205
 
 /* LUTs */
 #define FD1_LUT_DIF_ADJ			0x1000
@@ -2365,6 +2367,12 @@ static int fdp1_probe(struct platform_device *pdev)
 	case FD1_IP_H3:
 		dprintk(fdp1, "FDP1 Version R-Car H3\n");
 		break;
+	case FD1_IP_M3N:
+		dprintk(fdp1, "FDP1 Version R-Car M3N\n");
+		break;
+	case FD1_IP_E3:
+		dprintk(fdp1, "FDP1 Version R-Car E3\n");
+		break;
 	default:
 		dev_err(fdp1->dev, "FDP1 Unidentifiable (0x%08x)\n",
 			hw_version);
Signature · 1 line
2.20.1
author maintainer
#96 lore raw

PATCH 89/249 iommu: Fix a leak in iommu_insert_resv_region

In reply to Sasha Levin

From: Eric Auger <[email protected]>

[ Upstream commit ad0834dedaa15c3a176f783c0373f836e44b4700 ]

In case we expand an existing region, we unlink this latter and insert the larger one. In that case we should free the original region after the insertion. Also we can immediately return.

Fixes 6c65fb318e8b ("iommu: iommu_get_group_resv_regions")
---
 drivers/iommu/iommu.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)
drivers/iommu/iommu.c +5 -3
index 9f0a2844371c..30db41e9f15c 100644
@@ -225,18 +225,21 @@ static int iommu_insert_resv_region(struct iommu_resv_region *new,
 			pos = pos->next;
 		} else if ((start >= a) && (end <= b)) {
 			if (new->type == type)
-				goto done;
+				return 0;
 			else
 				pos = pos->next;
 		} else {
 			if (new->type == type) {
 				phys_addr_t new_start = min(a, start);
 				phys_addr_t new_end = max(b, end);
+				int ret;
 
 				list_del(&entry->list);
 				entry->start = new_start;
 				entry->length = new_end - new_start + 1;
-				iommu_insert_resv_region(entry, regions);
+				ret = iommu_insert_resv_region(entry, regions);
+				kfree(entry);
+				return ret;
 			} else {
 				pos = pos->next;
 			}
@@ -249,7 +252,6 @@ static int iommu_insert_resv_region(struct iommu_resv_region *new,
 		return -ENOMEM;
 
 	list_add_tail(&region->list, pos);
-done:
 	return 0;
 }
 
Signature · 6 lines
2.20.1

_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
author maintainer
#97 lore raw

PATCH 90/249 gpio: omap: fix lack of irqstatus_raw0 for OMAP4

In reply to Sasha Levin

From: Russell King <[email protected]>

[ Upstream commit 64ea3e9094a1f13b96c33244a3fb3a0f45690bd2 ]

Commit 384ebe1c2849 ("gpio/omap: Add DT support to GPIO driver") added the register definition tables to the gpio-omap driver. Subsequently to that commit, commit 4e962e8998cc ("gpio/omap: remove cpu_is_omapxxxx() checks from *_runtime_resume()") added definitions for irqstatus_raw* registers to the legacy OMAP4 definitions, but missed the DT definitions.

This causes an unintentional change of behaviour for the 1.101 errata workaround on OMAP4 platforms. Fix this oversight.

Fixes 4e962e8998cc ("gpio/omap: remove cpu_is_omapxxxx() checks from *_runtime_resume()")
---
 drivers/gpio/gpio-omap.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/gpio/gpio-omap.c +2 -0
index 9276ef616430..7632c98aa3a4 100644
@@ -1453,6 +1453,8 @@ static struct omap_gpio_reg_offs omap4_gpio_regs = {
 	.clr_dataout =		OMAP4_GPIO_CLEARDATAOUT,
 	.irqstatus =		OMAP4_GPIO_IRQSTATUS0,
 	.irqstatus2 =		OMAP4_GPIO_IRQSTATUS1,
+	.irqstatus_raw0 =	OMAP4_GPIO_IRQSTATUSRAW0,
+	.irqstatus_raw1 =	OMAP4_GPIO_IRQSTATUSRAW1,
 	.irqenable =		OMAP4_GPIO_IRQSTATUSSET0,
 	.irqenable2 =		OMAP4_GPIO_IRQSTATUSSET1,
 	.set_irqenable =	OMAP4_GPIO_IRQSTATUSSET0,
Signature · 1 line
2.20.1
author maintainer
#98 lore raw

PATCH 91/249 gpio: omap: ensure irq is enabled before wakeup

In reply to Sasha Levin

From: Russell King <[email protected]>

[ Upstream commit c859e0d479b3b4f6132fc12637c51e01492f31f6 ]

Documentation states:

NOTE: There must be a correlation between the wake-up enable and interrupt-enable registers. If a GPIO pin has a wake-up configured on it, it must also have the corresponding interrupt enabled (on one of the two interrupt lines).

Ensure that this condition is always satisfied by enabling the detection events after enabling the interrupt, and disabling the detection before disabling the interrupt. This ensures interrupt/wakeup events can not happen until both the wakeup and interrupt enables correlate.

If we do any clearing, clear between the interrupt enable/disable and trigger setting.

---
 drivers/gpio/gpio-omap.c | 15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)
drivers/gpio/gpio-omap.c +8 -7
index 7632c98aa3a4..746aa9caf934 100644
@@ -829,9 +829,9 @@ static void omap_gpio_irq_shutdown(struct irq_data *d)
 
 	raw_spin_lock_irqsave(&bank->lock, flags);
 	bank->irq_usage &= ~(BIT(offset));
-	omap_set_gpio_irqenable(bank, offset, 0);
-	omap_clear_gpio_irqstatus(bank, offset);
 	omap_set_gpio_triggering(bank, offset, IRQ_TYPE_NONE);
+	omap_clear_gpio_irqstatus(bank, offset);
+	omap_set_gpio_irqenable(bank, offset, 0);
 	if (!LINE_USED(bank->mod_usage, offset))
 		omap_clear_gpio_debounce(bank, offset);
 	omap_disable_gpio_module(bank, offset);
@@ -867,8 +867,8 @@ static void omap_gpio_mask_irq(struct irq_data *d)
 	unsigned long flags;
 
 	raw_spin_lock_irqsave(&bank->lock, flags);
-	omap_set_gpio_irqenable(bank, offset, 0);
 	omap_set_gpio_triggering(bank, offset, IRQ_TYPE_NONE);
+	omap_set_gpio_irqenable(bank, offset, 0);
 	raw_spin_unlock_irqrestore(&bank->lock, flags);
 }
 
@@ -880,9 +880,6 @@ static void omap_gpio_unmask_irq(struct irq_data *d)
 	unsigned long flags;
 
 	raw_spin_lock_irqsave(&bank->lock, flags);
-	if (trigger)
-		omap_set_gpio_triggering(bank, offset, trigger);
-
 	omap_set_gpio_irqenable(bank, offset, 1);
 
 	/*
@@ -890,9 +887,13 @@ static void omap_gpio_unmask_irq(struct irq_data *d)
 	 * is cleared, thus after the handler has run. OMAP4 needs this done
 	 * after enabing the interrupt to clear the wakeup status.
 	 */
-	if (bank->level_mask & BIT(offset))
+	if (bank->regs->leveldetect0 && bank->regs->wkup_en &&
+	    trigger & (IRQ_TYPE_LEVEL_HIGH | IRQ_TYPE_LEVEL_LOW))
 		omap_clear_gpio_irqstatus(bank, offset);
 
+	if (trigger)
+		omap_set_gpio_triggering(bank, offset, trigger);
+
 	raw_spin_unlock_irqrestore(&bank->lock, flags);
 }
 
Signature · 1 line
2.20.1
author maintainer
#99 lore raw

PATCH 92/249 regmap: fix bulk writes on paged registers

In reply to Sasha Levin

From: Srinivas Kandagatla <[email protected]>

[ Upstream commit db057679de3e9e6a03c1bcd5aee09b0d25fd9f5b ]

On buses like SlimBus and SoundWire which does not support gather_writes yet in regmap, A bulk write on paged register would be silently ignored after programming page.
This is because local variable 'ret' value in regmap_raw_write_impl() gets reset to 0 once page register is written successfully and the code below checks for 'ret' value to be -ENOTSUPP before linearising the write buffer to send to bus->write().

Fix this by resetting the 'ret' value to -ENOTSUPP in cases where gather_writes() is not supported or single register write is not possible.

---
 drivers/base/regmap/regmap.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/base/regmap/regmap.c +2 -0
index f1025452bb39..19f57ccfbe1d 100644
@@ -1637,6 +1637,8 @@ static int _regmap_raw_write_impl(struct regmap *map, unsigned int reg,
 					     map->format.reg_bytes +
 					     map->format.pad_bytes,
 					     val, val_len);
+	else
+		ret = -ENOTSUPP;
 
 	/* If that didn't work fall back on linearising by hand. */
 	if (ret == -ENOTSUPP) {
Signature · 1 line
2.20.1
author maintainer
#100 lore raw

PATCH 93/249 gpio: omap: Fix lost edge wake-up interrupts

In reply to Sasha Levin

From: Tony Lindgren <[email protected]>

[ Upstream commit a522f1d0c381c42f3ace13b8bbeeccabdd6d2e5c ]

If an edge interrupt triggers while entering idle just before we save GPIO datain register to saved_datain, the triggered GPIO will not be noticed on wake-up. This is because the saved_datain and GPIO datain are the same on wake-up in omap_gpio_unidle(). Let's fix this by ignoring any pending edge interrupts for saved_datain.

This issue affects only idle states where the GPIO module internal wake-up path is operational. For deeper idle states where the GPIO module gets powered off, Linux generic wakeirqs must be used for the padconf wake-up events with pinctrl-single driver. For examples, please see "interrupts-extended" dts usage in many drivers.

This issue can be somewhat easily reproduced by pinging an idle system with smsc911x Ethernet interface configured IRQ_TYPE_EDGE_FALLING. At some point the smsc911x interrupts will just stop triggering. Also if WLCORE WLAN is used with EDGE interrupt like it's documentation specifies, we can see lost interrupts without this patch.

Note that in the long run we may be able to cancel entering idle by returning an error in gpio_omap_cpu_notifier() on pending interrupts. But let's fix the bug first.

Also note that because of the recent clean-up efforts this patch does not apply directly to older kernels. This does fix a long term issue though, and can be backported as needed.

Cc Aaro Koskinen <[email protected]>
Cc Grygorii Strashko <[email protected]>
Cc Keerthy <[email protected]>
Cc Ladislav Michl <[email protected]>
Cc Peter Ujfalusi <[email protected]>
Cc Russell King <[email protected]>
Cc Tero Kristo <[email protected]>
---
 drivers/gpio/gpio-omap.c | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)
drivers/gpio/gpio-omap.c +11 -1
index 746aa9caf934..8591c410ecaa 100644
@@ -1275,13 +1275,23 @@ static void omap_gpio_idle(struct gpio_bank *bank, bool may_lose_context)
 {
 	struct device *dev = bank->chip.parent;
 	void __iomem *base = bank->base;
-	u32 nowake;
+	u32 mask, nowake;
 
 	bank->saved_datain = readl_relaxed(base + bank->regs->datain);
 
 	if (!bank->enabled_non_wakeup_gpios)
 		goto update_gpio_context_count;
 
+	/* Check for pending EDGE_FALLING, ignore EDGE_BOTH */
+	mask = bank->enabled_non_wakeup_gpios & bank->context.fallingdetect;
+	mask &= ~bank->context.risingdetect;
+	bank->saved_datain |= mask;
+
+	/* Check for pending EDGE_RISING, ignore EDGE_BOTH */
+	mask = bank->enabled_non_wakeup_gpios & bank->context.risingdetect;
+	mask &= ~bank->context.fallingdetect;
+	bank->saved_datain &= ~mask;
+
 	if (!may_lose_context)
 		goto update_gpio_context_count;
 
Signature · 1 line
2.20.1
author maintainer
#101 lore raw

PATCH 94/249 media: davinci: vpif_capture: fix memory leak in vpif_probe()

In reply to Sasha Levin

From: Young Xiao <[email protected]>

[ Upstream commit 64f883cd98c6d43013fb0cea788b63e50ebc068c ]

If vpif_probe() fails on v4l2_device_register() and vpif_probe_complete(), then memory allocated at initialize_vpif() for global vpif_obj.dev[i] become unreleased.

The patch adds deallocation of vpif_obj.dev[i] on the error path.

---
 drivers/media/platform/davinci/vpif_capture.c | 16 ++++++++++++++--
 1 file changed, 14 insertions(+), 2 deletions(-)
drivers/media/platform/davinci/vpif_capture.c +14 -2
index 61809d2050fa..f0f7ef638c56 100644
@@ -1376,6 +1376,14 @@ static int initialize_vpif(void)
 	return err;
 }
 
+static inline void free_vpif_objs(void)
+{
+	int i;
+
+	for (i = 0; i < VPIF_CAPTURE_MAX_DEVICES; i++)
+		kfree(vpif_obj.dev[i]);
+}
+
 static int vpif_async_bound(struct v4l2_async_notifier *notifier,
 			    struct v4l2_subdev *subdev,
 			    struct v4l2_async_subdev *asd)
@@ -1645,7 +1653,7 @@ static __init int vpif_probe(struct platform_device *pdev)
 	err = v4l2_device_register(vpif_dev, &vpif_obj.v4l2_dev);
 	if (err) {
 		v4l2_err(vpif_dev->driver, "Error registering v4l2 device\n");
-		goto cleanup;
+		goto vpif_free;
 	}
 
 	while ((res = platform_get_resource(pdev, IORESOURCE_IRQ, res_idx))) {
@@ -1692,7 +1700,9 @@ static __init int vpif_probe(struct platform_device *pdev)
 				  "registered sub device %s\n",
 				   subdevdata->name);
 		}
-		vpif_probe_complete();
+		err = vpif_probe_complete();
+		if (err)
+			goto probe_subdev_out;
 	} else {
 		vpif_obj.notifier.ops = &vpif_async_ops;
 		err = v4l2_async_notifier_register(&vpif_obj.v4l2_dev,
@@ -1711,6 +1721,8 @@ static __init int vpif_probe(struct platform_device *pdev)
 	kfree(vpif_obj.sd);
 vpif_unregister:
 	v4l2_device_unregister(&vpif_obj.v4l2_dev);
+vpif_free:
+	free_vpif_objs();
 cleanup:
 	v4l2_async_notifier_cleanup(&vpif_obj.notifier);
 
Signature · 1 line
2.20.1
author maintainer
#102 lore raw

PATCH 95/249 bpf: silence warning messages in core

In reply to Sasha Levin

From: Valdis Klētnieks <[email protected]>

[ Upstream commit aee450cbe482a8c2f6fa5b05b178ef8b8ff107ca ]

Compiling kernel/bpf/core.c with W=1 causes a flood of warnings:

kernel/bpf/core.c:1198:65: warning: initialized field overwritten [-Woverride-init]
1198 | #define BPF_INSN_3_TBL(x, y, z) [BPF_##x | BPF_##y | BPF_##z] = true
| ^~~~
kernel/bpf/core.c:1087:2: note: in expansion of macro 'BPF_INSN_3_TBL'

 1087 |  INSN_3(ALU, ADD,  X),   \
      |  ^~~~~~
kernel/bpf/core.c:1202:3: note: in expansion of macro 'BPF_INSN_MAP'
 1202 |   BPF_INSN_MAP(BPF_INSN_2_TBL, BPF_INSN_3_TBL),
      |   ^~~~~~~~~~~~

kernel/bpf/core.c:1198:65: note: (near initialization for 'public_insntable[12]')
1198 | #define BPF_INSN_3_TBL(x, y, z) [BPF_##x | BPF_##y | BPF_##z] = true
| ^~~~
kernel/bpf/core.c:1087:2: note: in expansion of macro 'BPF_INSN_3_TBL'

 1087 |  INSN_3(ALU, ADD,  X),   \
      |  ^~~~~~
kernel/bpf/core.c:1202:3: note: in expansion of macro 'BPF_INSN_MAP'
 1202 |   BPF_INSN_MAP(BPF_INSN_2_TBL, BPF_INSN_3_TBL),
      |   ^~~~~~~~~~~~

98 copies of the above.

The attached patch silences the warnings, because we *know* we're overwriting the default initializer. That leaves bpf/core.c with only 6 other warnings, which become more visible in comparison.

---
 kernel/bpf/Makefile | 1 +
 1 file changed, 1 insertion(+)
kernel/bpf/Makefile +1 -0
index 4c2fa3ac56f6..29d781061cd5 100644
@@ -1,5 +1,6 @@
 # SPDX-License-Identifier: GPL-2.0
 obj-y := core.o
+CFLAGS_core.o += $(call cc-disable-warning, override-init)
 
 obj-$(CONFIG_BPF_SYSCALL) += syscall.o verifier.o inode.o helpers.o tnum.o
 obj-$(CONFIG_BPF_SYSCALL) += hashtab.o arraymap.o percpu_freelist.o bpf_lru_list.o lpm_trie.o map_in_map.o
Signature · 1 line
2.20.1
author maintainer
#103 lore raw

PATCH 96/249 media: s5p-mfc: fix reading min scratch buffer size on MFC v6/v7

In reply to Sasha Levin

From: Marek Szyprowski <[email protected]>

[ Upstream commit be22203aec440c1761ce8542c2636ac6c8951e3a ]

MFC v6 and v7 has no register to read min scratch buffer size, so it has to be read conditionally only if hardware supports it. This fixes following NULL pointer exception on SoCs with MFC v6/v7:

8<--- cut here ---

Unable to handle kernel NULL pointer dereference at virtual address 00000000
pgd = f25837f9
[00000000] *pgd=bd93d835
Internal error: Oops: 17 [#1] PREEMPT SMP ARM
Modules linked in: btmrvl_sdio btmrvl bluetooth mwifiex_sdio mwifiex ecdh_generic ecc
Hardware name: SAMSUNG EXYNOS (Flattened Device Tree)
PC is at s5p_mfc_get_min_scratch_buf_size+0x30/0x3c
LR is at s5p_mfc_get_min_scratch_buf_size+0x28/0x3c

...
[<c074f998>] (s5p_mfc_get_min_scratch_buf_size) from [<c0745bc0>] (s5p_mfc_irq+0x814/0xa5c) [<c0745bc0>] (s5p_mfc_irq) from [<c019a218>] (__handle_irq_event_percpu+0x64/0x3f8) [<c019a218>] (__handle_irq_event_percpu) from [<c019a5d8>] (handle_irq_event_percpu+0x2c/0x7c) [<c019a5d8>] (handle_irq_event_percpu) from [<c019a660>] (handle_irq_event+0x38/0x5c) [<c019a660>] (handle_irq_event) from [<c019ebc4>] (handle_fasteoi_irq+0xc4/0x180) [<c019ebc4>] (handle_fasteoi_irq) from [<c0199270>] (generic_handle_irq+0x24/0x34) [<c0199270>] (generic_handle_irq) from [<c0199888>] (__handle_domain_irq+0x7c/0xec) [<c0199888>] (__handle_domain_irq) from [<c04ac298>] (gic_handle_irq+0x58/0x9c) [<c04ac298>] (gic_handle_irq) from [<c0101ab0>] (__irq_svc+0x70/0xb0)

Exception stack(0xe73ddc60 to 0xe73ddca8)

...
[<c0101ab0>] (__irq_svc) from [<c01967d8>] (console_unlock+0x5a8/0x6a8) [<c01967d8>] (console_unlock) from [<c01981d0>] (vprintk_emit+0x118/0x2d8) [<c01981d0>] (vprintk_emit) from [<c01983b0>] (vprintk_default+0x20/0x28) [<c01983b0>] (vprintk_default) from [<c01989b4>] (printk+0x30/0x54) [<c01989b4>] (printk) from [<c07500b8>] (s5p_mfc_init_decode_v6+0x1d4/0x284) [<c07500b8>] (s5p_mfc_init_decode_v6) from [<c07230d0>] (vb2_start_streaming+0x24/0x150) [<c07230d0>] (vb2_start_streaming) from [<c0724e4c>] (vb2_core_streamon+0x11c/0x15c) [<c0724e4c>] (vb2_core_streamon) from [<c07478b8>] (vidioc_streamon+0x64/0xa0) [<c07478b8>] (vidioc_streamon) from [<c0709640>] (__video_do_ioctl+0x28c/0x45c) [<c0709640>] (__video_do_ioctl) from [<c0709bc8>] (video_usercopy+0x260/0x8a4) [<c0709bc8>] (video_usercopy) from [<c02b3820>] (do_vfs_ioctl+0xb0/0x9fc) [<c02b3820>] (do_vfs_ioctl) from [<c02b41a0>] (ksys_ioctl+0x34/0x58) [<c02b41a0>] (ksys_ioctl) from [<c0101000>] (ret_fast_syscall+0x0/0x28)

Exception stack(0xe73ddfa8 to 0xe73ddff0)
...
---[ end trace 376cf5ba6e0bee93 ]---
---
 drivers/media/platform/s5p-mfc/s5p_mfc.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/media/platform/s5p-mfc/s5p_mfc.c +2 -1
index 4e936b95018a..481088a83212 100644
@@ -523,7 +523,8 @@ static void s5p_mfc_handle_seq_done(struct s5p_mfc_ctx *ctx,
 				dev);
 		ctx->mv_count = s5p_mfc_hw_call(dev->mfc_ops, get_mv_count,
 				dev);
-		ctx->scratch_buf_size = s5p_mfc_hw_call(dev->mfc_ops,
+		if (FW_HAS_E_MIN_SCRATCH_BUF(dev))
+			ctx->scratch_buf_size = s5p_mfc_hw_call(dev->mfc_ops,
 						get_min_scratch_buf_size, dev);
 		if (ctx->img_width == 0 || ctx->img_height == 0)
 			ctx->state = MFCINST_ERROR;
Signature · 1 line
2.20.1
author maintainer
#104 lore raw

PATCH 97/249 selinux: fix empty write to keycreate file

In reply to Sasha Levin

From: Ondrej Mosnacek <[email protected]>

[ Upstream commit 464c258aa45b09f16aa0f05847ed8895873262d9 ]

When sid == 0 (we are resetting keycreate_sid to the default value), we should skip the KEY__CREATE check.

Before this patch, doing a zero-sized write to /proc/self/keycreate would check if the current task can create unlabeled keys (which would usually fail with -EACCESS and generate an AVC). Now it skips the check and correctly sets the task's keycreate_sid to 0.

Bug report: https://bugzilla.redhat.com/show_bug.cgi?id=1719067

Tested using the reproducer from the report above.

Fixes 4eb582cf1fbd ("[PATCH] keys: add a way to store the appropriate context for newly-created keys")
---
 security/selinux/hooks.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)
security/selinux/hooks.c +6 -5
index 94de51628fdc..3ec7ac70c313 100644
@@ -6351,11 +6351,12 @@ static int selinux_setprocattr(const char *name, void *value, size_t size)
 	} else if (!strcmp(name, "fscreate")) {
 		tsec->create_sid = sid;
 	} else if (!strcmp(name, "keycreate")) {
-		error = avc_has_perm(&selinux_state,
-				     mysid, sid, SECCLASS_KEY, KEY__CREATE,
-				     NULL);
-		if (error)
-			goto abort_change;
+		if (sid) {
+			error = avc_has_perm(&selinux_state, mysid, sid,
+					     SECCLASS_KEY, KEY__CREATE, NULL);
+			if (error)
+				goto abort_change;
+		}
 		tsec->keycreate_sid = sid;
 	} else if (!strcmp(name, "sockcreate")) {
 		tsec->sockcreate_sid = sid;
Signature · 1 line
2.20.1
author maintainer
#105 lore raw

PATCH 98/249 crypto: testmgr - add some more preemption points

In reply to Sasha Levin

From: Eric Biggers <[email protected]>

[ Upstream commit e63e1b0dd0003dc31f73d875907432be3a2abe5d ]

Call cond_resched() after each fuzz test iteration. This avoids stall warnings if fuzz_iterations is set very high for testing purposes.

While we're at it, also call cond_resched() after finishing testing each test vector.

---
 crypto/testmgr.c | 6 ++++++
 1 file changed, 6 insertions(+)
crypto/testmgr.c +6 -0
index 658a7eeebab2..292d28caf00f 100644
@@ -1279,6 +1279,7 @@ static int test_hash_vec(const char *driver, const struct hash_testvec *vec,
 						req, tsgl, hashstate);
 			if (err)
 				return err;
+			cond_resched();
 		}
 	}
 #endif
@@ -1493,6 +1494,7 @@ static int __alg_test_hash(const struct hash_testvec *vecs,
 		err = test_hash_vec(driver, &vecs[i], i, req, tsgl, hashstate);
 		if (err)
 			goto out;
+		cond_resched();
 	}
 	err = test_hash_vs_generic_impl(driver, generic_driver, maxkeysize, req,
 					tsgl, hashstate);
@@ -1755,6 +1757,7 @@ static int test_aead_vec(const char *driver, int enc,
 						&cfg, req, tsgls);
 			if (err)
 				return err;
+			cond_resched();
 		}
 	}
 #endif
@@ -1994,6 +1997,7 @@ static int test_aead(const char *driver, int enc,
 				    tsgls);
 		if (err)
 			return err;
+		cond_resched();
 	}
 	return 0;
 }
@@ -2336,6 +2340,7 @@ static int test_skcipher_vec(const char *driver, int enc,
 						    &cfg, req, tsgls);
 			if (err)
 				return err;
+			cond_resched();
 		}
 	}
 #endif
@@ -2535,6 +2540,7 @@ static int test_skcipher(const char *driver, int enc,
 					tsgls);
 		if (err)
 			return err;
+		cond_resched();
 	}
 	return 0;
 }
Signature · 1 line
2.20.1
author maintainer
#106 lore raw

PATCH 99/249 x86/cpu: Add Ice Lake NNPI to Intel family

In reply to Sasha Levin

From: Rajneesh Bhardwaj <[email protected]>

[ Upstream commit e32d045cd4ba06b59878323e434bad010e78e658 ]

Add the CPUID model number of Ice Lake Neural Network Processor for Deep Learning Inference (ICL-NNPI) to the Intel family list. Ice Lake NNPI uses model number 0x9D and this will be documented in a future version of Intel Software Development Manual.

---
 arch/x86/include/asm/intel-family.h | 1 +
 1 file changed, 1 insertion(+)
arch/x86/include/asm/intel-family.h +1 -0
index 310118805f57..f60ddd655c78 100644
@@ -56,6 +56,7 @@
 #define INTEL_FAM6_ICELAKE_XEON_D	0x6C
 #define INTEL_FAM6_ICELAKE_DESKTOP	0x7D
 #define INTEL_FAM6_ICELAKE_MOBILE	0x7E
+#define INTEL_FAM6_ICELAKE_NNPI		0x9D
 
 /* "Small Core" Processors (Atom) */
 
Signature · 1 line
2.20.1
author maintainer
#107 lore raw

PATCH 100/249 ASoC: meson: axg-tdm: fix sample clock inversion

In reply to Sasha Levin

From: Jerome Brunet <[email protected]>

[ Upstream commit cb36ff785e868992e96e8b9e5a0c2822b680a9e2 ]

The content of SND_SOC_DAIFMT_FORMAT_MASK is a number, not a bitfield, so the test to check if the format is i2s is wrong. Because of this the clock setting may be wrong. For example, the sample clock gets inverted in DSP B mode, when it should not.

Fix the lrclk invert helper function

Fixes 1a11d88f499c ("ASoC: meson: add tdm formatter base driver")
---
 sound/soc/meson/axg-tdm.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
sound/soc/meson/axg-tdm.h +1 -1
index e578b6f40a07..5774ce0916d4 100644
@@ -40,7 +40,7 @@ struct axg_tdm_iface {
 
 static inline bool axg_tdm_lrclk_invert(unsigned int fmt)
 {
-	return (fmt & SND_SOC_DAIFMT_I2S) ^
+	return ((fmt & SND_SOC_DAIFMT_FORMAT_MASK) == SND_SOC_DAIFMT_I2S) ^
 		!!(fmt & (SND_SOC_DAIFMT_IB_IF | SND_SOC_DAIFMT_NB_IF));
 }
 
Signature · 1 line
2.20.1
author maintainer
#108 lore raw

PATCH 101/249 rcu: Force inlining of rcu_read_lock()

In reply to Sasha Levin

From: Waiman Long <[email protected]>

[ Upstream commit 6da9f775175e516fc7229ceaa9b54f8f56aa7924 ]

When debugging options are turned on, the rcu_read_lock() function might not be inlined. This results in lockdep's print_lock() function printing "rcu_read_lock+0x0/0x70" instead of rcu_read_lock()'s caller. For example:

[   10.579995] =============================
[   10.584033] WARNING: suspicious RCU usage
[   10.588074] 4.18.0.memcg_v2+ #1 Not tainted
[   10.593162] -----------------------------
[   10.597203] include/linux/rcupdate.h:281 Illegal context switch in
RCU read-side critical section!
[   10.606220]
[   10.606220] other info that might help us debug this:
[   10.606220]
[   10.614280]
[   10.614280] rcu_scheduler_active = 2, debug_locks = 1
[   10.620853] 3 locks held by systemd/1:
[   10.624632]  #0: (____ptrval____) (&type->i_mutex_dir_key#5){.+.+}, at: lookup_slow+0x42/0x70
[   10.633232]  #1: (____ptrval____) (rcu_read_lock){....}, at: rcu_read_lock+0x0/0x70
[   10.640954]  #2: (____ptrval____) (rcu_read_lock){....}, at: rcu_read_lock+0x0/0x70

These "rcu_read_lock+0x0/0x70" strings are not providing any useful information. This commit therefore forces inlining of the rcu_read_lock() function so that rcu_read_lock()'s caller is instead shown.

---
 include/linux/rcupdate.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
include/linux/rcupdate.h +1 -1
index b25d20822e75..3508f4508a11 100644
@@ -586,7 +586,7 @@ static inline void rcu_preempt_sleep_check(void) { }
  * read-side critical sections may be preempted and they may also block, but
  * only when acquiring spinlocks that are subject to priority inheritance.
  */
-static inline void rcu_read_lock(void)
+static __always_inline void rcu_read_lock(void)
 {
 	__rcu_read_lock();
 	__acquire(RCU);
Signature · 1 line
2.20.1
author maintainer
#109 lore raw

PATCH 102/249 perf/x86: Add Intel Ice Lake NNPI uncore support

In reply to Sasha Levin

From: Rajneesh Bhardwaj <[email protected]>

[ Upstream commit 5f4318c1b1d23a9290e4def78ee76017c288bf60 ]

Intel Ice Lake uncore support already included IMC PCI ID but ICL-NNPI CPUID is missing so add it to fix the probe function.

Fixes e39875d15ad6 ("perf/x86: add Intel Icelake uncore support")
Cc Dave Hansen <[email protected]>
Cc Andy Shevchenko <[email protected]>
Cc "H. Peter Anvin" <[email protected]>
Cc Kan Liang <[email protected]>
Cc Qiuxu Zhuo <[email protected]>
Cc Srinivas Pandruvada <[email protected]>
Cc Len Brown <[email protected]>
Cc Linux PM <[email protected]>
---
 arch/x86/events/intel/uncore.c | 1 +
 1 file changed, 1 insertion(+)
arch/x86/events/intel/uncore.c +1 -0
index 9e3fbd47cb56..089bfcdf2f7f 100644
@@ -1400,6 +1400,7 @@ static const struct x86_cpu_id intel_uncore_match[] __initconst = {
 	X86_UNCORE_MODEL_MATCH(INTEL_FAM6_KABYLAKE_MOBILE, skl_uncore_init),
 	X86_UNCORE_MODEL_MATCH(INTEL_FAM6_KABYLAKE_DESKTOP, skl_uncore_init),
 	X86_UNCORE_MODEL_MATCH(INTEL_FAM6_ICELAKE_MOBILE, icl_uncore_init),
+	X86_UNCORE_MODEL_MATCH(INTEL_FAM6_ICELAKE_NNPI, icl_uncore_init),
 	{},
 };
 
Signature · 1 line
2.20.1
author maintainer
#110 lore raw

PATCH 103/249 x86/cpufeatures: Add FDP_EXCPTN_ONLY and ZERO_FCS_FDS

In reply to Sasha Levin

From: Aaron Lewis <[email protected]>

[ Upstream commit cbb99c0f588737ec98c333558922ce47e9a95827 ]

Add the CPUID enumeration for Intel's de-feature bits to accommodate passing these de-features through to kvm guests.

These de-features are (from SDM vol 1, section 8.1.8):
- X86_FEATURE_FDP_EXCPTN_ONLY: If CPUID.(EAX=07H,ECX=0H):EBX[bit 6] = 1, the data pointer (FDP) is updated only for the x87 non-control instructions that incur unmasked x87 exceptions.
- X86_FEATURE_ZERO_FCS_FDS: If CPUID.(EAX=07H,ECX=0H):EBX[bit 13] = 1, the processor deprecates FCS and FDS; it saves each as 0000H.

---
 arch/x86/include/asm/cpufeatures.h | 2 ++
 1 file changed, 2 insertions(+)
arch/x86/include/asm/cpufeatures.h +2 -0
index 75f27ee2c263..1017b9c7dfe0 100644
@@ -239,12 +239,14 @@
 #define X86_FEATURE_BMI1		( 9*32+ 3) /* 1st group bit manipulation extensions */
 #define X86_FEATURE_HLE			( 9*32+ 4) /* Hardware Lock Elision */
 #define X86_FEATURE_AVX2		( 9*32+ 5) /* AVX2 instructions */
+#define X86_FEATURE_FDP_EXCPTN_ONLY	( 9*32+ 6) /* "" FPU data pointer updated only on x87 exceptions */
 #define X86_FEATURE_SMEP		( 9*32+ 7) /* Supervisor Mode Execution Protection */
 #define X86_FEATURE_BMI2		( 9*32+ 8) /* 2nd group bit manipulation extensions */
 #define X86_FEATURE_ERMS		( 9*32+ 9) /* Enhanced REP MOVSB/STOSB instructions */
 #define X86_FEATURE_INVPCID		( 9*32+10) /* Invalidate Processor Context ID */
 #define X86_FEATURE_RTM			( 9*32+11) /* Restricted Transactional Memory */
 #define X86_FEATURE_CQM			( 9*32+12) /* Cache QoS Monitoring */
+#define X86_FEATURE_ZERO_FCS_FDS	( 9*32+13) /* "" Zero out FPU CS and FPU DS */
 #define X86_FEATURE_MPX			( 9*32+14) /* Memory Protection Extension */
 #define X86_FEATURE_RDT_A		( 9*32+15) /* Resource Director Technology Allocation */
 #define X86_FEATURE_AVX512F		( 9*32+16) /* AVX-512 Foundation */
Signature · 1 line
2.20.1
author maintainer
#111 lore raw

PATCH 104/249 qed: iWARP - Fix tc for MPA ll2 connection

In reply to Sasha Levin

From: Michal Kalderon <[email protected]>

[ Upstream commit cb94d52b93c74fe1f2595734fabeda9f8ae891ee ]

The driver needs to assign a lossless traffic class for the MPA ll2 connection to ensure no packets are dropped when returning from the driver as they will never be re-transmitted by the peer.

Fixes ae3488ff37dc ("qed: Add ll2 connection for processing unaligned MPA packets")
---
 drivers/net/ethernet/qlogic/qed/qed_iwarp.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/net/ethernet/qlogic/qed/qed_iwarp.c +2 -0
index ded556b7bab5..eeea8683d99b 100644
@@ -2708,6 +2708,8 @@ qed_iwarp_ll2_start(struct qed_hwfn *p_hwfn,
 	data.input.rx_num_desc = n_ooo_bufs * 2;
 	data.input.tx_num_desc = data.input.rx_num_desc;
 	data.input.tx_max_bds_per_packet = QED_IWARP_MAX_BDS_PER_FPDU;
+	data.input.tx_tc = PKT_LB_TC;
+	data.input.tx_dest = QED_LL2_TX_DEST_LB;
 	data.p_connection_handle = &iwarp_info->ll2_mpa_handle;
 	data.input.secondary_queue = true;
 	data.cbs = &cbs;
Signature · 1 line
2.20.1
author maintainer
#112 lore raw

PATCH 105/249 net: hns3: fix for dereferencing before null checking

In reply to Sasha Levin

From: Yunsheng Lin <[email protected]>

[ Upstream commit 757188005f905664b0186b88cf26a7e844190a63 ]

The netdev is dereferenced before null checking in the function hns3_setup_tc.

This patch moves the dereferencing after the null checking.

Fixes 76ad4f0ee747 ("net: hns3: Add support of HNS3 Ethernet Driver for hip08 SoC")
---
 drivers/net/ethernet/hisilicon/hns3/hns3_enet.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
drivers/net/ethernet/hisilicon/hns3/hns3_enet.c +5 -2
index 5611b990ac34..d18ad7b48a31 100644
@@ -1514,12 +1514,12 @@ static void hns3_nic_get_stats64(struct net_device *netdev,
 static int hns3_setup_tc(struct net_device *netdev, void *type_data)
 {
 	struct tc_mqprio_qopt_offload *mqprio_qopt = type_data;
-	struct hnae3_handle *h = hns3_get_handle(netdev);
-	struct hnae3_knic_private_info *kinfo = &h->kinfo;
 	u8 *prio_tc = mqprio_qopt->qopt.prio_tc_map;
+	struct hnae3_knic_private_info *kinfo;
 	u8 tc = mqprio_qopt->qopt.num_tc;
 	u16 mode = mqprio_qopt->mode;
 	u8 hw = mqprio_qopt->qopt.hw;
+	struct hnae3_handle *h;
 
 	if (!((hw == TC_MQPRIO_HW_OFFLOAD_TCS &&
 	       mode == TC_MQPRIO_MODE_CHANNEL) || (!hw && tc == 0)))
@@ -1531,6 +1531,9 @@ static int hns3_setup_tc(struct net_device *netdev, void *type_data)
 	if (!netdev)
 		return -EINVAL;
 
+	h = hns3_get_handle(netdev);
+	kinfo = &h->kinfo;
+
 	return (kinfo->dcb_ops && kinfo->dcb_ops->setup_tc) ?
 		kinfo->dcb_ops->setup_tc(h, tc, prio_tc) : -EOPNOTSUPP;
 }
Signature · 1 line
2.20.1
author maintainer
#113 lore raw

PATCH 106/249 net: hns3: fix for skb leak when doing selftest

In reply to Sasha Levin

From: Yunsheng Lin <[email protected]>

[ Upstream commit 8f9eed1a8791b83eb1c54c261d68424717e4111e ]

If hns3_nic_net_xmit does not return NETDEV_TX_BUSY when doing a loopback selftest, the skb is not freed in hns3_clean_tx_ring or hns3_nic_net_xmit, which causes skb not freed problem.

This patch fixes it by freeing skb when hns3_nic_net_xmit does not return NETDEV_TX_OK.

Fixes c39c4d98dc65 ("net: hns3: Add mac loopback selftest support in hns3 driver")
---
 drivers/net/ethernet/hisilicon/hns3/hns3_ethtool.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/net/ethernet/hisilicon/hns3/hns3_ethtool.c +4 -2
index d1588ea6132c..24fce343e7fc 100644
@@ -243,11 +243,13 @@ static int hns3_lp_run_test(struct net_device *ndev, enum hnae3_loop mode)
 
 		skb_get(skb);
 		tx_ret = hns3_nic_net_xmit(skb, ndev);
-		if (tx_ret == NETDEV_TX_OK)
+		if (tx_ret == NETDEV_TX_OK) {
 			good_cnt++;
-		else
+		} else {
+			kfree_skb(skb);
 			netdev_err(ndev, "hns3_lb_run_test xmit failed: %d\n",
 				   tx_ret);
+		}
 	}
 	if (good_cnt != HNS3_NIC_LB_TEST_PKT_NUM) {
 		ret_val = HNS3_NIC_LB_TEST_TX_CNT_ERR;
Signature · 1 line
2.20.1
author maintainer
#114 lore raw

PATCH 107/249 net: hns3: delay ring buffer clearing during reset

In reply to Sasha Levin

From: Yunsheng Lin <[email protected]>

[ Upstream commit 3a30964a2eef6aabd3ab18b979ea0eacf1147731 ]

The driver may not be able to disable the ring through firmware when downing the netdev during reset process, which may cause hardware accessing freed buffer problem.

This patch delays the ring buffer clearing to reset uninit process because hardware will not access the ring buffer after hardware reset is completed.

Fixes bb6b94a896d4 ("net: hns3: Add reset interface implementation in client")
---
 .../net/ethernet/hisilicon/hns3/hns3_enet.c   | 19 ++++++++++++++-----
 1 file changed, 14 insertions(+), 5 deletions(-)
drivers/net/ethernet/hisilicon/hns3/hns3_enet.c +14 -5
index d18ad7b48a31..e0d3e2f9801d 100644
@@ -28,7 +28,7 @@
 #define hns3_tx_bd_count(S)	DIV_ROUND_UP(S, HNS3_MAX_BD_SIZE)
 
 static void hns3_clear_all_ring(struct hnae3_handle *h);
-static void hns3_force_clear_all_rx_ring(struct hnae3_handle *h);
+static void hns3_force_clear_all_ring(struct hnae3_handle *h);
 static void hns3_remove_hw_addr(struct net_device *netdev);
 
 static const char hns3_driver_name[] = "hns3";
@@ -491,7 +491,12 @@ static void hns3_nic_net_down(struct net_device *netdev)
 	/* free irq resources */
 	hns3_nic_uninit_irq(priv);
 
-	hns3_clear_all_ring(priv->ae_handle);
+	/* delay ring buffer clearing to hns3_reset_notify_uninit_enet
+	 * during reset process, because driver may not be able
+	 * to disable the ring through firmware when downing the netdev.
+	 */
+	if (!hns3_nic_resetting(netdev))
+		hns3_clear_all_ring(priv->ae_handle);
 }
 
 static int hns3_nic_net_stop(struct net_device *netdev)
@@ -3883,7 +3888,7 @@ static void hns3_client_uninit(struct hnae3_handle *handle, bool reset)
 
 	hns3_del_all_fd_rules(netdev, true);
 
-	hns3_force_clear_all_rx_ring(handle);
+	hns3_force_clear_all_ring(handle);
 
 	hns3_uninit_phy(netdev);
 
@@ -4055,7 +4060,7 @@ static void hns3_force_clear_rx_ring(struct hns3_enet_ring *ring)
 	}
 }
 
-static void hns3_force_clear_all_rx_ring(struct hnae3_handle *h)
+static void hns3_force_clear_all_ring(struct hnae3_handle *h)
 {
 	struct net_device *ndev = h->kinfo.netdev;
 	struct hns3_nic_priv *priv = netdev_priv(ndev);
@@ -4063,6 +4068,9 @@ static void hns3_force_clear_all_rx_ring(struct hnae3_handle *h)
 	u32 i;
 
 	for (i = 0; i < h->kinfo.num_tqps; i++) {
+		ring = priv->ring_data[i].ring;
+		hns3_clear_tx_ring(ring);
+
 		ring = priv->ring_data[i + h->kinfo.num_tqps].ring;
 		hns3_force_clear_rx_ring(ring);
 	}
@@ -4297,7 +4305,8 @@ static int hns3_reset_notify_uninit_enet(struct hnae3_handle *handle)
 		return 0;
 	}
 
-	hns3_force_clear_all_rx_ring(handle);
+	hns3_clear_all_ring(handle);
+	hns3_force_clear_all_ring(handle);
 
 	hns3_nic_uninit_vector_data(priv);
 
Signature · 1 line
2.20.1
author maintainer
#115 lore raw

PATCH 108/249 block: null_blk: fix race condition for null_del_dev

In reply to Sasha Levin

From: Bob Liu <[email protected]>

[ Upstream commit 7602843fd873cae43a444b83b14dfdd114a9659c ]

Dulicate call of null_del_dev() will trigger null pointer error like below. The reason is a race condition between nullb_device_power_store() and nullb_group_drop_item().

CPU#0 CPU#1
---------------- -----------------
do_rmdir()

Quoted 2 lines

configfs_rmdir()
client_drop_item()

     >nullb_group_drop_item()
                                nullb_device_power_store()
				>null_del_dev()

      >test_and_clear_bit(NULLB_DEV_FL_UP
       >null_del_dev()
       ^^^^^
       Duplicated null_dev_dev() triger null pointer error

				>clear_bit(NULLB_DEV_FL_UP

The fix could be keep the sequnce of clear NULLB_DEV_FL_UP and null_del_dev().

[  698.613600] BUG: unable to handle kernel NULL pointer dereference at 0000000000000018
[  698.613608] #PF error: [normal kernel read fault]
[  698.613611] PGD 0 P4D 0
[  698.613619] Oops: 0000 [#1] SMP PTI
[  698.613627] CPU: 3 PID: 6382 Comm: rmdir Not tainted 5.0.0+ #35
[  698.613631] Hardware name: LENOVO 20LJS2EV08/20LJS2EV08, BIOS R0SET33W (1.17 ) 07/18/2018
[  698.613644] RIP: 0010:null_del_dev+0xc/0x110 [null_blk]
[  698.613649] Code: 00 00 00 5b 41 5c 41 5d 41 5e 41 5f 5d c3 0f 0b eb 97 e8 47 bb 2a e8 0f 1f 80 00 00 00 00 0f 1f 44 00 00 55 48 89 e5 41 54 53 <8b> 77 18 48 89 fb 4c 8b 27 48 c7 c7 40 57 1e c1 e8 bf c7 cb e8 48
[  698.613654] RSP: 0018:ffffb887888bfde0 EFLAGS: 00010286
[  698.613659] RAX: 0000000000000000 RBX: ffff9d436d92bc00 RCX: ffff9d43a9184681
[  698.613663] RDX: ffffffffc11e5c30 RSI: 0000000068be6540 RDI: 0000000000000000
[  698.613667] RBP: ffffb887888bfdf0 R08: 0000000000000001 R09: 0000000000000000
[  698.613671] R10: ffffb887888bfdd8 R11: 0000000000000f16 R12: ffff9d436d92bc08
[  698.613675] R13: ffff9d436d94e630 R14: ffffffffc11e5088 R15: ffffffffc11e5000
[  698.613680] FS:  00007faa68be6540(0000) GS:ffff9d43d14c0000(0000) knlGS:0000000000000000
[  698.613685] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  698.613689] CR2: 0000000000000018 CR3: 000000042f70c002 CR4: 00000000003606e0
[  698.613693] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[  698.613697] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[  698.613700] Call Trace:
[  698.613712]  nullb_group_drop_item+0x50/0x70 [null_blk]
[  698.613722]  client_drop_item+0x29/0x40
[  698.613728]  configfs_rmdir+0x1ed/0x300
[  698.613738]  vfs_rmdir+0xb2/0x130
[  698.613743]  do_rmdir+0x1c7/0x1e0
[  698.613750]  __x64_sys_rmdir+0x17/0x20
[  698.613759]  do_syscall_64+0x5a/0x110
[  698.613768]  entry_SYSCALL_64_after_hwframe+0x44/0xa9
---
 drivers/block/null_blk_main.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)
drivers/block/null_blk_main.c +6 -5
index 447d635c79a2..2a4f8bc4f930 100644
@@ -327,11 +327,12 @@ static ssize_t nullb_device_power_store(struct config_item *item,
 		set_bit(NULLB_DEV_FL_CONFIGURED, &dev->flags);
 		dev->power = newp;
 	} else if (dev->power && !newp) {
-		mutex_lock(&lock);
-		dev->power = newp;
-		null_del_dev(dev->nullb);
-		mutex_unlock(&lock);
-		clear_bit(NULLB_DEV_FL_UP, &dev->flags);
+		if (test_and_clear_bit(NULLB_DEV_FL_UP, &dev->flags)) {
+			mutex_lock(&lock);
+			dev->power = newp;
+			null_del_dev(dev->nullb);
+			mutex_unlock(&lock);
+		}
 		clear_bit(NULLB_DEV_FL_CONFIGURED, &dev->flags);
 	}
 
Signature · 1 line
2.20.1
author maintainer
#116 lore raw

PATCH 109/249 blkcg, writeback: dead memcgs shouldn't contribute to writeback ownership arbitration

In reply to Sasha Levin

From: Tejun Heo <[email protected]>

[ Upstream commit 6631142229005e1b1c311a09efe9fb3cfdac8559 ]

wbc_account_io() collects information on cgroup ownership of writeback pages to determine which cgroup should own the inode. Pages can stay associated with dead memcgs but we want to avoid attributing IOs to dead blkcgs as much as possible as the association is likely to be stale. However, currently, pages associated with dead memcgs contribute to the accounting delaying and/or confusing the arbitration.

Fix it by ignoring pages associated with dead memcgs.

---
 fs/fs-writeback.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
fs/fs-writeback.c +7 -1
index e41cbe8e81b9..9ebfb1b28430 100644
@@ -715,6 +715,7 @@ void wbc_detach_inode(struct writeback_control *wbc)
 void wbc_account_io(struct writeback_control *wbc, struct page *page,
 		    size_t bytes)
 {
+	struct cgroup_subsys_state *css;
 	int id;
 
 	/*
@@ -726,7 +727,12 @@ void wbc_account_io(struct writeback_control *wbc, struct page *page,
 	if (!wbc->wb)
 		return;
 
-	id = mem_cgroup_css_from_page(page)->id;
+	css = mem_cgroup_css_from_page(page);
+	/* dead cgroups shouldn't contribute to inode ownership arbitration */
+	if (!(css->flags & CSS_ONLINE))
+		return;
+
+	id = css->id;
 
 	if (id == wbc->wb_id) {
 		wbc->wb_bytes += bytes;
Signature · 1 line
2.20.1
author maintainer
#117 lore raw

PATCH 110/249 xfrm: fix sa selector validation

In reply to Sasha Levin

From: Nicolas Dichtel <[email protected]>

[ Upstream commit b8d6d0079757cbd1b69724cfd1c08e2171c68cee ]

After commit b38ff4075a80, the following command does not work anymore:

$ ip xfrm state add src 10.125.0.2 dst 10.125.0.1 proto esp spi 34 reqid 1 \
  mode tunnel enc 'cbc(aes)' 0xb0abdba8b782ad9d364ec81e3a7d82a1 auth-trunc \
  'hmac(sha1)' 0xe26609ebd00acb6a4d51fca13e49ea78a72c73e6 96 flag align4

In fact, the selector is not mandatory, allow the user to provide an empty selector.

Fixes b38ff4075a80 ("xfrm: Fix xfrm sel prefix length validation")
CC Anirudh Gupta <[email protected]>
---
 net/xfrm/xfrm_user.c | 3 +++
 1 file changed, 3 insertions(+)
net/xfrm/xfrm_user.c +3 -0
index 76ad7e201626..b88ba45ff1ac 100644
@@ -167,6 +167,9 @@ static int verify_newsa_info(struct xfrm_usersa_info *p,
 	}
 
 	switch (p->sel.family) {
+	case AF_UNSPEC:
+		break;
+
 	case AF_INET:
 		if (p->sel.prefixlen_d > 32 || p->sel.prefixlen_s > 32)
 			goto out;
Signature · 1 line
2.20.1
author maintainer
#118 lore raw

PATCH 111/249 sched/core: Add __sched tag for io_schedule()

In reply to Sasha Levin

From: Gao Xiang <[email protected]>

[ Upstream commit e3b929b0a184edb35531153c5afcaebb09014f9d ]

Non-inline io_schedule() was introduced in:

commit 10ab56434f2f ("sched/core: Separate out io_schedule_prepare() and io_schedule_finish()")

Keep in line with io_schedule_timeout(), otherwise "/proc/<pid>/wchan" will report io_schedule() rather than its callers when waiting for IO.

Cc Andrew Morton <[email protected]>
Cc Linus Torvalds <[email protected]>
Cc Miao Xie <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Thomas Gleixner <[email protected]>
Fixes 10ab56434f2f ("sched/core: Separate out io_schedule_prepare() and io_schedule_finish()")
---
 kernel/sched/core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/sched/core.c +1 -1
index 874c427742a9..4d5962232a55 100644
@@ -5123,7 +5123,7 @@ long __sched io_schedule_timeout(long timeout)
 }
 EXPORT_SYMBOL(io_schedule_timeout);
 
-void io_schedule(void)
+void __sched io_schedule(void)
 {
 	int token;
 
Signature · 1 line
2.20.1
author maintainer
#119 lore raw

PATCH 112/249 perf/x86/intel: Add more Icelake CPUIDs

In reply to Sasha Levin

From: Kan Liang <[email protected]>

[ Upstream commit faaeff98666c24376cebd0b106504d05a36881d1 ]

Add new model number for Icelake desktop and server to perf.

The data source encoding for Icelake server is the same as Skylake server.

---
 arch/x86/events/intel/core.c | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)
arch/x86/events/intel/core.c +9 -4
index a5436cee20b1..b6cae65aa7ef 100644
@@ -4439,6 +4439,7 @@ __init int intel_pmu_init(void)
 	struct event_constraint *c;
 	unsigned int unused;
 	struct extra_reg *er;
+	bool pmem = false;
 	int version, i;
 	char *name;
 
@@ -4890,9 +4891,10 @@ __init int intel_pmu_init(void)
 		name = "knights-landing";
 		break;
 
+	case INTEL_FAM6_SKYLAKE_X:
+		pmem = true;
 	case INTEL_FAM6_SKYLAKE_MOBILE:
 	case INTEL_FAM6_SKYLAKE_DESKTOP:
-	case INTEL_FAM6_SKYLAKE_X:
 	case INTEL_FAM6_KABYLAKE_MOBILE:
 	case INTEL_FAM6_KABYLAKE_DESKTOP:
 		x86_add_quirk(intel_pebs_isolation_quirk);
@@ -4925,8 +4927,7 @@ __init int intel_pmu_init(void)
 		x86_pmu.cpu_events = hsw_events_attrs;
 		mem_attr = hsw_mem_events_attrs;
 		tsx_attr = hsw_tsx_events_attrs;
-		intel_pmu_pebs_data_source_skl(
-			boot_cpu_data.x86_model == INTEL_FAM6_SKYLAKE_X);
+		intel_pmu_pebs_data_source_skl(pmem);
 
 		if (boot_cpu_has(X86_FEATURE_TSX_FORCE_ABORT)) {
 			x86_pmu.flags |= PMU_FL_TFA;
@@ -4940,7 +4941,11 @@ __init int intel_pmu_init(void)
 		name = "skylake";
 		break;
 
+	case INTEL_FAM6_ICELAKE_X:
+	case INTEL_FAM6_ICELAKE_XEON_D:
+		pmem = true;
 	case INTEL_FAM6_ICELAKE_MOBILE:
+	case INTEL_FAM6_ICELAKE_DESKTOP:
 		x86_pmu.late_ack = true;
 		memcpy(hw_cache_event_ids, skl_hw_cache_event_ids, sizeof(hw_cache_event_ids));
 		memcpy(hw_cache_extra_regs, skl_hw_cache_extra_regs, sizeof(hw_cache_extra_regs));
@@ -4963,7 +4968,7 @@ __init int intel_pmu_init(void)
 		x86_pmu.cpu_events = get_icl_events_attrs();
 		x86_pmu.rtm_abort_event = X86_CONFIG(.event=0xca, .umask=0x02);
 		x86_pmu.lbr_pt_coexist = true;
-		intel_pmu_pebs_data_source_skl(false);
+		intel_pmu_pebs_data_source_skl(pmem);
 		pr_cont("Icelake events, ");
 		name = "icelake";
 		break;
Signature · 1 line
2.20.1
author maintainer
#120 lore raw

PATCH 113/249 sched/fair: Fix "runnable_avg_yN_inv" not used warnings

In reply to Sasha Levin

From: Qian Cai <[email protected]>

[ Upstream commit 509466b7d480bc5d22e90b9fbe6122ae0e2fbe39 ]

runnable_avg_yN_inv[] is only used in kernel/sched/pelt.c but was included in several other places because they need other macros all came from kernel/sched/sched-pelt.h which was generated by Documentation/scheduler/sched-pelt. As the result, it causes compilation a lot of warnings,

kernel/sched/sched-pelt.h:4:18: warning: 'runnable_avg_yN_inv' defined but not used [-Wunused-const-variable=] kernel/sched/sched-pelt.h:4:18: warning: 'runnable_avg_yN_inv' defined but not used [-Wunused-const-variable=] kernel/sched/sched-pelt.h:4:18: warning: 'runnable_avg_yN_inv' defined but not used [-Wunused-const-variable=]
...

Silence it by appending the __maybe_unused attribute for it, so all generated variables and macros can still be kept in the same file.

---
 Documentation/scheduler/sched-pelt.c | 3 ++-
 kernel/sched/sched-pelt.h            | 2 +-
 2 files changed, 3 insertions(+), 2 deletions(-)
Documentation/scheduler/sched-pelt.c +2 -1
index e4219139386a..7238b355919c 100644
@@ -20,7 +20,8 @@ void calc_runnable_avg_yN_inv(void)
 	int i;
 	unsigned int x;
 
-	printf("static const u32 runnable_avg_yN_inv[] = {");
+	/* To silence -Wunused-but-set-variable warnings. */
+	printf("static const u32 runnable_avg_yN_inv[] __maybe_unused = {");
 	for (i = 0; i < HALFLIFE; i++) {
 		x = ((1UL<<32)-1)*pow(y, i);
 
kernel/sched/sched-pelt.h +1 -1
index a26473674fb7..c529706bed11 100644
@@ -1,7 +1,7 @@
 /* SPDX-License-Identifier: GPL-2.0 */
 /* Generated by Documentation/scheduler/sched-pelt; do not modify. */
 
-static const u32 runnable_avg_yN_inv[] = {
+static const u32 runnable_avg_yN_inv[] __maybe_unused = {
 	0xffffffff, 0xfa83b2da, 0xf5257d14, 0xefe4b99a, 0xeac0c6e6, 0xe5b906e6,
 	0xe0ccdeeb, 0xdbfbb796, 0xd744fcc9, 0xd2a81d91, 0xce248c14, 0xc9b9bd85,
 	0xc5672a10, 0xc12c4cc9, 0xbd08a39e, 0xb8fbaf46, 0xb504f333, 0xb123f581,
Signature · 1 line
2.20.1
author maintainer
#121 lore raw

PATCH 114/249 perf/x86/intel: Disable check_msr for real HW

In reply to Sasha Levin

From: Jiri Olsa <[email protected]>

[ Upstream commit d0e1a507bdc761a14906f03399d933ea639a1756 ]

Tom Vaden reported false failure of the check_msr() function, because some servers can do POST tracing and enable LBR tracing during bootup.

Kan confirmed that check_msr patch was to fix a bug report in guest, so it's ok to disable it for real HW.

---
 arch/x86/events/intel/core.c | 8 ++++++++
 1 file changed, 8 insertions(+)
arch/x86/events/intel/core.c +8 -0
index b6cae65aa7ef..f0c14665893b 100644
@@ -20,6 +20,7 @@
 #include <asm/intel-family.h>
 #include <asm/apic.h>
 #include <asm/cpu_device_id.h>
+#include <asm/hypervisor.h>
 
 #include "../perf_event.h"
 
@@ -4054,6 +4055,13 @@ static bool check_msr(unsigned long msr, u64 mask)
 {
 	u64 val_old, val_new, val_tmp;
 
+	/*
+	 * Disable the check for real HW, so we don't
+	 * mess with potentionaly enabled registers:
+	 */
+	if (hypervisor_is_type(X86_HYPER_NATIVE))
+		return true;
+
 	/*
 	 * Read the current value, change it and read it back to see if it
 	 * matches, this is needed to detect certain hardware emulators
Signature · 1 line
2.20.1
author maintainer
#122 lore raw

PATCH 115/249 perf/x86/intel/uncore: Handle invalid event coding for free-running counter

In reply to Sasha Levin

From: Kan Liang <[email protected]>

[ Upstream commit 543ac280b3576c0009e8c0fcd4d6bfc9978d7bd0 ]

Counting with invalid event coding for free-running counter may cause OOPs, e.g. uncore_iio_free_running_0/event=1/.

Current code only validate the event with free-running event format, event=0xff,umask=0xXY. Non-free-running event format never be checked for the PMU with free-running counters.

Add generic hw_config() to check and reject the invalid event coding for free-running PMU.

Cc Linus Torvalds <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Thomas Gleixner <[email protected]>
Fixes 0f519f0352e3 ("perf/x86/intel/uncore: Support IIO free-running counters on SKX")
---
 arch/x86/events/intel/uncore.h       | 10 ++++++++++
 arch/x86/events/intel/uncore_snbep.c |  1 +
 2 files changed, 11 insertions(+)
arch/x86/events/intel/uncore.h +10 -0
index 79eb2e21e4f0..28499e39679f 100644
@@ -419,6 +419,16 @@ static inline bool is_freerunning_event(struct perf_event *event)
 	       (((cfg >> 8) & 0xff) >= UNCORE_FREERUNNING_UMASK_START);
 }
 
+/* Check and reject invalid config */
+static inline int uncore_freerunning_hw_config(struct intel_uncore_box *box,
+					       struct perf_event *event)
+{
+	if (is_freerunning_event(event))
+		return 0;
+
+	return -EINVAL;
+}
+
 static inline void uncore_disable_box(struct intel_uncore_box *box)
 {
 	if (box->pmu->type->ops->disable_box)
arch/x86/events/intel/uncore_snbep.c +1 -0
index b10e04387f38..8e4e8e423839 100644
@@ -3585,6 +3585,7 @@ static struct uncore_event_desc skx_uncore_iio_freerunning_events[] = {
 
 static struct intel_uncore_ops skx_uncore_iio_freerunning_ops = {
 	.read_counter		= uncore_msr_read_counter,
+	.hw_config		= uncore_freerunning_hw_config,
 };
 
 static struct attribute *skx_uncore_iio_freerunning_formats_attr[] = {
Signature · 1 line
2.20.1
author maintainer
#123 lore raw

PATCH 116/249 integrity: Fix __integrity_init_keyring() section mismatch

In reply to Sasha Levin

From: Geert Uytterhoeven <[email protected]>

[ Upstream commit 8c655784e2cf59cb6140759b8b546d98261d1ad9 ]

With gcc-4.6.3:

WARNING: vmlinux.o(.text.unlikely+0x24c64): Section mismatch in reference from the function __integrity_init_keyring() to the function .init.text:set_platform_trusted_keys() The function __integrity_init_keyring() references
the function __init set_platform_trusted_keys().
This is often because __integrity_init_keyring lacks a __init
annotation or the annotation of set_platform_trusted_keys is wrong.

Indeed, if the compiler decides not to inline __integrity_init_keyring(), a warning is issued.

Fix this by adding the missing __init annotation.

Fixes 9dc92c45177ab70e ("integrity: Define a trusted platform keyring")
---
 security/integrity/digsig.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
security/integrity/digsig.c +3 -2
index 4582bc26770a..868ade3e8970 100644
@@ -69,8 +69,9 @@ int integrity_digsig_verify(const unsigned int id, const char *sig, int siglen,
 	return -EOPNOTSUPP;
 }
 
-static int __integrity_init_keyring(const unsigned int id, key_perm_t perm,
-				    struct key_restriction *restriction)
+static int __init __integrity_init_keyring(const unsigned int id,
+					   key_perm_t perm,
+					   struct key_restriction *restriction)
 {
 	const struct cred *cred = current_cred();
 	int err = 0;
Signature · 1 line
2.20.1
author maintainer
#124 lore raw

PATCH 117/249 x86/atomic: Fix smp_mb__{before,after}_atomic()

In reply to Sasha Levin

From: Peter Zijlstra <[email protected]>

[ Upstream commit 69d927bba39517d0980462efc051875b7f4db185 ]

Recent probing at the Linux Kernel Memory Model uncovered a 'surprise'. Strongly ordered architectures where the atomic RmW primitive implies full memory ordering and smp_mb__{before,after}_atomic() are a simple barrier() (such as x86) fail for:

	*x = 1;
	atomic_inc(u);
	smp_mb__after_atomic();
	r0 = *y;

Because, while the atomic_inc() implies memory order, it (surprisingly) does not provide a compiler barrier. This then allows the compiler to re-order like so:

	atomic_inc(u);
	*x = 1;
	smp_mb__after_atomic();
	r0 = *y;

Which the CPU is then allowed to re-order (under TSO rules) like:

	atomic_inc(u);
	r0 = *y;
	*x = 1;

And this very much was not intended. Therefore strengthen the atomic RmW ops to include a compiler barrier.

NOTE: atomic_{or,and,xor} and the bitops already had the compiler barrier.

---
 Documentation/atomic_t.txt         | 3 +++
 arch/x86/include/asm/atomic.h      | 8 ++++----
 arch/x86/include/asm/atomic64_64.h | 8 ++++----
 arch/x86/include/asm/barrier.h     | 4 ++--
 4 files changed, 13 insertions(+), 10 deletions(-)
Documentation/atomic_t.txt +3 -0
index dca3fb0554db..65bb09a29324 100644
@@ -194,6 +194,9 @@ These helper barriers exist because architectures have varying implicit
 ordering on their SMP atomic primitives. For example our TSO architectures
 provide full ordered atomics and these barriers are no-ops.
 
+NOTE: when the atomic RmW ops are fully ordered, they should also imply a
+compiler barrier.
+
 Thus:
 
   atomic_fetch_add();
arch/x86/include/asm/atomic.h +4 -4
index ea3d95275b43..115127c7ad28 100644
@@ -54,7 +54,7 @@ static __always_inline void arch_atomic_add(int i, atomic_t *v)
 {
 	asm volatile(LOCK_PREFIX "addl %1,%0"
 		     : "+m" (v->counter)
-		     : "ir" (i));
+		     : "ir" (i) : "memory");
 }
 
 /**
@@ -68,7 +68,7 @@ static __always_inline void arch_atomic_sub(int i, atomic_t *v)
 {
 	asm volatile(LOCK_PREFIX "subl %1,%0"
 		     : "+m" (v->counter)
-		     : "ir" (i));
+		     : "ir" (i) : "memory");
 }
 
 /**
@@ -95,7 +95,7 @@ static __always_inline bool arch_atomic_sub_and_test(int i, atomic_t *v)
 static __always_inline void arch_atomic_inc(atomic_t *v)
 {
 	asm volatile(LOCK_PREFIX "incl %0"
-		     : "+m" (v->counter));
+		     : "+m" (v->counter) :: "memory");
 }
 #define arch_atomic_inc arch_atomic_inc
 
@@ -108,7 +108,7 @@ static __always_inline void arch_atomic_inc(atomic_t *v)
 static __always_inline void arch_atomic_dec(atomic_t *v)
 {
 	asm volatile(LOCK_PREFIX "decl %0"
-		     : "+m" (v->counter));
+		     : "+m" (v->counter) :: "memory");
 }
 #define arch_atomic_dec arch_atomic_dec
 
arch/x86/include/asm/atomic64_64.h +4 -4
index dadc20adba21..5e86c0d68ac1 100644
@@ -45,7 +45,7 @@ static __always_inline void arch_atomic64_add(long i, atomic64_t *v)
 {
 	asm volatile(LOCK_PREFIX "addq %1,%0"
 		     : "=m" (v->counter)
-		     : "er" (i), "m" (v->counter));
+		     : "er" (i), "m" (v->counter) : "memory");
 }
 
 /**
@@ -59,7 +59,7 @@ static inline void arch_atomic64_sub(long i, atomic64_t *v)
 {
 	asm volatile(LOCK_PREFIX "subq %1,%0"
 		     : "=m" (v->counter)
-		     : "er" (i), "m" (v->counter));
+		     : "er" (i), "m" (v->counter) : "memory");
 }
 
 /**
@@ -87,7 +87,7 @@ static __always_inline void arch_atomic64_inc(atomic64_t *v)
 {
 	asm volatile(LOCK_PREFIX "incq %0"
 		     : "=m" (v->counter)
-		     : "m" (v->counter));
+		     : "m" (v->counter) : "memory");
 }
 #define arch_atomic64_inc arch_atomic64_inc
 
@@ -101,7 +101,7 @@ static __always_inline void arch_atomic64_dec(atomic64_t *v)
 {
 	asm volatile(LOCK_PREFIX "decq %0"
 		     : "=m" (v->counter)
-		     : "m" (v->counter));
+		     : "m" (v->counter) : "memory");
 }
 #define arch_atomic64_dec arch_atomic64_dec
 
arch/x86/include/asm/barrier.h +2 -2
index 14de0432d288..84f848c2541a 100644
@@ -80,8 +80,8 @@ do {									\
 })
 
 /* Atomic operations are already serializing on x86 */
-#define __smp_mb__before_atomic()	barrier()
-#define __smp_mb__after_atomic()	barrier()
+#define __smp_mb__before_atomic()	do { } while (0)
+#define __smp_mb__after_atomic()	do { } while (0)
 
 #include <asm-generic/barrier.h>
 
Signature · 1 line
2.20.1
author maintainer
#125 lore raw

PATCH 118/249 perf evsel: Make perf_evsel__name() accept a NULL argument

In reply to Sasha Levin

From: Arnaldo Carvalho de Melo <[email protected]>

[ Upstream commit fdbdd7e8580eac9bdafa532746c865644d125e34 ]

In which case it simply returns "unknown", like when it can't figure out the evsel->name value.

This makes this code more robust and fixes a problem in 'perf trace' where a NULL evsel was being passed to a routine that only used the evsel for printing its name when a invalid syscall id was passed.

---
 tools/perf/util/evsel.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
tools/perf/util/evsel.c +7 -1
index 4a5947625c5c..2c46f9aa416c 100644
@@ -589,6 +589,9 @@ const char *perf_evsel__name(struct perf_evsel *evsel)
 {
 	char bf[128];
 
+	if (!evsel)
+		goto out_unknown;
+
 	if (evsel->name)
 		return evsel->name;
 
@@ -628,7 +631,10 @@ const char *perf_evsel__name(struct perf_evsel *evsel)
 
 	evsel->name = strdup(bf);
 
-	return evsel->name ?: "unknown";
+	if (evsel->name)
+		return evsel->name;
+out_unknown:
+	return "unknown";
 }
 
 const char *perf_evsel__group_name(struct perf_evsel *evsel)
Signature · 1 line
2.20.1
author maintainer
#126 lore raw

PATCH 119/249 vhost_net: disable zerocopy by default

In reply to Sasha Levin

From: Jason Wang <[email protected]>

[ Upstream commit 098eadce3c622c07b328d0a43dda379b38cf7c5e ]

Vhost_net was known to suffer from HOL[1] issues which is not easy to fix. Several downstream disable the feature by default. What's more, the datapath was split and datacopy path got the support of batching and XDP support recently which makes it faster than zerocopy part for small packets transmission.

It looks to me that disable zerocopy by default is more appropriate. It cold be enabled by default again in the future if we fix the above issues.

[1] https://patchwork.kernel.org/patch/3787671/

---
 drivers/vhost/net.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/vhost/net.c +1 -1
index d57ebdd616d9..247e5585af5d 100644
@@ -35,7 +35,7 @@
 
 #include "vhost.h"
 
-static int experimental_zcopytx = 1;
+static int experimental_zcopytx = 0;
 module_param(experimental_zcopytx, int, 0444);
 MODULE_PARM_DESC(experimental_zcopytx, "Enable Zero Copy TX;"
 		                       " 1 -Enable; 0 - Disable");
Signature · 1 line
2.20.1
author maintainer
#127 lore raw

PATCH 119/249 vhost_net: disable zerocopy by default

In reply to Sasha Levin

From: Jason Wang <[email protected]>

[ Upstream commit 098eadce3c622c07b328d0a43dda379b38cf7c5e ]

Vhost_net was known to suffer from HOL[1] issues which is not easy to fix. Several downstream disable the feature by default. What's more, the datapath was split and datacopy path got the support of batching and XDP support recently which makes it faster than zerocopy part for small packets transmission.

It looks to me that disable zerocopy by default is more appropriate. It cold be enabled by default again in the future if we fix the above issues.

[1] https://patchwork.kernel.org/patch/3787671/

---
 drivers/vhost/net.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/vhost/net.c +1 -1
index d57ebdd616d9..247e5585af5d 100644
@@ -35,7 +35,7 @@
 
 #include "vhost.h"
 
-static int experimental_zcopytx = 1;
+static int experimental_zcopytx = 0;
 module_param(experimental_zcopytx, int, 0444);
 MODULE_PARM_DESC(experimental_zcopytx, "Enable Zero Copy TX;"
 		                       " 1 -Enable; 0 - Disable");
Signature · 1 line
2.20.1
author maintainer
#128 lore raw

PATCH 120/249 iavf: allow null RX descriptors

In reply to Sasha Levin

From: Mitch Williams <[email protected]>

[ Upstream commit efa14c3985828da3163f5372137cb64d992b0f79 ]

In some circumstances, the hardware can hand us a null receive descriptor, with no data attached but otherwise valid. Unfortunately, the driver was ill-equipped to handle such an event, and would stop processing packets at that point.

To fix this, use the Descriptor Done bit instead of the size to determine whether or not a descriptor is ready to be processed. Add some checks to allow for unused buffers.

---
 drivers/net/ethernet/intel/iavf/iavf_txrx.c | 21 ++++++++++++++++++---
 1 file changed, 18 insertions(+), 3 deletions(-)
drivers/net/ethernet/intel/iavf/iavf_txrx.c +18 -3
index 06d1509d57f7..c97b9ecf026a 100644
@@ -1236,6 +1236,9 @@ static void iavf_add_rx_frag(struct iavf_ring *rx_ring,
 	unsigned int truesize = SKB_DATA_ALIGN(size + iavf_rx_offset(rx_ring));
 #endif
 
+	if (!size)
+		return;
+
 	skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, rx_buffer->page,
 			rx_buffer->page_offset, size, truesize);
 
@@ -1260,6 +1263,9 @@ static struct iavf_rx_buffer *iavf_get_rx_buffer(struct iavf_ring *rx_ring,
 {
 	struct iavf_rx_buffer *rx_buffer;
 
+	if (!size)
+		return NULL;
+
 	rx_buffer = &rx_ring->rx_bi[rx_ring->next_to_clean];
 	prefetchw(rx_buffer->page);
 
@@ -1299,6 +1305,8 @@ static struct sk_buff *iavf_construct_skb(struct iavf_ring *rx_ring,
 	unsigned int headlen;
 	struct sk_buff *skb;
 
+	if (!rx_buffer)
+		return NULL;
 	/* prefetch first cache line of first page */
 	prefetch(va);
 #if L1_CACHE_BYTES < 128
@@ -1363,6 +1371,8 @@ static struct sk_buff *iavf_build_skb(struct iavf_ring *rx_ring,
 #endif
 	struct sk_buff *skb;
 
+	if (!rx_buffer)
+		return NULL;
 	/* prefetch first cache line of first page */
 	prefetch(va);
 #if L1_CACHE_BYTES < 128
@@ -1398,6 +1408,9 @@ static struct sk_buff *iavf_build_skb(struct iavf_ring *rx_ring,
 static void iavf_put_rx_buffer(struct iavf_ring *rx_ring,
 			       struct iavf_rx_buffer *rx_buffer)
 {
+	if (!rx_buffer)
+		return;
+
 	if (iavf_can_reuse_rx_page(rx_buffer)) {
 		/* hand second half of page back to the ring */
 		iavf_reuse_rx_page(rx_ring, rx_buffer);
@@ -1496,11 +1509,12 @@ static int iavf_clean_rx_irq(struct iavf_ring *rx_ring, int budget)
 		 * verified the descriptor has been written back.
 		 */
 		dma_rmb();
+#define IAVF_RXD_DD BIT(IAVF_RX_DESC_STATUS_DD_SHIFT)
+		if (!iavf_test_staterr(rx_desc, IAVF_RXD_DD))
+			break;
 
 		size = (qword & IAVF_RXD_QW1_LENGTH_PBUF_MASK) >>
 		       IAVF_RXD_QW1_LENGTH_PBUF_SHIFT;
-		if (!size)
-			break;
 
 		iavf_trace(clean_rx_irq, rx_ring, rx_desc, skb);
 		rx_buffer = iavf_get_rx_buffer(rx_ring, size);
@@ -1516,7 +1530,8 @@ static int iavf_clean_rx_irq(struct iavf_ring *rx_ring, int budget)
 		/* exit if we failed to retrieve a buffer */
 		if (!skb) {
 			rx_ring->rx_stats.alloc_buff_failed++;
-			rx_buffer->pagecnt_bias++;
+			if (rx_buffer)
+				rx_buffer->pagecnt_bias++;
 			break;
 		}
 
Signature · 1 line
2.20.1
author maintainer
#129 lore raw

PATCH 121/249 ipoib: correcly show a VF hardware address

In reply to Sasha Levin

From: Denis Kirjanov <[email protected]>

[ Upstream commit 64d701c608fea362881e823b666327f5d28d7ffd ]

in the case of IPoIB with SRIOV enabled hardware
ip link show command incorrecly prints
0 instead of a VF hardware address.

Before:
11: ib1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 2044 qdisc pfifo_fast state UP mode DEFAULT group default qlen 256
link/infiniband
80:00:00:66:fe:80:00:00:00:00:00:00:24:8a:07:03:00:a4:3e:7c brd 00:ff:ff:ff:ff:12:40:1b:ff:ff:00:00:00:00:00:00:ff:ff:ff:ff

    vf 0 MAC 00:00:00:00:00:00, spoof checking off, link-state disable,

trust off, query_rss off
...
After:
11: ib1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 2044 qdisc pfifo_fast state UP mode DEFAULT group default qlen 256
link/infiniband
80:00:00:66:fe:80:00:00:00:00:00:00:24:8a:07:03:00:a4:3e:7c brd 00:ff:ff:ff:ff:12:40:1b:ff:ff:00:00:00:00:00:00:ff:ff:ff:ff
vf 0 link/infiniband
80:00:00:66:fe:80:00:00:00:00:00:00:24:8a:07:03:00:a4:3e:7c brd 00:ff:ff:ff:ff:12:40:1b:ff:ff:00:00:00:00:00:00:ff:ff:ff:ff, spoof checking off, link-state disable, trust off, query_rss off

v1->v2: just copy an address without modifing ifla_vf_mac v2->v3: update the changelog

---
 drivers/infiniband/ulp/ipoib/ipoib_main.c | 1 +
 1 file changed, 1 insertion(+)
drivers/infiniband/ulp/ipoib/ipoib_main.c +1 -0
index 9b5e11d3fb85..04ea7db08e87 100644
@@ -1998,6 +1998,7 @@ static int ipoib_get_vf_config(struct net_device *dev, int vf,
 		return err;
 
 	ivf->vf = vf;
+	memcpy(ivf->mac, dev->dev_addr, dev->addr_len);
 
 	return 0;
 }
Signature · 1 line
2.20.1
author maintainer
#130 lore raw

PATCH 122/249 ASoC: rsnd: fixup mod ID calculation in rsnd_ctu_probe_

In reply to Sasha Levin

From: Nilkanth Ahirrao <[email protected]>

[ Upstream commit ac28ec07ae1c5c1e18ed6855eb105a328418da88 ]

commit c16015f36cc1 ("ASoC: rsnd: add .get_id/.get_id_sub") introduces rsnd_ctu_id which calcualates and gives the main Device id of the CTU by dividing the id by 4. rsnd_mod_id uses this interface to get the CTU main Device id. But this commit forgets to revert the main Device id calcution previously done in rsnd_ctu_probe_ which also divides the id by 4. This path corrects the same to get the correct main Device id.

The issue is observered when rsnd_ctu_probe_ is done for CTU1

Fixes c16015f36cc1 ("ASoC: rsnd: add .get_id/.get_id_sub")
---
 sound/soc/sh/rcar/ctu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
sound/soc/sh/rcar/ctu.c +1 -1
index 8cb06dab234e..7647b3d4c0ba 100644
@@ -108,7 +108,7 @@ static int rsnd_ctu_probe_(struct rsnd_mod *mod,
 			   struct rsnd_dai_stream *io,
 			   struct rsnd_priv *priv)
 {
-	return rsnd_cmd_attach(io, rsnd_mod_id(mod) / 4);
+	return rsnd_cmd_attach(io, rsnd_mod_id(mod));
 }
 
 static void rsnd_ctu_value_init(struct rsnd_dai_stream *io,
Signature · 1 line
2.20.1
author maintainer
#131 lore raw

PATCH 123/249 tools build: Fix the zstd test in the test-all.c common case feature test

In reply to Sasha Levin

From: Arnaldo Carvalho de Melo <[email protected]>

[ Upstream commit 3469fa84c1631face938efc42b3f488a2c2504e0 ]

We were renanimg 'main' to 'main_zstd' but then using 'main_libzstd();' in the main() for test-all.c, causing this:

  $ cat /tmp/build/perf/feature/test-all.make.output
  test-all.c: In function ‘main’:
  test-all.c:236:2: error: implicit declaration of function ‘main_test_libzstd’; did you mean ‘main_test_zstd’? [-Werror=implicit-function-declaration]
    main_test_libzstd();
    ^~~~~~~~~~~~~~~~~
    main_test_zstd
  cc1: all warnings being treated as errors
  $

I.e. what was supposed to be the fast path feature test was _always_ failing, duh, fix it.

Cc Adrian Hunter <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Alexey Budankov <[email protected]>
Fixes 3b1c5d965971 ("tools build: Implement libzstd feature check, LIBZSTD_DIR and NO_LIBZSTD defines")
---
 tools/build/feature/test-all.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/build/feature/test-all.c +1 -1
index a59c53705093..939ac2fcc783 100644
@@ -182,7 +182,7 @@
 # include "test-disassembler-four-args.c"
 #undef main
 
-#define main main_test_zstd
+#define main main_test_libzstd
 # include "test-libzstd.c"
 #undef main
 
Signature · 1 line
2.20.1
author maintainer
#132 lore raw

PATCH 124/249 bpf: fix callees pruning callers

In reply to Sasha Levin

From: Alexei Starovoitov <[email protected]>

[ Upstream commit eea1c227b9e9bad295e8ef984004a9acf12bb68c ]

The commit 7640ead93924 partially resolved the issue of callees incorrectly pruning the callers.
With introduction of bounded loops and jmps_processed heuristic single verifier state may contain multiple branches and calls.
It's possible that new verifier state (for future pruning) will be allocated inside callee. Then callee will exit (still within the same verifier state). It will go back to the caller and there R6-R9 registers will be read and will trigger mark_reg_read. But the reg->live for all frames but the top frame is not set to LIVE_NONE. Hence mark_reg_read will fail to propagate liveness into parent and future walking will incorrectly conclude that the states are equivalent because LIVE_READ is not set. In other words the rule for parent/live should be:
whenever register parentage chain is set the reg->live should be set to LIVE_NONE. is_state_visited logic already follows this rule for spilled registers.

Fixes 7640ead93924 ("bpf: verifier: make sure callees don't prune with caller differences")
Fixes f4d7e40a5b71 ("bpf: introduce function calls (verification)")
---
 kernel/bpf/verifier.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)
kernel/bpf/verifier.c +6 -5
index a5c369e60343..11528bdaa9dc 100644
@@ -6456,17 +6456,18 @@ static int is_state_visited(struct bpf_verifier_env *env, int insn_idx)
 	 * the state of the call instruction (with WRITTEN set), and r0 comes
 	 * from callee with its full parentage chain, anyway.
 	 */
-	for (j = 0; j <= cur->curframe; j++)
-		for (i = j < cur->curframe ? BPF_REG_6 : 0; i < BPF_REG_FP; i++)
-			cur->frame[j]->regs[i].parent = &new->frame[j]->regs[i];
 	/* clear write marks in current state: the writes we did are not writes
 	 * our child did, so they don't screen off its reads from us.
 	 * (There are no read marks in current state, because reads always mark
 	 * their parent and current state never has children yet.  Only
 	 * explored_states can get read marks.)
 	 */
-	for (i = 0; i < BPF_REG_FP; i++)
-		cur->frame[cur->curframe]->regs[i].live = REG_LIVE_NONE;
+	for (j = 0; j <= cur->curframe; j++) {
+		for (i = j < cur->curframe ? BPF_REG_6 : 0; i < BPF_REG_FP; i++)
+			cur->frame[j]->regs[i].parent = &new->frame[j]->regs[i];
+		for (i = 0; i < BPF_REG_FP; i++)
+			cur->frame[j]->regs[i].live = REG_LIVE_NONE;
+	}
 
 	/* all stack frames are accessible from callee, clear them all */
 	for (j = 0; j <= cur->curframe; j++) {
Signature · 1 line
2.20.1
author maintainer
#133 lore raw

PATCH 125/249 perf build: Handle slang being in /usr/include and in /usr/include/slang/

In reply to Sasha Levin

From: Arnaldo Carvalho de Melo <[email protected]>

[ Upstream commit 78d6ccce03e86de34c7000bcada493ed0679e350 ]

In some distros slang.h may be in a /usr/include 'slang' subdir, so use the if slang is not explicitely disabled (by using NO_SLANG=1) and its feature test for the common case (having /usr/include/slang.h) failed, use the results for the test that checks if it is in slang/slang.h.

Change the only file in perf that includes slang.h to use HAVE_SLANG_INCLUDE_SUBDIR and forget about this for good.

On a rhel6 system now we have:

  $ /tmp/build/perf/perf -vv | grep slang
                libslang: [ on  ]  # HAVE_SLANG_SUPPORT
  $ ldd /tmp/build/perf/perf | grep libslang
  	libslang.so.2 => /usr/lib64/libslang.so.2 (0x00007fa2d5a8d000)
  $ grep slang /tmp/build/perf/FEATURE-DUMP
  feature-libslang=0
  feature-libslang-include-subdir=1
  $ cat /etc/redhat-release
  CentOS release 6.10 (Final)
  $

While on fedora:29:

  $ /tmp/build/perf/perf -vv | grep slang
                libslang: [ on  ]  # HAVE_SLANG_SUPPORT
  $ ldd /tmp/build/perf/perf | grep slang
  	libslang.so.2 => /lib64/libslang.so.2 (0x00007f8eb11a7000)
  $ grep slang /tmp/build/perf/FEATURE-DUMP
  feature-libslang=1
  feature-libslang-include-subdir=1
  $
  $ cat /etc/fedora-release
  Fedora release 29 (Twenty Nine)
  $

The feature-libslang-include-subdir=1 line is because the 'gettid()' test was added to test-all.c as the new glibc has an implementation for that, so we soon should have it not failing, i.e. should be the common case soon. Perhaps I should move it out till it becomes the norm...

Cc Adrian Hunter <[email protected]>
Cc Florian Fainelli <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Fixes 1955c8cf5e26 ("perf tools: Don't hardcode host include path for libslang")
---
 tools/perf/Makefile.config | 11 ++++++++---
 tools/perf/ui/libslang.h   |  5 +++++
 2 files changed, 13 insertions(+), 3 deletions(-)
tools/perf/Makefile.config +8 -3
index 85fbcd265351..17b81bc403e4 100644
@@ -637,9 +637,14 @@ endif
 
 ifndef NO_SLANG
   ifneq ($(feature-libslang), 1)
-    msg := $(warning slang not found, disables TUI support. Please install slang-devel, libslang-dev or libslang2-dev);
-    NO_SLANG := 1
-  else
+    ifneq ($(feature-libslang-include-subdir), 1)
+      msg := $(warning slang not found, disables TUI support. Please install slang-devel, libslang-dev or libslang2-dev);
+      NO_SLANG := 1
+    else
+      CFLAGS += -DHAVE_SLANG_INCLUDE_SUBDIR
+    endif
+  endif
+  ifndef NO_SLANG
     # Fedora has /usr/include/slang/slang.h, but ubuntu /usr/include/slang.h
     CFLAGS += -I/usr/include/slang
     CFLAGS += -DHAVE_SLANG_SUPPORT
tools/perf/ui/libslang.h +5 -0
index c0686cda39a5..991e692b9b46 100644
@@ -10,7 +10,12 @@
 #ifndef HAVE_LONG_LONG
 #define HAVE_LONG_LONG __GLIBC_HAVE_LONG_LONG
 #endif
+
+#ifdef HAVE_SLANG_INCLUDE_SUBDIR
+#include <slang/slang.h>
+#else
 #include <slang.h>
+#endif
 
 #if SLANG_VERSION < 20104
 #define slsmg_printf(msg, args...) \
Signature · 1 line
2.20.1
author maintainer
#134 lore raw

PATCH 126/249 PCI: Add missing link delays required by the PCIe spec

In reply to Sasha Levin

From: Mika Westerberg <[email protected]>

[ Upstream commit c2bf1fc212f7e6f25ace1af8f0b3ac061ea48ba5 ]

Currently Linux does not follow PCIe spec regarding the required delays after reset. A concrete example is a Thunderbolt add-in-card that consists of a PCIe switch and two PCIe endpoints:

  +-1b.0-[01-6b]----00.0-[02-6b]--+-00.0-[03]----00.0 TBT controller
                                  +-01.0-[04-36]-- DS hotplug port
                                  +-02.0-[37]----00.0 xHCI controller
                                  \-04.0-[38-6b]-- DS hotplug port

The root port (1b.0) and the PCIe switch downstream ports are all PCIe gen3 so they support 8GT/s link speeds.

We wait for the PCIe hierarchy to enter D3cold (runtime):

pcieport 0000:00:1b.0: power state changed by ACPI to D3cold

When it wakes up from D3cold, according to the PCIe 4.0 section 5.8 the PCIe switch is put to reset and its power is re-applied. This means that we must follow the rules in PCIe 4.0 section 6.6.1.

For the PCIe gen3 ports we are dealing with here, the following applies:

With a Downstream Port that supports Link speeds greater than 5.0 GT/s, software must wait a minimum of 100 ms after Link training completes before sending a Configuration Request to the device immediately below that Port. Software can determine when Link training completes by polling the Data Link Layer Link Active bit or by setting up an associated interrupt (see Section 6.7.3.3).

Translating this into the above topology we would need to do this (DLLLA stands for Data Link Layer Link Active):

pcieport 0000:00:1b.0: wait for 100ms after DLLLA is set before access to 0000:01:00.0 pcieport 0000:02:00.0: wait for 100ms after DLLLA is set before access to 0000:03:00.0 pcieport 0000:02:02.0: wait for 100ms after DLLLA is set before access to 0000:37:00.0

I've instrumented the kernel with additional logging so we can see the actual delays the kernel performs:

pcieport 0000:00:1b.0: power state changed by ACPI to D0
pcieport 0000:00:1b.0: waiting for D3cold delay of 100 ms
pcieport 0000:00:1b.0: waking up bus
pcieport 0000:00:1b.0: waiting for D3hot delay of 10 ms
pcieport 0000:00:1b.0: restoring config space at offset 0x2c (was 0x60, writing 0x60)
...
pcieport 0000:00:1b.0: PME# disabled
pcieport 0000:01:00.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
...
pcieport 0000:01:00.0: PME# disabled
pcieport 0000:02:00.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
...
pcieport 0000:02:00.0: PME# disabled
pcieport 0000:02:01.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
...
pcieport 0000:02:01.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100407) pcieport 0000:02:01.0: PME# disabled
pcieport 0000:02:02.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
...
pcieport 0000:02:02.0: PME# disabled
pcieport 0000:02:04.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
...
pcieport 0000:02:04.0: PME# disabled
pcieport 0000:02:01.0: PME# enabled
pcieport 0000:02:01.0: waiting for D3hot delay of 10 ms
pcieport 0000:02:04.0: PME# enabled
pcieport 0000:02:04.0: waiting for D3hot delay of 10 ms
thunderbolt 0000:03:00.0: restoring config space at offset 0x14 (was 0x0, writing 0x8a040000)
...
thunderbolt 0000:03:00.0: PME# disabled
xhci_hcd 0000:37:00.0: restoring config space at offset 0x10 (was 0x0, writing 0x73f00000)
...
xhci_hcd 0000:37:00.0: PME# disabled

For the switch upstream port (01:00.0) we wait for 100ms but not taking into account the DLLLA requirement. We then wait 10ms for D3hot -> D0 transition of the root port and the two downstream hotplug ports. This means that we deviate from what the spec requires.

Performing the same check for system sleep (s2idle) transitions we can see following when resuming from s2idle:

pcieport 0000:00:1b.0: power state changed by ACPI to D0
pcieport 0000:00:1b.0: restoring config space at offset 0x2c (was 0x60, writing 0x60)
...
pcieport 0000:01:00.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
...
pcieport 0000:02:02.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff) pcieport 0000:02:02.0: restoring config space at offset 0x2c (was 0x0, writing 0x0) pcieport 0000:02:01.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff) pcieport 0000:02:04.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff) pcieport 0000:02:02.0: restoring config space at offset 0x28 (was 0x0, writing 0x0) pcieport 0000:02:00.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff) pcieport 0000:02:02.0: restoring config space at offset 0x24 (was 0x10001, writing 0x1fff1) pcieport 0000:02:01.0: restoring config space at offset 0x2c (was 0x0, writing 0x60) pcieport 0000:02:02.0: restoring config space at offset 0x20 (was 0x0, writing 0x73f073f0) pcieport 0000:02:04.0: restoring config space at offset 0x2c (was 0x0, writing 0x60) pcieport 0000:02:01.0: restoring config space at offset 0x28 (was 0x0, writing 0x60) pcieport 0000:02:00.0: restoring config space at offset 0x2c (was 0x0, writing 0x0) pcieport 0000:02:02.0: restoring config space at offset 0x1c (was 0x101, writing 0x1f1) pcieport 0000:02:04.0: restoring config space at offset 0x28 (was 0x0, writing 0x60) pcieport 0000:02:01.0: restoring config space at offset 0x24 (was 0x10001, writing 0x1ff10001) pcieport 0000:02:00.0: restoring config space at offset 0x28 (was 0x0, writing 0x0) pcieport 0000:02:02.0: restoring config space at offset 0x18 (was 0x0, writing 0x373702) pcieport 0000:02:04.0: restoring config space at offset 0x24 (was 0x10001, writing 0x49f12001) pcieport 0000:02:01.0: restoring config space at offset 0x20 (was 0x0, writing 0x73e05c00) pcieport 0000:02:00.0: restoring config space at offset 0x24 (was 0x10001, writing 0x1fff1) pcieport 0000:02:04.0: restoring config space at offset 0x20 (was 0x0, writing 0x89f07400) pcieport 0000:02:01.0: restoring config space at offset 0x1c (was 0x101, writing 0x5151) pcieport 0000:02:00.0: restoring config space at offset 0x20 (was 0x0, writing 0x8a008a00) pcieport 0000:02:02.0: restoring config space at offset 0xc (was 0x10000, writing 0x10020) pcieport 0000:02:04.0: restoring config space at offset 0x1c (was 0x101, writing 0x6161) pcieport 0000:02:01.0: restoring config space at offset 0x18 (was 0x0, writing 0x360402) pcieport 0000:02:00.0: restoring config space at offset 0x1c (was 0x101, writing 0x1f1) pcieport 0000:02:04.0: restoring config space at offset 0x18 (was 0x0, writing 0x6b3802) pcieport 0000:02:02.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100407) pcieport 0000:02:00.0: restoring config space at offset 0x18 (was 0x0, writing 0x30302) pcieport 0000:02:01.0: restoring config space at offset 0xc (was 0x10000, writing 0x10020) pcieport 0000:02:04.0: restoring config space at offset 0xc (was 0x10000, writing 0x10020) pcieport 0000:02:00.0: restoring config space at offset 0xc (was 0x10000, writing 0x10020) pcieport 0000:02:01.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100407) pcieport 0000:02:04.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100407) pcieport 0000:02:00.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100407) xhci_hcd 0000:37:00.0: restoring config space at offset 0x10 (was 0x0, writing 0x73f00000)
...
thunderbolt 0000:03:00.0: restoring config space at offset 0x14 (was 0x0, writing 0x8a040000)

This is even worse. None of the mandatory delays are performed. If this would be S3 instead of s2idle then according to PCI FW spec 3.2 section 4.6.8. there is a specific _DSM that allows the OS to skip the delays but this platform does not provide the _DSM and does not go to S3 anyway so no firmware is involved that could already handle these delays.

In this particular Intel Coffee Lake platform these delays are not actually needed because there is an additional delay as part of the ACPI power resource that is used to turn on power to the hierarchy but since that additional delay is not required by any of standards (PCIe, ACPI) it is not present in the Intel Ice Lake, for example where missing the mandatory delays causes pciehp to start tearing down the stack too early (links are not yet trained).

For this reason, change the PCIe portdrv PM resume hooks so that they perform the mandatory delays before the downstream component gets resumed. We perform the delays before port services are resumed because otherwise pciehp might find that the link is not up (even if it is just training) and tears-down the hierarchy.

---
 drivers/pci/pci.c               | 29 ++++++++++-----
 drivers/pci/pci.h               |  1 +
 drivers/pci/pcie/portdrv_core.c | 66 +++++++++++++++++++++++++++++++++
 3 files changed, 86 insertions(+), 10 deletions(-)
drivers/pci/pci.c +19 -10
index 8abc843b1615..87a1f902fa8e 100644
@@ -1004,15 +1004,10 @@ static void __pci_start_power_transition(struct pci_dev *dev, pci_power_t state)
 	if (state == PCI_D0) {
 		pci_platform_power_transition(dev, PCI_D0);
 		/*
-		 * Mandatory power management transition delays, see
-		 * PCI Express Base Specification Revision 2.0 Section
-		 * 6.6.1: Conventional Reset.  Do not delay for
-		 * devices powered on/off by corresponding bridge,
-		 * because have already delayed for the bridge.
+		 * Mandatory power management transition delays are
+		 * handled in the PCIe portdrv resume hooks.
 		 */
 		if (dev->runtime_d3cold) {
-			if (dev->d3cold_delay && !dev->imm_ready)
-				msleep(dev->d3cold_delay);
 			/*
 			 * When powering on a bridge from D3cold, the
 			 * whole hierarchy may be powered on into
@@ -4568,14 +4563,16 @@ static int pci_pm_reset(struct pci_dev *dev, int probe)
 
 	return pci_dev_wait(dev, "PM D3->D0", PCIE_RESET_READY_POLL_MS);
 }
+
 /**
- * pcie_wait_for_link - Wait until link is active or inactive
+ * pcie_wait_for_link_delay - Wait until link is active or inactive
  * @pdev: Bridge device
  * @active: waiting for active or inactive?
+ * @delay: Delay to wait after link has become active (in ms)
  *
  * Use this to wait till link becomes active or inactive.
  */
-bool pcie_wait_for_link(struct pci_dev *pdev, bool active)
+bool pcie_wait_for_link_delay(struct pci_dev *pdev, bool active, int delay)
 {
 	int timeout = 1000;
 	bool ret;
@@ -4612,13 +4609,25 @@ bool pcie_wait_for_link(struct pci_dev *pdev, bool active)
 		timeout -= 10;
 	}
 	if (active && ret)
-		msleep(100);
+		msleep(delay);
 	else if (ret != active)
 		pci_info(pdev, "Data Link Layer Link Active not %s in 1000 msec\n",
 			active ? "set" : "cleared");
 	return ret == active;
 }
 
+/**
+ * pcie_wait_for_link - Wait until link is active or inactive
+ * @pdev: Bridge device
+ * @active: waiting for active or inactive?
+ *
+ * Use this to wait till link becomes active or inactive.
+ */
+bool pcie_wait_for_link(struct pci_dev *pdev, bool active)
+{
+	return pcie_wait_for_link_delay(pdev, active, 100);
+}
+
 void pci_reset_secondary_bus(struct pci_dev *dev)
 {
 	u16 ctrl;
drivers/pci/pci.h +1 -0
index 9cb99380c61e..59802b3def4b 100644
@@ -493,6 +493,7 @@ static inline int pci_dev_specific_disable_acs_redir(struct pci_dev *dev)
 void pcie_do_recovery(struct pci_dev *dev, enum pci_channel_state state,
 		      u32 service);
 
+bool pcie_wait_for_link_delay(struct pci_dev *pdev, bool active, int delay);
 bool pcie_wait_for_link(struct pci_dev *pdev, bool active);
 #ifdef CONFIG_PCIEASPM
 void pcie_aspm_init_link_state(struct pci_dev *pdev);
drivers/pci/pcie/portdrv_core.c +66 -0

90 lines not on this page · raw message

Signature · 1 line
2.20.1
author maintainer
#135 lore raw

PATCH 127/249 net: netsec: initialize tx ring on ndo_open

In reply to Sasha Levin

From: Ilias Apalodimas <[email protected]>

[ Upstream commit 39e3622edeffa63c2871153d8743c5825b139968 ]

Since we changed the Tx ring handling and now depends on bit31 to figure out the owner of the descriptor, we should initialize this every time the device goes down-up instead of doing it once on driver init. If the value is not correctly initialized the device won't have any available descriptors

Changes since v1:
- Typo fixes

Fixes 35e07d234739 ("net: socionext: remove mmio reads on Tx")
---
 drivers/net/ethernet/socionext/netsec.c | 32 ++++++++++++++-----------
 1 file changed, 18 insertions(+), 14 deletions(-)
drivers/net/ethernet/socionext/netsec.c +18 -14
index cba5881b2746..a10ef700f16d 100644
@@ -1029,7 +1029,6 @@ static void netsec_free_dring(struct netsec_priv *priv, int id)
 static int netsec_alloc_dring(struct netsec_priv *priv, enum ring_id id)
 {
 	struct netsec_desc_ring *dring = &priv->desc_ring[id];
-	int i;
 
 	dring->vaddr = dma_alloc_coherent(priv->dev, DESC_SZ * DESC_NUM,
 					  &dring->desc_dma, GFP_KERNEL);
@@ -1040,19 +1039,6 @@ static int netsec_alloc_dring(struct netsec_priv *priv, enum ring_id id)
 	if (!dring->desc)
 		goto err;
 
-	if (id == NETSEC_RING_TX) {
-		for (i = 0; i < DESC_NUM; i++) {
-			struct netsec_de *de;
-
-			de = dring->vaddr + (DESC_SZ * i);
-			/* de->attr is not going to be accessed by the NIC
-			 * until netsec_set_tx_de() is called.
-			 * No need for a dma_wmb() here
-			 */
-			de->attr = 1U << NETSEC_TX_SHIFT_OWN_FIELD;
-		}
-	}
-
 	return 0;
 err:
 	netsec_free_dring(priv, id);
@@ -1060,6 +1046,23 @@ static int netsec_alloc_dring(struct netsec_priv *priv, enum ring_id id)
 	return -ENOMEM;
 }
 
+static void netsec_setup_tx_dring(struct netsec_priv *priv)
+{
+	struct netsec_desc_ring *dring = &priv->desc_ring[NETSEC_RING_TX];
+	int i;
+
+	for (i = 0; i < DESC_NUM; i++) {
+		struct netsec_de *de;
+
+		de = dring->vaddr + (DESC_SZ * i);
+		/* de->attr is not going to be accessed by the NIC
+		 * until netsec_set_tx_de() is called.
+		 * No need for a dma_wmb() here
+		 */
+		de->attr = 1U << NETSEC_TX_SHIFT_OWN_FIELD;
+	}
+}
+
 static int netsec_setup_rx_dring(struct netsec_priv *priv)
 {
 	struct netsec_desc_ring *dring = &priv->desc_ring[NETSEC_RING_RX];
@@ -1361,6 +1364,7 @@ static int netsec_netdev_open(struct net_device *ndev)
 
 	pm_runtime_get_sync(priv->dev);
 
+	netsec_setup_tx_dring(priv);
 	ret = netsec_setup_rx_dring(priv);
 	if (ret) {
 		netif_err(priv, probe, priv->ndev,
Signature · 1 line
2.20.1
author maintainer
#136 lore raw

PATCH 128/249 x86/cacheinfo: Fix a -Wtype-limits warning

In reply to Sasha Levin

From: Qian Cai <[email protected]>

[ Upstream commit 1b7aebf0487613033aff26420e32fa2076d52846 ]

cpuinfo_x86.x86_model is an unsigned type, so comparing against zero will generate a compilation warning:

arch/x86/kernel/cpu/cacheinfo.c: In function 'cacheinfo_amd_init_llc_id':
arch/x86/kernel/cpu/cacheinfo.c:662:19: warning: comparison is always true \
due to limited range of data type [-Wtype-limits]

Remove the unnecessary lower bound check.

[ bp: Massage. ]

Fixes 68091ee7ac3c ("x86/CPU/AMD: Calculate last level cache ID from number of sharing threads")
Cc "Gustavo A. R. Silva" <[email protected]>
Cc "H. Peter Anvin" <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Masami Hiramatsu <[email protected]>
Cc Pu Wen <[email protected]>
Cc Suravee Suthikulpanit <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc x86-ml <[email protected]>
---
 arch/x86/kernel/cpu/cacheinfo.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
arch/x86/kernel/cpu/cacheinfo.c +1 -2
index 395d46f78582..c7503be92f35 100644
@@ -658,8 +658,7 @@ void cacheinfo_amd_init_llc_id(struct cpuinfo_x86 *c, int cpu, u8 node_id)
 	if (c->x86 < 0x17) {
 		/* LLC is at the node level. */
 		per_cpu(cpu_llc_id, cpu) = node_id;
-	} else if (c->x86 == 0x17 &&
-		   c->x86_model >= 0 && c->x86_model <= 0x1F) {
+	} else if (c->x86 == 0x17 && c->x86_model <= 0x1F) {
 		/*
 		 * LLC is at the core complex level.
 		 * Core complex ID is ApicId[3] for these processors.
Signature · 1 line
2.20.1
author maintainer
#137 lore raw

PATCH 129/249 blk-iolatency: only account submitted bios

In reply to Sasha Levin

From: Dennis Zhou <[email protected]>

[ Upstream commit a3fb01ba5af066521f3f3421839e501bb2c71805 ]

As is, iolatency recognizes done_bio and cleanup as ending paths. If a request is marked REQ_NOWAIT and fails to get a request, the bio is cleaned up via rq_qos_cleanup() and ended in bio_wouldblock_error(). This results in underflowing the inflight counter. Fix this by only accounting bios that were actually submitted.

---
 block/blk-iolatency.c | 4 ++++
 1 file changed, 4 insertions(+)
block/blk-iolatency.c +4 -0
index d22e61bced86..c91b84bb9d0a 100644
@@ -600,6 +600,10 @@ static void blkcg_iolatency_done_bio(struct rq_qos *rqos, struct bio *bio)
 	if (!blkg || !bio_flagged(bio, BIO_TRACKED))
 		return;
 
+	/* We didn't actually submit this bio, don't account it. */
+	if (bio->bi_status == BLK_STS_AGAIN)
+		return;
+
 	iolat = blkg_to_lat(bio->bi_blkg);
 	if (!iolat)
 		return;
Signature · 1 line
2.20.1
#138 lore raw

Re: [PATCH AUTOSEL 5.2 129/249] blk-iolatency: only account submitted bios

Quoted 36 lines · Sasha Levin

From: Dennis Zhou <[email protected]>

[ Upstream commit a3fb01ba5af066521f3f3421839e501bb2c71805 ]

As is, iolatency recognizes done_bio and cleanup as ending paths. If a request is marked REQ_NOWAIT and fails to get a request, the bio is cleaned up via rq_qos_cleanup() and ended in bio_wouldblock_error(). This results in underflowing the inflight counter. Fix this by only accounting bios that were actually submitted.

---
 block/blk-iolatency.c | 4 ++++
 1 file changed, 4 insertions(+)
block/blk-iolatency.c +4 -0
index d22e61bced86..c91b84bb9d0a 100644
@@ -600,6 +600,10 @@ static void blkcg_iolatency_done_bio(struct rq_qos *rqos, struct bio *bio)
 	if (!blkg || !bio_flagged(bio, BIO_TRACKED))
 		return;
 
+	/* We didn't actually submit this bio, don't account it. */
+	if (bio->bi_status == BLK_STS_AGAIN)
+		return;
+
 	iolat = blkg_to_lat(bio->bi_blkg);
 	if (!iolat)
 		return;
Signature · 1 line
2.20.1

Hi Sasha,

If you're going to pick this up, c9b3007feca0 ("blk-iolatency: fix STS_AGAIN handling") fixes this patch, so please pick that up too.

Thanks,
Dennis

author maintainer Applied
#139 lore raw

Re: [PATCH AUTOSEL 5.2 129/249] blk-iolatency: only account submitted bios

Quoted 40 lines · Dennis Zhou
Quoted 34 lines · Sasha Levin

From: Dennis Zhou <[email protected]>

[ Upstream commit a3fb01ba5af066521f3f3421839e501bb2c71805 ]

As is, iolatency recognizes done_bio and cleanup as ending paths. If a request is marked REQ_NOWAIT and fails to get a request, the bio is cleaned up via rq_qos_cleanup() and ended in bio_wouldblock_error(). This results in underflowing the inflight counter. Fix this by only accounting bios that were actually submitted.

---
 block/blk-iolatency.c | 4 ++++
 1 file changed, 4 insertions(+)
block/blk-iolatency.c +4 -1
index d22e61bced86..c91b84bb9d0a 100644
@@ -600,6 +600,10 @@ static void blkcg_iolatency_done_bio(struct rq_qos *rqos, struct bio *bio)
 	if (!blkg || !bio_flagged(bio, BIO_TRACKED))
 		return;
 
+	/* We didn't actually submit this bio, don't account it. */
+	if (bio->bi_status == BLK_STS_AGAIN)
+		return;
+
 	iolat = blkg_to_lat(bio->bi_blkg);
 	if (!iolat)
 		return;
--
Quoted 2 lines

2.20.1

Hi Sasha,

Quoted 2 lines

If you're going to pick this up, c9b3007feca0 ("blk-iolatency: fix STS_AGAIN handling") fixes this patch, so please pick that up too.

I've picked it up, thanks!

--
Thanks,
Sasha

author maintainer
#140 lore raw

PATCH 130/249 ACPICA: Clear status of GPEs on first direct enable

In reply to Sasha Levin

From: "Rafael J. Wysocki" <[email protected]>

[ Upstream commit 44758bafa53602f2581a6857bb20b55d4d8ad5b2 ]

ACPI GPEs (other than the EC one) can be enabled in two situations. First, the GPEs with existing _Lxx and _Exx methods are enabled implicitly by ACPICA during system initialization. Second, the GPEs without these methods (like GPEs listed by _PRW objects for wakeup devices) need to be enabled directly by the code that is going to use them (e.g. ACPI power management or device drivers).

In the former case, if the status of a given GPE is set to start with, its handler method (either _Lxx or _Exx) needs to be invoked to take care of the events (possibly) signaled before the GPE was enabled. In the latter case, however, the first caller of acpi_enable_gpe() for a given GPE should not be expected to care about any events that might be signaled through it earlier. In that case, it is better to clear the status of the GPE before enabling it, to prevent stale events from triggering unwanted actions (like spurious system resume, for example).

For this reason, modify acpi_ev_add_gpe_reference() to take an additional boolean argument indicating whether or not the GPE status needs to be cleared when its reference counter changes from zero to one and make acpi_enable_gpe() pass TRUE to it through that new argument.

Fixes 18996f2db918 ("ACPICA: Events: Stop unconditionally clearing ACPI IRQs during suspend/resume")
---
 drivers/acpi/acpica/acevents.h | 3 ++-
 drivers/acpi/acpica/evgpe.c    | 8 +++++++-
 drivers/acpi/acpica/evgpeblk.c | 2 +-
 drivers/acpi/acpica/evxface.c  | 2 +-
 drivers/acpi/acpica/evxfgpe.c  | 2 +-
 5 files changed, 12 insertions(+), 5 deletions(-)
drivers/acpi/acpica/acevents.h +2 -1
index 831660179662..c8652f91054e 100644
@@ -69,7 +69,8 @@ acpi_status
 acpi_ev_mask_gpe(struct acpi_gpe_event_info *gpe_event_info, u8 is_masked);
 
 acpi_status
-acpi_ev_add_gpe_reference(struct acpi_gpe_event_info *gpe_event_info);
+acpi_ev_add_gpe_reference(struct acpi_gpe_event_info *gpe_event_info,
+			  u8 clear_on_enable);
 
 acpi_status
 acpi_ev_remove_gpe_reference(struct acpi_gpe_event_info *gpe_event_info);
drivers/acpi/acpica/evgpe.c +7 -1
index 62d3aa74277b..344feba29063 100644
@@ -146,6 +146,7 @@ acpi_ev_mask_gpe(struct acpi_gpe_event_info *gpe_event_info, u8 is_masked)
  * FUNCTION:    acpi_ev_add_gpe_reference
  *
  * PARAMETERS:  gpe_event_info          - Add a reference to this GPE
+ *              clear_on_enable         - Clear GPE status before enabling it
  *
  * RETURN:      Status
  *
@@ -155,7 +156,8 @@ acpi_ev_mask_gpe(struct acpi_gpe_event_info *gpe_event_info, u8 is_masked)
  ******************************************************************************/
 
 acpi_status
-acpi_ev_add_gpe_reference(struct acpi_gpe_event_info *gpe_event_info)
+acpi_ev_add_gpe_reference(struct acpi_gpe_event_info *gpe_event_info,
+			  u8 clear_on_enable)
 {
 	acpi_status status = AE_OK;
 
@@ -170,6 +172,10 @@ acpi_ev_add_gpe_reference(struct acpi_gpe_event_info *gpe_event_info)
 
 		/* Enable on first reference */
 
+		if (clear_on_enable) {
+			(void)acpi_hw_clear_gpe(gpe_event_info);
+		}
+
 		status = acpi_ev_update_gpe_enable_mask(gpe_event_info);
 		if (ACPI_SUCCESS(status)) {
 			status = acpi_ev_enable_gpe(gpe_event_info);
drivers/acpi/acpica/evgpeblk.c +1 -1
index 328d1d6123ad..fb15e9e2373b 100644
@@ -453,7 +453,7 @@ acpi_ev_initialize_gpe_block(struct acpi_gpe_xrupt_info *gpe_xrupt_info,
 				continue;
 			}
 
-			status = acpi_ev_add_gpe_reference(gpe_event_info);
+			status = acpi_ev_add_gpe_reference(gpe_event_info, FALSE);
 			if (ACPI_FAILURE(status)) {
 				ACPI_EXCEPTION((AE_INFO, status,
 					"Could not enable GPE 0x%02X",
drivers/acpi/acpica/evxface.c +1 -1
index 3df00eb6621b..279ef0557aa3 100644
@@ -971,7 +971,7 @@ acpi_remove_gpe_handler(acpi_handle gpe_device,
 	      ACPI_GPE_DISPATCH_METHOD) ||
 	     (ACPI_GPE_DISPATCH_TYPE(handler->original_flags) ==
 	      ACPI_GPE_DISPATCH_NOTIFY)) && handler->originally_enabled) {
-		(void)acpi_ev_add_gpe_reference(gpe_event_info);
+		(void)acpi_ev_add_gpe_reference(gpe_event_info, FALSE);
 		if (ACPI_GPE_IS_POLLING_NEEDED(gpe_event_info)) {
 
 			/* Poll edge triggered GPEs to handle existing events */
drivers/acpi/acpica/evxfgpe.c +1 -1
index 30a083902f52..710488ec59e9 100644
@@ -108,7 +108,7 @@ acpi_status acpi_enable_gpe(acpi_handle gpe_device, u32 gpe_number)
 	if (gpe_event_info) {
 		if (ACPI_GPE_DISPATCH_TYPE(gpe_event_info->flags) !=
 		    ACPI_GPE_DISPATCH_NONE) {
-			status = acpi_ev_add_gpe_reference(gpe_event_info);
+			status = acpi_ev_add_gpe_reference(gpe_event_info, TRUE);
 			if (ACPI_SUCCESS(status) &&
 			    ACPI_GPE_IS_POLLING_NEEDED(gpe_event_info)) {
 
Signature · 1 line
2.20.1
author maintainer
#141 lore raw

PATCH 131/249 spi: fix ctrl->num_chipselect constraint

In reply to Sasha Levin

From: Tudor Ambarus <[email protected]>

[ Upstream commit f9481b08220d7dc1ff21e296a330ee8b721b44e4 ]

at91sam9g25ek showed the following error at probe:
atmel_spi f0000000.spi: Using dma0chan2 (tx) and dma0chan3 (rx) for DMA transfers
atmel_spi: probe of f0000000.spi failed with error -22

Commit 0a919ae49223 ("spi: Don't call spi_get_gpio_descs() before device name is set") moved the calling of spi_get_gpio_descs() after ctrl->dev is set,
but didn't move the !ctrl->num_chipselect check. When there are
chip selects in the device tree, the spi-atmel driver lets the
SPI core discover them when registering the SPI master.
The ctrl->num_chipselect is thus expected to be set by spi_get_gpio_descs().

Move the !ctlr->num_chipselect after spi_get_gpio_descs() as it was before the aforementioned commit. While touching this block, get rid of the explicit comparison with 0 and update the commenting style.

Fixes 0a919ae49223 ("spi: Don't call spi_get_gpio_descs() before device name is set")
---
 drivers/spi/spi.c | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)
drivers/spi/spi.c +7 -5
index 5e4654032bfa..29916e446143 100644
@@ -2286,11 +2286,6 @@ int spi_register_controller(struct spi_controller *ctlr)
 	if (status)
 		return status;
 
-	/* even if it's just one always-selected device, there must
-	 * be at least one chipselect
-	 */
-	if (ctlr->num_chipselect == 0)
-		return -EINVAL;
 	if (ctlr->bus_num >= 0) {
 		/* devices with a fixed bus num must check-in with the num */
 		mutex_lock(&board_lock);
@@ -2361,6 +2356,13 @@ int spi_register_controller(struct spi_controller *ctlr)
 		}
 	}
 
+	/*
+	 * Even if it's just one always-selected device, there must
+	 * be at least one chipselect.
+	 */
+	if (!ctlr->num_chipselect)
+		return -EINVAL;
+
 	status = device_add(&ctlr->dev);
 	if (status < 0) {
 		/* free bus id */
Signature · 1 line
2.20.1
author maintainer
#142 lore raw

PATCH 132/249 EDAC/sysfs: Drop device references properly

In reply to Sasha Levin

From: Greg KH <[email protected]>

[ Upstream commit 7adc05d2dc3af95e4e1534841d58f736262142cd ]

Do put_device() if device_add() fails.

[ bp: do device_del() for the successfully created devices in
edac_create_csrow_objects(), on the unwind path. ]

---
 drivers/edac/edac_mc_sysfs.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)
drivers/edac/edac_mc_sysfs.c +7 -3
index 464174685589..bf9273437e3f 100644
@@ -443,7 +443,8 @@ static int edac_create_csrow_objects(struct mem_ctl_info *mci)
 		csrow = mci->csrows[i];
 		if (!nr_pages_per_csrow(csrow))
 			continue;
-		put_device(&mci->csrows[i]->dev);
+
+		device_del(&mci->csrows[i]->dev);
 	}
 
 	return err;
@@ -645,9 +646,11 @@ static int edac_create_dimm_object(struct mem_ctl_info *mci,
 	dev_set_drvdata(&dimm->dev, dimm);
 	pm_runtime_forbid(&mci->dev);
 
-	err =  device_add(&dimm->dev);
+	err = device_add(&dimm->dev);
+	if (err)
+		put_device(&dimm->dev);
 
-	edac_dbg(0, "creating rank/dimm device %s\n", dev_name(&dimm->dev));
+	edac_dbg(0, "created rank/dimm device %s\n", dev_name(&dimm->dev));
 
 	return err;
 }
@@ -928,6 +931,7 @@ int edac_create_sysfs_mci_device(struct mem_ctl_info *mci,
 	err = device_add(&mci->dev);
 	if (err < 0) {
 		edac_dbg(1, "failure: create device %s\n", dev_name(&mci->dev));
+		put_device(&mci->dev);
 		goto out;
 	}
 
Signature · 1 line
2.20.1
author maintainer
#143 lore raw

PATCH 133/249 EDAC/sysfs: Fix memory leak when creating a csrow object

In reply to Sasha Levin

From: Pan Bian <[email protected]>

[ Upstream commit 585fb3d93d32dbe89e718b85009f9c322cc554cd ]

In edac_create_csrow_object(), the reference to the object is not released when adding the device to the device hierarchy fails (device_add()). This may result in a memory leak.

---
 drivers/edac/edac_mc_sysfs.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
drivers/edac/edac_mc_sysfs.c +7 -1
index bf9273437e3f..7c01e1cc030c 100644
@@ -404,6 +404,8 @@ static inline int nr_pages_per_csrow(struct csrow_info *csrow)
 static int edac_create_csrow_object(struct mem_ctl_info *mci,
 				    struct csrow_info *csrow, int index)
 {
+	int err;
+
 	csrow->dev.type = &csrow_attr_type;
 	csrow->dev.groups = csrow_dev_groups;
 	device_initialize(&csrow->dev);
@@ -415,7 +417,11 @@ static int edac_create_csrow_object(struct mem_ctl_info *mci,
 	edac_dbg(0, "creating (virtual) csrow node %s\n",
 		 dev_name(&csrow->dev));
 
-	return device_add(&csrow->dev);
+	err = device_add(&csrow->dev);
+	if (err)
+		put_device(&csrow->dev);
+
+	return err;
 }
 
 /* Create a CSROW object under specifed edac_mc_device */
Signature · 1 line
2.20.1
author maintainer
#144 lore raw

PATCH 134/249 nvme: fix possible io failures when removing multipathed ns

In reply to Sasha Levin

From: Anton Eidelman <[email protected]>

[ Upstream commit 2181e455612a8db2761eabbf126640552a451e96 ]

When a shared namespace is removed, we call blk_cleanup_queue() when the device can still be accessed as the current path and this can result in submission to a dying queue. Hence, direct_make_request() called by our mpath device may fail (propagating the failure to userspace). Instead, we want to failover this I/O to a different path if one exists. Thus, before we cleanup the request queue, we make sure that the device is cleared from the current path nor it can be selected again as such.

Fix this by:
- clear the ns from the head->list and synchronize rcu to make sure there is no concurrent path search that restores it as the current path
- clear the mpath current path in order to trigger a subsequent path search and sync srcu to wait for any ongoing request submissions
- safely continue to namespace removal and blk_cleanup_queue

Signed-off-by Anton Eidelman <anton at lightbitslabs.com>
Signed-off-by Sagi Grimberg <sagi at grimberg.me>
Signed-off-by Christoph Hellwig <hch at lst.de>
Signed-off-by Sasha Levin <sashal at kernel.org>
---
 drivers/nvme/host/core.c | 14 ++++++++------
 1 file changed, 8 insertions(+), 6 deletions(-)
drivers/nvme/host/core.c +8 -6
index 120fb593d1da..22c68e3b71d5 100644
@@ -3344,6 +3344,14 @@ static void nvme_ns_remove(struct nvme_ns *ns)
 		return;
 
 	nvme_fault_inject_fini(ns);
+
+	mutex_lock(&ns->ctrl->subsys->lock);
+	list_del_rcu(&ns->siblings);
+	mutex_unlock(&ns->ctrl->subsys->lock);
+	synchronize_rcu(); /* guarantee not available in head->list */
+	nvme_mpath_clear_current_path(ns);
+	synchronize_srcu(&ns->head->srcu); /* wait for concurrent submissions */
+
 	if (ns->disk && ns->disk->flags & GENHD_FL_UP) {
 		del_gendisk(ns->disk);
 		blk_cleanup_queue(ns->queue);
@@ -3351,16 +3359,10 @@ static void nvme_ns_remove(struct nvme_ns *ns)
 			blk_integrity_unregister(ns->disk);
 	}
 
-	mutex_lock(&ns->ctrl->subsys->lock);
-	list_del_rcu(&ns->siblings);
-	nvme_mpath_clear_current_path(ns);
-	mutex_unlock(&ns->ctrl->subsys->lock);
-
 	down_write(&ns->ctrl->namespaces_rwsem);
 	list_del_init(&ns->list);
 	up_write(&ns->ctrl->namespaces_rwsem);
 
-	synchronize_srcu(&ns->head->srcu);
 	nvme_mpath_check_last_path(ns);
 	nvme_put_ns(ns);
 }
Signature · 1 line
2.20.1
author maintainer
#145 lore raw

PATCH 135/249 nvme-pci: properly report state change failure in nvme_reset_work

In reply to Sasha Levin

From: Minwoo Im <[email protected]>

[ Upstream commit cee6c269b016ba89c62e34d6bccb103ee2c7de4f ]

If the state change to NVME_CTRL_CONNECTING fails, the dmesg is going to be like:

  [  293.689160] nvme nvme0: failed to mark controller CONNECTING
  [  293.689160] nvme nvme0: Removing after probe failure status: 0

Even it prints the first line to indicate the situation, the second line is not proper because the status is 0 which means normally success of the previous operation.

This patch makes it indicate the proper error value when it fails.

  [   25.932367] nvme nvme0: failed to mark controller CONNECTING
  [   25.932369] nvme nvme0: Removing after probe failure status: -16

This situation is able to be easily reproduced by:
root at target:~# rmmod nvme && modprobe nvme && rmmod nvme

Signed-off-by Minwoo Im <minwoo.im.dev at gmail.com>
Reviewed-by Chaitanya Kulkarni <chaitanya.kulkarni at wdc.com>
Signed-off-by Christoph Hellwig <hch at lst.de>
Signed-off-by Sasha Levin <sashal at kernel.org>
---
 drivers/nvme/host/pci.c | 1 +
 1 file changed, 1 insertion(+)
drivers/nvme/host/pci.c +1 -0
index 524d6bd6d095..385ba7a1e23b 100644
@@ -2528,6 +2528,7 @@ static void nvme_reset_work(struct work_struct *work)
 	if (!nvme_change_ctrl_state(&dev->ctrl, NVME_CTRL_CONNECTING)) {
 		dev_warn(dev->ctrl.device,
 			"failed to mark controller CONNECTING\n");
+		result = -EBUSY;
 		goto out;
 	}
 
Signature · 1 line
2.20.1
author maintainer
#146 lore raw

PATCH 136/249 nvme-pci: set the errno on ctrl state change error

In reply to Sasha Levin

From: Chaitanya Kulkarni <[email protected]>

[ Upstream commit e71afda49335620e3d9adf56015676db33a3bd86 ]

This patch removes the confusing assignment of the variable result at the time of declaration and sets the value in error cases next to the places where the actual error is happening.

Here we also set the result value to -ENODEV when we fail at the final ctrl state transition in nvme_reset_work(). Without this assignment result will hold 0 from nvme_setup_io_queue() and on failure 0 will be passed to he nvme_remove_dead_ctrl() from final state transition.

Signed-off-by Chaitanya Kulkarni <chaitanya.kulkarni at wdc.com>
Signed-off-by Christoph Hellwig <hch at lst.de>
Signed-off-by Sasha Levin <sashal at kernel.org>
---
 drivers/nvme/host/pci.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
drivers/nvme/host/pci.c +5 -2
index 385ba7a1e23b..544d095d44e5 100644
@@ -2480,11 +2480,13 @@ static void nvme_reset_work(struct work_struct *work)
 	struct nvme_dev *dev =
 		container_of(work, struct nvme_dev, ctrl.reset_work);
 	bool was_suspend = !!(dev->ctrl.ctrl_config & NVME_CC_SHN_NORMAL);
-	int result = -ENODEV;
+	int result;
 	enum nvme_ctrl_state new_state = NVME_CTRL_LIVE;
 
-	if (WARN_ON(dev->ctrl.state != NVME_CTRL_RESETTING))
+	if (WARN_ON(dev->ctrl.state != NVME_CTRL_RESETTING)) {
+		result = -ENODEV;
 		goto out;
+	}
 
 	/*
 	 * If we're called to reset a live controller first shut it down before
@@ -2589,6 +2591,7 @@ static void nvme_reset_work(struct work_struct *work)
 	if (!nvme_change_ctrl_state(&dev->ctrl, new_state)) {
 		dev_warn(dev->ctrl.device,
 			"failed to mark controller state %d\n", new_state);
+		result = -ENODEV;
 		goto out;
 	}
 
Signature · 1 line
2.20.1
author maintainer
#147 lore raw

PATCH 137/249 lightnvm: pblk: fix freeing of merged pages

In reply to Sasha Levin

From: Heiner Litz <[email protected]>

[ Upstream commit 510fd8ea98fcb586c01aef93d87c060a159ac30a ]

bio_add_pc_page() may merge pages when a bio is padded due to a flush. Fix iteration over the bio to free the correct pages in case of a merge.

---
 drivers/lightnvm/pblk-core.c | 18 ++++++++++--------
 1 file changed, 10 insertions(+), 8 deletions(-)
drivers/lightnvm/pblk-core.c +10 -8
index 773537804319..f546e6f28b8a 100644
@@ -323,14 +323,16 @@ void pblk_free_rqd(struct pblk *pblk, struct nvm_rq *rqd, int type)
 void pblk_bio_free_pages(struct pblk *pblk, struct bio *bio, int off,
 			 int nr_pages)
 {
-	struct bio_vec bv;
-	int i;
-
-	WARN_ON(off + nr_pages != bio->bi_vcnt);
-
-	for (i = off; i < nr_pages + off; i++) {
-		bv = bio->bi_io_vec[i];
-		mempool_free(bv.bv_page, &pblk->page_bio_pool);
+	struct bio_vec *bv;
+	struct page *page;
+	int i, e, nbv = 0;
+
+	for (i = 0; i < bio->bi_vcnt; i++) {
+		bv = &bio->bi_io_vec[i];
+		page = bv->bv_page;
+		for (e = 0; e < bv->bv_len; e += PBLK_EXPOSED_PAGE_SIZE, nbv++)
+			if (nbv >= off)
+				mempool_free(page++, &pblk->page_bio_pool);
 	}
 }
 
Signature · 1 line
2.20.1
author maintainer
#148 lore raw

PATCH 138/249 lightnvm: fix uninitialized pointer in nvm_remove_tgt()

In reply to Sasha Levin

From: Geert Uytterhoeven <[email protected]>

[ Upstream commit 2f5af4ab7de14bd35f3435e6a47300276bbb6c17 ]

With gcc 4.1:

drivers/lightnvm/core.c: In function ‘nvm_remove_tgt’:
drivers/lightnvm/core.c:510: warning: ‘t’ is used uninitialized in this function

Indeed, if no NVM devices have been registered, t will be an uninitialized pointer, and may be dereferenced later. A call to nvm_remove_tgt() can be triggered from userspace by issuing the NVM_DEV_REMOVE ioctl on the lightnvm control device.

Fix this by preinitializing t to NULL.

Fixes 843f2edbdde085b4 ("lightnvm: do not remove instance under global lock")
---
 drivers/lightnvm/core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/lightnvm/core.c +1 -1
index 7d555b110ecd..a600934fdd9c 100644
@@ -478,7 +478,7 @@ static void __nvm_remove_target(struct nvm_target *t, bool graceful)
  */
 static int nvm_remove_tgt(struct nvm_ioctl_remove *remove)
 {
-	struct nvm_target *t;
+	struct nvm_target *t = NULL;
 	struct nvm_dev *dev;
 
 	down_read(&nvm_lock);
Signature · 1 line
2.20.1
author maintainer
#149 lore raw

PATCH 139/249 nvme-pci: adjust irq max_vector using num_possible_cpus()

In reply to Sasha Levin

From: Minwoo Im <[email protected]>

[ Upstream commit dad77d63903e91a2e97a0c984cabe5d36e91ba60 ]

If the "irq_queues" are greater than num_possible_cpus(), nvme_calc_irq_sets() can have irq set_size for HCTX_TYPE_DEFAULT greater than it can be afforded.
2039 affd->set_size[HCTX_TYPE_DEFAULT] = nrirqs - nr_read_queues;

It might cause a WARN() from the irq_build_affinity_masks() like [1]:
220 if (nr_present < numvecs)
221 WARN_ON(nr_present + nr_others < numvecs);

This patch prevents it from the WARN() by adjusting the max_vector value from the nvme_setup_irqs().

[1] WARN messages when modprobe nvme write_queues=32 poll_queues=0:
root at target:~/nvme# nproc
8
root at target:~/nvme# modprobe nvme write_queues=32 poll_queues=0

[   17.925326] nvme nvme0: pci function 0000:00:04.0
[   17.940601] WARNING: CPU: 3 PID: 1030 at kernel/irq/affinity.c:221 irq_create_affinity_masks+0x222/0x330
[   17.940602] Modules linked in: nvme nvme_core [last unloaded: nvme]
[   17.940605] CPU: 3 PID: 1030 Comm: kworker/u17:4 Tainted: G        W         5.1.0+ #156
[   17.940605] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.12.1-0-ga5cab58e9a3f-prebuilt.qemu.org 04/01/2014
[   17.940608] Workqueue: nvme-reset-wq nvme_reset_work [nvme]
[   17.940609] RIP: 0010:irq_create_affinity_masks+0x222/0x330
[   17.940611] Code: 4c 8d 4c 24 28 4c 8d 44 24 30 e8 c9 fa ff ff 89 44 24 18 e8 c0 38 fa ff 8b 44 24 18 44 8b 54 24 1c 5a 44 01 d0 41 39 c4 76 02 <0f> 0b 48 89 df 44 01 e5 e8 f1 ce 10 00 48 8b 34 24 44 89 f0 44 01
[   17.940611] RSP: 0018:ffffc90002277c50 EFLAGS: 00010216
[   17.940612] RAX: 0000000000000008 RBX: ffff88807ca48860 RCX: 0000000000000000
[   17.940612] RDX: ffff88807bc03800 RSI: 0000000000000020 RDI: 0000000000000000
[   17.940613] RBP: 0000000000000001 R08: ffffc90002277c78 R09: ffffc90002277c70
[   17.940613] R10: 0000000000000008 R11: 0000000000000001 R12: 0000000000000020
[   17.940614] R13: 0000000000025d08 R14: 0000000000000001 R15: ffff88807bc03800
[   17.940614] FS:  0000000000000000(0000) GS:ffff88807db80000(0000) knlGS:0000000000000000
[   17.940616] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[   17.940617] CR2: 00005635e583f790 CR3: 000000000240a000 CR4: 00000000000006e0
[   17.940617] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[   17.940618] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[   17.940618] Call Trace:
[   17.940622]  __pci_enable_msix_range+0x215/0x540
[   17.940623]  ? kernfs_put+0x117/0x160
[   17.940625]  pci_alloc_irq_vectors_affinity+0x74/0x110
[   17.940626]  nvme_reset_work+0xc30/0x1397 [nvme]
[   17.940628]  ? __switch_to_asm+0x34/0x70
[   17.940628]  ? __switch_to_asm+0x40/0x70
[   17.940629]  ? __switch_to_asm+0x34/0x70
[   17.940630]  ? __switch_to_asm+0x40/0x70
[   17.940630]  ? __switch_to_asm+0x34/0x70
[   17.940631]  ? __switch_to_asm+0x40/0x70
[   17.940632]  ? nvme_irq_check+0x30/0x30 [nvme]
[   17.940633]  process_one_work+0x20b/0x3e0
[   17.940634]  worker_thread+0x1f9/0x3d0
[   17.940635]  ? cancel_delayed_work+0xa0/0xa0
[   17.940636]  kthread+0x117/0x120
[   17.940637]  ? kthread_stop+0xf0/0xf0
[   17.940638]  ret_from_fork+0x3a/0x50
[   17.940639] ---[ end trace aca8a131361cd42a ]---
[   17.942124] nvme nvme0: 7/1/0 default/read/poll queues
Signed-off-by Minwoo Im <minwoo.im.dev at gmail.com>
Signed-off-by Christoph Hellwig <hch at lst.de>
Signed-off-by Sasha Levin <sashal at kernel.org>
---
 drivers/nvme/host/pci.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
drivers/nvme/host/pci.c +5 -1
index 544d095d44e5..f5bc1c30cef5 100644
@@ -2068,6 +2068,7 @@ static int nvme_setup_irqs(struct nvme_dev *dev, unsigned int nr_io_queues)
 		.priv		= dev,
 	};
 	unsigned int irq_queues, this_p_queues;
+	unsigned int nr_cpus = num_possible_cpus();
 
 	/*
 	 * Poll queues don't need interrupts, but we need at least one IO
@@ -2078,7 +2079,10 @@ static int nvme_setup_irqs(struct nvme_dev *dev, unsigned int nr_io_queues)
 		this_p_queues = nr_io_queues - 1;
 		irq_queues = 1;
 	} else {
-		irq_queues = nr_io_queues - this_p_queues + 1;
+		if (nr_cpus < nr_io_queues - this_p_queues)
+			irq_queues = nr_cpus + 1;
+		else
+			irq_queues = nr_io_queues - this_p_queues + 1;
 	}
 	dev->io_queues[HCTX_TYPE_POLL] = this_p_queues;
 
Signature · 1 line
2.20.1
author maintainer
#150 lore raw

PATCH 140/249 arm64: Do not enable IRQs for ct_user_exit

In reply to Sasha Levin

From: Julien Thierry <[email protected]>

[ Upstream commit 9034f6251572a4744597c51dea5ab73a55f2b938 ]

For el0_dbg and el0_error, DAIF bits get explicitly cleared before calling ct_user_exit.

When context tracking is disabled, DAIF gets set (almost) immediately after. When context tracking is enabled, among the first things done is disabling IRQs.

What is actually needed is:
- PSR.D = 0 so the system can be debugged (should be already the case)
- PSR.A = 0 so async error can be handled during context tracking

Do not clear PSR.I in those two locations.

---
 arch/arm64/kernel/entry.S | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
arch/arm64/kernel/entry.S +2 -2
index 2df8d0a1d980..dbe467686332 100644
@@ -859,7 +859,7 @@ el0_dbg:
 	mov	x1, x25
 	mov	x2, sp
 	bl	do_debug_exception
-	enable_daif
+	enable_da_f
 	ct_user_exit
 	b	ret_to_user
 el0_inv:
@@ -911,7 +911,7 @@ el0_error_naked:
 	enable_dbg
 	mov	x0, sp
 	bl	do_serror
-	enable_daif
+	enable_da_f
 	ct_user_exit
 	b	ret_to_user
 ENDPROC(el0_error)
Signature · 1 line
2.20.1
author maintainer
#151 lore raw

PATCH 141/249 ipsec: select crypto ciphers for xfrm_algo

In reply to Sasha Levin

From: Arnd Bergmann <[email protected]>

[ Upstream commit 597179b0ba550bd83fab1a9d57c42a9343c58514 ]

kernelci.org reports failed builds on arc because of what looks like an old missed 'select' statement:

net/xfrm/xfrm_algo.o: In function `xfrm_probe_algs':
xfrm_algo.c:(.text+0x1e8): undefined reference to `crypto_has_ahash'

I don't see this in randconfig builds on other architectures, but it's fairly clear we want to select the hash code for it, like we do for all its other users. As Herbert points out, CRYPTO_BLKCIPHER is also required even though it has not popped up in build tests.

Fixes 17bc19702221 ("ipsec: Use skcipher and ahash when probing algorithms")
---
 net/xfrm/Kconfig | 2 ++
 1 file changed, 2 insertions(+)
net/xfrm/Kconfig +2 -0
index c967fc3c38c8..51bb6018f3bf 100644
@@ -15,6 +15,8 @@ config XFRM_ALGO
 	tristate
 	select XFRM
 	select CRYPTO
+	select CRYPTO_HASH
+	select CRYPTO_BLKCIPHER
 
 if INET
 config XFRM_USER
Signature · 1 line
2.20.1
author maintainer
#152 lore raw

PATCH 142/249 media: staging: davinci: fix memory leaks and check for allocation failure

In reply to Sasha Levin

From: Colin Ian King <[email protected]>

[ Upstream commit a84e355ecd3ed9759d7aaa40170aab78e2a68a06 ]

There are three error return paths that don't kfree params causing a memory leak. Fix this by adding an error return path that kfree's params before returning. Also add a check to see params failed to be allocated.

---
 drivers/staging/media/davinci_vpfe/dm365_ipipe.c | 15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)
drivers/staging/media/davinci_vpfe/dm365_ipipe.c +10 -5
index 30e2edc0cec5..b88855c7ffe8 100644
@@ -1251,10 +1251,10 @@ static int ipipe_s_config(struct v4l2_subdev *sd, struct vpfe_ipipe_config *cfg)
 	struct vpfe_ipipe_device *ipipe = v4l2_get_subdevdata(sd);
 	unsigned int i;
 	int rval = 0;
+	struct ipipe_module_params *params;
 
 	for (i = 0; i < ARRAY_SIZE(ipipe_modules); i++) {
 		const struct ipipe_module_if *module_if;
-		struct ipipe_module_params *params;
 		void *from, *to;
 		size_t size;
 
@@ -1265,25 +1265,30 @@ static int ipipe_s_config(struct v4l2_subdev *sd, struct vpfe_ipipe_config *cfg)
 		from = *(void **)((void *)cfg + module_if->config_offset);
 
 		params = kmalloc(sizeof(*params), GFP_KERNEL);
+		if (!params)
+			return -ENOMEM;
 		to = (void *)params + module_if->param_offset;
 		size = module_if->param_size;
 
 		if (to && from && size) {
 			if (copy_from_user(to, (void __user *)from, size)) {
 				rval = -EFAULT;
-				break;
+				goto error_free;
 			}
 			rval = module_if->set(ipipe, to);
 			if (rval)
-				goto error;
+				goto error_free;
 		} else if (to && !from && size) {
 			rval = module_if->set(ipipe, NULL);
 			if (rval)
-				goto error;
+				goto error_free;
 		}
 		kfree(params);
 	}
-error:
+	return rval;
+
+error_free:
+	kfree(params);
 	return rval;
 }
 
Signature · 1 line
2.20.1
author maintainer
#153 lore raw

PATCH 143/249 ipvs: defer hook registration to avoid leaks

In reply to Sasha Levin

From: Julian Anastasov <[email protected]>

[ Upstream commit cf47a0b882a4e5f6b34c7949d7b293e9287f1972 ]

syzkaller reports for memory leak when registering hooks [1]

As we moved the nf_unregister_net_hooks() call into __ip_vs_dev_cleanup(), defer the nf_register_net_hooks() call, so that hooks are allocated and freed from same pernet_operations (ipvs_core_dev_ops).

[1]

BUG: memory leak

unreferenced object 0xffff88810acd8a80 (size 96):
comm "syz-executor073", pid 7254, jiffies 4294950560 (age 22.250s) hex dump (first 32 bytes):
02 00 00 00 00 00 00 00 50 8b bb 82 ff ff ff ff ........P....... 00 00 00 00 00 00 00 00 00 77 bb 82 ff ff ff ff .........w......
backtrace:

   [<0000000013db61f1>] kmemleak_alloc_recursive include/linux/kmemleak.h:55 [inline]
   [<0000000013db61f1>] slab_post_alloc_hook mm/slab.h:439 [inline]
   [<0000000013db61f1>] slab_alloc_node mm/slab.c:3269 [inline]
   [<0000000013db61f1>] kmem_cache_alloc_node_trace+0x15b/0x2a0 mm/slab.c:3597
   [<000000001a27307d>] __do_kmalloc_node mm/slab.c:3619 [inline]
   [<000000001a27307d>] __kmalloc_node+0x38/0x50 mm/slab.c:3627
   [<0000000025054add>] kmalloc_node include/linux/slab.h:590 [inline]
   [<0000000025054add>] kvmalloc_node+0x4a/0xd0 mm/util.c:431
   [<0000000050d1bc00>] kvmalloc include/linux/mm.h:637 [inline]
   [<0000000050d1bc00>] kvzalloc include/linux/mm.h:645 [inline]
   [<0000000050d1bc00>] allocate_hook_entries_size+0x3b/0x60 net/netfilter/core.c:61
   [<00000000e8abe142>] nf_hook_entries_grow+0xae/0x270 net/netfilter/core.c:128
   [<000000004b94797c>] __nf_register_net_hook+0x9a/0x170 net/netfilter/core.c:337
   [<00000000d1545cbc>] nf_register_net_hook+0x34/0xc0 net/netfilter/core.c:464
   [<00000000876c9b55>] nf_register_net_hooks+0x53/0xc0 net/netfilter/core.c:480
   [<000000002ea868e0>] __ip_vs_init+0xe8/0x170 net/netfilter/ipvs/ip_vs_core.c:2280
   [<000000002eb2d451>] ops_init+0x4c/0x140 net/core/net_namespace.c:130
   [<000000000284ec48>] setup_net+0xde/0x230 net/core/net_namespace.c:316
   [<00000000a70600fa>] copy_net_ns+0xf0/0x1e0 net/core/net_namespace.c:439
   [<00000000ff26c15e>] create_new_namespaces+0x141/0x2a0 kernel/nsproxy.c:107
   [<00000000b103dc79>] copy_namespaces+0xa1/0xe0 kernel/nsproxy.c:165
   [<000000007cc008a2>] copy_process.part.0+0x11fd/0x2150 kernel/fork.c:2035
   [<00000000c344af7c>] copy_process kernel/fork.c:1800 [inline]
   [<00000000c344af7c>] _do_fork+0x121/0x4f0 kernel/fork.c:2369
Reported-by [email protected]
Fixes 719c7d563c17 ("ipvs: Fix use-after-free in ip_vs_in")
---
 net/netfilter/ipvs/ip_vs_core.c | 21 ++++++++++++++-------
 1 file changed, 14 insertions(+), 7 deletions(-)
net/netfilter/ipvs/ip_vs_core.c +14 -7
index 7138556b206b..d5103a9eb302 100644
@@ -2245,7 +2245,6 @@ static const struct nf_hook_ops ip_vs_ops[] = {
 static int __net_init __ip_vs_init(struct net *net)
 {
 	struct netns_ipvs *ipvs;
-	int ret;
 
 	ipvs = net_generic(net, ip_vs_net_id);
 	if (ipvs == NULL)
@@ -2277,17 +2276,11 @@ static int __net_init __ip_vs_init(struct net *net)
 	if (ip_vs_sync_net_init(ipvs) < 0)
 		goto sync_fail;
 
-	ret = nf_register_net_hooks(net, ip_vs_ops, ARRAY_SIZE(ip_vs_ops));
-	if (ret < 0)
-		goto hook_fail;
-
 	return 0;
 /*
  * Error handling
  */
 
-hook_fail:
-	ip_vs_sync_net_cleanup(ipvs);
 sync_fail:
 	ip_vs_conn_net_cleanup(ipvs);
 conn_fail:
@@ -2317,6 +2310,19 @@ static void __net_exit __ip_vs_cleanup(struct net *net)
 	net->ipvs = NULL;
 }
 
+static int __net_init __ip_vs_dev_init(struct net *net)
+{
+	int ret;
+
+	ret = nf_register_net_hooks(net, ip_vs_ops, ARRAY_SIZE(ip_vs_ops));
+	if (ret < 0)
+		goto hook_fail;
+	return 0;
+
+hook_fail:
+	return ret;
+}
+
 static void __net_exit __ip_vs_dev_cleanup(struct net *net)
 {
 	struct netns_ipvs *ipvs = net_ipvs(net);
@@ -2336,6 +2342,7 @@ static struct pernet_operations ipvs_core_ops = {
 };
 
 static struct pernet_operations ipvs_core_dev_ops = {
+	.init = __ip_vs_dev_init,
 	.exit = __ip_vs_dev_cleanup,
 };
 
Signature · 1 line
2.20.1
author maintainer
#154 lore raw

PATCH 144/249 media: s5p-mfc: Make additional clocks optional

In reply to Sasha Levin

From: Marek Szyprowski <[email protected]>

[ Upstream commit e08efef8fe7db87206314c19b341612c719f891a ]

Since the beginning the second clock ('special', 'sclk') was optional and it is not available on some variants of Exynos SoCs (i.e. Exynos5420 with v7 of MFC hardware).

However commit 1bce6fb3edf1 ("[media] s5p-mfc: Rework clock handling") made handling of all specified clocks mandatory. This patch restores original behavior of the driver and fixes its operation on Exynos5420 SoCs.

Fixes 1bce6fb3edf1 ("[media] s5p-mfc: Rework clock handling")
---
 drivers/media/platform/s5p-mfc/s5p_mfc_pm.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/media/platform/s5p-mfc/s5p_mfc_pm.c +5 -0
index 2e62f8721fa5..7d52431c2c83 100644
@@ -34,6 +34,11 @@ int s5p_mfc_init_pm(struct s5p_mfc_dev *dev)
 	for (i = 0; i < pm->num_clocks; i++) {
 		pm->clocks[i] = devm_clk_get(pm->device, pm->clk_names[i]);
 		if (IS_ERR(pm->clocks[i])) {
+			/* additional clocks are optional */
+			if (i && PTR_ERR(pm->clocks[i]) == -ENOENT) {
+				pm->clocks[i] = NULL;
+				continue;
+			}
 			mfc_err("Failed to get clock: %s\n",
 				pm->clk_names[i]);
 			return PTR_ERR(pm->clocks[i]);
Signature · 7 lines
2.20.1


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
author maintainer
#155 lore raw

PATCH 145/249 media: i2c: fix warning same module names

In reply to Sasha Levin

From: Anders Roxell <[email protected]>

[ Upstream commit b2ce5617dad254230551feda3599f2cc68e53ad8 ]

When building with CONFIG_VIDEO_ADV7511 and CONFIG_DRM_I2C_ADV7511 enabled as loadable modules, we see the following warning:

drivers/gpu/drm/bridge/adv7511/adv7511.ko
drivers/media/i2c/adv7511.ko

Rework so that the file is named adv7511-v4l2.c.

---
 drivers/media/i2c/Makefile                      | 2 +-
 drivers/media/i2c/{adv7511.c => adv7511-v4l2.c} | 5 +++++
 2 files changed, 6 insertions(+), 1 deletion(-)
 rename drivers/media/i2c/{adv7511.c => adv7511-v4l2.c} (99%)
drivers/media/i2c/Makefile +1 -1
index d8ad9dad495d..fd4ea86dedd5 100644
@@ -35,7 +35,7 @@ obj-$(CONFIG_VIDEO_ADV748X) += adv748x/
 obj-$(CONFIG_VIDEO_ADV7604) += adv7604.o
 obj-$(CONFIG_VIDEO_ADV7842) += adv7842.o
 obj-$(CONFIG_VIDEO_AD9389B) += ad9389b.o
-obj-$(CONFIG_VIDEO_ADV7511) += adv7511.o
+obj-$(CONFIG_VIDEO_ADV7511) += adv7511-v4l2.o
 obj-$(CONFIG_VIDEO_VPX3220) += vpx3220.o
 obj-$(CONFIG_VIDEO_VS6624)  += vs6624.o
 obj-$(CONFIG_VIDEO_BT819) += bt819.o
drivers/media/i2c/adv7511-v4l2.c renamed from drivers/media/i2c/adv7511.c +5 -0
similarity index 99%
rename from drivers/media/i2c/adv7511.c
rename to drivers/media/i2c/adv7511-v4l2.c
index cec5ebb1c9e6..2ad6bdf1a9fc 100644
@@ -5,6 +5,11 @@
  * Copyright 2013 Cisco Systems, Inc. and/or its affiliates. All rights reserved.
  */
 
+/*
+ * This file is named adv7511-v4l2.c so it doesn't conflict with the Analog
+ * Device ADV7511 (config fragment CONFIG_DRM_I2C_ADV7511).
+ */
+
 
 #include <linux/kernel.h>
 #include <linux/module.h>
Signature · 1 line
2.20.1
author maintainer
#156 lore raw

PATCH 146/249 ntp: Limit TAI-UTC offset

In reply to Sasha Levin

From: Miroslav Lichvar <[email protected]>

[ Upstream commit d897a4ab11dc8a9fda50d2eccc081a96a6385998 ]

Don't allow the TAI-UTC offset of the system clock to be set by adjtimex() to a value larger than 100000 seconds.

This prevents an overflow in the conversion to int, prevents the CLOCK_TAI clock from getting too far ahead of the CLOCK_REALTIME clock, and it is still large enough to allow leap seconds to be inserted at the maximum rate currently supported by the kernel (once per day) for the next ~270 years, however unlikely it is that someone can survive a catastrophic event which slowed down the rotation of the Earth so much.

---
 kernel/time/ntp.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
kernel/time/ntp.c +3 -1
index 8de4f789dc1b..65eb796610dc 100644
@@ -43,6 +43,7 @@ static u64			tick_length_base;
 #define MAX_TICKADJ		500LL		/* usecs */
 #define MAX_TICKADJ_SCALED \
 	(((MAX_TICKADJ * NSEC_PER_USEC) << NTP_SCALE_SHIFT) / NTP_INTERVAL_FREQ)
+#define MAX_TAI_OFFSET		100000
 
 /*
  * phase-lock loop variables
@@ -691,7 +692,8 @@ static inline void process_adjtimex_modes(const struct __kernel_timex *txc,
 		time_constant = max(time_constant, 0l);
 	}
 
-	if (txc->modes & ADJ_TAI && txc->constant >= 0)
+	if (txc->modes & ADJ_TAI &&
+			txc->constant >= 0 && txc->constant <= MAX_TAI_OFFSET)
 		*time_tai = txc->constant;
 
 	if (txc->modes & ADJ_OFFSET)
Signature · 1 line
2.20.1
author maintainer
#157 lore raw

PATCH 147/249 timer_list: Guard procfs specific code

In reply to Sasha Levin

From: Nathan Huckleberry <[email protected]>

[ Upstream commit a9314773a91a1d3b36270085246a6715a326ff00 ]

With CONFIG_PROC_FS=n the following warning is emitted:

kernel/time/timer_list.c:361:36: warning: unused variable 'timer_list_sops' [-Wunused-const-variable]
static const struct seq_operations timer_list_sops = {

Add #ifdef guard around procfs specific code.

---
 kernel/time/timer_list.c | 36 +++++++++++++++++++-----------------
 1 file changed, 19 insertions(+), 17 deletions(-)
kernel/time/timer_list.c +19 -17
index 98ba50dcb1b2..acb326f5f50a 100644
@@ -282,23 +282,6 @@ static inline void timer_list_header(struct seq_file *m, u64 now)
 	SEQ_printf(m, "\n");
 }
 
-static int timer_list_show(struct seq_file *m, void *v)
-{
-	struct timer_list_iter *iter = v;
-
-	if (iter->cpu == -1 && !iter->second_pass)
-		timer_list_header(m, iter->now);
-	else if (!iter->second_pass)
-		print_cpu(m, iter->cpu, iter->now);
-#ifdef CONFIG_GENERIC_CLOCKEVENTS
-	else if (iter->cpu == -1 && iter->second_pass)
-		timer_list_show_tickdevices_header(m);
-	else
-		print_tickdevice(m, tick_get_device(iter->cpu), iter->cpu);
-#endif
-	return 0;
-}
-
 void sysrq_timer_list_show(void)
 {
 	u64 now = ktime_to_ns(ktime_get());
@@ -317,6 +300,24 @@ void sysrq_timer_list_show(void)
 	return;
 }
 
+#ifdef CONFIG_PROC_FS
+static int timer_list_show(struct seq_file *m, void *v)
+{
+	struct timer_list_iter *iter = v;
+
+	if (iter->cpu == -1 && !iter->second_pass)
+		timer_list_header(m, iter->now);
+	else if (!iter->second_pass)
+		print_cpu(m, iter->cpu, iter->now);
+#ifdef CONFIG_GENERIC_CLOCKEVENTS
+	else if (iter->cpu == -1 && iter->second_pass)
+		timer_list_show_tickdevices_header(m);
+	else
+		print_tickdevice(m, tick_get_device(iter->cpu), iter->cpu);
+#endif
+	return 0;
+}
+
 static void *move_iter(struct timer_list_iter *iter, loff_t offset)
 {
 	for (; offset; offset--) {
@@ -376,3 +377,4 @@ static int __init init_timer_list_procfs(void)
 	return 0;
 }
 __initcall(init_timer_list_procfs);
+#endif
Signature · 1 line
2.20.1
author maintainer
#158 lore raw

PATCH 148/249 media: mt9m111: fix fw-node refactoring

In reply to Sasha Levin

From: Robert Jarzmik <[email protected]>

[ Upstream commit 8d4e29a51a954b43e06d916772fa4f50b7e5bbd6 ]

In the patch refactoring the fw-node, the mt9m111 was broken for all platform_data based platforms, which were the first aim of this driver. Only the devicetree platform are still functional, probably because the testing was done on these.

The result is that -EINVAL is systematically return for such platforms, what this patch fixes.

[Sakari Ailus: Rework this to resolve a merge conflict and use dev_fwnode]

Fixes 98480d65c48c ("media: mt9m111: allow to setup pixclk polarity")
---
 drivers/media/i2c/mt9m111.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)
drivers/media/i2c/mt9m111.c +5 -3
index 362c3b93636e..5a642b5ad076 100644
@@ -1245,9 +1245,11 @@ static int mt9m111_probe(struct i2c_client *client,
 	if (!mt9m111)
 		return -ENOMEM;
 
-	ret = mt9m111_probe_fw(client, mt9m111);
-	if (ret)
-		return ret;
+	if (dev_fwnode(&client->dev)) {
+		ret = mt9m111_probe_fw(client, mt9m111);
+		if (ret)
+			return ret;
+	}
 
 	mt9m111->clk = v4l2_clk_get(&client->dev, "mclk");
 	if (IS_ERR(mt9m111->clk))
Signature · 1 line
2.20.1
author maintainer
#159 lore raw

PATCH 149/249 ASoC: soc-core: call snd_soc_unbind_card() under mutex_lock;

In reply to Sasha Levin

From: Kuninori Morimoto <[email protected]>

[ Upstream commit b545542a0b866f7975254e41c595836e9bc0ff2f ]

commit 34ac3c3eb8f0c07 ("ASoC: core: lock client_mutex while removing link components") added mutex_lock() at soc_remove_link_components().

Is is called from snd_soc_unbind_card()

snd_soc_unbind_card()
=> soc_remove_link_components()
soc_cleanup_card_resources()
soc_remove_dai_links()
=> soc_remove_link_components()

And, there are 2 way to call it.

(1)
snd_soc_unregister_component()
** mutex_lock()
snd_soc_component_del_unlocked()
=> snd_soc_unbind_card()
** mutex_unlock()

(2)
snd_soc_unregister_card()
=> snd_soc_unbind_card()

(1) case is already using mutex_lock() when it calles snd_soc_unbind_card(), thus, we will get lockdep warning.

commit 495f926c68ddb90 ("ASoC: core: Fix deadlock in snd_soc_instantiate_card()") tried to fixup it, but still not enough. We still have lockdep warning when we try unbind/bind.

We need mutex_lock() under snd_soc_unregister_card() instead of snd_remove_link_components()/snd_soc_unbind_card().

Fixes 34ac3c3eb8f0c07 ("ASoC: core: lock client_mutex while removing link components")
Fixes 495f926c68ddb90 ("ASoC: core: Fix deadlock in snd_soc_instantiate_card()")
---
 sound/soc/soc-core.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
sound/soc/soc-core.c +2 -2
index 41c0cfaf2db5..9138fcb15cd3 100644
@@ -2837,14 +2837,12 @@ static void snd_soc_unbind_card(struct snd_soc_card *card, bool unregister)
 		snd_soc_dapm_shutdown(card);
 		snd_soc_flush_all_delayed_work(card);
 
-		mutex_lock(&client_mutex);
 		/* remove all components used by DAI links on this card */
 		for_each_comp_order(order) {
 			for_each_card_rtds(card, rtd) {
 				soc_remove_link_components(card, rtd, order);
 			}
 		}
-		mutex_unlock(&client_mutex);
 
 		soc_cleanup_card_resources(card);
 		if (!unregister)
@@ -2863,7 +2861,9 @@ static void snd_soc_unbind_card(struct snd_soc_card *card, bool unregister)
  */
 int snd_soc_unregister_card(struct snd_soc_card *card)
 {
+	mutex_lock(&client_mutex);
 	snd_soc_unbind_card(card, true);
+	mutex_unlock(&client_mutex);
 	dev_dbg(card->dev, "ASoC: Unregistered card '%s'\n", card->name);
 
 	return 0;
Signature · 1 line
2.20.1
author maintainer
#160 lore raw

PATCH 150/249 acpi/arm64: ignore 5.1 FADTs that are reported as 5.0

In reply to Sasha Levin

From: Ard Biesheuvel <[email protected]>

[ Upstream commit 2af22f3ec3ca452f1e79b967f634708ff01ced8a ]

Some Qualcomm Snapdragon based laptops built to run Microsoft Windows are clearly ACPI 5.1 based, given that that is the first ACPI revision that supports ARM, and introduced the FADT 'arm_boot_flags' field, which has a non-zero field on those systems.

So in these cases, infer from the ARM boot flags that the FADT must be 5.1 or later, and treat it as 5.1.

---
 arch/arm64/kernel/acpi.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)
arch/arm64/kernel/acpi.c +7 -3
index 2804330c95dc..3a58e9db5cfe 100644
@@ -152,10 +152,14 @@ static int __init acpi_fadt_sanity_check(void)
 	 */
 	if (table->revision < 5 ||
 	   (table->revision == 5 && fadt->minor_revision < 1)) {
-		pr_err("Unsupported FADT revision %d.%d, should be 5.1+\n",
+		pr_err(FW_BUG "Unsupported FADT revision %d.%d, should be 5.1+\n",
 		       table->revision, fadt->minor_revision);
-		ret = -EINVAL;
-		goto out;
+
+		if (!fadt->arm_boot_flags) {
+			ret = -EINVAL;
+			goto out;
+		}
+		pr_err("FADT has ARM boot flags set, assuming 5.1\n");
 	}
 
 	if (!(fadt->flags & ACPI_FADT_HW_REDUCED)) {
Signature · 1 line
2.20.1
author maintainer
#161 lore raw

PATCH 151/249 media: coda: fix mpeg2 sequence number handling

In reply to Sasha Levin

From: Philipp Zabel <[email protected]>

[ Upstream commit 56d159a4ec6d8da7313aac6fcbb95d8fffe689ba ]

Sequence number handling assumed that the BIT processor frame number starts counting at 1, but this is not true for the MPEG-2 decoder, which starts at 0. Fix the sequence counter offset detection to handle this.

---
 drivers/media/platform/coda/coda-bit.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
drivers/media/platform/coda/coda-bit.c +4 -1
index 976f6aa69f41..1eeed34f300d 100644
@@ -1739,6 +1739,7 @@ static int __coda_start_decoding(struct coda_ctx *ctx)
 		v4l2_err(&dev->v4l2_dev, "CODA_COMMAND_SEQ_INIT timeout\n");
 		return ret;
 	}
+	ctx->sequence_offset = ~0U;
 	ctx->initialized = 1;
 
 	/* Update kfifo out pointer from coda bitstream read pointer */
@@ -2151,7 +2152,9 @@ static void coda_finish_decode(struct coda_ctx *ctx)
 		v4l2_err(&dev->v4l2_dev,
 			 "decoded frame index out of range: %d\n", decoded_idx);
 	} else {
-		val = coda_read(dev, CODA_RET_DEC_PIC_FRAME_NUM) - 1;
+		val = coda_read(dev, CODA_RET_DEC_PIC_FRAME_NUM);
+		if (ctx->sequence_offset == -1)
+			ctx->sequence_offset = val;
 		val -= ctx->sequence_offset;
 		spin_lock(&ctx->buffer_meta_lock);
 		if (!list_empty(&ctx->buffer_meta_list)) {
Signature · 1 line
2.20.1
author maintainer
#162 lore raw

PATCH 152/249 media: coda: fix last buffer handling in V4L2_ENC_CMD_STOP

In reply to Sasha Levin

From: Marco Felsch <[email protected]>

[ Upstream commit f3775f89852d167990b0d718587774cf00d22ac2 ]

coda_encoder_cmd() is racy, as the last scheduled picture run worker can still be in-flight while the ENC_CMD_STOP command is issued. Depending on the exact timing the sequence numbers might already be changed, but the last buffer might not have been put on the destination queue yet.

In this case the current implementation would prematurely wake the destination queue with last_buffer_dequeued=true, causing userspace to call streamoff before the last buffer is handled.

Close this race window by synchronizing with the pic_run_worker before doing the sequence check.

---
 drivers/media/platform/coda/coda-common.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/media/platform/coda/coda-common.c +2 -0
index 6238047273f2..68a585d3af91 100644
@@ -1024,6 +1024,8 @@ static int coda_encoder_cmd(struct file *file, void *fh,
 	/* Set the stream-end flag on this context */
 	ctx->bit_stream_param |= CODA_BIT_STREAM_END_FLAG;
 
+	flush_work(&ctx->pic_run_work);
+
 	/* If there is no buffer in flight, wake up */
 	if (!ctx->streamon_out || ctx->qsequence == ctx->osequence) {
 		dst_vq = v4l2_m2m_get_vq(ctx->fh.m2m_ctx,
Signature · 1 line
2.20.1
author maintainer
#163 lore raw

PATCH 153/249 media: coda: increment sequence offset for the last returned frame

In reply to Sasha Levin

From: Philipp Zabel <[email protected]>

[ Upstream commit b3b7d96817cdb8b6fc353867705275dce8f41ccc ]

If no more frames are decoded in bitstream end mode, and a previously decoded frame has been returned, the firmware still increments the frame number. To avoid a sequence number mismatch after decoder restart, increment the sequence_offset correction parameter.

---
 drivers/media/platform/coda/coda-bit.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/media/platform/coda/coda-bit.c +3 -0
index 1eeed34f300d..8c9743e067cf 100644
@@ -2147,6 +2147,9 @@ static void coda_finish_decode(struct coda_ctx *ctx)
 		else if (ctx->display_idx < 0)
 			ctx->hold = true;
 	} else if (decoded_idx == -2) {
+		if (ctx->display_idx >= 0 &&
+		    ctx->display_idx < ctx->num_internal_frames)
+			ctx->sequence_offset++;
 		/* no frame was decoded, we still return remaining buffers */
 	} else if (decoded_idx < 0 || decoded_idx >= ctx->num_internal_frames) {
 		v4l2_err(&dev->v4l2_dev,
Signature · 1 line
2.20.1
author maintainer
#164 lore raw

PATCH 154/249 media: vimc: cap: check v4l2_fill_pixfmt return value

In reply to Sasha Levin

From: André Almeida <[email protected]>

[ Upstream commit 77ae46e11df5c96bb4582633851f838f5d954df4 ]

v4l2_fill_pixfmt() returns -EINVAL if the pixelformat used as parameter is invalid or if the user is trying to use a multiplanar format with the singleplanar API. Currently, the vimc_cap_try_fmt_vid_cap() returns such value, but vimc_cap_s_fmt_vid_cap() is ignoring it. Fix that and returns an error value if vimc_cap_try_fmt_vid_cap() has failed.

---
 drivers/media/platform/vimc/vimc-capture.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
drivers/media/platform/vimc/vimc-capture.c +4 -1
index 946dc0908566..664855708fdf 100644
@@ -142,12 +142,15 @@ static int vimc_cap_s_fmt_vid_cap(struct file *file, void *priv,
 				  struct v4l2_format *f)
 {
 	struct vimc_cap_device *vcap = video_drvdata(file);
+	int ret;
 
 	/* Do not change the format while stream is on */
 	if (vb2_is_busy(&vcap->queue))
 		return -EBUSY;
 
-	vimc_cap_try_fmt_vid_cap(file, priv, f);
+	ret = vimc_cap_try_fmt_vid_cap(file, priv, f);
+	if (ret)
+		return ret;
 
 	dev_dbg(vcap->dev, "%s: format update: "
 		"old:%dx%d (0x%x, %d, %d, %d, %d) "
Signature · 1 line
2.20.1
author maintainer
#165 lore raw

PATCH 155/249 media: hdpvr: fix locking and a missing msleep

In reply to Sasha Levin

From: Hans Verkuil <[email protected]>

[ Upstream commit 6bc5a4a1927556ff9adce1aa95ea408c95453225 ]

This driver has three locking issues:

- The wait_event_interruptible() condition calls hdpvr_get_next_buffer(dev) which uses a mutex, which is not allowed. Rewrite with list_empty_careful() that doesn't need locking.

- In hdpvr_read() the call to hdpvr_stop_streaming() didn't lock io_mutex, but it should have since stop_streaming expects that.

- In hdpvr_device_release() io_mutex was locked when calling flush_work(), but there it shouldn't take that mutex since the work done by flush_work() also wants to lock that mutex.

There are also two other changes (suggested by Keith):

- msecs_to_jiffies(4000); (a NOP) should have been msleep(4000).
- Change v4l2_dbg to v4l2_info to always log if streaming had to be restarted.

---
 drivers/media/usb/hdpvr/hdpvr-video.c | 17 +++++++++++------
 1 file changed, 11 insertions(+), 6 deletions(-)
drivers/media/usb/hdpvr/hdpvr-video.c +11 -6
index 7580fc5f2f12..6a6405b80797 100644
@@ -435,7 +435,7 @@ static ssize_t hdpvr_read(struct file *file, char __user *buffer, size_t count,
 	/* wait for the first buffer */
 	if (!(file->f_flags & O_NONBLOCK)) {
 		if (wait_event_interruptible(dev->wait_data,
-					     hdpvr_get_next_buffer(dev)))
+					     !list_empty_careful(&dev->rec_buff_list)))
 			return -ERESTARTSYS;
 	}
 
@@ -461,10 +461,17 @@ static ssize_t hdpvr_read(struct file *file, char __user *buffer, size_t count,
 				goto err;
 			}
 			if (!err) {
-				v4l2_dbg(MSG_INFO, hdpvr_debug, &dev->v4l2_dev,
-					"timeout: restart streaming\n");
+				v4l2_info(&dev->v4l2_dev,
+					  "timeout: restart streaming\n");
+				mutex_lock(&dev->io_mutex);
 				hdpvr_stop_streaming(dev);
-				msecs_to_jiffies(4000);
+				mutex_unlock(&dev->io_mutex);
+				/*
+				 * The FW needs about 4 seconds after streaming
+				 * stopped before it is ready to restart
+				 * streaming.
+				 */
+				msleep(4000);
 				err = hdpvr_start_streaming(dev);
 				if (err) {
 					ret = err;
@@ -1127,9 +1134,7 @@ static void hdpvr_device_release(struct video_device *vdev)
 	struct hdpvr_device *dev = video_get_drvdata(vdev);
 
 	hdpvr_delete(dev);
-	mutex_lock(&dev->io_mutex);
 	flush_work(&dev->worker);
-	mutex_unlock(&dev->io_mutex);
 
 	v4l2_device_unregister(&dev->v4l2_dev);
 	v4l2_ctrl_handler_free(&dev->hdl);
Signature · 1 line
2.20.1
author maintainer
#166 lore raw

PATCH 156/249 net: stmmac: sun8i: force select external PHY when no internal one

In reply to Sasha Levin

From: Icenowy Zheng <[email protected]>

[ Upstream commit 0fec7e72ae1391bb2d7527efb54fe6ae88acabce ]

The PHY selection bit also exists on SoCs without an internal PHY; if it's set to 1 (internal PHY, default value) then the MAC will not make use of any PHY on such SoCs.

This problem appears when adapting for H6, which has no real internal PHY (the "internal PHY" on H6 is not on-die, but on a co-packaged AC200 chip, connected via RMII interface at GPIO bank A).

Force the PHY selection bit to 0 when the SOC doesn't have an internal PHY, to address the problem of a wrong default value.

---
 drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c +5 -0
index a69c34f605b1..98a15ba8be9f 100644
@@ -884,6 +884,11 @@ static int sun8i_dwmac_set_syscon(struct stmmac_priv *priv)
 		 * address. No need to mask it again.
 		 */
 		reg |= 1 << H3_EPHY_ADDR_SHIFT;
+	} else {
+		/* For SoCs without internal PHY the PHY selection bit should be
+		 * set to 0 (external PHY).
+		 */
+		reg &= ~H3_EPHY_SELECT;
 	}
 
 	if (!of_property_read_u32(node, "allwinner,tx-delay-ps", &val)) {
Signature · 1 line
2.20.1