Skip to contentSkip to conversation

bcache: only clear BTREE_NODE_dirty bit when it is set

PATCH In review Sasha Levin 2 participants 25 messages Sep 3, 2019 – Sep 12, 2019

Subsystems (10)
lore
Share & tools
lore

Sasha Levin

author maintainer
#0 lore raw

PATCH 1/23 bcache: only clear BTREE_NODE_dirty bit when it is set

From: Coly Li <[email protected]>

In bch_btree_cache_free() and btree_node_free(), BTREE_NODE_dirty is always set no matter btree node is dirty or not. The code looks like this,

	if (btree_node_dirty(b))
		btree_complete_write(b, btree_current_write(b));
	clear_bit(BTREE_NODE_dirty, &b->flags);

Indeed if btree_node_dirty(b) returns false, it means BTREE_NODE_dirty bit is cleared, then it is unnecessary to clear the bit again.

This patch only clears BTREE_NODE_dirty when btree_node_dirty(b) is true (the bit is set), to save a few CPU cycles.

---
 drivers/md/bcache/btree.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)
drivers/md/bcache/btree.c +6 -5
index 773f5fdad25fb..3fbadf2058a65 100644
@@ -778,10 +778,10 @@ void bch_btree_cache_free(struct cache_set *c)
 	while (!list_empty(&c->btree_cache)) {
 		b = list_first_entry(&c->btree_cache, struct btree, list);
 
-		if (btree_node_dirty(b))
+		if (btree_node_dirty(b)) {
 			btree_complete_write(b, btree_current_write(b));
-		clear_bit(BTREE_NODE_dirty, &b->flags);
-
+			clear_bit(BTREE_NODE_dirty, &b->flags);
+		}
 		mca_data_free(b);
 	}
 
@@ -1069,9 +1069,10 @@ static void btree_node_free(struct btree *b)
 
 	mutex_lock(&b->write_lock);
 
-	if (btree_node_dirty(b))
+	if (btree_node_dirty(b)) {
 		btree_complete_write(b, btree_current_write(b));
-	clear_bit(BTREE_NODE_dirty, &b->flags);
+		clear_bit(BTREE_NODE_dirty, &b->flags);
+	}
 
 	mutex_unlock(&b->write_lock);
 
Signature · 1 line
2.20.1
author maintainer
#1 lore raw

PATCH 2/23 bcache: add comments for mutex_lock(&b->write_lock)

From: Coly Li <[email protected]>

When accessing or modifying BTREE_NODE_dirty bit, it is not always necessary to acquire b->write_lock. In bch_btree_cache_free() and mca_reap() acquiring b->write_lock is necessary, and this patch adds comments to explain why mutex_lock(&b->write_lock) is necessary for checking or clearing BTREE_NODE_dirty bit there.

---
 drivers/md/bcache/btree.c | 10 ++++++++++
 1 file changed, 10 insertions(+)
drivers/md/bcache/btree.c +10 -0
index 3fbadf2058a65..9788b2ee6638f 100644
@@ -655,6 +655,11 @@ static int mca_reap(struct btree *b, unsigned int min_order, bool flush)
 		up(&b->io_mutex);
 	}
 
+	/*
+	 * BTREE_NODE_dirty might be cleared in btree_flush_btree() by
+	 * __bch_btree_node_write(). To avoid an extra flush, acquire
+	 * b->write_lock before checking BTREE_NODE_dirty bit.
+	 */
 	mutex_lock(&b->write_lock);
 	if (btree_node_dirty(b))
 		__bch_btree_node_write(b, &cl);
@@ -778,6 +783,11 @@ void bch_btree_cache_free(struct cache_set *c)
 	while (!list_empty(&c->btree_cache)) {
 		b = list_first_entry(&c->btree_cache, struct btree, list);
 
+		/*
+		 * This function is called by cache_set_free(), no I/O
+		 * request on cache now, it is unnecessary to acquire
+		 * b->write_lock before clearing BTREE_NODE_dirty anymore.
+		 */
 		if (btree_node_dirty(b)) {
 			btree_complete_write(b, btree_current_write(b));
 			clear_bit(BTREE_NODE_dirty, &b->flags);
Signature · 1 line
2.20.1
author maintainer
#2 lore raw

PATCH 3/23 bcache: fix race in btree_flush_write()

In reply to Sasha Levin

From: Coly Li <[email protected]>

There is a race between mca_reap(), btree_node_free() and journal code btree_flush_write(), which results very rare and strange deadlock or panic and are very hard to reproduce.

Let me explain how the race happens. In btree_flush_write() one btree node with oldest journal pin is selected, then it is flushed to cache device, the select-and-flush is a two steps operation. Between these two steps, there are something may happen inside the race window,
- The selected btree node was reaped by mca_reap() and allocated to other requesters for other btree node.
- The slected btree node was selected, flushed and released by mca shrink callback bch_mca_scan().
When btree_flush_write() tries to flush the selected btree node, firstly b->write_lock is held by mutex_lock(). If the race happens and the memory of selected btree node is allocated to other btree node, if that btree node's write_lock is held already, a deadlock very probably happens here. A worse case is the memory of the selected btree node is released, then all references to this btree node (e.g. b->write_lock) will trigger NULL pointer deference panic.

This race was introduced in commit cafe56359144 ("bcache: A block layer cache"), and enlarged by commit c4dc2497d50d ("bcache: fix high CPU occupancy during journal"), which selected 128 btree nodes and flushed them one-by-one in a quite long time period.

Such race is not easy to reproduce before. On a Lenovo SR650 server with 48 Xeon cores, and configure 1 NVMe SSD as cache device, a MD raid0 device assembled by 3 NVMe SSDs as backing device, this race can be observed around every 10,000 times btree_flush_write() gets called. Both deadlock and kernel panic all happened as aftermath of the race.

The idea of the fix is to add a btree flag BTREE_NODE_journal_flush. It is set when selecting btree nodes, and cleared after btree nodes flushed. Then when mca_reap() selects a btree node with this bit set, this btree node will be skipped. Since mca_reap() only reaps btree node without BTREE_NODE_journal_flush flag, such race is avoided.

Once corner case should be noticed, that is btree_node_free(). It might be called in some error handling code path. For example the following code piece from btree_split(),

        2149 err_free2:
        2150         bkey_put(b->c, &n2->key);
        2151         btree_node_free(n2);
        2152         rw_unlock(true, n2);
        2153 err_free1:
        2154         bkey_put(b->c, &n1->key);
        2155         btree_node_free(n1);
        2156         rw_unlock(true, n1);

At line 2151 and 2155, the btree node n2 and n1 are released without mac_reap(), so BTREE_NODE_journal_flush also needs to be checked here. If btree_node_free() is called directly in such error handling path, and the selected btree node has BTREE_NODE_journal_flush bit set, just delay for 1 us and retry again. In this case this btree node won't be skipped, just retry until the BTREE_NODE_journal_flush bit cleared, and free the btree node memory.

Fixes cafe56359144 ("bcache: A block layer cache")
---
 drivers/md/bcache/btree.c   | 28 +++++++++++++++++++++++++++-
 drivers/md/bcache/btree.h   |  2 ++
 drivers/md/bcache/journal.c |  7 +++++++
 3 files changed, 36 insertions(+), 1 deletion(-)
drivers/md/bcache/btree.c +27 -1
index 9788b2ee6638f..5cf3247e8afb2 100644
@@ -35,7 +35,7 @@
 #include <linux/rcupdate.h>
 #include <linux/sched/clock.h>
 #include <linux/rculist.h>
-
+#include <linux/delay.h>
 #include <trace/events/bcache.h>
 
 /*
@@ -655,12 +655,25 @@ static int mca_reap(struct btree *b, unsigned int min_order, bool flush)
 		up(&b->io_mutex);
 	}
 
+retry:
 	/*
 	 * BTREE_NODE_dirty might be cleared in btree_flush_btree() by
 	 * __bch_btree_node_write(). To avoid an extra flush, acquire
 	 * b->write_lock before checking BTREE_NODE_dirty bit.
 	 */
 	mutex_lock(&b->write_lock);
+	/*
+	 * If this btree node is selected in btree_flush_write() by journal
+	 * code, delay and retry until the node is flushed by journal code
+	 * and BTREE_NODE_journal_flush bit cleared by btree_flush_write().
+	 */
+	if (btree_node_journal_flush(b)) {
+		pr_debug("bnode %p is flushing by journal, retry", b);
+		mutex_unlock(&b->write_lock);
+		udelay(1);
+		goto retry;
+	}
+
 	if (btree_node_dirty(b))
 		__bch_btree_node_write(b, &cl);
 	mutex_unlock(&b->write_lock);
@@ -1077,7 +1090,20 @@ static void btree_node_free(struct btree *b)
 
 	BUG_ON(b == b->c->root);
 
+retry:
 	mutex_lock(&b->write_lock);
+	/*
+	 * If the btree node is selected and flushing in btree_flush_write(),
+	 * delay and retry until the BTREE_NODE_journal_flush bit cleared,
+	 * then it is safe to free the btree node here. Otherwise this btree
+	 * node will be in race condition.
+	 */
+	if (btree_node_journal_flush(b)) {
+		mutex_unlock(&b->write_lock);
+		pr_debug("bnode %p journal_flush set, retry", b);
+		udelay(1);
+		goto retry;
+	}
 
 	if (btree_node_dirty(b)) {
 		btree_complete_write(b, btree_current_write(b));
drivers/md/bcache/btree.h +2 -0
index d1c72ef64edf5..76cfd121a4861 100644
@@ -158,11 +158,13 @@ enum btree_flags {
 	BTREE_NODE_io_error,
 	BTREE_NODE_dirty,
 	BTREE_NODE_write_idx,
+	BTREE_NODE_journal_flush,
 };
 
 BTREE_FLAG(io_error);
 BTREE_FLAG(dirty);
 BTREE_FLAG(write_idx);
+BTREE_FLAG(journal_flush);
 
 static inline struct btree_write *btree_current_write(struct btree *b)
 {
drivers/md/bcache/journal.c +7 -0
index cae2aff5e27ae..33556acdcf9cd 100644
@@ -405,6 +405,7 @@ static void btree_flush_write(struct cache_set *c)
 retry:
 	best = NULL;
 
+	mutex_lock(&c->bucket_lock);
 	for_each_cached_btree(b, c, i)
 		if (btree_current_write(b)->journal) {
 			if (!best)
@@ -417,9 +418,14 @@ static void btree_flush_write(struct cache_set *c)
 		}
 
 	b = best;
+	if (b)
+		set_btree_node_journal_flush(b);
+	mutex_unlock(&c->bucket_lock);
+
 	if (b) {
 		mutex_lock(&b->write_lock);
 		if (!btree_current_write(b)->journal) {
+			clear_bit(BTREE_NODE_journal_flush, &b->flags);
 			mutex_unlock(&b->write_lock);
 			/* We raced */
 			atomic_long_inc(&c->retry_flush_write);
@@ -427,6 +433,7 @@ static void btree_flush_write(struct cache_set *c)
 		}
 
 		__bch_btree_node_write(b, NULL);
+		clear_bit(BTREE_NODE_journal_flush, &b->flags);
 		mutex_unlock(&b->write_lock);
 	}
 }
Signature · 1 line
2.20.1
author maintainer
#3 lore raw

PATCH 4/23 iwlwifi: add new cards for 22000 and fix struct name

In reply to Sasha Levin

From: Ihab Zhaika <[email protected]>

add few PCI ID'S for 22000 and fix the wrong name for one of the structs

---
 .../net/wireless/intel/iwlwifi/cfg/22000.c    | 20 ++++++++++++----
 .../net/wireless/intel/iwlwifi/iwl-config.h   |  5 ++--
 drivers/net/wireless/intel/iwlwifi/pcie/drv.c | 23 +++++++++++++------
 .../net/wireless/intel/iwlwifi/pcie/trans.c   |  4 ++--
 4 files changed, 37 insertions(+), 15 deletions(-)
drivers/net/wireless/intel/iwlwifi/cfg/22000.c +16 -4
index a9c846c59289e..650ca46efc48f 100644
@@ -241,6 +241,18 @@ const struct iwl_cfg iwl_ax101_cfg_qu_hr = {
 	.max_tx_agg_size = IEEE80211_MAX_AMPDU_BUF_HT,
 };
 
+const struct iwl_cfg iwl_ax201_cfg_qu_hr = {
+	.name = "Intel(R) Wi-Fi 6 AX201 160MHz",
+	.fw_name_pre = IWL_22000_QU_B_HR_B_FW_PRE,
+	IWL_DEVICE_22500,
+	/*
+	 * This device doesn't support receiving BlockAck with a large bitmap
+	 * so we need to restrict the size of transmitted aggregation to the
+	 * HT size; mac80211 would otherwise pick the HE max (256) by default.
+	 */
+	.max_tx_agg_size = IEEE80211_MAX_AMPDU_BUF_HT,
+};
+
 const struct iwl_cfg iwl_ax101_cfg_quz_hr = {
 	.name = "Intel(R) Wi-Fi 6 AX101",
 	.fw_name_pre = IWL_QUZ_A_HR_B_FW_PRE,
@@ -424,12 +436,12 @@ const struct iwl_cfg iwlax210_2ax_cfg_so_jf_a0 = {
 };
 
 const struct iwl_cfg iwlax210_2ax_cfg_so_hr_a0 = {
-	.name = "Intel(R) Wi-Fi 6 AX201 160MHz",
+	.name = "Intel(R) Wi-Fi 7 AX210 160MHz",
 	.fw_name_pre = IWL_22000_SO_A_HR_B_FW_PRE,
 	IWL_DEVICE_AX210,
 };
 
-const struct iwl_cfg iwlax210_2ax_cfg_so_gf_a0 = {
+const struct iwl_cfg iwlax211_2ax_cfg_so_gf_a0 = {
 	.name = "Intel(R) Wi-Fi 7 AX211 160MHz",
 	.fw_name_pre = IWL_22000_SO_A_GF_A_FW_PRE,
 	.uhb_supported = true,
@@ -443,8 +455,8 @@ const struct iwl_cfg iwlax210_2ax_cfg_ty_gf_a0 = {
 	IWL_DEVICE_AX210,
 };
 
-const struct iwl_cfg iwlax210_2ax_cfg_so_gf4_a0 = {
-	.name = "Intel(R) Wi-Fi 7 AX210 160MHz",
+const struct iwl_cfg iwlax411_2ax_cfg_so_gf4_a0 = {
+	.name = "Intel(R) Wi-Fi 7 AX411 160MHz",
 	.fw_name_pre = IWL_22000_SO_A_GF4_A_FW_PRE,
 	IWL_DEVICE_AX210,
 };
drivers/net/wireless/intel/iwlwifi/iwl-config.h +3 -2
index f3e69edf89071..29aaf649c13c3 100644
@@ -562,6 +562,7 @@ extern const struct iwl_cfg iwl_ax101_cfg_qu_hr;
 extern const struct iwl_cfg iwl_ax101_cfg_quz_hr;
 extern const struct iwl_cfg iwl22000_2ax_cfg_hr;
 extern const struct iwl_cfg iwl_ax200_cfg_cc;
+extern const struct iwl_cfg iwl_ax201_cfg_qu_hr;
 extern const struct iwl_cfg killer1650s_2ax_cfg_qu_b0_hr_b0;
 extern const struct iwl_cfg killer1650i_2ax_cfg_qu_b0_hr_b0;
 extern const struct iwl_cfg killer1650x_2ax_cfg;
@@ -580,9 +581,9 @@ extern const struct iwl_cfg iwl9560_2ac_cfg_qnj_jf_b0;
 extern const struct iwl_cfg iwl22000_2ax_cfg_qnj_hr_a0;
 extern const struct iwl_cfg iwlax210_2ax_cfg_so_jf_a0;
 extern const struct iwl_cfg iwlax210_2ax_cfg_so_hr_a0;
-extern const struct iwl_cfg iwlax210_2ax_cfg_so_gf_a0;
+extern const struct iwl_cfg iwlax211_2ax_cfg_so_gf_a0;
 extern const struct iwl_cfg iwlax210_2ax_cfg_ty_gf_a0;
-extern const struct iwl_cfg iwlax210_2ax_cfg_so_gf4_a0;
+extern const struct iwl_cfg iwlax411_2ax_cfg_so_gf4_a0;
 #endif /* CPTCFG_IWLMVM || CPTCFG_IWLFMAC */
 
 #endif /* __IWL_CONFIG_H__ */
drivers/net/wireless/intel/iwlwifi/pcie/drv.c +16 -7
index cd035061cdd55..2f3ee5769fdd3 100644
@@ -897,6 +897,7 @@ static const struct pci_device_id iwl_hw_card_ids[] = {
 	{IWL_PCI_DEVICE(0x02F0, 0x0310, iwl_ax101_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0x02F0, 0x1651, killer1650s_2ax_cfg_qu_b0_hr_b0)},
 	{IWL_PCI_DEVICE(0x02F0, 0x1652, killer1650i_2ax_cfg_qu_b0_hr_b0)},
+	{IWL_PCI_DEVICE(0x02F0, 0x2074, iwl_ax201_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0x02F0, 0x4070, iwl_ax101_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0x06F0, 0x0070, iwl_ax101_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0x06F0, 0x0074, iwl_ax101_cfg_qu_hr)},
@@ -905,6 +906,7 @@ static const struct pci_device_id iwl_hw_card_ids[] = {
 	{IWL_PCI_DEVICE(0x06F0, 0x0310, iwl_ax101_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0x06F0, 0x1651, killer1650s_2ax_cfg_qu_b0_hr_b0)},
 	{IWL_PCI_DEVICE(0x06F0, 0x1652, killer1650i_2ax_cfg_qu_b0_hr_b0)},
+	{IWL_PCI_DEVICE(0x06F0, 0x2074, iwl_ax201_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0x06F0, 0x4070, iwl_ax101_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0x2720, 0x0000, iwl_ax101_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0x2720, 0x0040, iwl_ax101_cfg_qu_hr)},
@@ -918,6 +920,7 @@ static const struct pci_device_id iwl_hw_card_ids[] = {
 	{IWL_PCI_DEVICE(0x2720, 0x1080, iwl_ax101_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0x2720, 0x1651, killer1650s_2ax_cfg_qu_b0_hr_b0)},
 	{IWL_PCI_DEVICE(0x2720, 0x1652, killer1650i_2ax_cfg_qu_b0_hr_b0)},
+	{IWL_PCI_DEVICE(0x2720, 0x2074, iwl_ax201_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0x2720, 0x4070, iwl_ax101_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0x34F0, 0x0040, iwl_ax101_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0x34F0, 0x0070, iwl_ax101_cfg_qu_hr)},
@@ -927,6 +930,7 @@ static const struct pci_device_id iwl_hw_card_ids[] = {
 	{IWL_PCI_DEVICE(0x34F0, 0x0310, iwl_ax101_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0x34F0, 0x1651, killer1650s_2ax_cfg_qu_b0_hr_b0)},
 	{IWL_PCI_DEVICE(0x34F0, 0x1652, killer1650i_2ax_cfg_qu_b0_hr_b0)},
+	{IWL_PCI_DEVICE(0x34F0, 0x2074, iwl_ax201_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0x34F0, 0x4070, iwl_ax101_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0x43F0, 0x0040, iwl_ax101_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0x43F0, 0x0070, iwl_ax101_cfg_qu_hr)},
@@ -935,6 +939,7 @@ static const struct pci_device_id iwl_hw_card_ids[] = {
 	{IWL_PCI_DEVICE(0x43F0, 0x007C, iwl_ax101_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0x43F0, 0x1651, killer1650s_2ax_cfg_qu_b0_hr_b0)},
 	{IWL_PCI_DEVICE(0x43F0, 0x1652, killer1650i_2ax_cfg_qu_b0_hr_b0)},
+	{IWL_PCI_DEVICE(0x43F0, 0x2074, iwl_ax201_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0x43F0, 0x4070, iwl_ax101_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0xA0F0, 0x0000, iwl_ax101_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0xA0F0, 0x0040, iwl_ax101_cfg_qu_hr)},
@@ -946,6 +951,7 @@ static const struct pci_device_id iwl_hw_card_ids[] = {
 	{IWL_PCI_DEVICE(0xA0F0, 0x0A10, iwl_ax101_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0xA0F0, 0x1651, killer1650s_2ax_cfg_qu_b0_hr_b0)},
 	{IWL_PCI_DEVICE(0xA0F0, 0x1652, killer1650i_2ax_cfg_qu_b0_hr_b0)},
+	{IWL_PCI_DEVICE(0xA0F0, 0x2074, iwl_ax201_cfg_qu_hr)},
 	{IWL_PCI_DEVICE(0xA0F0, 0x4070, iwl_ax101_cfg_qu_hr)},
 
 	{IWL_PCI_DEVICE(0x2723, 0x0080, iwl_ax200_cfg_cc)},
@@ -958,13 +964,16 @@ static const struct pci_device_id iwl_hw_card_ids[] = {
 	{IWL_PCI_DEVICE(0x2723, 0x4080, iwl_ax200_cfg_cc)},
 	{IWL_PCI_DEVICE(0x2723, 0x4088, iwl_ax200_cfg_cc)},
 
-	{IWL_PCI_DEVICE(0x2725, 0x0090, iwlax210_2ax_cfg_so_hr_a0)},
-	{IWL_PCI_DEVICE(0x7A70, 0x0090, iwlax210_2ax_cfg_so_hr_a0)},
-	{IWL_PCI_DEVICE(0x7A70, 0x0310, iwlax210_2ax_cfg_so_hr_a0)},
-	{IWL_PCI_DEVICE(0x2725, 0x0020, iwlax210_2ax_cfg_so_hr_a0)},
-	{IWL_PCI_DEVICE(0x2725, 0x0310, iwlax210_2ax_cfg_so_hr_a0)},
-	{IWL_PCI_DEVICE(0x2725, 0x0A10, iwlax210_2ax_cfg_so_hr_a0)},
-	{IWL_PCI_DEVICE(0x2725, 0x00B0, iwlax210_2ax_cfg_so_hr_a0)},
+	{IWL_PCI_DEVICE(0x2725, 0x0090, iwlax211_2ax_cfg_so_gf_a0)},
+	{IWL_PCI_DEVICE(0x2725, 0x0020, iwlax210_2ax_cfg_ty_gf_a0)},
+	{IWL_PCI_DEVICE(0x2725, 0x0310, iwlax210_2ax_cfg_ty_gf_a0)},
+	{IWL_PCI_DEVICE(0x2725, 0x0510, iwlax210_2ax_cfg_ty_gf_a0)},
+	{IWL_PCI_DEVICE(0x2725, 0x0A10, iwlax210_2ax_cfg_ty_gf_a0)},
+	{IWL_PCI_DEVICE(0x2725, 0x00B0, iwlax411_2ax_cfg_so_gf4_a0)},
+	{IWL_PCI_DEVICE(0x7A70, 0x0090, iwlax211_2ax_cfg_so_gf_a0)},
+	{IWL_PCI_DEVICE(0x7A70, 0x0310, iwlax211_2ax_cfg_so_gf_a0)},
+	{IWL_PCI_DEVICE(0x7A70, 0x0510, iwlax211_2ax_cfg_so_gf_a0)},
+	{IWL_PCI_DEVICE(0x7A70, 0x0A10, iwlax211_2ax_cfg_so_gf_a0)},
 
 #endif /* CONFIG_IWLMVM */
 
drivers/net/wireless/intel/iwlwifi/pcie/trans.c +2 -2
index 199eddea82a9a..51a3f77474e66 100644
@@ -3569,10 +3569,10 @@ struct iwl_trans *iwl_trans_pcie_alloc(struct pci_dev *pdev,
 			trans->cfg = &iwlax210_2ax_cfg_so_jf_a0;
 		} else if (CSR_HW_RF_ID_TYPE_CHIP_ID(trans->hw_rf_id) ==
 			   CSR_HW_RF_ID_TYPE_CHIP_ID(CSR_HW_RF_ID_TYPE_GF)) {
-			trans->cfg = &iwlax210_2ax_cfg_so_gf_a0;
+			trans->cfg = &iwlax211_2ax_cfg_so_gf_a0;
 		} else if (CSR_HW_RF_ID_TYPE_CHIP_ID(trans->hw_rf_id) ==
 			   CSR_HW_RF_ID_TYPE_CHIP_ID(CSR_HW_RF_ID_TYPE_GF4)) {
-			trans->cfg = &iwlax210_2ax_cfg_so_gf4_a0;
+			trans->cfg = &iwlax411_2ax_cfg_so_gf4_a0;
 		}
 	} else if (cfg == &iwl_ax101_cfg_qu_hr) {
 		if ((CSR_HW_RF_ID_TYPE_CHIP_ID(trans->hw_rf_id) ==
Signature · 1 line
2.20.1
author maintainer
#4 lore raw

PATCH 5/23 iwlwifi: add new cards for 22000 and change wrong structs

In reply to Sasha Levin

From: Ihab Zhaika <[email protected]>

add few PCI ID'S for 22000 and chainge few cards structs names

---
 .../net/wireless/intel/iwlwifi/cfg/22000.c    |  36 ++++++
 .../net/wireless/intel/iwlwifi/iwl-config.h   |   3 +
 drivers/net/wireless/intel/iwlwifi/pcie/drv.c | 122 +++++++++---------
 .../net/wireless/intel/iwlwifi/pcie/trans.c   |   5 +-
 4 files changed, 103 insertions(+), 63 deletions(-)
drivers/net/wireless/intel/iwlwifi/cfg/22000.c +36 -0
index 650ca46efc48f..e40fa12212b75 100644
@@ -265,6 +265,42 @@ const struct iwl_cfg iwl_ax101_cfg_quz_hr = {
 	.max_tx_agg_size = IEEE80211_MAX_AMPDU_BUF_HT,
 };
 
+const struct iwl_cfg iwl_ax201_cfg_quz_hr = {
+		.name = "Intel(R) Wi-Fi 6 AX201 160MHz",
+		.fw_name_pre = IWL_QUZ_A_HR_B_FW_PRE,
+		IWL_DEVICE_22500,
+		/*
+         * This device doesn't support receiving BlockAck with a large bitmap
+         * so we need to restrict the size of transmitted aggregation to the
+         * HT size; mac80211 would otherwise pick the HE max (256) by default.
+         */
+		.max_tx_agg_size = IEEE80211_MAX_AMPDU_BUF_HT,
+};
+
+const struct iwl_cfg iwl_ax1650s_cfg_quz_hr = {
+		.name = "Killer(R) Wi-Fi 6 AX1650s 160MHz Wireless Network Adapter (201D2W)",
+		.fw_name_pre = IWL_QUZ_A_HR_B_FW_PRE,
+		IWL_DEVICE_22500,
+		/*
+         * This device doesn't support receiving BlockAck with a large bitmap
+         * so we need to restrict the size of transmitted aggregation to the
+         * HT size; mac80211 would otherwise pick the HE max (256) by default.
+         */
+		.max_tx_agg_size = IEEE80211_MAX_AMPDU_BUF_HT,
+};
+
+const struct iwl_cfg iwl_ax1650i_cfg_quz_hr = {
+		.name = "Killer(R) Wi-Fi 6 AX1650i 160MHz Wireless Network Adapter (201NGW)",
+		.fw_name_pre = IWL_QUZ_A_HR_B_FW_PRE,
+		IWL_DEVICE_22500,
+		/*
+         * This device doesn't support receiving BlockAck with a large bitmap
+         * so we need to restrict the size of transmitted aggregation to the
+         * HT size; mac80211 would otherwise pick the HE max (256) by default.
+         */
+		.max_tx_agg_size = IEEE80211_MAX_AMPDU_BUF_HT,
+};
+
 const struct iwl_cfg iwl_ax200_cfg_cc = {
 	.name = "Intel(R) Wi-Fi 6 AX200 160MHz",
 	.fw_name_pre = IWL_CC_A_FW_PRE,
drivers/net/wireless/intel/iwlwifi/iwl-config.h +3 -0
index 29aaf649c13c3..dbe6437249f07 100644
@@ -563,6 +563,9 @@ extern const struct iwl_cfg iwl_ax101_cfg_quz_hr;
 extern const struct iwl_cfg iwl22000_2ax_cfg_hr;
 extern const struct iwl_cfg iwl_ax200_cfg_cc;
 extern const struct iwl_cfg iwl_ax201_cfg_qu_hr;
+extern const struct iwl_cfg iwl_ax201_cfg_quz_hr;
+extern const struct iwl_cfg iwl_ax1650i_cfg_quz_hr;
+extern const struct iwl_cfg iwl_ax1650s_cfg_quz_hr;
 extern const struct iwl_cfg killer1650s_2ax_cfg_qu_b0_hr_b0;
 extern const struct iwl_cfg killer1650i_2ax_cfg_qu_b0_hr_b0;
 extern const struct iwl_cfg killer1650x_2ax_cfg;
drivers/net/wireless/intel/iwlwifi/pcie/drv.c +61 -61

222 lines not on this page · raw message

drivers/net/wireless/intel/iwlwifi/pcie/trans.c +3 -2
index 51a3f77474e66..38ab24d962446 100644
@@ -3601,8 +3601,9 @@ struct iwl_trans *iwl_trans_pcie_alloc(struct pci_dev *pdev,
 	} else if (CSR_HW_RF_ID_TYPE_CHIP_ID(trans->hw_rf_id) ==
 		   CSR_HW_RF_ID_TYPE_CHIP_ID(CSR_HW_RF_ID_TYPE_HR) &&
 		   ((trans->cfg != &iwl_ax200_cfg_cc &&
-		    trans->cfg != &killer1650x_2ax_cfg &&
-		    trans->cfg != &killer1650w_2ax_cfg) ||
+		     trans->cfg != &killer1650x_2ax_cfg &&
+		     trans->cfg != &killer1650w_2ax_cfg &&
+		     trans->cfg != &iwl_ax201_cfg_quz_hr) ||
 		    trans->hw_rev == CSR_HW_REV_TYPE_QNJ_B0)) {
 		u32 hw_status;
 
Signature · 1 line
2.20.1
author maintainer
#5 lore raw

PATCH 6/23 iwlwifi: add new cards for 9000 and 20000 series

In reply to Sasha Levin

From: Ihab Zhaika <[email protected]>

add two new PCI ID's for 9000 and 20000 series

---
 drivers/net/wireless/intel/iwlwifi/pcie/drv.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/net/wireless/intel/iwlwifi/pcie/drv.c +2 -0
index 02af9073793af..09cf5f4fccb0f 100644
@@ -604,6 +604,7 @@ static const struct pci_device_id iwl_hw_card_ids[] = {
 	{IWL_PCI_DEVICE(0x2526, 0x40A4, iwl9460_2ac_cfg)},
 	{IWL_PCI_DEVICE(0x2526, 0x4234, iwl9560_2ac_cfg_soc)},
 	{IWL_PCI_DEVICE(0x2526, 0x42A4, iwl9462_2ac_cfg_soc)},
+	{IWL_PCI_DEVICE(0x2526, 0x6014, iwl9260_2ac_160_cfg)},
 	{IWL_PCI_DEVICE(0x2526, 0x8014, iwl9260_2ac_160_cfg)},
 	{IWL_PCI_DEVICE(0x2526, 0x8010, iwl9260_2ac_160_cfg)},
 	{IWL_PCI_DEVICE(0x2526, 0xA014, iwl9260_2ac_160_cfg)},
@@ -971,6 +972,7 @@ static const struct pci_device_id iwl_hw_card_ids[] = {
 	{IWL_PCI_DEVICE(0x7A70, 0x0310, iwlax211_2ax_cfg_so_gf_a0)},
 	{IWL_PCI_DEVICE(0x7A70, 0x0510, iwlax211_2ax_cfg_so_gf_a0)},
 	{IWL_PCI_DEVICE(0x7A70, 0x0A10, iwlax211_2ax_cfg_so_gf_a0)},
+	{IWL_PCI_DEVICE(0x7AF0, 0x0090, iwlax211_2ax_cfg_so_gf_a0)},
 	{IWL_PCI_DEVICE(0x7AF0, 0x0310, iwlax211_2ax_cfg_so_gf_a0)},
 	{IWL_PCI_DEVICE(0x7AF0, 0x0510, iwlax211_2ax_cfg_so_gf_a0)},
 	{IWL_PCI_DEVICE(0x7AF0, 0x0A10, iwlax211_2ax_cfg_so_gf_a0)},
Signature · 1 line
2.20.1
author maintainer
#6 lore raw

PATCH 7/23 iwlwifi: change 0x02F0 fw from qu to quz

In reply to Sasha Levin

From: Ihab Zhaika <[email protected]>

change the fw of 0x02F0 platform from qu to quz

---
 .../net/wireless/intel/iwlwifi/cfg/22000.c    |  88 +++++++++++++++
 .../net/wireless/intel/iwlwifi/iwl-config.h   |   6 ++
 drivers/net/wireless/intel/iwlwifi/pcie/drv.c | 100 +++++++++---------
 3 files changed, 144 insertions(+), 50 deletions(-)
drivers/net/wireless/intel/iwlwifi/cfg/22000.c +88 -0
index e40fa12212b75..93526dfaf7912 100644
@@ -82,6 +82,7 @@
 #define IWL_22000_HR_A0_FW_PRE		"iwlwifi-QuQnj-a0-hr-a0-"
 #define IWL_QU_B_JF_B_FW_PRE		"iwlwifi-Qu-b0-jf-b0-"
 #define IWL_QUZ_A_HR_B_FW_PRE		"iwlwifi-QuZ-a0-hr-b0-"
+#define IWL_QUZ_A_JF_B_FW_PRE		"iwlwifi-QuZ-a0-jf-b0-"
 #define IWL_QNJ_B_JF_B_FW_PRE		"iwlwifi-QuQnj-b0-jf-b0-"
 #define IWL_CC_A_FW_PRE			"iwlwifi-cc-a0-"
 #define IWL_22000_SO_A_JF_B_FW_PRE	"iwlwifi-so-a0-jf-b0-"
@@ -106,6 +107,8 @@
 	IWL_22000_HR_A0_FW_PRE __stringify(api) ".ucode"
 #define IWL_QUZ_A_HR_B_MODULE_FIRMWARE(api) \
 	IWL_QUZ_A_HR_B_FW_PRE __stringify(api) ".ucode"
+#define IWL_QUZ_A_JF_B_MODULE_FIRMWARE(api) \
+	IWL_QUZ_A_JF_B_FW_PRE __stringify(api) ".ucode"
 #define IWL_QU_B_JF_B_MODULE_FIRMWARE(api) \
 	IWL_QU_B_JF_B_FW_PRE __stringify(api) ".ucode"
 #define IWL_QNJ_B_JF_B_MODULE_FIRMWARE(api)		\
@@ -381,6 +384,90 @@ const struct iwl_cfg iwl9560_2ac_cfg_qnj_jf_b0 = {
 	.max_tx_agg_size = IEEE80211_MAX_AMPDU_BUF_HT,
 };
 
+const struct iwl_cfg iwl9560_2ac_cfg_quz_a0_jf_b0_soc = {
+	.name = "Intel(R) Wireless-AC 9560 160MHz",
+	.fw_name_pre = IWL_QUZ_A_JF_B_FW_PRE,
+	IWL_DEVICE_22500,
+	/*
+	 * This device doesn't support receiving BlockAck with a large bitmap
+	 * so we need to restrict the size of transmitted aggregation to the
+	 * HT size; mac80211 would otherwise pick the HE max (256) by default.
+	 */
+	.max_tx_agg_size = IEEE80211_MAX_AMPDU_BUF_HT,
+	.integrated = true,
+	.soc_latency = 5000,
+};
+
+const struct iwl_cfg iwl9560_2ac_160_cfg_quz_a0_jf_b0_soc = {
+	.name = "Intel(R) Wireless-AC 9560 160MHz",
+	.fw_name_pre = IWL_QUZ_A_JF_B_FW_PRE,
+	IWL_DEVICE_22500,
+	/*
+	 * This device doesn't support receiving BlockAck with a large bitmap
+	 * so we need to restrict the size of transmitted aggregation to the
+	 * HT size; mac80211 would otherwise pick the HE max (256) by default.
+	 */
+	.max_tx_agg_size = IEEE80211_MAX_AMPDU_BUF_HT,
+	.integrated = true,
+	.soc_latency = 5000,
+};
+
+const struct iwl_cfg iwl9461_2ac_cfg_quz_a0_jf_b0_soc = {
+	.name = "Intel(R) Dual Band Wireless AC 9461",
+	.fw_name_pre = IWL_QUZ_A_JF_B_FW_PRE,
+	IWL_DEVICE_22500,
+	/*
+	 * This device doesn't support receiving BlockAck with a large bitmap
+	 * so we need to restrict the size of transmitted aggregation to the
+	 * HT size; mac80211 would otherwise pick the HE max (256) by default.
+	 */
+	.max_tx_agg_size = IEEE80211_MAX_AMPDU_BUF_HT,
+	.integrated = true,
+	.soc_latency = 5000,
+};
+
+const struct iwl_cfg iwl9462_2ac_cfg_quz_a0_jf_b0_soc = {
+	.name = "Intel(R) Dual Band Wireless AC 9462",
+	.fw_name_pre = IWL_QUZ_A_JF_B_FW_PRE,
+	IWL_DEVICE_22500,
+	/*
+	 * This device doesn't support receiving BlockAck with a large bitmap
+	 * so we need to restrict the size of transmitted aggregation to the
+	 * HT size; mac80211 would otherwise pick the HE max (256) by default.
+	 */
+	.max_tx_agg_size = IEEE80211_MAX_AMPDU_BUF_HT,
+	.integrated = true,
+	.soc_latency = 5000,
+};
+
+const struct iwl_cfg iwl9560_killer_s_2ac_cfg_quz_a0_jf_b0_soc = {
+	.name = "Killer (R) Wireless-AC 1550s Wireless Network Adapter (9560NGW)",
+	.fw_name_pre = IWL_QUZ_A_JF_B_FW_PRE,
+	IWL_DEVICE_22500,
+	/*
+	 * This device doesn't support receiving BlockAck with a large bitmap
+	 * so we need to restrict the size of transmitted aggregation to the
+	 * HT size; mac80211 would otherwise pick the HE max (256) by default.
+	 */
+	.max_tx_agg_size = IEEE80211_MAX_AMPDU_BUF_HT,
+	.integrated = true,
+	.soc_latency = 5000,
+};
+
+const struct iwl_cfg iwl9560_killer_i_2ac_cfg_quz_a0_jf_b0_soc = {
+	.name = "Killer (R) Wireless-AC 1550i Wireless Network Adapter (9560NGW)",
+	.fw_name_pre = IWL_QUZ_A_JF_B_FW_PRE,
+	IWL_DEVICE_22500,
+	/*
+	 * This device doesn't support receiving BlockAck with a large bitmap
+	 * so we need to restrict the size of transmitted aggregation to the
+	 * HT size; mac80211 would otherwise pick the HE max (256) by default.
+	 */
+	.max_tx_agg_size = IEEE80211_MAX_AMPDU_BUF_HT,
+	.integrated = true,
+	.soc_latency = 5000,
+};
+
 const struct iwl_cfg killer1550i_2ac_cfg_qu_b0_jf_b0 = {
 	.name = "Killer (R) Wireless-AC 1550i Wireless Network Adapter (9560NGW)",
 	.fw_name_pre = IWL_QU_B_JF_B_FW_PRE,
@@ -505,6 +592,7 @@ MODULE_FIRMWARE(IWL_22000_HR_B_QNJ_MODULE_FIRMWARE(IWL_22000_UCODE_API_MAX));
 MODULE_FIRMWARE(IWL_22000_HR_A0_QNJ_MODULE_FIRMWARE(IWL_22000_UCODE_API_MAX));
 MODULE_FIRMWARE(IWL_QU_B_JF_B_MODULE_FIRMWARE(IWL_22000_UCODE_API_MAX));
 MODULE_FIRMWARE(IWL_QUZ_A_HR_B_MODULE_FIRMWARE(IWL_22000_UCODE_API_MAX));
+MODULE_FIRMWARE(IWL_QUZ_A_JF_B_MODULE_FIRMWARE(IWL_22000_UCODE_API_MAX));
 MODULE_FIRMWARE(IWL_QNJ_B_JF_B_MODULE_FIRMWARE(IWL_22000_UCODE_API_MAX));
 MODULE_FIRMWARE(IWL_CC_A_MODULE_FIRMWARE(IWL_22000_UCODE_API_MAX));
 MODULE_FIRMWARE(IWL_22000_SO_A_JF_B_MODULE_FIRMWARE(IWL_22000_UCODE_API_MAX));
drivers/net/wireless/intel/iwlwifi/iwl-config.h +6 -0
index dbe6437249f07..bc267bd2c3b0e 100644
@@ -540,14 +540,20 @@ extern const struct iwl_cfg iwl9260_killer_2ac_cfg;
 extern const struct iwl_cfg iwl9270_2ac_cfg;
 extern const struct iwl_cfg iwl9460_2ac_cfg;
 extern const struct iwl_cfg iwl9560_2ac_cfg;
+extern const struct iwl_cfg iwl9560_2ac_cfg_quz_a0_jf_b0_soc;
 extern const struct iwl_cfg iwl9560_2ac_160_cfg;
+extern const struct iwl_cfg iwl9560_2ac_160_cfg_quz_a0_jf_b0_soc;
 extern const struct iwl_cfg iwl9460_2ac_cfg_soc;
 extern const struct iwl_cfg iwl9461_2ac_cfg_soc;
+extern const struct iwl_cfg iwl9461_2ac_cfg_quz_a0_jf_b0_soc;
 extern const struct iwl_cfg iwl9462_2ac_cfg_soc;
+extern const struct iwl_cfg iwl9462_2ac_cfg_quz_a0_jf_b0_soc;
 extern const struct iwl_cfg iwl9560_2ac_cfg_soc;
 extern const struct iwl_cfg iwl9560_2ac_160_cfg_soc;
 extern const struct iwl_cfg iwl9560_killer_2ac_cfg_soc;
 extern const struct iwl_cfg iwl9560_killer_s_2ac_cfg_soc;
+extern const struct iwl_cfg iwl9560_killer_i_2ac_cfg_quz_a0_jf_b0_soc;
+extern const struct iwl_cfg iwl9560_killer_s_2ac_cfg_quz_a0_jf_b0_soc;
 extern const struct iwl_cfg iwl9460_2ac_cfg_shared_clk;
 extern const struct iwl_cfg iwl9461_2ac_cfg_shared_clk;
 extern const struct iwl_cfg iwl9462_2ac_cfg_shared_clk;
drivers/net/wireless/intel/iwlwifi/pcie/drv.c +50 -50

106 lines not on this page · raw message

Signature · 1 line
2.20.1
author maintainer
#7 lore raw

PATCH 8/23 iwlwifi: pcie: add support for qu c-step devices

In reply to Sasha Levin

From: Luca Coelho <[email protected]>

Add support for C-step devices. Currently we don't have a nice way of matching the step and choosing the proper configuration, so we need to switch the config structs one by one.

---
 .../net/wireless/intel/iwlwifi/cfg/22000.c    | 53 +++++++++++++++++++
 .../net/wireless/intel/iwlwifi/iwl-config.h   |  7 +++
 drivers/net/wireless/intel/iwlwifi/iwl-csr.h  |  2 +
 drivers/net/wireless/intel/iwlwifi/pcie/drv.c | 21 ++++++++
 4 files changed, 83 insertions(+)
drivers/net/wireless/intel/iwlwifi/cfg/22000.c +53 -0
index 93526dfaf7912..1f500cddb3a75 100644
@@ -80,7 +80,9 @@
 #define IWL_22000_QU_B_HR_B_FW_PRE	"iwlwifi-Qu-b0-hr-b0-"
 #define IWL_22000_HR_B_FW_PRE		"iwlwifi-QuQnj-b0-hr-b0-"
 #define IWL_22000_HR_A0_FW_PRE		"iwlwifi-QuQnj-a0-hr-a0-"
+#define IWL_QU_C_HR_B_FW_PRE		"iwlwifi-Qu-c0-hr-b0-"
 #define IWL_QU_B_JF_B_FW_PRE		"iwlwifi-Qu-b0-jf-b0-"
+#define IWL_QU_C_JF_B_FW_PRE		"iwlwifi-Qu-c0-jf-b0-"
 #define IWL_QUZ_A_HR_B_FW_PRE		"iwlwifi-QuZ-a0-hr-b0-"
 #define IWL_QUZ_A_JF_B_FW_PRE		"iwlwifi-QuZ-a0-jf-b0-"
 #define IWL_QNJ_B_JF_B_FW_PRE		"iwlwifi-QuQnj-b0-jf-b0-"
@@ -109,6 +111,8 @@
 	IWL_QUZ_A_HR_B_FW_PRE __stringify(api) ".ucode"
 #define IWL_QUZ_A_JF_B_MODULE_FIRMWARE(api) \
 	IWL_QUZ_A_JF_B_FW_PRE __stringify(api) ".ucode"
+#define IWL_QU_C_HR_B_MODULE_FIRMWARE(api) \
+	IWL_QU_C_HR_B_FW_PRE __stringify(api) ".ucode"
 #define IWL_QU_B_JF_B_MODULE_FIRMWARE(api) \
 	IWL_QU_B_JF_B_FW_PRE __stringify(api) ".ucode"
 #define IWL_QNJ_B_JF_B_MODULE_FIRMWARE(api)		\
@@ -256,6 +260,30 @@ const struct iwl_cfg iwl_ax201_cfg_qu_hr = {
 	.max_tx_agg_size = IEEE80211_MAX_AMPDU_BUF_HT,
 };
 
+const struct iwl_cfg iwl_ax101_cfg_qu_c0_hr_b0 = {
+	.name = "Intel(R) Wi-Fi 6 AX101",
+	.fw_name_pre = IWL_QU_C_HR_B_FW_PRE,
+	IWL_DEVICE_22500,
+	/*
+	 * This device doesn't support receiving BlockAck with a large bitmap
+	 * so we need to restrict the size of transmitted aggregation to the
+	 * HT size; mac80211 would otherwise pick the HE max (256) by default.
+	 */
+	.max_tx_agg_size = IEEE80211_MAX_AMPDU_BUF_HT,
+};
+
+const struct iwl_cfg iwl_ax201_cfg_qu_c0_hr_b0 = {
+	.name = "Intel(R) Wi-Fi 6 AX201 160MHz",
+	.fw_name_pre = IWL_QU_C_HR_B_FW_PRE,
+	IWL_DEVICE_22500,
+	/*
+	 * This device doesn't support receiving BlockAck with a large bitmap
+	 * so we need to restrict the size of transmitted aggregation to the
+	 * HT size; mac80211 would otherwise pick the HE max (256) by default.
+	 */
+	.max_tx_agg_size = IEEE80211_MAX_AMPDU_BUF_HT,
+};
+
 const struct iwl_cfg iwl_ax101_cfg_quz_hr = {
 	.name = "Intel(R) Wi-Fi 6 AX101",
 	.fw_name_pre = IWL_QUZ_A_HR_B_FW_PRE,
@@ -372,6 +400,30 @@ const struct iwl_cfg iwl9560_2ac_160_cfg_qu_b0_jf_b0 = {
 	IWL_DEVICE_22500,
 };
 
+const struct iwl_cfg iwl9461_2ac_cfg_qu_c0_jf_b0 = {
+	.name = "Intel(R) Wireless-AC 9461",
+	.fw_name_pre = IWL_QU_C_JF_B_FW_PRE,
+	IWL_DEVICE_22500,
+};
+
+const struct iwl_cfg iwl9462_2ac_cfg_qu_c0_jf_b0 = {
+	.name = "Intel(R) Wireless-AC 9462",
+	.fw_name_pre = IWL_QU_C_JF_B_FW_PRE,
+	IWL_DEVICE_22500,
+};
+
+const struct iwl_cfg iwl9560_2ac_cfg_qu_c0_jf_b0 = {
+	.name = "Intel(R) Wireless-AC 9560",
+	.fw_name_pre = IWL_QU_C_JF_B_FW_PRE,
+	IWL_DEVICE_22500,
+};
+
+const struct iwl_cfg iwl9560_2ac_160_cfg_qu_c0_jf_b0 = {
+	.name = "Intel(R) Wireless-AC 9560 160MHz",
+	.fw_name_pre = IWL_QU_C_JF_B_FW_PRE,
+	IWL_DEVICE_22500,
+};
+
 const struct iwl_cfg iwl9560_2ac_cfg_qnj_jf_b0 = {
 	.name = "Intel(R) Wireless-AC 9560 160MHz",
 	.fw_name_pre = IWL_QNJ_B_JF_B_FW_PRE,
@@ -590,6 +642,7 @@ MODULE_FIRMWARE(IWL_22000_HR_A_F0_QNJ_MODULE_FIRMWARE(IWL_22000_UCODE_API_MAX));
 MODULE_FIRMWARE(IWL_22000_HR_B_F0_QNJ_MODULE_FIRMWARE(IWL_22000_UCODE_API_MAX));
 MODULE_FIRMWARE(IWL_22000_HR_B_QNJ_MODULE_FIRMWARE(IWL_22000_UCODE_API_MAX));
 MODULE_FIRMWARE(IWL_22000_HR_A0_QNJ_MODULE_FIRMWARE(IWL_22000_UCODE_API_MAX));
+MODULE_FIRMWARE(IWL_QU_C_HR_B_MODULE_FIRMWARE(IWL_22000_UCODE_API_MAX));
 MODULE_FIRMWARE(IWL_QU_B_JF_B_MODULE_FIRMWARE(IWL_22000_UCODE_API_MAX));
 MODULE_FIRMWARE(IWL_QUZ_A_HR_B_MODULE_FIRMWARE(IWL_22000_UCODE_API_MAX));
 MODULE_FIRMWARE(IWL_QUZ_A_JF_B_MODULE_FIRMWARE(IWL_22000_UCODE_API_MAX));
drivers/net/wireless/intel/iwlwifi/iwl-config.h +7 -0
index bc267bd2c3b0e..1c1bf1b281cd9 100644
@@ -565,10 +565,13 @@ extern const struct iwl_cfg iwl22000_2ac_cfg_hr;
 extern const struct iwl_cfg iwl22000_2ac_cfg_hr_cdb;
 extern const struct iwl_cfg iwl22000_2ac_cfg_jf;
 extern const struct iwl_cfg iwl_ax101_cfg_qu_hr;
+extern const struct iwl_cfg iwl_ax101_cfg_qu_c0_hr_b0;
 extern const struct iwl_cfg iwl_ax101_cfg_quz_hr;
 extern const struct iwl_cfg iwl22000_2ax_cfg_hr;
 extern const struct iwl_cfg iwl_ax200_cfg_cc;
 extern const struct iwl_cfg iwl_ax201_cfg_qu_hr;
+extern const struct iwl_cfg iwl_ax201_cfg_qu_hr;
+extern const struct iwl_cfg iwl_ax201_cfg_qu_c0_hr_b0;
 extern const struct iwl_cfg iwl_ax201_cfg_quz_hr;
 extern const struct iwl_cfg iwl_ax1650i_cfg_quz_hr;
 extern const struct iwl_cfg iwl_ax1650s_cfg_quz_hr;
@@ -580,6 +583,10 @@ extern const struct iwl_cfg iwl9461_2ac_cfg_qu_b0_jf_b0;
 extern const struct iwl_cfg iwl9462_2ac_cfg_qu_b0_jf_b0;
 extern const struct iwl_cfg iwl9560_2ac_cfg_qu_b0_jf_b0;
 extern const struct iwl_cfg iwl9560_2ac_160_cfg_qu_b0_jf_b0;
+extern const struct iwl_cfg iwl9461_2ac_cfg_qu_c0_jf_b0;
+extern const struct iwl_cfg iwl9462_2ac_cfg_qu_c0_jf_b0;
+extern const struct iwl_cfg iwl9560_2ac_cfg_qu_c0_jf_b0;
+extern const struct iwl_cfg iwl9560_2ac_160_cfg_qu_c0_jf_b0;
 extern const struct iwl_cfg killer1550i_2ac_cfg_qu_b0_jf_b0;
 extern const struct iwl_cfg killer1550s_2ac_cfg_qu_b0_jf_b0;
 extern const struct iwl_cfg iwl22000_2ax_cfg_jf;
drivers/net/wireless/intel/iwlwifi/iwl-csr.h +2 -0
index 93da96a7247c3..cb4c5514a5560 100644
@@ -328,6 +328,8 @@ enum {
 #define CSR_HW_REV_TYPE_NONE		(0x00001F0)
 #define CSR_HW_REV_TYPE_QNJ		(0x0000360)
 #define CSR_HW_REV_TYPE_QNJ_B0		(0x0000364)
+#define CSR_HW_REV_TYPE_QU_B0		(0x0000334)
+#define CSR_HW_REV_TYPE_QU_C0		(0x0000338)
 #define CSR_HW_REV_TYPE_QUZ		(0x0000354)
 #define CSR_HW_REV_TYPE_HR_CDB		(0x0000340)
 #define CSR_HW_REV_TYPE_SO		(0x0000370)
drivers/net/wireless/intel/iwlwifi/pcie/drv.c +21 -0
index fe645380bd2fa..ea2a03d4bf55c 100644
@@ -1039,6 +1039,27 @@ static int iwl_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 		}
 		iwl_trans->cfg = cfg;
 	}
+
+	/*
+	 * This is a hack to switch from Qu B0 to Qu C0.  We need to
+	 * do this for all cfgs that use Qu B0.  All this code is in
+	 * urgent need for a refactor, but for now this is the easiest
+	 * thing to do to support Qu C-step.
+	 */
+	if (iwl_trans->hw_rev == CSR_HW_REV_TYPE_QU_C0) {
+		if (iwl_trans->cfg == &iwl_ax101_cfg_qu_hr)
+			iwl_trans->cfg = &iwl_ax101_cfg_qu_c0_hr_b0;
+		else if (iwl_trans->cfg == &iwl_ax201_cfg_qu_hr)
+			iwl_trans->cfg = &iwl_ax201_cfg_qu_c0_hr_b0;
+		else if (iwl_trans->cfg == &iwl9461_2ac_cfg_qu_b0_jf_b0)
+			iwl_trans->cfg = &iwl9461_2ac_cfg_qu_c0_jf_b0;
+		else if (iwl_trans->cfg == &iwl9462_2ac_cfg_qu_b0_jf_b0)
+			iwl_trans->cfg = &iwl9462_2ac_cfg_qu_c0_jf_b0;
+		else if (iwl_trans->cfg == &iwl9560_2ac_cfg_qu_b0_jf_b0)
+			iwl_trans->cfg = &iwl9560_2ac_cfg_qu_c0_jf_b0;
+		else if (iwl_trans->cfg == &iwl9560_2ac_160_cfg_qu_b0_jf_b0)
+			iwl_trans->cfg = &iwl9560_2ac_160_cfg_qu_c0_jf_b0;
+	}
 #endif
 
 	pci_set_drvdata(pdev, iwl_trans);
Signature · 1 line
2.20.1
author maintainer
#8 lore raw

PATCH 9/23 IB/rdmavt: Add new completion inline

In reply to Sasha Levin

From: Mike Marciniszyn <[email protected]>

There is opencoded send completion logic all over all the drivers.

We need to convert to this routine to enforce ordering
issues for completions. This routine fixes an ordering
issue where the read of the SWQE fields necessary for creating the completion can race with a post send if the post send catches a send queue at the edge of being full. Is is possible in that situation to read SWQE fields that are being written.

This new routine insures that SWQE fields are read prior to advancing the index that post send uses to determine queue fullness.

---
 include/rdma/rdmavt_qp.h | 72 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 72 insertions(+)
include/rdma/rdmavt_qp.h +72 -0
index 68e38c20afc04..6014f17669071 100644
@@ -737,6 +737,78 @@ static inline void rvt_put_qp_swqe(struct rvt_qp *qp, struct rvt_swqe *wqe)
 		atomic_dec(&ibah_to_rvtah(wqe->ud_wr.ah)->refcount);
 }
 
+/**
+ * rvt_qp_sqwe_incr - increment ring index
+ * @qp: the qp
+ * @val: the starting value
+ *
+ * Return: the new value wrapping as appropriate
+ */
+static inline u32
+rvt_qp_swqe_incr(struct rvt_qp *qp, u32 val)
+{
+	if (++val >= qp->s_size)
+		val = 0;
+	return val;
+}
+
+/**
+ * rvt_qp_complete_swqe - insert send completion
+ * @qp - the qp
+ * @wqe - the send wqe
+ * @opcode - wc operation (driver dependent)
+ * @status - completion status
+ *
+ * Update the s_last information, and then insert a send
+ * completion into the completion
+ * queue if the qp indicates it should be done.
+ *
+ * See IBTA 10.7.3.1 for info on completion
+ * control.
+ *
+ * Return: new last
+ */
+static inline u32
+rvt_qp_complete_swqe(struct rvt_qp *qp,
+		     struct rvt_swqe *wqe,
+		     enum ib_wc_opcode opcode,
+		     enum ib_wc_status status)
+{
+	bool need_completion;
+	u64 wr_id;
+	u32 byte_len, last;
+	int flags = wqe->wr.send_flags;
+
+	rvt_put_qp_swqe(qp, wqe);
+
+	need_completion =
+		!(flags & RVT_SEND_RESERVE_USED) &&
+		(!(qp->s_flags & RVT_S_SIGNAL_REQ_WR) ||
+		(flags & IB_SEND_SIGNALED) ||
+		status != IB_WC_SUCCESS);
+	if (need_completion) {
+		wr_id = wqe->wr.wr_id;
+		byte_len = wqe->length;
+		/* above fields required before writing s_last */
+	}
+	last = rvt_qp_swqe_incr(qp, qp->s_last);
+	/* see rvt_qp_is_avail() */
+	smp_store_release(&qp->s_last, last);
+	if (need_completion) {
+		struct ib_wc w = {
+			.wr_id = wr_id,
+			.status = status,
+			.opcode = opcode,
+			.qp = &qp->ibqp,
+			.byte_len = byte_len,
+		};
+
+		rvt_cq_enter(ibcq_to_rvtcq(qp->ibqp.send_cq), &w,
+			     status != IB_WC_SUCCESS);
+	}
+	return last;
+}
+
 extern const int  ib_rvt_state_ops[];
 
 struct rvt_dev_info;
Signature · 1 line
2.20.1
author maintainer
#9 lore raw

PATCH 10/23 IB/{rdmavt, qib, hfi1}: Convert to new completion API

In reply to Sasha Levin

From: Mike Marciniszyn <[email protected]>

Convert all completions to use the new completion routine that fixes a race between post send and completion where fields from a SWQE can be read after SWQE has been freed.

This patch also addresses issues reported in
https://marc.info/?l=linux-kernel&m=155656897409107&w=2.

The reserved operation path has no need for any barrier.

The barrier for the other path is addressed by the smp_load_acquire() barrier.

---
 drivers/infiniband/hw/hfi1/rc.c    | 26 ++++-----------------
 drivers/infiniband/hw/qib/qib_rc.c | 26 ++++-----------------
 drivers/infiniband/sw/rdmavt/qp.c  | 31 ++++++++-----------------
 include/rdma/rdmavt_qp.h           | 36 ------------------------------
 4 files changed, 17 insertions(+), 102 deletions(-)
drivers/infiniband/hw/hfi1/rc.c +4 -22
index 7c8cfb149da09..235bdbc706acc 100644
@@ -1830,23 +1830,14 @@ void hfi1_rc_send_complete(struct rvt_qp *qp, struct hfi1_opa_header *opah)
 	}
 
 	while (qp->s_last != qp->s_acked) {
-		u32 s_last;
-
 		wqe = rvt_get_swqe_ptr(qp, qp->s_last);
 		if (cmp_psn(wqe->lpsn, qp->s_sending_psn) >= 0 &&
 		    cmp_psn(qp->s_sending_psn, qp->s_sending_hpsn) <= 0)
 			break;
 		trdma_clean_swqe(qp, wqe);
 		rvt_qp_wqe_unreserve(qp, wqe);
-		s_last = qp->s_last;
-		trace_hfi1_qp_send_completion(qp, wqe, s_last);
-		if (++s_last >= qp->s_size)
-			s_last = 0;
-		qp->s_last = s_last;
-		/* see post_send() */
-		barrier();
-		rvt_put_qp_swqe(qp, wqe);
-		rvt_qp_swqe_complete(qp,
+		trace_hfi1_qp_send_completion(qp, wqe, qp->s_last);
+		rvt_qp_complete_swqe(qp,
 				     wqe,
 				     ib_hfi1_wc_opcode[wqe->wr.opcode],
 				     IB_WC_SUCCESS);
@@ -1890,19 +1881,10 @@ struct rvt_swqe *do_rc_completion(struct rvt_qp *qp,
 	trace_hfi1_rc_completion(qp, wqe->lpsn);
 	if (cmp_psn(wqe->lpsn, qp->s_sending_psn) < 0 ||
 	    cmp_psn(qp->s_sending_psn, qp->s_sending_hpsn) > 0) {
-		u32 s_last;
-
 		trdma_clean_swqe(qp, wqe);
-		rvt_put_qp_swqe(qp, wqe);
 		rvt_qp_wqe_unreserve(qp, wqe);
-		s_last = qp->s_last;
-		trace_hfi1_qp_send_completion(qp, wqe, s_last);
-		if (++s_last >= qp->s_size)
-			s_last = 0;
-		qp->s_last = s_last;
-		/* see post_send() */
-		barrier();
-		rvt_qp_swqe_complete(qp,
+		trace_hfi1_qp_send_completion(qp, wqe, qp->s_last);
+		rvt_qp_complete_swqe(qp,
 				     wqe,
 				     ib_hfi1_wc_opcode[wqe->wr.opcode],
 				     IB_WC_SUCCESS);
drivers/infiniband/hw/qib/qib_rc.c +4 -22
index 2ac4c67f5ba1a..8d9a94d6f6856 100644
@@ -921,20 +921,11 @@ void qib_rc_send_complete(struct rvt_qp *qp, struct ib_header *hdr)
 		rvt_add_retry_timer(qp);
 
 	while (qp->s_last != qp->s_acked) {
-		u32 s_last;
-
 		wqe = rvt_get_swqe_ptr(qp, qp->s_last);
 		if (qib_cmp24(wqe->lpsn, qp->s_sending_psn) >= 0 &&
 		    qib_cmp24(qp->s_sending_psn, qp->s_sending_hpsn) <= 0)
 			break;
-		s_last = qp->s_last;
-		if (++s_last >= qp->s_size)
-			s_last = 0;
-		qp->s_last = s_last;
-		/* see post_send() */
-		barrier();
-		rvt_put_qp_swqe(qp, wqe);
-		rvt_qp_swqe_complete(qp,
+		rvt_qp_complete_swqe(qp,
 				     wqe,
 				     ib_qib_wc_opcode[wqe->wr.opcode],
 				     IB_WC_SUCCESS);
@@ -972,21 +963,12 @@ static struct rvt_swqe *do_rc_completion(struct rvt_qp *qp,
 	 * is finished.
 	 */
 	if (qib_cmp24(wqe->lpsn, qp->s_sending_psn) < 0 ||
-	    qib_cmp24(qp->s_sending_psn, qp->s_sending_hpsn) > 0) {
-		u32 s_last;
-
-		rvt_put_qp_swqe(qp, wqe);
-		s_last = qp->s_last;
-		if (++s_last >= qp->s_size)
-			s_last = 0;
-		qp->s_last = s_last;
-		/* see post_send() */
-		barrier();
-		rvt_qp_swqe_complete(qp,
+	    qib_cmp24(qp->s_sending_psn, qp->s_sending_hpsn) > 0)
+		rvt_qp_complete_swqe(qp,
 				     wqe,
 				     ib_qib_wc_opcode[wqe->wr.opcode],
 				     IB_WC_SUCCESS);
-	} else
+	else
 		this_cpu_inc(*ibp->rvp.rc_delayed_comp);
 
 	qp->s_retry = qp->s_retry_cnt;
drivers/infiniband/sw/rdmavt/qp.c +9 -22
index c5a50614a6c63..cb9e171d7e7bf 100644
@@ -1856,10 +1856,9 @@ static inline int rvt_qp_is_avail(
 
 	/* see rvt_qp_wqe_unreserve() */
 	smp_mb__before_atomic();
-	reserved_used = atomic_read(&qp->s_reserved_used);
 	if (unlikely(reserved_op)) {
 		/* see rvt_qp_wqe_unreserve() */
-		smp_mb__before_atomic();
+		reserved_used = atomic_read(&qp->s_reserved_used);
 		if (reserved_used >= rdi->dparms.reserved_operations)
 			return -ENOMEM;
 		return 0;
@@ -1867,14 +1866,13 @@ static inline int rvt_qp_is_avail(
 	/* non-reserved operations */
 	if (likely(qp->s_avail))
 		return 0;
-	slast = READ_ONCE(qp->s_last);
+	/* See rvt_qp_complete_swqe() */
+	slast = smp_load_acquire(&qp->s_last);
 	if (qp->s_head >= slast)
 		avail = qp->s_size - (qp->s_head - slast);
 	else
 		avail = slast - qp->s_head;
 
-	/* see rvt_qp_wqe_unreserve() */
-	smp_mb__before_atomic();
 	reserved_used = atomic_read(&qp->s_reserved_used);
 	avail =  avail - 1 -
 		(rdi->dparms.reserved_operations - reserved_used);
@@ -2667,27 +2665,16 @@ void rvt_send_complete(struct rvt_qp *qp, struct rvt_swqe *wqe,
 		       enum ib_wc_status status)
 {
 	u32 old_last, last;
-	struct rvt_dev_info *rdi = ib_to_rvt(qp->ibqp.device);
+	struct rvt_dev_info *rdi;
 
 	if (!(ib_rvt_state_ops[qp->state] & RVT_PROCESS_OR_FLUSH_SEND))
 		return;
+	rdi = ib_to_rvt(qp->ibqp.device);
 
-	last = qp->s_last;
-	old_last = last;
-	trace_rvt_qp_send_completion(qp, wqe, last);
-	if (++last >= qp->s_size)
-		last = 0;
-	trace_rvt_qp_send_completion(qp, wqe, last);
-	qp->s_last = last;
-	/* See post_send() */
-	barrier();
-	rvt_put_qp_swqe(qp, wqe);
-
-	rvt_qp_swqe_complete(qp,
-			     wqe,
-			     rdi->wc_opcode[wqe->wr.opcode],
-			     status);
-
+	old_last = qp->s_last;
+	trace_rvt_qp_send_completion(qp, wqe, old_last);
+	last = rvt_qp_complete_swqe(qp, wqe, rdi->wc_opcode[wqe->wr.opcode],
+				    status);
 	if (qp->s_acked == old_last)
 		qp->s_acked = last;
 	if (qp->s_cur == old_last)
include/rdma/rdmavt_qp.h +0 -36
index 6014f17669071..84d0f36afc2f7 100644
@@ -565,42 +565,6 @@ static inline void rvt_qp_wqe_unreserve(
 
 extern const enum ib_wc_opcode ib_rvt_wc_opcode[];
 
-/**
- * rvt_qp_swqe_complete() - insert send completion
- * @qp - the qp
- * @wqe - the send wqe
- * @status - completion status
- *
- * Insert a send completion into the completion
- * queue if the qp indicates it should be done.
- *
- * See IBTA 10.7.3.1 for info on completion
- * control.
- */
-static inline void rvt_qp_swqe_complete(
-	struct rvt_qp *qp,
-	struct rvt_swqe *wqe,
-	enum ib_wc_opcode opcode,
-	enum ib_wc_status status)
-{
-	if (unlikely(wqe->wr.send_flags & RVT_SEND_RESERVE_USED))
-		return;
-	if (!(qp->s_flags & RVT_S_SIGNAL_REQ_WR) ||
-	    (wqe->wr.send_flags & IB_SEND_SIGNALED) ||
-	     status != IB_WC_SUCCESS) {
-		struct ib_wc wc;
-
-		memset(&wc, 0, sizeof(wc));
-		wc.wr_id = wqe->wr.wr_id;
-		wc.status = status;
-		wc.opcode = opcode;
-		wc.qp = &qp->ibqp;
-		wc.byte_len = wqe->length;
-		rvt_cq_enter(ibcq_to_rvtcq(qp->ibqp.send_cq), &wc,
-			     status != IB_WC_SUCCESS);
-	}
-}
-
 /*
  * Compare the lower 24 bits of the msn values.
  * Returns an integer <, ==, or > than zero.
Signature · 1 line
2.20.1
author maintainer
#10 lore raw

PATCH 11/23 IB/hfi1: Unreserve a flushed OPFN request

In reply to Sasha Levin

From: Kaike Wan <[email protected]>

When an OPFN request is flushed, the request is completed without unreserving itself from the send queue. Subsequently, when a new request is post sent, the following warning will be triggered:

WARNING: CPU: 4 PID: 8130 at rdmavt/qp.c:1761 rvt_post_send+0x72a/0x880 [rdmavt]
Call Trace:
[<ffffffffbbb61e41>] dump_stack+0x19/0x1b
[<ffffffffbb497688>] __warn+0xd8/0x100
[<ffffffffbb4977cd>] warn_slowpath_null+0x1d/0x20
[<ffffffffc01c941a>] rvt_post_send+0x72a/0x880 [rdmavt]
[<ffffffffbb4dcabe>] ? account_entity_dequeue+0xae/0xd0
[<ffffffffbb61d645>] ? __kmalloc+0x55/0x230
[<ffffffffc04e1a4c>] ib_uverbs_post_send+0x37c/0x5d0 [ib_uverbs]
[<ffffffffc04e5e36>] ? rdma_lookup_put_uobject+0x26/0x60 [ib_uverbs]
[<ffffffffc04dbce6>] ib_uverbs_write+0x286/0x460 [ib_uverbs]
[<ffffffffbb6f9457>] ? security_file_permission+0x27/0xa0
[<ffffffffbb641650>] vfs_write+0xc0/0x1f0
[<ffffffffbb64246f>] SyS_write+0x7f/0xf0
[<ffffffffbbb74ddb>] system_call_fastpath+0x22/0x27

This patch fixes the problem by moving rvt_qp_wqe_unreserve() into rvt_qp_complete_swqe() to simplify the code and make it less error-prone.

---
 drivers/infiniband/hw/hfi1/rc.c | 2 --
 include/rdma/rdmavt_qp.h        | 9 ++++-----
 2 files changed, 4 insertions(+), 7 deletions(-)
drivers/infiniband/hw/hfi1/rc.c +0 -2
index 235bdbc706acc..5c0d90418e8c4 100644
@@ -1835,7 +1835,6 @@ void hfi1_rc_send_complete(struct rvt_qp *qp, struct hfi1_opa_header *opah)
 		    cmp_psn(qp->s_sending_psn, qp->s_sending_hpsn) <= 0)
 			break;
 		trdma_clean_swqe(qp, wqe);
-		rvt_qp_wqe_unreserve(qp, wqe);
 		trace_hfi1_qp_send_completion(qp, wqe, qp->s_last);
 		rvt_qp_complete_swqe(qp,
 				     wqe,
@@ -1882,7 +1881,6 @@ struct rvt_swqe *do_rc_completion(struct rvt_qp *qp,
 	if (cmp_psn(wqe->lpsn, qp->s_sending_psn) < 0 ||
 	    cmp_psn(qp->s_sending_psn, qp->s_sending_hpsn) > 0) {
 		trdma_clean_swqe(qp, wqe);
-		rvt_qp_wqe_unreserve(qp, wqe);
 		trace_hfi1_qp_send_completion(qp, wqe, qp->s_last);
 		rvt_qp_complete_swqe(qp,
 				     wqe,
include/rdma/rdmavt_qp.h +4 -5
index 84d0f36afc2f7..85544777587db 100644
@@ -540,7 +540,7 @@ static inline void rvt_qp_wqe_reserve(
 /**
  * rvt_qp_wqe_unreserve - clean reserved operation
  * @qp - the rvt qp
- * @wqe - the send wqe
+ * @flags - send wqe flags
  *
  * This decrements the reserve use count.
  *
@@ -552,11 +552,9 @@ static inline void rvt_qp_wqe_reserve(
  * the compiler does not juggle the order of the s_last
  * ring index and the decrementing of s_reserved_used.
  */
-static inline void rvt_qp_wqe_unreserve(
-	struct rvt_qp *qp,
-	struct rvt_swqe *wqe)
+static inline void rvt_qp_wqe_unreserve(struct rvt_qp *qp, int flags)
 {
-	if (unlikely(wqe->wr.send_flags & RVT_SEND_RESERVE_USED)) {
+	if (unlikely(flags & RVT_SEND_RESERVE_USED)) {
 		atomic_dec(&qp->s_reserved_used);
 		/* insure no compiler re-order up to s_last change */
 		smp_mb__after_atomic();
@@ -743,6 +741,7 @@ rvt_qp_complete_swqe(struct rvt_qp *qp,
 	u32 byte_len, last;
 	int flags = wqe->wr.send_flags;
 
+	rvt_qp_wqe_unreserve(qp, flags);
 	rvt_put_qp_swqe(qp, wqe);
 
 	need_completion =
Signature · 1 line
2.20.1
author maintainer
#11 lore raw

PATCH 12/23 drm/i915: Disable SAMPLER_STATE prefetching on all Gen11 steppings.

In reply to Sasha Levin

From: Kenneth Graunke <[email protected]>

[ Upstream commit 248f883db61283b4f5a1c92a5e27277377b09f16 ]

The Demand Prefetch workaround (binding table prefetching) only applies to Icelake A0/B0. But the Sampler Prefetch workaround needs to be applied to all Gen11 steppings, according to a programming note in the SARCHKMD documentation.

Using the Intel Gallium driver, I have seen intermittent failures in the dEQP-GLES31.functional.copy_image.non_compressed.* tests. After applying this workaround, the tests reliably pass.

v2: Remove the overlap with a pre-production w/a

BSpec: 9663

---
 drivers/gpu/drm/i915/intel_workarounds.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/intel_workarounds.c +6 -2
index 841b8e515f4d6..2fb70fab2d1c6 100644
@@ -1167,8 +1167,12 @@ rcs_engine_wa_init(struct intel_engine_cs *engine, struct i915_wa_list *wal)
 		if (IS_ICL_REVID(i915, ICL_REVID_A0, ICL_REVID_B0))
 			wa_write_or(wal,
 				    GEN7_SARCHKMD,
-				    GEN7_DISABLE_DEMAND_PREFETCH |
-				    GEN7_DISABLE_SAMPLER_PREFETCH);
+				    GEN7_DISABLE_DEMAND_PREFETCH);
+
+		/* Wa_1606682166:icl */
+		wa_write_or(wal,
+			    GEN7_SARCHKMD,
+			    GEN7_DISABLE_SAMPLER_PREFETCH);
 	}
 
 	if (IS_GEN_RANGE(i915, 9, 11)) {
Signature · 6 lines
2.20.1

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
author maintainer
#12 lore raw

PATCH 13/23 drm/i915/userptr: Acquire the page lock around set_page_dirty()

In reply to Sasha Levin

From: Chris Wilson <[email protected]>

[ Upstream commit aa56a292ce623734ddd30f52d73f527d1f3529b5 ]

set_page_dirty says:

For pages with a mapping this should be done under the page lock for the benefit of asynchronous memory errors who prefer a consistent dirty state. This rule can be broken in some special cases, but should be better not to.

Under those rules, it is only safe for us to use the plain set_page_dirty calls for shmemfs/anonymous memory. Userptr may be used with real mappings and so needs to use the locked version (set_page_dirty_lock).

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=203317

Fixes 5cc9ed4b9a7a ("drm/i915: Introduce mapping of user pages into video memory (userptr) ioctl")

References: 6dcc693bc57f ("ext4: warn when page is dirtied without buffers")

---
 drivers/gpu/drm/i915/i915_gem_userptr.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_gem_userptr.c +9 -1
index 8079ea3af1039..b1fc15c7f5997 100644
@@ -678,7 +678,15 @@ i915_gem_userptr_put_pages(struct drm_i915_gem_object *obj,
 
 	for_each_sgt_page(page, sgt_iter, pages) {
 		if (obj->mm.dirty)
-			set_page_dirty(page);
+			/*
+			 * As this may not be anonymous memory (e.g. shmem)
+			 * but exist on a real mapping, we have to lock
+			 * the page in order to dirty it -- holding
+			 * the page reference is not sufficient to
+			 * prevent the inode from being truncated.
+			 * Play safe and take the lock.
+			 */
+			set_page_dirty_lock(page);
 
 		mark_page_accessed(page);
 		put_page(page);
Signature · 6 lines
2.20.1

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#13 lore raw

Re: [PATCH AUTOSEL 5.2 13/23] drm/i915/userptr: Acquire the page lock around set_page_dirty()

Den 03-09-2019 kl. 19:24, skrev Sasha Levin:

Quoted 42 lines

From: Chris Wilson <[email protected]>

[ Upstream commit aa56a292ce623734ddd30f52d73f527d1f3529b5 ]

set_page_dirty says:

For pages with a mapping this should be done under the page lock for the benefit of asynchronous memory errors who prefer a consistent dirty state. This rule can be broken in some special cases, but should be better not to.

Under those rules, it is only safe for us to use the plain set_page_dirty calls for shmemfs/anonymous memory. Userptr may be used with real mappings and so needs to use the locked version (set_page_dirty_lock).

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=203317

Fixes 5cc9ed4b9a7a ("drm/i915: Introduce mapping of user pages into video memory (userptr) ioctl")

References: 6dcc693bc57f ("ext4: warn when page is dirtied without buffers")

---
  drivers/gpu/drm/i915/i915_gem_userptr.c | 10 +++++++++-
  1 file changed, 9 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_gem_userptr.c +2 -1
index 8079ea3af1039..b1fc15c7f5997 100644
@@ -678,7 +678,15 @@ i915_gem_userptr_put_pages(struct drm_i915_gem_object *obj,
 
  	for_each_sgt_page(page, sgt_iter, pages) {
  		if (obj->mm.dirty)
-			set_page_dirty(page);
+			/*
+			 * As this may not be anonymous memory (e.g. shmem)

drivers/gpu/drm/i915/i915_gem_userptr.c i915_gem_userptr_put_pages(struct drm_i915_gem_object *obj, · line 692

Quoted 10 lines
drivers/gpu/drm/i915/i915_gem_userptr.c +7 -0
+			 * but exist on a real mapping, we have to lock
+			 * the page in order to dirty it -- holding
+			 * the page reference is not sufficient to
+			 * prevent the inode from being truncated.
+			 * Play safe and take the lock.
+			 */
+			set_page_dirty_lock(page);
 
  		mark_page_accessed(page);
  		put_page(page);

Please drop this one from all 5.2 and 4.19 stable queues

It has now been reverted in Linus tree:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=505a8ec7e11ae5236c4a154a1e24ef49a8349600

author maintainer
#14 lore raw

Re: [PATCH AUTOSEL 5.2 13/23] drm/i915/userptr: Acquire the page lock around set_page_dirty()

Quoted 55 lines · Thomas Backlund

Den 03-09-2019 kl. 19:24, skrev Sasha Levin:

Quoted 40 lines

From: Chris Wilson <[email protected]>

[ Upstream commit aa56a292ce623734ddd30f52d73f527d1f3529b5 ]

set_page_dirty says:

For pages with a mapping this should be done under the page lock for the benefit of asynchronous memory errors who prefer a consistent dirty state. This rule can be broken in some special cases, but should be better not to.

Under those rules, it is only safe for us to use the plain set_page_dirty calls for shmemfs/anonymous memory. Userptr may be used with real mappings and so needs to use the locked version (set_page_dirty_lock).

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=203317

Fixes 5cc9ed4b9a7a ("drm/i915: Introduce mapping of user pages into video memory (userptr) ioctl")

References: 6dcc693bc57f ("ext4: warn when page is dirtied without buffers")

---
 drivers/gpu/drm/i915/i915_gem_userptr.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_gem_userptr.c +2 -1
index 8079ea3af1039..b1fc15c7f5997 100644
@@ -678,7 +678,15 @@ i915_gem_userptr_put_pages(struct drm_i915_gem_object *obj,
 	for_each_sgt_page(page, sgt_iter, pages) {
 		if (obj->mm.dirty)
-			set_page_dirty(page);
+			/*
+			 * As this may not be anonymous memory (e.g. shmem)

drivers/gpu/drm/i915/i915_gem_userptr.c i915_gem_userptr_put_pages(struct drm_i915_gem_object *obj, · line 691

Quoted 10 lines
drivers/gpu/drm/i915/i915_gem_userptr.c +7 -0
+			 * but exist on a real mapping, we have to lock
+			 * the page in order to dirty it -- holding
+			 * the page reference is not sufficient to
+			 * prevent the inode from being truncated.
+			 * Play safe and take the lock.
+			 */
+			set_page_dirty_lock(page);
 		mark_page_accessed(page);
 		put_page(page);
 

Please drop this one from all 5.2 and 4.19 stable queues

Quoted 2 lines

It has now been reverted in Linus tree:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=505a8ec7e11ae5236c4a154a1e24ef49a8349600

Now dropped, thank you.

--
Thanks,
Sasha
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

author maintainer
#15 lore raw

PATCH 14/23 drm/i915: Make sure cdclk is high enough for DP audio on VLV/CHV

In reply to Sasha Levin

From: Ville Syrjälä <[email protected]>

[ Upstream commit a8f196a0fa6391a436f63f360a1fb57031fdf26c ]

On VLV/CHV there is some kind of linkage between the cdclk frequency and the DP link frequency. The spec says:
"For DP audio configuration, cdclk frequency shall be set to
meet the following requirements:

 DP Link Frequency(MHz) | Cdclk frequency(MHz)
 270                    | 320 or higher
 162                    | 200 or higher"

I suspect that would more accurately be expressed as "cdclk >= DP link clock", and in any case we can express it like that in the code because of the limited set of cdclk (200, 266, 320, 400 MHz) and link frequencies (162 and 270 MHz) we support.

Without this we can end up in a situation where the cdclk is too low and enabling DP audio will kill the pipe. Happens eg. with 2560x1440 modes where the 266MHz cdclk is sufficient to pump the pixels (241.5 MHz dotclock) but is too low for the DP audio due to the link frequency being 270 MHz.

v2: Spell out the cdclk and link frequencies we actually support

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=111149

---
 drivers/gpu/drm/i915/intel_cdclk.c | 11 +++++++++++
 1 file changed, 11 insertions(+)
drivers/gpu/drm/i915/intel_cdclk.c +11 -0
index ae40a8679314e..fd5236da039fb 100644
@@ -2269,6 +2269,17 @@ int intel_crtc_compute_min_cdclk(const struct intel_crtc_state *crtc_state)
 	if (crtc_state->has_audio && INTEL_GEN(dev_priv) >= 9)
 		min_cdclk = max(2 * 96000, min_cdclk);
 
+	/*
+	 * "For DP audio configuration, cdclk frequency shall be set to
+	 *  meet the following requirements:
+	 *  DP Link Frequency(MHz) | Cdclk frequency(MHz)
+	 *  270                    | 320 or higher
+	 *  162                    | 200 or higher"
+	 */
+	if ((IS_VALLEYVIEW(dev_priv) || IS_CHERRYVIEW(dev_priv)) &&
+	    intel_crtc_has_dp_encoder(crtc_state) && crtc_state->has_audio)
+		min_cdclk = max(crtc_state->port_clock, min_cdclk);
+
 	/*
 	 * On Valleyview some DSI panels lose (v|h)sync when the clock is lower
 	 * than 320000KHz.
Signature · 1 line
2.20.1
author maintainer
#16 lore raw

PATCH 15/23 mmc: sdhci-sprd: Fix the incorrect soft reset operation when runtime resuming

In reply to Sasha Levin

From: Baolin Wang <[email protected]>

[ Upstream commit c6303c5d52d5ec3e5bce2e6a5480fa2a1baa45e6 ]

The SD host controller specification defines 3 types software reset:
software reset for data line, software reset for command line and software reset for all. Software reset for all means this reset affects the entire Host controller except for the card detection circuit.

In sdhci_runtime_resume_host() we always do a software "reset for all", which causes the Spreadtrum variant controller to work abnormally after resuming. To fix the problem, let's do a software reset for the data and the command part, rather than "for all".

However, as sdhci_runtime_resume() is a common sdhci function and we don't want to change the behaviour for other variants, let's introduce a new in-parameter for it. This enables the caller to decide if a "reset for all" shall be done or not.

Fixes fb8bd90f83c4 ("mmc: sdhci-sprd: Add Spreadtrum's initial host controller")
---
 drivers/mmc/host/sdhci-acpi.c      | 2 +-
 drivers/mmc/host/sdhci-esdhc-imx.c | 2 +-
 drivers/mmc/host/sdhci-of-at91.c   | 2 +-
 drivers/mmc/host/sdhci-pci-core.c  | 4 ++--
 drivers/mmc/host/sdhci-pxav3.c     | 2 +-
 drivers/mmc/host/sdhci-s3c.c       | 2 +-
 drivers/mmc/host/sdhci-sprd.c      | 2 +-
 drivers/mmc/host/sdhci-xenon.c     | 2 +-
 drivers/mmc/host/sdhci.c           | 4 ++--
 drivers/mmc/host/sdhci.h           | 2 +-
 10 files changed, 12 insertions(+), 12 deletions(-)
drivers/mmc/host/sdhci-acpi.c +1 -1
index b3a130a9ee233..1604f512c7bd1 100644
@@ -883,7 +883,7 @@ static int sdhci_acpi_runtime_resume(struct device *dev)
 
 	sdhci_acpi_byt_setting(&c->pdev->dev);
 
-	return sdhci_runtime_resume_host(c->host);
+	return sdhci_runtime_resume_host(c->host, 0);
 }
 
 #endif
drivers/mmc/host/sdhci-esdhc-imx.c +1 -1
index c391510e9ef40..776a942162488 100644
@@ -1705,7 +1705,7 @@ static int sdhci_esdhc_runtime_resume(struct device *dev)
 		esdhc_pltfm_set_clock(host, imx_data->actual_clock);
 	}
 
-	err = sdhci_runtime_resume_host(host);
+	err = sdhci_runtime_resume_host(host, 0);
 	if (err)
 		goto disable_ipg_clk;
 
drivers/mmc/host/sdhci-of-at91.c +1 -1
index e377b9bc55a46..d4e7e8b7be772 100644
@@ -289,7 +289,7 @@ static int sdhci_at91_runtime_resume(struct device *dev)
 	}
 
 out:
-	return sdhci_runtime_resume_host(host);
+	return sdhci_runtime_resume_host(host, 0);
 }
 #endif /* CONFIG_PM */
 
drivers/mmc/host/sdhci-pci-core.c +2 -2
index 4154ee11b47dc..267b90374fa48 100644
@@ -167,7 +167,7 @@ static int sdhci_pci_runtime_suspend_host(struct sdhci_pci_chip *chip)
 
 err_pci_runtime_suspend:
 	while (--i >= 0)
-		sdhci_runtime_resume_host(chip->slots[i]->host);
+		sdhci_runtime_resume_host(chip->slots[i]->host, 0);
 	return ret;
 }
 
@@ -181,7 +181,7 @@ static int sdhci_pci_runtime_resume_host(struct sdhci_pci_chip *chip)
 		if (!slot)
 			continue;
 
-		ret = sdhci_runtime_resume_host(slot->host);
+		ret = sdhci_runtime_resume_host(slot->host, 0);
 		if (ret)
 			return ret;
 	}
drivers/mmc/host/sdhci-pxav3.c +1 -1
index 3ddecf4792958..e55037ceda734 100644
@@ -554,7 +554,7 @@ static int sdhci_pxav3_runtime_resume(struct device *dev)
 	if (!IS_ERR(pxa->clk_core))
 		clk_prepare_enable(pxa->clk_core);
 
-	return sdhci_runtime_resume_host(host);
+	return sdhci_runtime_resume_host(host, 0);
 }
 #endif
 
drivers/mmc/host/sdhci-s3c.c +1 -1
index 8e4a8ba33f050..f5753aef71511 100644
@@ -745,7 +745,7 @@ static int sdhci_s3c_runtime_resume(struct device *dev)
 	clk_prepare_enable(busclk);
 	if (ourhost->cur_clk >= 0)
 		clk_prepare_enable(ourhost->clk_bus[ourhost->cur_clk]);
-	ret = sdhci_runtime_resume_host(host);
+	ret = sdhci_runtime_resume_host(host, 0);
 	return ret;
 }
 #endif
drivers/mmc/host/sdhci-sprd.c +1 -1
index 06f84a4d79e00..f3261068adfbc 100644
@@ -470,7 +470,7 @@ static int sdhci_sprd_runtime_resume(struct device *dev)
 		return ret;
 	}
 
-	sdhci_runtime_resume_host(host);
+	sdhci_runtime_resume_host(host, 1);
 
 	return 0;
 }
drivers/mmc/host/sdhci-xenon.c +1 -1
index 8a18f14cf842d..1dea1ba66f7b4 100644
@@ -638,7 +638,7 @@ static int xenon_runtime_resume(struct device *dev)
 		priv->restore_needed = false;
 	}
 
-	ret = sdhci_runtime_resume_host(host);
+	ret = sdhci_runtime_resume_host(host, 0);
 	if (ret)
 		goto out;
 	return 0;
drivers/mmc/host/sdhci.c +2 -2
index 59acf8e3331ee..a5dc5aae973e6 100644
@@ -3320,7 +3320,7 @@ int sdhci_runtime_suspend_host(struct sdhci_host *host)
 }
 EXPORT_SYMBOL_GPL(sdhci_runtime_suspend_host);
 
-int sdhci_runtime_resume_host(struct sdhci_host *host)
+int sdhci_runtime_resume_host(struct sdhci_host *host, int soft_reset)
 {
 	struct mmc_host *mmc = host->mmc;
 	unsigned long flags;
@@ -3331,7 +3331,7 @@ int sdhci_runtime_resume_host(struct sdhci_host *host)
 			host->ops->enable_dma(host);
 	}
 
-	sdhci_init(host, 0);
+	sdhci_init(host, soft_reset);
 
 	if (mmc->ios.power_mode != MMC_POWER_UNDEFINED &&
 	    mmc->ios.power_mode != MMC_POWER_OFF) {
drivers/mmc/host/sdhci.h +1 -1
index 199712e7adbb3..d2c7c9c436c97 100644
@@ -781,7 +781,7 @@ void sdhci_adma_write_desc(struct sdhci_host *host, void **desc,
 int sdhci_suspend_host(struct sdhci_host *host);
 int sdhci_resume_host(struct sdhci_host *host);
 int sdhci_runtime_suspend_host(struct sdhci_host *host);
-int sdhci_runtime_resume_host(struct sdhci_host *host);
+int sdhci_runtime_resume_host(struct sdhci_host *host, int soft_reset);
 #endif
 
 void sdhci_cqe_enable(struct mmc_host *mmc);
Signature · 1 line
2.20.1
author maintainer
#17 lore raw

PATCH 16/23 usb: chipidea: imx: add imx7ulp support

In reply to Sasha Levin

From: Peter Chen <[email protected]>

In this commit, we add CI_HDRC_PMQOS to avoid system entering idle, at imx7ulp, if the system enters idle, the DMA will stop, so the USB transfer can't work at this case.

---
 drivers/usb/chipidea/ci_hdrc_imx.c | 28 +++++++++++++++++++++++++++-
 drivers/usb/chipidea/usbmisc_imx.c |  4 ++++
 include/linux/usb/chipidea.h       |  1 +
 3 files changed, 32 insertions(+), 1 deletion(-)
drivers/usb/chipidea/ci_hdrc_imx.c +27 -1
index ceec8d5985d46..a76708501236d 100644
@@ -13,6 +13,7 @@
 #include <linux/usb/of.h>
 #include <linux/clk.h>
 #include <linux/pinctrl/consumer.h>
+#include <linux/pm_qos.h>
 
 #include "ci.h"
 #include "ci_hdrc_imx.h"
@@ -63,6 +64,11 @@ static const struct ci_hdrc_imx_platform_flag imx7d_usb_data = {
 	.flags = CI_HDRC_SUPPORTS_RUNTIME_PM,
 };
 
+static const struct ci_hdrc_imx_platform_flag imx7ulp_usb_data = {
+	.flags = CI_HDRC_SUPPORTS_RUNTIME_PM |
+		CI_HDRC_PMQOS,
+};
+
 static const struct of_device_id ci_hdrc_imx_dt_ids[] = {
 	{ .compatible = "fsl,imx23-usb", .data = &imx23_usb_data},
 	{ .compatible = "fsl,imx28-usb", .data = &imx28_usb_data},
@@ -72,6 +78,7 @@ static const struct of_device_id ci_hdrc_imx_dt_ids[] = {
 	{ .compatible = "fsl,imx6sx-usb", .data = &imx6sx_usb_data},
 	{ .compatible = "fsl,imx6ul-usb", .data = &imx6ul_usb_data},
 	{ .compatible = "fsl,imx7d-usb", .data = &imx7d_usb_data},
+	{ .compatible = "fsl,imx7ulp-usb", .data = &imx7ulp_usb_data},
 	{ /* sentinel */ }
 };
 MODULE_DEVICE_TABLE(of, ci_hdrc_imx_dt_ids);
@@ -93,6 +100,8 @@ struct ci_hdrc_imx_data {
 	struct clk *clk_ahb;
 	struct clk *clk_per;
 	/* --------------------------------- */
+	struct pm_qos_request pm_qos_req;
+	const struct ci_hdrc_imx_platform_flag *plat_data;
 };
 
 /* Common functions shared by usbmisc drivers */
@@ -309,6 +318,8 @@ static int ci_hdrc_imx_probe(struct platform_device *pdev)
 	if (!data)
 		return -ENOMEM;
 
+	data->plat_data = imx_platform_flag;
+	pdata.flags |= imx_platform_flag->flags;
 	platform_set_drvdata(pdev, data);
 	data->usbmisc_data = usbmisc_get_init_data(dev);
 	if (IS_ERR(data->usbmisc_data))
@@ -369,6 +380,11 @@ static int ci_hdrc_imx_probe(struct platform_device *pdev)
 			}
 		}
 	}
+
+	if (pdata.flags & CI_HDRC_PMQOS)
+		pm_qos_add_request(&data->pm_qos_req,
+			PM_QOS_CPU_DMA_LATENCY, 0);
+
 	ret = imx_get_clks(dev);
 	if (ret)
 		goto disable_hsic_regulator;
@@ -396,7 +412,6 @@ static int ci_hdrc_imx_probe(struct platform_device *pdev)
 		usb_phy_init(pdata.usb_phy);
 	}
 
-	pdata.flags |= imx_platform_flag->flags;
 	if (pdata.flags & CI_HDRC_SUPPORTS_RUNTIME_PM)
 		data->supports_runtime_pm = true;
 
@@ -439,6 +454,8 @@ static int ci_hdrc_imx_probe(struct platform_device *pdev)
 disable_hsic_regulator:
 	if (data->hsic_pad_regulator)
 		ret = regulator_disable(data->hsic_pad_regulator);
+	if (pdata.flags & CI_HDRC_PMQOS)
+		pm_qos_remove_request(&data->pm_qos_req);
 	return ret;
 }
 
@@ -455,6 +472,8 @@ static int ci_hdrc_imx_remove(struct platform_device *pdev)
 	if (data->override_phy_control)
 		usb_phy_shutdown(data->phy);
 	imx_disable_unprepare_clks(&pdev->dev);
+	if (data->plat_data->flags & CI_HDRC_PMQOS)
+		pm_qos_remove_request(&data->pm_qos_req);
 	if (data->hsic_pad_regulator)
 		regulator_disable(data->hsic_pad_regulator);
 
@@ -480,6 +499,9 @@ static int __maybe_unused imx_controller_suspend(struct device *dev)
 	}
 
 	imx_disable_unprepare_clks(dev);
+	if (data->plat_data->flags & CI_HDRC_PMQOS)
+		pm_qos_remove_request(&data->pm_qos_req);
+
 	data->in_lpm = true;
 
 	return 0;
@@ -497,6 +519,10 @@ static int __maybe_unused imx_controller_resume(struct device *dev)
 		return 0;
 	}
 
+	if (data->plat_data->flags & CI_HDRC_PMQOS)
+		pm_qos_add_request(&data->pm_qos_req,
+			PM_QOS_CPU_DMA_LATENCY, 0);
+
 	ret = imx_prepare_enable_clks(dev);
 	if (ret)
 		return ret;
drivers/usb/chipidea/usbmisc_imx.c +4 -0
index d8b67e150b129..b7a5727d0c8a8 100644
@@ -763,6 +763,10 @@ static const struct of_device_id usbmisc_imx_dt_ids[] = {
 		.compatible = "fsl,imx7d-usbmisc",
 		.data = &imx7d_usbmisc_ops,
 	},
+	{
+		.compatible = "fsl,imx7ulp-usbmisc",
+		.data = &imx7d_usbmisc_ops,
+	},
 	{ /* sentinel */ }
 };
 MODULE_DEVICE_TABLE(of, usbmisc_imx_dt_ids);
include/linux/usb/chipidea.h +1 -0
index 911e05af671ea..edd89b7c8f184 100644
@@ -61,6 +61,7 @@ struct ci_hdrc_platform_data {
 #define CI_HDRC_OVERRIDE_PHY_CONTROL	BIT(12) /* Glue layer manages phy */
 #define CI_HDRC_REQUIRES_ALIGNED_DMA	BIT(13)
 #define CI_HDRC_IMX_IS_HSIC		BIT(14)
+#define CI_HDRC_PMQOS			BIT(15)
 	enum usb_dr_mode	dr_mode;
 #define CI_HDRC_CONTROLLER_RESET_EVENT		0
 #define CI_HDRC_CONTROLLER_STOPPED_EVENT	1
Signature · 1 line
2.20.1
author maintainer
#18 lore raw

PATCH 17/23 usb: chipidea: imx: fix EPROBE_DEFER support during driver probe

In reply to Sasha Levin

From: André Draszik <[email protected]>

If driver probe needs to be deferred, e.g. because ci_hdrc_add_device() isn't ready yet, this driver currently misbehaves badly:

    a) success is still reported to the driver core (meaning a 2nd
       probe attempt will never be done), leaving the driver in
       a dysfunctional state and the hardware unusable

    b) driver remove / shutdown OOPSes:
    [  206.786916] Unable to handle kernel paging request at virtual address fffffdff
    [  206.794148] pgd = 880b9f82
    [  206.796890] [fffffdff] *pgd=abf5e861, *pte=00000000, *ppte=00000000
    [  206.803179] Internal error: Oops: 37 [#1] PREEMPT SMP ARM
    [  206.808581] Modules linked in: wl18xx evbug
    [  206.813308] CPU: 1 PID: 1 Comm: systemd-shutdow Not tainted 4.19.35+gf345c93b4195 #1
    [  206.821053] Hardware name: Freescale i.MX7 Dual (Device Tree)
    [  206.826813] PC is at ci_hdrc_remove_device+0x4/0x20
    [  206.831699] LR is at ci_hdrc_imx_remove+0x20/0xe8
    [  206.836407] pc : [<805cd4b0>]    lr : [<805d62cc>]    psr: 20000013
    [  206.842678] sp : a806be40  ip : 00000001  fp : 80adbd3c
    [  206.847906] r10: 80b1b794  r9 : 80d5dfe0  r8 : a8192c44
    [  206.853136] r7 : 80db93a0  r6 : a8192c10  r5 : a8192c00  r4 : a93a4a00
    [  206.859668] r3 : 00000000  r2 : a8192ce4  r1 : ffffffff  r0 : fffffdfb
    [  206.866201] Flags: nzCv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment none
    [  206.873341] Control: 10c5387d  Table: a9e0c06a  DAC: 00000051
    [  206.879092] Process systemd-shutdow (pid: 1, stack limit = 0xb271353c)
    [  206.885624] Stack: (0xa806be40 to 0xa806c000)
    [  206.889992] be40: a93a4a00 805d62cc a8192c1c a8170e10 a8192c10 8049a490 80d04d08 00000000
    [  206.898179] be60: 00000000 80d0da2c fee1dead 00000000 a806a000 00000058 00000000 80148b08
    [  206.906366] be80: 01234567 80148d8c a9858600 00000000 00000000 00000000 00000000 80d04d08
    [  206.914553] bea0: 00000000 00000000 a82741e0 a9858600 00000024 00000002 a9858608 00000005
    [  206.922740] bec0: 0000001e 8022c058 00000000 00000000 a806bf14 a9858600 00000000 a806befc
    [  206.930927] bee0: a806bf78 00000000 7ee12c30 8022c18c a806bef8 a806befc 00000000 00000001
    [  206.939115] bf00: 00000000 00000024 a806bf14 00000005 7ee13b34 7ee12c68 00000004 7ee13f20
    [  206.947302] bf20: 00000010 7ee12c7c 00000005 7ee12d04 0000000a 76e7dc00 00000001 80d0f140
    [  206.955490] bf40: ab637880 a974de40 60000013 80d0f140 ab6378a0 80d04d08 a8080470 a9858600
    [  206.963677] bf60: a9858600 00000000 00000000 8022c24c 00000000 80144310 00000000 00000000
    [  206.971864] bf80: 80101204 80d04d08 00000000 80d04d08 00000000 00000000 00000003 00000058
    [  206.980051] bfa0: 80101204 80101000 00000000 00000000 fee1dead 28121969 01234567 00000000
    [  206.988237] bfc0: 00000000 00000000 00000003 00000058 00000000 00000000 00000000 00000000
    [  206.996425] bfe0: 0049ffb0 7ee13d58 0048a84b 76f245a6 60000030 fee1dead 00000000 00000000
    [  207.004622] [<805cd4b0>] (ci_hdrc_remove_device) from [<805d62cc>] (ci_hdrc_imx_remove+0x20/0xe8)
    [  207.013509] [<805d62cc>] (ci_hdrc_imx_remove) from [<8049a490>] (device_shutdown+0x16c/0x218)
    [  207.022050] [<8049a490>] (device_shutdown) from [<80148b08>] (kernel_restart+0xc/0x50)
    [  207.029980] [<80148b08>] (kernel_restart) from [<80148d8c>] (sys_reboot+0xf4/0x1f0)
    [  207.037648] [<80148d8c>] (sys_reboot) from [<80101000>] (ret_fast_syscall+0x0/0x54)
    [  207.045308] Exception stack(0xa806bfa8 to 0xa806bff0)
    [  207.050368] bfa0:                   00000000 00000000 fee1dead 28121969 01234567 00000000
    [  207.058554] bfc0: 00000000 00000000 00000003 00000058 00000000 00000000 00000000 00000000
    [  207.066737] bfe0: 0049ffb0 7ee13d58 0048a84b 76f245a6
    [  207.071799] Code: ebffffa8 e3a00000 e8bd8010 e92d4010 (e5904004)
    [  207.078021] ---[ end trace be47424e3fd46e9f ]---
    [  207.082647] Kernel panic - not syncing: Fatal exception
    [  207.087894] ---[ end Kernel panic - not syncing: Fatal exception ]---

    c) the error path in combination with driver removal causes
       imbalanced calls to the clk_*() and pm_()* APIs

a) happens because the original intended return value is overwritten (with 0) by the return code of regulator_disable() in ci_hdrc_imx_probe()'s error path
b) happens because ci_pdev is -EPROBE_DEFER, which causes ci_hdrc_remove_device() to OOPS

Fix a) by being more careful in ci_hdrc_imx_probe()'s error path and not overwriting the real error code

Fix b) by calling the respective cleanup functions during remove only when needed (when ci_pdev != NULL, i.e. when everything was initialised correctly). This also has the side effect of not causing imbalanced clk_*() and pm_*() API calls as part of the error code path.

Fixes 7c8e8909417e ("usb: chipidea: imx: add HSIC support")
Cc stable <[email protected]>
CC Peter Chen <[email protected]>
CC Greg Kroah-Hartman <[email protected]>
CC Shawn Guo <[email protected]>
CC Sascha Hauer <[email protected]>
CC Pengutronix Kernel Team <[email protected]>
CC Fabio Estevam <[email protected]>
CC NXP Linux Team <[email protected]>
---
 drivers/usb/chipidea/ci_hdrc_imx.c | 19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)
drivers/usb/chipidea/ci_hdrc_imx.c +12 -7
index a76708501236d..5faae96735e62 100644
@@ -453,9 +453,11 @@ static int ci_hdrc_imx_probe(struct platform_device *pdev)
 	imx_disable_unprepare_clks(dev);
 disable_hsic_regulator:
 	if (data->hsic_pad_regulator)
-		ret = regulator_disable(data->hsic_pad_regulator);
+		/* don't overwrite original ret (cf. EPROBE_DEFER) */
+		regulator_disable(data->hsic_pad_regulator);
 	if (pdata.flags & CI_HDRC_PMQOS)
 		pm_qos_remove_request(&data->pm_qos_req);
+	data->ci_pdev = NULL;
 	return ret;
 }
 
@@ -468,14 +470,17 @@ static int ci_hdrc_imx_remove(struct platform_device *pdev)
 		pm_runtime_disable(&pdev->dev);
 		pm_runtime_put_noidle(&pdev->dev);
 	}
-	ci_hdrc_remove_device(data->ci_pdev);
+	if (data->ci_pdev)
+		ci_hdrc_remove_device(data->ci_pdev);
 	if (data->override_phy_control)
 		usb_phy_shutdown(data->phy);
-	imx_disable_unprepare_clks(&pdev->dev);
-	if (data->plat_data->flags & CI_HDRC_PMQOS)
-		pm_qos_remove_request(&data->pm_qos_req);
-	if (data->hsic_pad_regulator)
-		regulator_disable(data->hsic_pad_regulator);
+	if (data->ci_pdev) {
+		imx_disable_unprepare_clks(&pdev->dev);
+		if (data->plat_data->flags & CI_HDRC_PMQOS)
+			pm_qos_remove_request(&data->pm_qos_req);
+		if (data->hsic_pad_regulator)
+			regulator_disable(data->hsic_pad_regulator);
+	}
 
 	return 0;
 }
Signature · 1 line
2.20.1
author maintainer
#19 lore raw

PATCH 18/23 virtio/s390: fix race on airq_areas[]

In reply to Sasha Levin

From: Halil Pasic <[email protected]>

[ Upstream commit 4f419eb14272e0698e8c55bb5f3f266cc2a21c81 ]

The access to airq_areas was racy ever since the adapter interrupts got introduced to virtio-ccw, but since commit 39c7dcb15892 ("virtio/s390:
make airq summary indicators DMA") this became an issue in practice as well. Namely before that commit the airq_info that got overwritten was still functional. After that commit however the two infos share a summary_indicator, which aggravates the situation. Which means auto-online mechanism occasionally hangs the boot with virtio_blk.

Fixes 96b14536d935 ("virtio-ccw: virtio-ccw adapter interrupt support.")
---
 drivers/s390/virtio/virtio_ccw.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/s390/virtio/virtio_ccw.c +3 -0
index 6a30768813219..8d47ad61bac3d 100644
@@ -132,6 +132,7 @@ struct airq_info {
 	struct airq_iv *aiv;
 };
 static struct airq_info *airq_areas[MAX_AIRQ_AREAS];
+static DEFINE_MUTEX(airq_areas_lock);
 
 #define CCW_CMD_SET_VQ 0x13
 #define CCW_CMD_VDEV_RESET 0x33
@@ -244,9 +245,11 @@ static unsigned long get_airq_indicator(struct virtqueue *vqs[], int nvqs,
 	unsigned long bit, flags;
 
 	for (i = 0; i < MAX_AIRQ_AREAS && !indicator_addr; i++) {
+		mutex_lock(&airq_areas_lock);
 		if (!airq_areas[i])
 			airq_areas[i] = new_airq_info();
 		info = airq_areas[i];
+		mutex_unlock(&airq_areas_lock);
 		if (!info)
 			return 0;
 		write_lock_irqsave(&info->lock, flags);
Signature · 1 line
2.20.1
author maintainer
#20 lore raw

PATCH 19/23 drm/i915: Support flags in whitlist WAs

In reply to Sasha Levin

From: John Harrison <[email protected]>

[ Upstream commit 6883eab274813d158bfcfb499aa225ece61c0f29 ]

Newer hardware adds flags to the whitelist work-around register. These allow per access direction privileges and ranges.

---
 drivers/gpu/drm/i915/i915_reg.h          | 7 +++++++
 drivers/gpu/drm/i915/intel_workarounds.c | 9 ++++++++-
 2 files changed, 15 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_reg.h +7 -0
index 13d6bd4e17b20..cf748b80e6401 100644
@@ -2510,6 +2510,13 @@ enum i915_power_well_id {
 #define   RING_WAIT_SEMAPHORE	(1 << 10) /* gen6+ */
 
 #define RING_FORCE_TO_NONPRIV(base, i) _MMIO(((base) + 0x4D0) + (i) * 4)
+#define   RING_FORCE_TO_NONPRIV_RW		(0 << 28)    /* CFL+ & Gen11+ */
+#define   RING_FORCE_TO_NONPRIV_RD		(1 << 28)
+#define   RING_FORCE_TO_NONPRIV_WR		(2 << 28)
+#define   RING_FORCE_TO_NONPRIV_RANGE_1		(0 << 0)     /* CFL+ & Gen11+ */
+#define   RING_FORCE_TO_NONPRIV_RANGE_4		(1 << 0)
+#define   RING_FORCE_TO_NONPRIV_RANGE_16	(2 << 0)
+#define   RING_FORCE_TO_NONPRIV_RANGE_64	(3 << 0)
 #define   RING_MAX_NONPRIV_SLOTS  12
 
 #define GEN7_TLB_RD_ADDR	_MMIO(0x4700)
drivers/gpu/drm/i915/intel_workarounds.c +8 -1
index 2fb70fab2d1c6..1db826b12774e 100644
@@ -981,7 +981,7 @@ bool intel_gt_verify_workarounds(struct drm_i915_private *i915,
 }
 
 static void
-whitelist_reg(struct i915_wa_list *wal, i915_reg_t reg)
+whitelist_reg_ext(struct i915_wa_list *wal, i915_reg_t reg, u32 flags)
 {
 	struct i915_wa wa = {
 		.reg = reg
@@ -990,9 +990,16 @@ whitelist_reg(struct i915_wa_list *wal, i915_reg_t reg)
 	if (GEM_DEBUG_WARN_ON(wal->count >= RING_MAX_NONPRIV_SLOTS))
 		return;
 
+	wa.reg.reg |= flags;
 	_wa_add(wal, &wa);
 }
 
+static void
+whitelist_reg(struct i915_wa_list *wal, i915_reg_t reg)
+{
+	whitelist_reg_ext(wal, reg, RING_FORCE_TO_NONPRIV_RW);
+}
+
 static void gen9_whitelist_build(struct i915_wa_list *w)
 {
 	/* WaVFEStateAfterPipeControlwithMediaStateClear:skl,bxt,glk,cfl */
Signature · 6 lines
2.20.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
author maintainer
#21 lore raw

PATCH 20/23 drm/i915: Support whitelist workarounds on all engines

In reply to Sasha Levin

From: John Harrison <[email protected]>

[ Upstream commit ebd2de47a19f1c17ae47f8331aae3cd436766663 ]

Newer hardware requires setting up whitelists on engines other than render. So, extend the whitelist code to support all engines.

---
 drivers/gpu/drm/i915/intel_workarounds.c | 65 +++++++++++++++++-------
 1 file changed, 47 insertions(+), 18 deletions(-)
drivers/gpu/drm/i915/intel_workarounds.c +47 -18
index 1db826b12774e..0b80fde927899 100644
@@ -1012,48 +1012,79 @@ static void gen9_whitelist_build(struct i915_wa_list *w)
 	whitelist_reg(w, GEN8_HDC_CHICKEN1);
 }
 
-static void skl_whitelist_build(struct i915_wa_list *w)
+static void skl_whitelist_build(struct intel_engine_cs *engine)
 {
+	struct i915_wa_list *w = &engine->whitelist;
+
+	if (engine->class != RENDER_CLASS)
+		return;
+
 	gen9_whitelist_build(w);
 
 	/* WaDisableLSQCROPERFforOCL:skl */
 	whitelist_reg(w, GEN8_L3SQCREG4);
 }
 
-static void bxt_whitelist_build(struct i915_wa_list *w)
+static void bxt_whitelist_build(struct intel_engine_cs *engine)
 {
-	gen9_whitelist_build(w);
+	if (engine->class != RENDER_CLASS)
+		return;
+
+	gen9_whitelist_build(&engine->whitelist);
 }
 
-static void kbl_whitelist_build(struct i915_wa_list *w)
+static void kbl_whitelist_build(struct intel_engine_cs *engine)
 {
+	struct i915_wa_list *w = &engine->whitelist;
+
+	if (engine->class != RENDER_CLASS)
+		return;
+
 	gen9_whitelist_build(w);
 
 	/* WaDisableLSQCROPERFforOCL:kbl */
 	whitelist_reg(w, GEN8_L3SQCREG4);
 }
 
-static void glk_whitelist_build(struct i915_wa_list *w)
+static void glk_whitelist_build(struct intel_engine_cs *engine)
 {
+	struct i915_wa_list *w = &engine->whitelist;
+
+	if (engine->class != RENDER_CLASS)
+		return;
+
 	gen9_whitelist_build(w);
 
 	/* WA #0862: Userspace has to set "Barrier Mode" to avoid hangs. */
 	whitelist_reg(w, GEN9_SLICE_COMMON_ECO_CHICKEN1);
 }
 
-static void cfl_whitelist_build(struct i915_wa_list *w)
+static void cfl_whitelist_build(struct intel_engine_cs *engine)
 {
-	gen9_whitelist_build(w);
+	if (engine->class != RENDER_CLASS)
+		return;
+
+	gen9_whitelist_build(&engine->whitelist);
 }
 
-static void cnl_whitelist_build(struct i915_wa_list *w)
+static void cnl_whitelist_build(struct intel_engine_cs *engine)
 {
+	struct i915_wa_list *w = &engine->whitelist;
+
+	if (engine->class != RENDER_CLASS)
+		return;
+
 	/* WaEnablePreemptionGranularityControlByUMD:cnl */
 	whitelist_reg(w, GEN8_CS_CHICKEN1);
 }
 
-static void icl_whitelist_build(struct i915_wa_list *w)
+static void icl_whitelist_build(struct intel_engine_cs *engine)
 {
+	struct i915_wa_list *w = &engine->whitelist;
+
+	if (engine->class != RENDER_CLASS)
+		return;
+
 	/* WaAllowUMDToModifyHalfSliceChicken7:icl */
 	whitelist_reg(w, GEN9_HALF_SLICE_CHICKEN7);
 
@@ -1069,24 +1100,22 @@ void intel_engine_init_whitelist(struct intel_engine_cs *engine)
 	struct drm_i915_private *i915 = engine->i915;
 	struct i915_wa_list *w = &engine->whitelist;
 
-	GEM_BUG_ON(engine->id != RCS0);
-
 	wa_init_start(w, "whitelist");
 
 	if (IS_GEN(i915, 11))
-		icl_whitelist_build(w);
+		icl_whitelist_build(engine);
 	else if (IS_CANNONLAKE(i915))
-		cnl_whitelist_build(w);
+		cnl_whitelist_build(engine);
 	else if (IS_COFFEELAKE(i915))
-		cfl_whitelist_build(w);
+		cfl_whitelist_build(engine);
 	else if (IS_GEMINILAKE(i915))
-		glk_whitelist_build(w);
+		glk_whitelist_build(engine);
 	else if (IS_KABYLAKE(i915))
-		kbl_whitelist_build(w);
+		kbl_whitelist_build(engine);
 	else if (IS_BROXTON(i915))
-		bxt_whitelist_build(w);
+		bxt_whitelist_build(engine);
 	else if (IS_SKYLAKE(i915))
-		skl_whitelist_build(w);
+		skl_whitelist_build(engine);
 	else if (INTEL_GEN(i915) <= 8)
 		return;
 	else
Signature · 6 lines
2.20.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
author maintainer
#22 lore raw

PATCH 21/23 drm/i915: whitelist PS_(DEPTH|INVOCATION)_COUNT

In reply to Sasha Levin

From: Lionel Landwerlin <[email protected]>

[ Upstream commit 6ce5bfe936ac31d5c52c4b1328d0bfda5f97e7ca ]

CFL:C0+ changed the status of those registers which are now blacklisted by default.

This is breaking a number of CTS tests on GL & Vulkan :

KHR-GL45.pipeline_statistics_query_tests_ARB.functional_fragment_shader_invocations (GL)

dEQP-VK.query_pool.statistics_query.fragment_shader_invocations.* (Vulkan)

v2: Only use one whitelist entry (Lionel)

Bspec: 14091

Cc [email protected] # 6883eab27481: drm/i915: Support flags in whitlist WAs
(cherry picked from commit 2c903da50f5a9522b134e488bd0f92646c46f3c0)
---
 drivers/gpu/drm/i915/intel_workarounds.c | 17 ++++++++++++++++-
 1 file changed, 16 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/intel_workarounds.c +16 -1
index 0b80fde927899..be3688908f0ce 100644
@@ -1061,10 +1061,25 @@ static void glk_whitelist_build(struct intel_engine_cs *engine)
 
 static void cfl_whitelist_build(struct intel_engine_cs *engine)
 {
+	struct i915_wa_list *w = &engine->whitelist;
+
 	if (engine->class != RENDER_CLASS)
 		return;
 
-	gen9_whitelist_build(&engine->whitelist);
+	gen9_whitelist_build(w);
+
+	/*
+	 * WaAllowPMDepthAndInvocationCountAccessFromUMD:cfl,whl,cml,aml
+	 *
+	 * This covers 4 register which are next to one another :
+	 *   - PS_INVOCATION_COUNT
+	 *   - PS_INVOCATION_COUNT_UDW
+	 *   - PS_DEPTH_COUNT
+	 *   - PS_DEPTH_COUNT_UDW
+	 */
+	whitelist_reg_ext(w, PS_INVOCATION_COUNT,
+			  RING_FORCE_TO_NONPRIV_RD |
+			  RING_FORCE_TO_NONPRIV_RANGE_4);
 }
 
 static void cnl_whitelist_build(struct intel_engine_cs *engine)
Signature · 6 lines
2.20.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
author maintainer
#23 lore raw

PATCH 22/23 drm/i915: Add whitelist workarounds for ICL

In reply to Sasha Levin

From: John Harrison <[email protected]>

[ Upstream commit 7b3d406310983a89ed7a1ecdd115efbe12b0ded5 ]

Updated whitelist table for ICL.

v2: Reduce changes to just those required for media driver until the selftest can be updated to support the new features of the other entries.

---
 drivers/gpu/drm/i915/intel_workarounds.c | 38 +++++++++++++++++-------
 1 file changed, 27 insertions(+), 11 deletions(-)
drivers/gpu/drm/i915/intel_workarounds.c +27 -11
index be3688908f0ce..efea5a18fa6db 100644
@@ -1097,17 +1097,33 @@ static void icl_whitelist_build(struct intel_engine_cs *engine)
 {
 	struct i915_wa_list *w = &engine->whitelist;
 
-	if (engine->class != RENDER_CLASS)
-		return;
-
-	/* WaAllowUMDToModifyHalfSliceChicken7:icl */
-	whitelist_reg(w, GEN9_HALF_SLICE_CHICKEN7);
-
-	/* WaAllowUMDToModifySamplerMode:icl */
-	whitelist_reg(w, GEN10_SAMPLER_MODE);
-
-	/* WaEnableStateCacheRedirectToCS:icl */
-	whitelist_reg(w, GEN9_SLICE_COMMON_ECO_CHICKEN1);
+	switch (engine->class) {
+	case RENDER_CLASS:
+		/* WaAllowUMDToModifyHalfSliceChicken7:icl */
+		whitelist_reg(w, GEN9_HALF_SLICE_CHICKEN7);
+
+		/* WaAllowUMDToModifySamplerMode:icl */
+		whitelist_reg(w, GEN10_SAMPLER_MODE);
+
+		/* WaEnableStateCacheRedirectToCS:icl */
+		whitelist_reg(w, GEN9_SLICE_COMMON_ECO_CHICKEN1);
+		break;
+
+	case VIDEO_DECODE_CLASS:
+		/* hucStatusRegOffset */
+		whitelist_reg_ext(w, _MMIO(0x2000 + engine->mmio_base),
+				  RING_FORCE_TO_NONPRIV_RD);
+		/* hucUKernelHdrInfoRegOffset */
+		whitelist_reg_ext(w, _MMIO(0x2014 + engine->mmio_base),
+				  RING_FORCE_TO_NONPRIV_RD);
+		/* hucStatus2RegOffset */
+		whitelist_reg_ext(w, _MMIO(0x23B0 + engine->mmio_base),
+				  RING_FORCE_TO_NONPRIV_RD);
+		break;
+
+	default:
+		break;
+	}
 }
 
 void intel_engine_init_whitelist(struct intel_engine_cs *engine)
Signature · 6 lines
2.20.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
author maintainer
#24 lore raw

PATCH 23/23 drm/i915/icl: whitelist PS_(DEPTH|INVOCATION)_COUNT

In reply to Sasha Levin

From: Lionel Landwerlin <[email protected]>

[ Upstream commit cf8f9aa1eda7d916bd23f6b8c226404deb11690c ]

The same tests failing on CFL+ platforms are also failing on ICL. Documentation doesn't list the WaAllowPMDepthAndInvocationCountAccessFromUMD workaround for ICL but applying it fixes the same tests as CFL.

v2: Use only one whitelist entry (Lionel)

Cc [email protected] # 6883eab27481: drm/i915: Support flags in whitlist WAs
(cherry picked from commit 3fe0107e45ab396342497e06b8924cdd485cde3b)
---
 drivers/gpu/drm/i915/intel_workarounds.c | 13 +++++++++++++
 1 file changed, 13 insertions(+)
drivers/gpu/drm/i915/intel_workarounds.c +13 -0
index efea5a18fa6db..edd57a5e0495f 100644
@@ -1107,6 +1107,19 @@ static void icl_whitelist_build(struct intel_engine_cs *engine)
 
 		/* WaEnableStateCacheRedirectToCS:icl */
 		whitelist_reg(w, GEN9_SLICE_COMMON_ECO_CHICKEN1);
+
+		/*
+		 * WaAllowPMDepthAndInvocationCountAccessFromUMD:icl
+		 *
+		 * This covers 4 register which are next to one another :
+		 *   - PS_INVOCATION_COUNT
+		 *   - PS_INVOCATION_COUNT_UDW
+		 *   - PS_DEPTH_COUNT
+		 *   - PS_DEPTH_COUNT_UDW
+		 */
+		whitelist_reg_ext(w, PS_INVOCATION_COUNT,
+				  RING_FORCE_TO_NONPRIV_RD |
+				  RING_FORCE_TO_NONPRIV_RANGE_4);
 		break;
 
 	case VIDEO_DECODE_CLASS:
Signature · 1 line
2.20.1