Skip to contentSkip to conversation

3.10.68-stable review

PATCH In review Greg Kroah-Hartman 3 participants 54 messages Feb 3, 2015 – Feb 4, 2015 Stable branch

lore
Share & tools
lore

Greg Kroah-Hartman

author maintainer
#0 lore raw

cover 3.10.68-stable review

This is the start of the stable review cycle for the 3.10.68 release. There are 51 patches in this series, all will be posted as a response to this one. If anyone has any issues with these being applied, please let me know.

Responses should be made by Thu Feb 5 23:17:14 UTC 2015. Anything received after that time might be too late.

The whole patch series can be found in one patch at:

	kernel.org/pub/linux/kernel/v3.0/stable-review/patch-3.10.68-rc1.gz

and the diffstat can be found below.

thanks,

greg k-h

-------------
Pseudo-Shortlog of commits:

Greg Kroah-Hartman <[email protected]>

    Linux 3.10.68-rc1

Nicholas Bellinger <[email protected]>
target: Drop arbitrary maximum I/O size limit

Sagi Grimberg <[email protected]>
iser-target: Fix implicit termination of connections

Sagi Grimberg <[email protected]>
iser-target: Handle ADDR_CHANGE event for listener cm_id

Sagi Grimberg <[email protected]>
iser-target: Fix connected_handler + teardown flow race

Sagi Grimberg <[email protected]>
iser-target: Parallelize CM connection establishment

Sagi Grimberg <[email protected]>
iser-target: Fix flush + disconnect completion handling

Sagi Grimberg <[email protected]>
iscsi,iser-target: Initiate termination only once

Nicholas Bellinger <[email protected]>

    vhost-scsi: Add missing virtio-scsi -> TCM attribute conversion

Hannes Reinecke <[email protected]>
tcm_loop: Fix wrong I_T nexus association

Nicholas Bellinger <[email protected]>
vhost-scsi: Take configfs group dependency during VHOST_SCSI_SET_ENDPOINT

Or Gerlitz <[email protected]>
ib_isert: Add max_send_sge=2 minimum for control PDU responses

Chris Moore <[email protected]>
IB/isert: Adjust CQ size to HW limits

Tejun Heo <[email protected]>
workqueue: fix subtle pool management issue which can stall whole worker_pool

Martin Kaiser <[email protected]>
gpio: squelch a compiler warning

Madper Xie <[email protected]>
efi-pstore: Make efi-pstore return a unique id

Rob Herring <[email protected]>
pstore/ram: avoid atomic accesses for ioremapped regions

Liu ShuoX <[email protected]>
pstore: Fix NULL pointer fault if get NULL prz in ramoops_get_next_prz

Liu ShuoX <[email protected]>
pstore: skip zero size persistent ram buffer in traverse

Liu ShuoX <[email protected]>
pstore: clarify clearing of _read_cnt in ramoops_context

Dan Carpenter <[email protected]>
pstore: d_alloc_name() doesn't return an ERR_PTR

Aruna Balakrishnaiah <[email protected]>
pstore: Fail to unlink if a driver has not defined pstore_erase

Steven Capper <[email protected]>
ARM: 8109/1: mm: Modify pte_write and pmd_write logic for LPAE

Steven Capper <[email protected]>

    ARM: 8108/1: mm: Introduce {pte,pmd}_isset and {pte,pmd}_isclear

Russell King <[email protected]>
ARM: DMA: ensure that old section mappings are flushed from the TLB

Laura Abbott <[email protected]>

    ARM: 7931/1: Correct virt_addr_valid

Russell King <[email protected]>
ARM: fix asm/memory.h build error

Chen Gang <[email protected]>
ARM: 7867/1: include: asm: use 'int' instead of 'unsigned long' for 'oldval' in atomic_cmpxchg().

Chen Gang <[email protected]>
ARM: 7866/1: include: asm: use 'long long' instead of 'u64' within atomic.h

Will Deacon <[email protected]>
ARM: lpae: fix definition of PTE_HWTABLE_PTRS

Cyril Chemparathy <[email protected]>
ARM: fix type of PHYS_PFN_OFFSET to unsigned long

Vitaly Andrianov <[email protected]>
ARM: LPAE: use phys_addr_t in alloc_init_pud()

Cyril Chemparathy <[email protected]>
ARM: LPAE: use signed arithmetic for mask definitions

Steve Capper <[email protected]>
ARM: mm: correct pte_same behaviour for LPAE.

Douglas Anderson <[email protected]>
ARM: 7829/1: Add ".text.unlikely" and ".text.hot" to arm unwind tables

Mugunthan V N <[email protected]>
drivers: net: cpsw: discard dual emac default vlan configuration

Ashay Jaiswal <[email protected]>
regulator: core: fix race condition in regulator_put()

Mika Westerberg <[email protected]>
spi/pxa2xx: Clear cur_chip pointer before starting next message

Joe Thornber <[email protected]>
dm cache: fix missing ERR_PTR returns and handling

Joe Thornber <[email protected]>
dm thin: don't allow messages to be sent to a pool target in READ_ONLY or FAIL mode

Johannes Berg <[email protected]>
nl80211: fix per-station group key get/del and memory leak

Trond Myklebust <[email protected]>
NFSv4.1: Fix an Oops in nfs41_walk_client_list

Peng Tao <[email protected]>
nfs: fix dio deadlock when O_DIRECT flag is flipped

Jochen Hein <[email protected]>

    Input: i8042 - add noloop quirk for Medion Akoya E7225 (MD98857)

Clemens Ladisch <[email protected]>
ALSA: seq-dummy: remove deadlock-causing events on close

Laurent Dufour <[email protected]>
powerpc/xmon: Fix another endiannes issue in RTAS call from xmon

Ahmed S. Darwish <[email protected]>
can: kvaser_usb: Fix state handling upon BUS_ERROR events

Ahmed S. Darwish <[email protected]>
can: kvaser_usb: Retry the first bulk transfer on -ETIMEDOUT

Ahmed S. Darwish <[email protected]>
can: kvaser_usb: Send correct context to URB completion

Ahmed S. Darwish <[email protected]>
can: kvaser_usb: Do not sleep in atomic context

Zidan Wang <[email protected]>

    ASoC: wm8960: Fix capture sample rate from 11250 to 11025

Andy Shevchenko <[email protected]>
spi: dw-mid: fix FIFO size

-------------

Diffstat:

 Makefile                                    |   4 +-
 arch/arm/include/asm/atomic.h               |  52 ++--
 arch/arm/include/asm/memory.h               |  36 +--
 arch/arm/include/asm/module.h               |   2 +
 arch/arm/include/asm/page.h                 |   2 +-
 arch/arm/include/asm/pgtable-3level-hwdef.h |   1 +
 arch/arm/include/asm/pgtable-3level.h       |  31 ++-
 arch/arm/include/asm/pgtable.h              |  14 +-
 arch/arm/kernel/head.S                      |   2 +-
 arch/arm/kernel/module.c                    |   8 +
 arch/arm/mm/dma-mapping.c                   |  11 +-
 arch/arm/mm/mmu.c                           |   3 +-
 arch/arm/mm/proc-v7-3level.S                |   9 +-
 arch/powerpc/xmon/xmon.c                    |   1 +
 drivers/firmware/efi/efi-pstore.c           |  18 +-
 drivers/gpio/gpiolib-of.c                   |   2 +-
 drivers/infiniband/ulp/isert/ib_isert.c     | 372 ++++++++++++++++++++--------
 drivers/infiniband/ulp/isert/ib_isert.h     |   7 +-
 drivers/input/serio/i8042-x86ia64io.h       |   8 +
 drivers/md/dm-cache-metadata.c              |   9 +-
 drivers/md/dm-thin.c                        |   6 +
 drivers/net/can/usb/kvaser_usb.c            |  28 ++-
 drivers/net/ethernet/ti/cpsw.c              |  22 ++
 drivers/regulator/core.c                    |   4 +-
 drivers/spi/spi-dw-mid.c                    |   1 -
 drivers/spi/spi-pxa2xx.c                    |   2 +-
 drivers/target/iscsi/iscsi_target_login.c   |   3 +
 drivers/target/loopback/tcm_loop.c          |  29 ++-
 drivers/target/loopback/tcm_loop.h          |   7 +-
 drivers/target/target_core_device.c         |   8 +-
 drivers/target/target_core_file.c           |  11 +-
 drivers/target/target_core_iblock.c         |   2 +-
 drivers/target/target_core_sbc.c            |  15 --
 drivers/target/target_core_spc.c            |   5 +-
 drivers/vhost/scsi.c                        |  47 +++-
 fs/nfs/direct.c                             |   6 +
 fs/nfs/nfs4client.c                         |   2 +-
 fs/pstore/inode.c                           |   5 +-
 fs/pstore/ram.c                             |  15 +-
 fs/pstore/ram_core.c                        |  54 +++-
 kernel/workqueue.c                          |  39 +--
 net/wireless/nl80211.c                      |   9 +-
 sound/core/seq/seq_dummy.c                  |  31 ---
 sound/soc/codecs/wm8960.c                   |   2 +-
 44 files changed, 623 insertions(+), 322 deletions(-)
author maintainer
#1 lore raw

PATCH 1/51 spi: dw-mid: fix FIFO size

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Andy Shevchenko <[email protected]>

commit 67bf9cda4b498b8cea4a40be67a470afe57d2e88 upstream.

The FIFO size is 40 accordingly to the specifications, but this means 0x40, i.e. 64 bytes. This patch fixes the typo and enables FIFO size autodetection for Intel MID devices.

Fixes 7063c0d942a1 (spi/dw_spi: add DMA support)
---
 drivers/spi/spi-dw-mid.c |    1 -
 1 file changed, 1 deletion(-)
drivers/spi/spi-dw-mid.c +0 -1
@@ -222,7 +222,6 @@ int dw_spi_mid_init(struct dw_spi *dws)
 	iounmap(clk_reg);
 
 	dws->num_cs = 16;
-	dws->fifo_len = 40;	/* FIFO has 40 words buffer */
 
 #ifdef CONFIG_SPI_DW_MID_DMA
 	dws->dma_priv = kzalloc(sizeof(struct mid_dma), GFP_KERNEL);
author maintainer
#2 lore raw

PATCH 2/51 ASoC: wm8960: Fix capture sample rate from 11250 to 11025

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Zidan Wang <[email protected]>

commit 22ee76daddb87f88d2336d1b4737ef27c4f307ac upstream.

wm8960 codec can't support sample rate 11250, it must be 11025.

---
 sound/soc/codecs/wm8960.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
sound/soc/codecs/wm8960.c +1 -1
@@ -555,7 +555,7 @@ static struct {
 	{ 22050, 2 },
 	{ 24000, 2 },
 	{ 16000, 3 },
-	{ 11250, 4 },
+	{ 11025, 4 },
 	{ 12000, 4 },
 	{  8000, 5 },
 };
author maintainer
#3 lore raw

PATCH 3/51 can: kvaser_usb: Do not sleep in atomic context

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: "Ahmed S. Darwish" <[email protected]>

commit ded5006667318c06df875609535176bd33f243a1 upstream.

Upon receiving a hardware event with the BUS_RESET flag set, the driver kills all of its anchored URBs and resets all of its transmit URB contexts.

Unfortunately it does so under the context of URB completion handler `kvaser_usb_read_bulk_callback()', which is often called in an atomic context.

While the device is flooded with many received error packets, usb_kill_urb() typically sleeps/reschedules till the transfer request of each killed URB in question completes, leading to the sleep in atomic bug. [3]

In v2 submission of the original driver patch [1], it was stated that the URBs kill and tx contexts reset was needed since we don't receive any tx acknowledgments later and thus such resources will be locked down forever. Fortunately this is no longer needed since an earlier bugfix in this patch series is now applied: all tx URB contexts are reset upon CAN channel close. [2]

Moreover, a BUS_RESET is now treated _exactly_ like a BUS_OFF event, which is the recommended handling method advised by the device manufacturer.

[1] http://article.gmane.org/gmane.linux.network/239442
http://www.webcitation.org/6Vr2yagAQ

[2] can: kvaser_usb: Reset all URB tx contexts upon channel close

    889b77f7fd2bcc922493d73a4c51d8a851505815

[3] Stacktrace:

 <IRQ>  [<ffffffff8158de87>] dump_stack+0x45/0x57
 [<ffffffff8158b60c>] __schedule_bug+0x41/0x4f
 [<ffffffff815904b1>] __schedule+0x5f1/0x700
 [<ffffffff8159360a>] ? _raw_spin_unlock_irqrestore+0xa/0x10
 [<ffffffff81590684>] schedule+0x24/0x70
 [<ffffffff8147d0a5>] usb_kill_urb+0x65/0xa0
 [<ffffffff81077970>] ? prepare_to_wait_event+0x110/0x110
 [<ffffffff8147d7d8>] usb_kill_anchored_urbs+0x48/0x80
 [<ffffffffa01f4028>] kvaser_usb_unlink_tx_urbs+0x18/0x50 [kvaser_usb]
 [<ffffffffa01f45d0>] kvaser_usb_rx_error+0xc0/0x400 [kvaser_usb]
 [<ffffffff8108b14a>] ? vprintk_default+0x1a/0x20
 [<ffffffffa01f5241>] kvaser_usb_read_bulk_callback+0x4c1/0x5f0 [kvaser_usb]
 [<ffffffff8147a73e>] __usb_hcd_giveback_urb+0x5e/0xc0
 [<ffffffff8147a8a1>] usb_hcd_giveback_urb+0x41/0x110
 [<ffffffffa0008748>] finish_urb+0x98/0x180 [ohci_hcd]
 [<ffffffff810cd1a7>] ? acct_account_cputime+0x17/0x20
 [<ffffffff81069f65>] ? local_clock+0x15/0x30
 [<ffffffffa000a36b>] ohci_work+0x1fb/0x5a0 [ohci_hcd]
 [<ffffffff814fbb31>] ? process_backlog+0xb1/0x130
 [<ffffffffa000cd5b>] ohci_irq+0xeb/0x270 [ohci_hcd]
 [<ffffffff81479fc1>] usb_hcd_irq+0x21/0x30
 [<ffffffff8108bfd3>] handle_irq_event_percpu+0x43/0x120
 [<ffffffff8108c0ed>] handle_irq_event+0x3d/0x60
 [<ffffffff8108ec84>] handle_fasteoi_irq+0x74/0x110
 [<ffffffff81004dfd>] handle_irq+0x1d/0x30
 [<ffffffff81004727>] do_IRQ+0x57/0x100
 [<ffffffff8159482a>] common_interrupt+0x6a/0x6a
---
 drivers/net/can/usb/kvaser_usb.c |    7 +------
 1 file changed, 1 insertion(+), 6 deletions(-)
drivers/net/can/usb/kvaser_usb.c +1 -6
@@ -654,11 +654,6 @@ static void kvaser_usb_rx_error(const st
 	priv = dev->nets[channel];
 	stats = &priv->netdev->stats;
 
-	if (status & M16C_STATE_BUS_RESET) {
-		kvaser_usb_unlink_tx_urbs(priv);
-		return;
-	}
-
 	skb = alloc_can_err_skb(priv->netdev, &cf);
 	if (!skb) {
 		stats->rx_dropped++;
@@ -669,7 +664,7 @@ static void kvaser_usb_rx_error(const st
 
 	netdev_dbg(priv->netdev, "Error status: 0x%02x\n", status);
 
-	if (status & M16C_STATE_BUS_OFF) {
+	if (status & (M16C_STATE_BUS_OFF | M16C_STATE_BUS_RESET)) {
 		cf->can_id |= CAN_ERR_BUSOFF;
 
 		priv->can.can_stats.bus_off++;
author maintainer
#4 lore raw

PATCH 4/51 can: kvaser_usb: Send correct context to URB completion

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: "Ahmed S. Darwish" <[email protected]>

commit 3803fa6977f1de15fda4e8646c8fec97c8045cae upstream.

Send expected argument to the URB completion hander: a CAN netdevice instead of the network interface private context `kvaser_usb_net_priv'.

This was discovered by having some garbage in the kernel log in place of the netdevice names: can0 and can1.

---
 drivers/net/can/usb/kvaser_usb.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/can/usb/kvaser_usb.c +1 -1
@@ -579,7 +579,7 @@ static int kvaser_usb_simple_msg_async(s
 			  usb_sndbulkpipe(dev->udev,
 					  dev->bulk_out->bEndpointAddress),
 			  buf, msg->len,
-			  kvaser_usb_simple_msg_callback, priv);
+			  kvaser_usb_simple_msg_callback, netdev);
 	usb_anchor_urb(urb, &priv->tx_submitted);
 
 	err = usb_submit_urb(urb, GFP_ATOMIC);
author maintainer
#5 lore raw

PATCH 5/51 can: kvaser_usb: Retry the first bulk transfer on -ETIMEDOUT

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: "Ahmed S. Darwish" <[email protected]>

commit 14c10c2a1dd8eb8e00b750b521753260befa2789 upstream.

On some x86 laptops, plugging a Kvaser device again after an unplug makes the firmware always ignore the very first command. For such a case, provide some room for retries instead of completely exiting the driver init code.

---
 drivers/net/can/usb/kvaser_usb.c |   12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)
drivers/net/can/usb/kvaser_usb.c +10 -2
@@ -1578,7 +1578,7 @@ static int kvaser_usb_probe(struct usb_i
 {
 	struct kvaser_usb *dev;
 	int err = -ENOMEM;
-	int i;
+	int i, retry = 3;
 
 	dev = devm_kzalloc(&intf->dev, sizeof(*dev), GFP_KERNEL);
 	if (!dev)
@@ -1596,7 +1596,15 @@ static int kvaser_usb_probe(struct usb_i
 
 	usb_set_intfdata(intf, dev);
 
-	err = kvaser_usb_get_software_info(dev);
+	/* On some x86 laptops, plugging a Kvaser device again after
+	 * an unplug makes the firmware always ignore the very first
+	 * command. For such a case, provide some room for retries
+	 * instead of completely exiting the driver.
+	 */
+	do {
+		err = kvaser_usb_get_software_info(dev);
+	} while (--retry && err == -ETIMEDOUT);
+
 	if (err) {
 		dev_err(&intf->dev,
 			"Cannot get software infos, error %d\n", err);
author maintainer
#6 lore raw

PATCH 6/51 can: kvaser_usb: Fix state handling upon BUS_ERROR events

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: "Ahmed S. Darwish" <[email protected]>

commit e638642b08c170d2021b706f0b1c4f4ae93d8cbd upstream.

While being in an ERROR_WARNING state, and receiving further bus error events with error counters still in the ERROR_WARNING range of 97-127 inclusive, the state handling code erroneously reverts back to ERROR_ACTIVE.

Per the CAN standard, only revert to ERROR_ACTIVE when the error counters are less than 96.

Moreover, in certain Kvaser models, the BUS_ERROR flag is always set along with undefined bits in the M16C status register. Thus use bitwise operators instead of full equality for checking that register against bus errors.

---
 drivers/net/can/usb/kvaser_usb.c |    7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)
drivers/net/can/usb/kvaser_usb.c +3 -4
@@ -690,9 +690,7 @@ static void kvaser_usb_rx_error(const st
 		}
 
 		new_state = CAN_STATE_ERROR_PASSIVE;
-	}
-
-	if (status == M16C_STATE_BUS_ERROR) {
+	} else if (status & M16C_STATE_BUS_ERROR) {
 		if ((priv->can.state < CAN_STATE_ERROR_WARNING) &&
 		    ((txerr >= 96) || (rxerr >= 96))) {
 			cf->can_id |= CAN_ERR_CRTL;
@@ -702,7 +700,8 @@ static void kvaser_usb_rx_error(const st
 
 			priv->can.can_stats.error_warning++;
 			new_state = CAN_STATE_ERROR_WARNING;
-		} else if (priv->can.state > CAN_STATE_ERROR_ACTIVE) {
+		} else if ((priv->can.state > CAN_STATE_ERROR_ACTIVE) &&
+			   ((txerr < 96) && (rxerr < 96))) {
 			cf->can_id |= CAN_ERR_PROT;
 			cf->data[2] = CAN_ERR_PROT_ACTIVE;
 
author maintainer
#7 lore raw

PATCH 7/51 powerpc/xmon: Fix another endiannes issue in RTAS call from xmon

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Laurent Dufour <[email protected]>

commit e6eb2eba494d6f99e69ca3c3748cd37a2544ab38 upstream.

The commit 3b8a3c010969 ("powerpc/pseries: Fix endiannes issue in RTAS call from xmon") was fixing an endianness issue in the call made from xmon to RTAS.

However, as Michael Ellerman noticed, this fix was not complete, the token value was not byte swapped. This lead to call an unexpected and most of the time unexisting RTAS function, which is silently ignored by RTAS.

This fix addresses this hole.

---
 arch/powerpc/xmon/xmon.c |    1 +
 1 file changed, 1 insertion(+)
arch/powerpc/xmon/xmon.c +1 -0
@@ -288,6 +288,7 @@ static inline void disable_surveillance(
 	args.token = rtas_token("set-indicator");
 	if (args.token == RTAS_UNKNOWN_SERVICE)
 		return;
+	args.token = cpu_to_be32(args.token);
 	args.nargs = cpu_to_be32(3);
 	args.nret = cpu_to_be32(1);
 	args.rets = &args.args[3];
author maintainer
#8 lore raw

PATCH 8/51 ALSA: seq-dummy: remove deadlock-causing events on close

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Clemens Ladisch <[email protected]>

commit 0767e95bb96d7fdddcd590fb809e6975d93aebc5 upstream.

When the last subscriber to a "Through" port has been removed, the subscribed destination ports might still be active, so it would be wrong to send "all sounds off" and "reset controller" events to them. The proper place for such a shutdown would be the closing of the actual MIDI port (and close_substream() in rawmidi.c already can do this).

This also fixes a deadlock when dummy_unuse() tries to send events to its own port that is already locked because it is being freed.

---
 sound/core/seq/seq_dummy.c |   31 -------------------------------
 1 file changed, 31 deletions(-)
sound/core/seq/seq_dummy.c +0 -31
@@ -82,36 +82,6 @@ struct snd_seq_dummy_port {
 static int my_client = -1;
 
 /*
- * unuse callback - send ALL_SOUNDS_OFF and RESET_CONTROLLERS events
- * to subscribers.
- * Note: this callback is called only after all subscribers are removed.
- */
-static int
-dummy_unuse(void *private_data, struct snd_seq_port_subscribe *info)
-{
-	struct snd_seq_dummy_port *p;
-	int i;
-	struct snd_seq_event ev;
-
-	p = private_data;
-	memset(&ev, 0, sizeof(ev));
-	if (p->duplex)
-		ev.source.port = p->connect;
-	else
-		ev.source.port = p->port;
-	ev.dest.client = SNDRV_SEQ_ADDRESS_SUBSCRIBERS;
-	ev.type = SNDRV_SEQ_EVENT_CONTROLLER;
-	for (i = 0; i < 16; i++) {
-		ev.data.control.channel = i;
-		ev.data.control.param = MIDI_CTL_ALL_SOUNDS_OFF;
-		snd_seq_kernel_client_dispatch(p->client, &ev, 0, 0);
-		ev.data.control.param = MIDI_CTL_RESET_CONTROLLERS;
-		snd_seq_kernel_client_dispatch(p->client, &ev, 0, 0);
-	}
-	return 0;
-}
-
-/*
  * event input callback - just redirect events to subscribers
  */
 static int
@@ -175,7 +145,6 @@ create_port(int idx, int type)
 		| SNDRV_SEQ_PORT_TYPE_PORT;
 	memset(&pcb, 0, sizeof(pcb));
 	pcb.owner = THIS_MODULE;
-	pcb.unuse = dummy_unuse;
 	pcb.event_input = dummy_input;
 	pcb.private_free = dummy_free;
 	pcb.private_data = rec;
author maintainer
#9 lore raw

PATCH 9/51 Input: i8042 - add noloop quirk for Medion Akoya E7225 (MD98857)

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Jochen Hein <[email protected]>

commit 1d90d6d5522befa8efa1a7ea406be65cf865ded4 upstream.

Without this the aux port does not get detected, and consequently the touchpad will not work.

With this patch the touchpad is detected:

$ dmesg | grep -E "(SYN|i8042|serio)"
pnp 00:03: Plug and Play ACPI device, IDs SYN1d22 PNP0f13 (active)
i8042: PNP: PS/2 Controller [PNP0303:PS2K,PNP0f13:PS2M] at 0x60,0x64 irq 1,12
serio: i8042 KBD port at 0x60,0x64 irq 1
serio: i8042 AUX port at 0x60,0x64 irq 12
input: AT Translated Set 2 keyboard as /devices/platform/i8042/serio0/input/input4
psmouse serio1: synaptics: Touchpad model: 1, fw: 8.1, id: 0x1e2b1, caps: 0xd00123/0x840300/0x126800, board id: 2863, fw id: 1473085
input: SynPS/2 Synaptics TouchPad as /devices/platform/i8042/serio1/input/input6

dmidecode excerpt for this laptop is:

Handle 0x0001, DMI type 1, 27 bytes
System Information
Manufacturer: Medion
Product Name: Akoya E7225
Version: 1.0

---
 drivers/input/serio/i8042-x86ia64io.h |    8 ++++++++
 1 file changed, 8 insertions(+)
drivers/input/serio/i8042-x86ia64io.h +8 -0
@@ -152,6 +152,14 @@ static const struct dmi_system_id __init
 		},
 	},
 	{
+		/* Medion Akoya E7225 */
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "Medion"),
+			DMI_MATCH(DMI_PRODUCT_NAME, "Akoya E7225"),
+			DMI_MATCH(DMI_PRODUCT_VERSION, "1.0"),
+		},
+	},
+	{
 		/* Blue FB5601 */
 		.matches = {
 			DMI_MATCH(DMI_SYS_VENDOR, "blue"),
author maintainer
#10 lore raw

PATCH 10/51 nfs: fix dio deadlock when O_DIRECT flag is flipped

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Peng Tao <[email protected]>

commit ee8a1a8b160a87dc3a9c81a86796aa4db85ea815 upstream.

We only support swap file calling nfs_direct_IO. However, application might be able to get to nfs_direct_IO if it toggles O_DIRECT flag during IO and it can deadlock because we grab inode->i_mutex in nfs_file_direct_write(). So return 0 for such case. Then the generic layer will fall back to buffer IO.

---
 fs/nfs/direct.c |    6 ++++++
 1 file changed, 6 insertions(+)
fs/nfs/direct.c +6 -0
@@ -123,6 +123,12 @@ static inline int put_dreq(struct nfs_di
  */
 ssize_t nfs_direct_IO(int rw, struct kiocb *iocb, const struct iovec *iov, loff_t pos, unsigned long nr_segs)
 {
+	struct inode *inode = iocb->ki_filp->f_mapping->host;
+
+	/* we only support swap file calling nfs_direct_IO */
+	if (!IS_SWAPFILE(inode))
+		return 0;
+
 #ifndef CONFIG_NFS_SWAP
 	dprintk("NFS: nfs_direct_IO (%s) off/no(%Ld/%lu) EINVAL\n",
 			iocb->ki_filp->f_path.dentry->d_name.name,
author maintainer
#11 lore raw

PATCH 11/51 NFSv4.1: Fix an Oops in nfs41_walk_client_list

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Trond Myklebust <[email protected]>

commit 3175e1dcec40fab1a444c010087f2068b6b04732 upstream.

If we start state recovery on a client that failed to initialise correctly, then we are very likely to Oops.

---
 fs/nfs/nfs4client.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/nfs/nfs4client.c +1 -1
@@ -462,7 +462,7 @@ int nfs41_walk_client_list(struct nfs_cl
 			prev = pos;
 
 			status = nfs_wait_client_init_complete(pos);
-			if (status == 0) {
+			if (pos->cl_cons_state == NFS_CS_SESSION_INITING) {
 				nfs4_schedule_lease_recovery(pos);
 				status = nfs4_wait_clnt_recover(pos);
 			}
author maintainer
#12 lore raw

PATCH 12/51 nl80211: fix per-station group key get/del and memory leak

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Johannes Berg <[email protected]>

commit 0fa7b39131576dd1baa6ca17fca53c65d7f62249 upstream.

In case userspace attempts to obtain key information for or delete a unicast key, this is currently erroneously rejected unless the driver sets the WIPHY_FLAG_IBSS_RSN flag. Apparently enough drivers do so it was never noticed.

Fix that, and while at it fix a potential memory leak: the error path in the get_key() function was placed after allocating a message but didn't free it - move it to a better place. Luckily admin permissions are needed to call this operation.

Fixes e31b82136d1ad ("cfg80211/mac80211: allow per-station GTKs")
---
 net/wireless/nl80211.c |    9 ++++-----
 1 file changed, 4 insertions(+), 5 deletions(-)
net/wireless/nl80211.c +4 -5
@@ -2629,6 +2629,9 @@ static int nl80211_get_key(struct sk_buf
 	if (!rdev->ops->get_key)
 		return -EOPNOTSUPP;
 
+	if (!pairwise && mac_addr && !(rdev->wiphy.flags & WIPHY_FLAG_IBSS_RSN))
+		return -ENOENT;
+
 	msg = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL);
 	if (!msg)
 		return -ENOMEM;
@@ -2648,10 +2651,6 @@ static int nl80211_get_key(struct sk_buf
 	    nla_put(msg, NL80211_ATTR_MAC, ETH_ALEN, mac_addr))
 		goto nla_put_failure;
 
-	if (pairwise && mac_addr &&
-	    !(rdev->wiphy.flags & WIPHY_FLAG_IBSS_RSN))
-		return -ENOENT;
-
 	err = rdev_get_key(rdev, dev, key_idx, pairwise, mac_addr, &cookie,
 			   get_key_callback);
 
@@ -2822,7 +2821,7 @@ static int nl80211_del_key(struct sk_buf
 	wdev_lock(dev->ieee80211_ptr);
 	err = nl80211_key_allowed(dev->ieee80211_ptr);
 
-	if (key.type == NL80211_KEYTYPE_PAIRWISE && mac_addr &&
+	if (key.type == NL80211_KEYTYPE_GROUP && mac_addr &&
 	    !(rdev->wiphy.flags & WIPHY_FLAG_IBSS_RSN))
 		err = -ENOENT;
 
author maintainer
#13 lore raw

PATCH 13/51 dm thin: dont allow messages to be sent to a pool target in READ_ONLY or FAIL mode

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Joe Thornber <[email protected]>

commit 2a7eaea02b99b6e267b1e89c79acc6e9a51cee3b upstream.

You can't modify the metadata in these modes. It's better to fail these messages immediately than let the block-manager deny write locks on metadata blocks. Otherwise these failed metadata changes will trigger 'needs_check' to get set in the metadata superblock -- requiring repair using the thin_check utility.

---
 drivers/md/dm-thin.c |    6 ++++++
 1 file changed, 6 insertions(+)
drivers/md/dm-thin.c +6 -0
@@ -2457,6 +2457,12 @@ static int pool_message(struct dm_target
 	struct pool_c *pt = ti->private;
 	struct pool *pool = pt->pool;
 
+	if (get_pool_mode(pool) >= PM_READ_ONLY) {
+		DMERR("%s: unable to service pool target messages in READ_ONLY or FAIL mode",
+		      dm_device_name(pool->pool_md));
+		return -EINVAL;
+	}
+
 	if (!strcasecmp(argv[0], "create_thin"))
 		r = process_create_thin_mesg(argc, argv, pool);
 
author maintainer
#14 lore raw

PATCH 14/51 dm cache: fix missing ERR_PTR returns and handling

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Joe Thornber <[email protected]>

commit 766a78882ddf79b162243649d7dfdbac1fb6fb88 upstream.

Commit 9b1cc9f251 ("dm cache: share cache-metadata object across inactive and active DM tables") mistakenly ignored the use of ERR_PTR returns. Restore missing IS_ERR checks and ERR_PTR returns where appropriate.

---
 drivers/md/dm-cache-metadata.c |    9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)
drivers/md/dm-cache-metadata.c +5 -4
@@ -648,7 +648,7 @@ static struct dm_cache_metadata *metadat
 	cmd = kzalloc(sizeof(*cmd), GFP_KERNEL);
 	if (!cmd) {
 		DMERR("could not allocate metadata struct");
-		return NULL;
+		return ERR_PTR(-ENOMEM);
 	}
 
 	atomic_set(&cmd->ref_count, 1);
@@ -710,7 +710,7 @@ static struct dm_cache_metadata *lookup_
 		return cmd;
 
 	cmd = metadata_open(bdev, data_block_size, may_format_device, policy_hint_size);
-	if (cmd) {
+	if (!IS_ERR(cmd)) {
 		mutex_lock(&table_lock);
 		cmd2 = lookup(bdev);
 		if (cmd2) {
@@ -745,9 +745,10 @@ struct dm_cache_metadata *dm_cache_metad
 {
 	struct dm_cache_metadata *cmd = lookup_or_open(bdev, data_block_size,
 						       may_format_device, policy_hint_size);
-	if (cmd && !same_params(cmd, data_block_size)) {
+
+	if (!IS_ERR(cmd) && !same_params(cmd, data_block_size)) {
 		dm_cache_metadata_close(cmd);
-		return NULL;
+		return ERR_PTR(-EINVAL);
 	}
 
 	return cmd;
author maintainer
#15 lore raw

PATCH 15/51 spi/pxa2xx: Clear cur_chip pointer before starting next message

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Mika Westerberg <[email protected]>

commit c957e8f084e0d21febcd6b8a0ea9631eccc92f36 upstream.

Once the current message is finished, the driver notifies SPI core about this by calling spi_finalize_current_message(). This function queues next message to be transferred. If there are more messages in the queue, it is possible that the driver is asked to transfer the next message at this point.

When spi_finalize_current_message() returns the driver clears the drv_data->cur_chip pointer to NULL. The problem is that if the driver already started the next message clearing drv_data->cur_chip will cause NULL pointer dereference which crashes the kernel like:

 BUG: unable to handle kernel NULL pointer dereference at 0000000000000048
 IP: [<ffffffffa0022bc8>] cs_deassert+0x18/0x70 [spi_pxa2xx_platform]
 PGD 78bb8067 PUD 37712067 PMD 0
 Oops: 0000 [#1] SMP
 Modules linked in:
 CPU: 1 PID: 11 Comm: ksoftirqd/1 Tainted: G           O   3.18.0-rc4-mjo #5
 Hardware name: Intel Corp. VALLEYVIEW B3 PLATFORM/NOTEBOOK, BIOS MNW2CRB1.X64.0071.R30.1408131301 08/13/2014
 task: ffff880077f9f290 ti: ffff88007a820000 task.ti: ffff88007a820000
 RIP: 0010:[<ffffffffa0022bc8>]  [<ffffffffa0022bc8>] cs_deassert+0x18/0x70 [spi_pxa2xx_platform]
 RSP: 0018:ffff88007a823d08  EFLAGS: 00010202
 RAX: 0000000000000008 RBX: ffff8800379a4430 RCX: 0000000000000026
 RDX: 0000000000000000 RSI: 0000000000000246 RDI: ffff8800379a4430
 RBP: ffff88007a823d18 R08: 00000000ffffffff R09: 000000007a9bc65a
 R10: 000000000000028f R11: 0000000000000005 R12: ffff880070123e98
 R13: ffff880070123de8 R14: 0000000000000100 R15: ffffc90004888000
 FS:  0000000000000000(0000) GS:ffff880079a80000(0000) knlGS:0000000000000000
 CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
 CR2: 0000000000000048 CR3: 000000007029b000 CR4: 00000000001007e0
 Stack:
  ffff88007a823d58 ffff8800379a4430 ffff88007a823d48 ffffffffa0022c89
  0000000000000000 ffff8800379a4430 0000000000000000 0000000000000006
  ffff88007a823da8 ffffffffa0023be0 ffff88007a823dd8 ffffffff81076204
 Call Trace:
  [<ffffffffa0022c89>] giveback+0x69/0xa0 [spi_pxa2xx_platform]
  [<ffffffffa0023be0>] pump_transfers+0x710/0x740 [spi_pxa2xx_platform]
  [<ffffffff81076204>] ? pick_next_task_fair+0x744/0x830
  [<ffffffff81049679>] tasklet_action+0xa9/0xe0
  [<ffffffff81049a0e>] __do_softirq+0xee/0x280
  [<ffffffff81049bc0>] run_ksoftirqd+0x20/0x40
  [<ffffffff810646df>] smpboot_thread_fn+0xff/0x1b0
  [<ffffffff810645e0>] ? SyS_setgroups+0x150/0x150
  [<ffffffff81060f9d>] kthread+0xcd/0xf0
  [<ffffffff81060ed0>] ? kthread_create_on_node+0x180/0x180
  [<ffffffff8187a82c>] ret_from_fork+0x7c/0xb0

Fix this by clearing drv_data->cur_chip before we call spi_finalize_current_message().

---
 drivers/spi/spi-pxa2xx.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/spi/spi-pxa2xx.c +1 -1
@@ -393,8 +393,8 @@ static void giveback(struct driver_data
 			cs_deassert(drv_data);
 	}
 
-	spi_finalize_current_message(drv_data->master);
 	drv_data->cur_chip = NULL;
+	spi_finalize_current_message(drv_data->master);
 }
 
 static void reset_sccr1(struct driver_data *drv_data)
author maintainer
#16 lore raw

PATCH 16/51 regulator: core: fix race condition in regulator_put()

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Ashay Jaiswal <[email protected]>

commit 83b0302d347a49f951e904184afe57ac3723476e upstream.

The regulator framework maintains a list of consumer regulators for a regulator device and protects it from concurrent access using the regulator device's mutex lock.

In the case of regulator_put() the consumer is removed and regulator device's parameters are updated without holding the regulator device's mutex. This would lead to a race condition between the regulator_put() and any function which traverses the consumer list or modifies regulator device's parameters.
Fix this race condition by holding the regulator device's mutex in case of regulator_put.

---
 drivers/regulator/core.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/regulator/core.c +3 -1
@@ -1410,7 +1410,7 @@ struct regulator *regulator_get_exclusiv
 }
 EXPORT_SYMBOL_GPL(regulator_get_exclusive);
 
-/* Locks held by regulator_put() */
+/* regulator_list_mutex lock held by regulator_put() */
 static void _regulator_put(struct regulator *regulator)
 {
 	struct regulator_dev *rdev;
@@ -1425,12 +1425,14 @@ static void _regulator_put(struct regula
 	/* remove any sysfs entries */
 	if (regulator->dev)
 		sysfs_remove_link(&rdev->dev.kobj, regulator->supply_name);
+	mutex_lock(&rdev->mutex);
 	kfree(regulator->supply_name);
 	list_del(&regulator->list);
 	kfree(regulator);
 
 	rdev->open_count--;
 	rdev->exclusive = 0;
+	mutex_unlock(&rdev->mutex);
 
 	module_put(rdev->owner);
 }
author maintainer
#17 lore raw

PATCH 17/51 drivers: net: cpsw: discard dual emac default vlan configuration

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Mugunthan V N <[email protected]>

commit 02a54164c52ed6eca3089a0d402170fbf34d6cf5 upstream.

In Dual EMAC, the default VLANs are used to segregate Rx packets between the ports, so adding the same default VLAN to the switch will affect the normal packet transfers. So returning error on addition of dual EMAC default VLANs.

Even if EMAC 0 default port VLAN is added to EMAC 1, it will lead to break dual EMAC port separations.

Fixes d9ba8f9e6298 (driver: net: ethernet: cpsw: dual emac interface implementation)
---
 drivers/net/ethernet/ti/cpsw.c |   22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)
drivers/net/ethernet/ti/cpsw.c +22 -0
@@ -1293,6 +1293,19 @@ static int cpsw_ndo_vlan_rx_add_vid(stru
 	if (vid == priv->data.default_vlan)
 		return 0;
 
+	if (priv->data.dual_emac) {
+		/* In dual EMAC, reserved VLAN id should not be used for
+		 * creating VLAN interfaces as this can break the dual
+		 * EMAC port separation
+		 */
+		int i;
+
+		for (i = 0; i < priv->data.slaves; i++) {
+			if (vid == priv->slaves[i].port_vlan)
+				return -EINVAL;
+		}
+	}
+
 	dev_info(priv->dev, "Adding vlanid %d to vlan filter\n", vid);
 	return cpsw_add_vlan_ale_entry(priv, vid);
 }
@@ -1306,6 +1319,15 @@ static int cpsw_ndo_vlan_rx_kill_vid(str
 	if (vid == priv->data.default_vlan)
 		return 0;
 
+	if (priv->data.dual_emac) {
+		int i;
+
+		for (i = 0; i < priv->data.slaves; i++) {
+			if (vid == priv->slaves[i].port_vlan)
+				return -EINVAL;
+		}
+	}
+
 	dev_info(priv->dev, "removing vlanid %d from vlan filter\n", vid);
 	ret = cpsw_ale_del_vlan(priv->ale, vid, 0);
 	if (ret != 0)
author maintainer
#18 lore raw

PATCH 18/51 ARM: 7829/1: Add ".text.unlikely" and ".text.hot" to arm unwind tables

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Douglas Anderson <[email protected]>

commit 849b882b52df0f276d9ffded01d85654aa0da422 upstream.

It appears that gcc may put some code in ".text.unlikely" or ".text.hot" sections. Right now those aren't accounted for in unwind tables. Add them.

I found some docs about this at:
http://gcc.gnu.org/onlinedocs/gcc-4.6.2/gcc.pdf

Without this, if you have slub_debug turned on, you can get messages that look like this:
unwind: Index not found 7f008c50

[wangkai: backport to 3.10
- adjust context
]

---
 arch/arm/include/asm/module.h |    2 ++
 arch/arm/kernel/module.c      |    8 ++++++++
 2 files changed, 10 insertions(+)
arch/arm/include/asm/module.h +2 -0
@@ -12,6 +12,8 @@ enum {
 	ARM_SEC_CORE,
 	ARM_SEC_EXIT,
 	ARM_SEC_DEVEXIT,
+	ARM_SEC_HOT,
+	ARM_SEC_UNLIKELY,
 	ARM_SEC_MAX,
 };
 
arch/arm/kernel/module.c +8 -0
@@ -296,6 +296,10 @@ int module_finalize(const Elf32_Ehdr *hd
 			maps[ARM_SEC_EXIT].unw_sec = s;
 		else if (strcmp(".ARM.exidx.devexit.text", secname) == 0)
 			maps[ARM_SEC_DEVEXIT].unw_sec = s;
+		else if (strcmp(".ARM.exidx.text.unlikely", secname) == 0)
+			maps[ARM_SEC_UNLIKELY].unw_sec = s;
+		else if (strcmp(".ARM.exidx.text.hot", secname) == 0)
+			maps[ARM_SEC_HOT].unw_sec = s;
 		else if (strcmp(".init.text", secname) == 0)
 			maps[ARM_SEC_INIT].txt_sec = s;
 		else if (strcmp(".devinit.text", secname) == 0)
@@ -306,6 +310,10 @@ int module_finalize(const Elf32_Ehdr *hd
 			maps[ARM_SEC_EXIT].txt_sec = s;
 		else if (strcmp(".devexit.text", secname) == 0)
 			maps[ARM_SEC_DEVEXIT].txt_sec = s;
+		else if (strcmp(".text.unlikely", secname) == 0)
+			maps[ARM_SEC_UNLIKELY].txt_sec = s;
+		else if (strcmp(".text.hot", secname) == 0)
+			maps[ARM_SEC_HOT].txt_sec = s;
 	}
 
 	for (i = 0; i < ARM_SEC_MAX; i++)
author maintainer
#19 lore raw

PATCH 19/51 ARM: mm: correct pte_same behaviour for LPAE.

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Steve Capper <[email protected]>

commit dde1b65110353517816bcbc58539463396202244 upstream.

For 3 levels of paging the PTE_EXT_NG bit will be set for user address ptes that are written to a page table but not for ptes created with mk_pte.

This can cause some comparison tests made by pte_same to fail spuriously and lead to other problems.

To correct this behaviour, we mask off PTE_EXT_NG for any pte that is present before running the comparison.

---
 arch/arm/include/asm/pgtable-3level.h |   17 +++++++++++++++++
 1 file changed, 17 insertions(+)
arch/arm/include/asm/pgtable-3level.h +17 -0
@@ -166,6 +166,23 @@ static inline pmd_t *pmd_offset(pud_t *p
 		clean_pmd_entry(pmdp);	\
 	} while (0)
 
+/*
+ * For 3 levels of paging the PTE_EXT_NG bit will be set for user address ptes
+ * that are written to a page table but not for ptes created with mk_pte.
+ *
+ * In hugetlb_no_page, a new huge pte (new_pte) is generated and passed to
+ * hugetlb_cow, where it is compared with an entry in a page table.
+ * This comparison test fails erroneously leading ultimately to a memory leak.
+ *
+ * To correct this behaviour, we mask off PTE_EXT_NG for any pte that is
+ * present before running the comparison.
+ */
+#define __HAVE_ARCH_PTE_SAME
+#define pte_same(pte_a,pte_b)	((pte_present(pte_a) ? pte_val(pte_a) & ~PTE_EXT_NG	\
+					: pte_val(pte_a))				\
+				== (pte_present(pte_b) ? pte_val(pte_b) & ~PTE_EXT_NG	\
+					: pte_val(pte_b)))
+
 #define set_pte_ext(ptep,pte,ext) cpu_set_pte_ext(ptep,__pte(pte_val(pte)|(ext)))
 
 #endif /* __ASSEMBLY__ */
author maintainer
#20 lore raw

PATCH 20/51 ARM: LPAE: use signed arithmetic for mask definitions

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Cyril Chemparathy <[email protected]>

commit 926edcc747e2efb3c9add7ed4dbc4e7a3a959d02 upstream.

This patch applies to PAGE_MASK, PMD_MASK, and PGDIR_MASK, where forcing unsigned long math truncates the mask at the 32-bits. This clearly does bad things on PAE systems.

This patch fixes this problem by defining these masks as signed quantities. We then rely on sign extension to do the right thing.

---
 arch/arm/include/asm/page.h           |    2 +-
 arch/arm/include/asm/pgtable-3level.h |    6 +++---
 2 files changed, 4 insertions(+), 4 deletions(-)
arch/arm/include/asm/page.h +1 -1
@@ -13,7 +13,7 @@
 /* PAGE_SHIFT determines the page size */
 #define PAGE_SHIFT		12
 #define PAGE_SIZE		(_AC(1,UL) << PAGE_SHIFT)
-#define PAGE_MASK		(~(PAGE_SIZE-1))
+#define PAGE_MASK		(~((1 << PAGE_SHIFT) - 1))
 
 #ifndef __ASSEMBLY__
 
arch/arm/include/asm/pgtable-3level.h +3 -3
@@ -48,16 +48,16 @@
 #define PMD_SHIFT		21
 
 #define PMD_SIZE		(1UL << PMD_SHIFT)
-#define PMD_MASK		(~(PMD_SIZE-1))
+#define PMD_MASK		(~((1 << PMD_SHIFT) - 1))
 #define PGDIR_SIZE		(1UL << PGDIR_SHIFT)
-#define PGDIR_MASK		(~(PGDIR_SIZE-1))
+#define PGDIR_MASK		(~((1 << PGDIR_SHIFT) - 1))
 
 /*
  * section address mask and size definitions.
  */
 #define SECTION_SHIFT		21
 #define SECTION_SIZE		(1UL << SECTION_SHIFT)
-#define SECTION_MASK		(~(SECTION_SIZE-1))
+#define SECTION_MASK		(~((1 << SECTION_SHIFT) - 1))
 
 #define USER_PTRS_PER_PGD	(PAGE_OFFSET / PGDIR_SIZE)
 
author maintainer
#21 lore raw

PATCH 21/51 ARM: LPAE: use phys_addr_t in alloc_init_pud()

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Vitaly Andrianov <[email protected]>

commit 20d6956d8cd2452cec0889ff040f18afc03c2e6b upstream.

This patch fixes the alloc_init_pud() function to use phys_addr_t instead of unsigned long when passing in the phys argument.

This is an extension to commit 97092e0c56830457af0639f6bd904537a150ea4a (ARM:
pgtable: use phys_addr_t for physical addresses), which applied similar changes elsewhere in the ARM memory management code.

---
 arch/arm/mm/mmu.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
arch/arm/mm/mmu.c +2 -1
@@ -685,7 +685,8 @@ static void __init alloc_init_pmd(pud_t
 }
 
 static void __init alloc_init_pud(pgd_t *pgd, unsigned long addr,
-	unsigned long end, unsigned long phys, const struct mem_type *type)
+				  unsigned long end, phys_addr_t phys,
+				  const struct mem_type *type)
 {
 	pud_t *pud = pud_offset(pgd, addr);
 	unsigned long next;
author maintainer
#22 lore raw

PATCH 22/51 ARM: fix type of PHYS_PFN_OFFSET to unsigned long

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Cyril Chemparathy <[email protected]>

commit 5b20c5b2f014ecc0a6310988af69cd7ede9e7c67 upstream.

On LPAE machines, PHYS_OFFSET evaluates to a phys_addr_t and this type is inherited by the PHYS_PFN_OFFSET definition as well. Consequently, the kernel build emits warnings of the form:

init/main.c: In function 'start_kernel':
init/main.c:588:7: warning: format '%lx' expects argument of type 'long unsigned int', but argument 2 has type 'phys_addr_t' [-Wformat]

This patch fixes this warning by pinning down the PFN type to unsigned long.

---
 arch/arm/include/asm/memory.h |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/arm/include/asm/memory.h +1 -1
@@ -207,7 +207,7 @@ static inline unsigned long __phys_to_vi
  * direct-mapped view.  We assume this is the first page
  * of RAM in the mem_map as well.
  */
-#define PHYS_PFN_OFFSET	(PHYS_OFFSET >> PAGE_SHIFT)
+#define PHYS_PFN_OFFSET	((unsigned long)(PHYS_OFFSET >> PAGE_SHIFT))
 
 /*
  * These are *only* valid on the kernel direct mapped RAM memory.
author maintainer
#23 lore raw

PATCH 23/51 ARM: lpae: fix definition of PTE_HWTABLE_PTRS

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Will Deacon <[email protected]>

commit e38a517578d6c0f764b0d0f6e26dcdf9f70c69d7 upstream.

For 2-level page tables, PTE_HWTABLE_PTRS describes the offset between Linux PTEs and hardware PTEs. On LPAE, there is no distinction (since we have 64-bit descriptors with plenty of space) so PTE_HWTABLE_PTRS should be 0. Unfortunately, it is wrongly defined as PTRS_PER_PTE, meaning that current pte table flushing is off by a page. Luckily, all current LPAE implementations are SMP, so the hardware walker can snoop L1.

This patch fixes the broken definition.

---
 arch/arm/include/asm/pgtable-3level.h |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/arm/include/asm/pgtable-3level.h +1 -1
@@ -33,7 +33,7 @@
 #define PTRS_PER_PMD		512
 #define PTRS_PER_PGD		4
 
-#define PTE_HWTABLE_PTRS	(PTRS_PER_PTE)
+#define PTE_HWTABLE_PTRS	(0)
 #define PTE_HWTABLE_OFF		(0)
 #define PTE_HWTABLE_SIZE	(PTRS_PER_PTE * sizeof(u64))
 
author maintainer
#24 lore raw

PATCH 24/51 ARM: 7866/1: include: asm: use long long instead of u64 within atomic.h

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Chen Gang <[email protected]>

commit 237f12337cfa2175474e4dd015bc07a25eb9080d upstream.

atomic* value is signed value, and atomic* functions need also process signed value (parameter value, and return value), so 32-bit arm need use 'long long' instead of 'u64'.

After replacement, it will also fix a bug for atomic64_add_negative():
"u64 is never less than 0".

The modifications are:

in vim, use "1,% s/\<u64\>/long long/g" command. remove '__aligned(8)' which is useless for 64-bit. be sure of 80 column limitation after replacement.

---
 arch/arm/include/asm/atomic.h |   49 +++++++++++++++++++++---------------------
 1 file changed, 25 insertions(+), 24 deletions(-)
arch/arm/include/asm/atomic.h +25 -24
@@ -238,15 +238,15 @@ static inline int __atomic_add_unless(at
 
 #ifndef CONFIG_GENERIC_ATOMIC64
 typedef struct {
-	u64 __aligned(8) counter;
+	long long counter;
 } atomic64_t;
 
 #define ATOMIC64_INIT(i) { (i) }
 
 #ifdef CONFIG_ARM_LPAE
-static inline u64 atomic64_read(const atomic64_t *v)
+static inline long long atomic64_read(const atomic64_t *v)
 {
-	u64 result;
+	long long result;
 
 	__asm__ __volatile__("@ atomic64_read\n"
 "	ldrd	%0, %H0, [%1]"
@@ -257,7 +257,7 @@ static inline u64 atomic64_read(const at
 	return result;
 }
 
-static inline void atomic64_set(atomic64_t *v, u64 i)
+static inline void atomic64_set(atomic64_t *v, long long i)
 {
 	__asm__ __volatile__("@ atomic64_set\n"
 "	strd	%2, %H2, [%1]"
@@ -266,9 +266,9 @@ static inline void atomic64_set(atomic64
 	);
 }
 #else
-static inline u64 atomic64_read(const atomic64_t *v)
+static inline long long atomic64_read(const atomic64_t *v)
 {
-	u64 result;
+	long long result;
 
 	__asm__ __volatile__("@ atomic64_read\n"
 "	ldrexd	%0, %H0, [%1]"
@@ -279,9 +279,9 @@ static inline u64 atomic64_read(const at
 	return result;
 }
 
-static inline void atomic64_set(atomic64_t *v, u64 i)
+static inline void atomic64_set(atomic64_t *v, long long i)
 {
-	u64 tmp;
+	long long tmp;
 
 	__asm__ __volatile__("@ atomic64_set\n"
 "1:	ldrexd	%0, %H0, [%2]\n"
@@ -294,9 +294,9 @@ static inline void atomic64_set(atomic64
 }
 #endif
 
-static inline void atomic64_add(u64 i, atomic64_t *v)
+static inline void atomic64_add(long long i, atomic64_t *v)
 {
-	u64 result;
+	long long result;
 	unsigned long tmp;
 
 	__asm__ __volatile__("@ atomic64_add\n"
@@ -311,9 +311,9 @@ static inline void atomic64_add(u64 i, a
 	: "cc");
 }
 
-static inline u64 atomic64_add_return(u64 i, atomic64_t *v)
+static inline long long atomic64_add_return(long long i, atomic64_t *v)
 {
-	u64 result;
+	long long result;
 	unsigned long tmp;
 
 	smp_mb();
@@ -334,9 +334,9 @@ static inline u64 atomic64_add_return(u6
 	return result;
 }
 
-static inline void atomic64_sub(u64 i, atomic64_t *v)
+static inline void atomic64_sub(long long i, atomic64_t *v)
 {
-	u64 result;
+	long long result;
 	unsigned long tmp;
 
 	__asm__ __volatile__("@ atomic64_sub\n"
@@ -351,9 +351,9 @@ static inline void atomic64_sub(u64 i, a
 	: "cc");
 }
 
-static inline u64 atomic64_sub_return(u64 i, atomic64_t *v)
+static inline long long atomic64_sub_return(long long i, atomic64_t *v)
 {
-	u64 result;
+	long long result;
 	unsigned long tmp;
 
 	smp_mb();
@@ -374,9 +374,10 @@ static inline u64 atomic64_sub_return(u6
 	return result;
 }
 
-static inline u64 atomic64_cmpxchg(atomic64_t *ptr, u64 old, u64 new)
+static inline long long atomic64_cmpxchg(atomic64_t *ptr, long long old,
+					long long new)
 {
-	u64 oldval;
+	long long oldval;
 	unsigned long res;
 
 	smp_mb();
@@ -398,9 +399,9 @@ static inline u64 atomic64_cmpxchg(atomi
 	return oldval;
 }
 
-static inline u64 atomic64_xchg(atomic64_t *ptr, u64 new)
+static inline long long atomic64_xchg(atomic64_t *ptr, long long new)
 {
-	u64 result;
+	long long result;
 	unsigned long tmp;
 
 	smp_mb();
@@ -419,9 +420,9 @@ static inline u64 atomic64_xchg(atomic64
 	return result;
 }
 
-static inline u64 atomic64_dec_if_positive(atomic64_t *v)
+static inline long long atomic64_dec_if_positive(atomic64_t *v)
 {
-	u64 result;
+	long long result;
 	unsigned long tmp;
 
 	smp_mb();
@@ -445,9 +446,9 @@ static inline u64 atomic64_dec_if_positi
 	return result;
 }
 
-static inline int atomic64_add_unless(atomic64_t *v, u64 a, u64 u)
+static inline int atomic64_add_unless(atomic64_t *v, long long a, long long u)
 {
-	u64 val;
+	long long val;
 	unsigned long tmp;
 	int ret = 1;
 
author maintainer
#25 lore raw

PATCH 25/51 ARM: 7867/1: include: asm: use int instead of unsigned long for oldval in atomic_cmpxchg().

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Chen Gang <[email protected]>

commit 4dcc1cf7316a26e112f5c9fcca531ff98ef44700 upstream.

For atomic_cmpxchg(), the type of 'oldval' need be 'int' to match the type of "*ptr" (used by 'ldrex' instruction) and 'old' (used by 'teq' instruction).

---
 arch/arm/include/asm/atomic.h |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
arch/arm/include/asm/atomic.h +2 -1
@@ -114,7 +114,8 @@ static inline int atomic_sub_return(int
 
 static inline int atomic_cmpxchg(atomic_t *ptr, int old, int new)
 {
-	unsigned long oldval, res;
+	int oldval;
+	unsigned long res;
 
 	smp_mb();
 
author maintainer
#26 lore raw

PATCH 26/51 ARM: fix asm/memory.h build error

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Russell King <[email protected]>

commit b713aa0b15015a65ad5421543b80df86de043d62 upstream.

Jason Gunthorpe reports a build failure when ARM_PATCH_PHYS_VIRT is not defined:

In file included from arch/arm/include/asm/page.h:163:0,
from include/linux/mm_types.h:16,
from include/linux/sched.h:24,
from arch/arm/kernel/asm-offsets.c:13:
arch/arm/include/asm/memory.h: In function '__virt_to_phys':
arch/arm/include/asm/memory.h:244:40: error: 'PHYS_OFFSET' undeclared (first use in this function) arch/arm/include/asm/memory.h:244:40: note: each undeclared identifier is reported only once for each function it appears in arch/arm/include/asm/memory.h: In function '__phys_to_virt':
arch/arm/include/asm/memory.h:249:13: error: 'PHYS_OFFSET' undeclared (first use in this function)

Fixes ca5a45c06cd4 ("ARM: mm: use phys_addr_t appropriately in p2v and v2p conversions")

[hpy: Backported to 3.10:
- adjust the context
- MPU is not supported by 3.10, so ignore fix to MPU compared with the original patch.]

---
 arch/arm/include/asm/memory.h |   31 +++++++++++++++----------------
 arch/arm/kernel/head.S        |    2 +-
 2 files changed, 16 insertions(+), 17 deletions(-)
arch/arm/include/asm/memory.h +15 -16
@@ -98,23 +98,19 @@
 #define TASK_UNMAPPED_BASE	UL(0x00000000)
 #endif
 
-#ifndef PHYS_OFFSET
-#define PHYS_OFFSET 		UL(CONFIG_DRAM_BASE)
-#endif
-
 #ifndef END_MEM
 #define END_MEM     		(UL(CONFIG_DRAM_BASE) + CONFIG_DRAM_SIZE)
 #endif
 
 #ifndef PAGE_OFFSET
-#define PAGE_OFFSET		(PHYS_OFFSET)
+#define PAGE_OFFSET		PLAT_PHYS_OFFSET
 #endif
 
 /*
  * The module can be at any place in ram in nommu mode.
  */
 #define MODULES_END		(END_MEM)
-#define MODULES_VADDR		(PHYS_OFFSET)
+#define MODULES_VADDR		PAGE_OFFSET
 
 #define XIP_VIRT_ADDR(physaddr)  (physaddr)
 
@@ -141,6 +137,16 @@
 #define page_to_phys(page)	(__pfn_to_phys(page_to_pfn(page)))
 #define phys_to_page(phys)	(pfn_to_page(__phys_to_pfn(phys)))
 
+/*
+ * PLAT_PHYS_OFFSET is the offset (from zero) of the start of physical
+ * memory.  This is used for XIP and NoMMU kernels, or by kernels which
+ * have their own mach/memory.h.  Assembly code must always use
+ * PLAT_PHYS_OFFSET and not PHYS_OFFSET.
+ */
+#ifndef PLAT_PHYS_OFFSET
+#define PLAT_PHYS_OFFSET	UL(CONFIG_PHYS_OFFSET)
+#endif
+
 #ifndef __ASSEMBLY__
 
 /*
@@ -183,22 +189,15 @@ static inline unsigned long __phys_to_vi
 	return t;
 }
 #else
+
+#define PHYS_OFFSET	PLAT_PHYS_OFFSET
+
 #define __virt_to_phys(x)	((x) - PAGE_OFFSET + PHYS_OFFSET)
 #define __phys_to_virt(x)	((x) - PHYS_OFFSET + PAGE_OFFSET)
-#endif
-#endif
-#endif /* __ASSEMBLY__ */
 
-#ifndef PHYS_OFFSET
-#ifdef PLAT_PHYS_OFFSET
-#define PHYS_OFFSET	PLAT_PHYS_OFFSET
-#else
-#define PHYS_OFFSET	UL(CONFIG_PHYS_OFFSET)
 #endif
 #endif
 
-#ifndef __ASSEMBLY__
-
 /*
  * PFNs are used to describe any physical page; this means
  * PFN 0 == physical address 0.
arch/arm/kernel/head.S +1 -1
@@ -109,7 +109,7 @@ ENTRY(stext)
 	sub	r4, r3, r4			@ (PHYS_OFFSET - PAGE_OFFSET)
 	add	r8, r8, r4			@ PHYS_OFFSET
 #else
-	ldr	r8, =PHYS_OFFSET		@ always constant in this case
+	ldr	r8, =PLAT_PHYS_OFFSET		@ always constant in this case
 #endif
 
 	/*
author maintainer
#27 lore raw

PATCH 27/51 ARM: 7931/1: Correct virt_addr_valid

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Laura Abbott <[email protected]>

commit efea3403d4b7c6d1dd5d5ac3234c161e8b314d66 upstream.

The definition of virt_addr_valid is that virt_addr_valid should return true if and only if virt_to_page returns a valid pointer. The current definition of virt_addr_valid only checks against the virtual address range. There's no guarantee that just because a virtual address falls bewteen PAGE_OFFSET and high_memory the associated physical memory has a valid backing struct page. Follow the example of other architectures and convert to pfn_valid to verify that the virtual address is actually valid. The check for an address between PAGE_OFFSET and high_memory is still necessary as vmalloc/highmem addresses are not valid with virt_to_page.

---
 arch/arm/include/asm/memory.h |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
arch/arm/include/asm/memory.h +2 -1
@@ -274,7 +274,8 @@ static inline __deprecated void *bus_to_
 #define ARCH_PFN_OFFSET		PHYS_PFN_OFFSET
 
 #define virt_to_page(kaddr)	pfn_to_page(__pa(kaddr) >> PAGE_SHIFT)
-#define virt_addr_valid(kaddr)	((unsigned long)(kaddr) >= PAGE_OFFSET && (unsigned long)(kaddr) < (unsigned long)high_memory)
+#define virt_addr_valid(kaddr)	(((unsigned long)(kaddr) >= PAGE_OFFSET && (unsigned long)(kaddr) < (unsigned long)high_memory) \
+					&& pfn_valid(__pa(kaddr) >> PAGE_SHIFT) )
 
 #endif
 
author maintainer
#28 lore raw

PATCH 28/51 ARM: DMA: ensure that old section mappings are flushed from the TLB

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Russell King <[email protected]>

commit 6b076991dca9817e75c37e2f0db6d52611ea42fa upstream.

When setting up the CMA region, we must ensure that the old section mappings are flushed from the TLB before replacing them with page tables, otherwise we can suffer from mismatched aliases if the CPU speculatively prefetches from these mappings at an inopportune time.

A mismatched alias can occur when the TLB contains a section mapping, but a subsequent prefetch causes it to load a page table mapping, resulting in the possibility of the TLB containing two matching mappings for the same virtual address region.

---
 arch/arm/mm/dma-mapping.c |   11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)
arch/arm/mm/dma-mapping.c +10 -1
@@ -429,12 +429,21 @@ void __init dma_contiguous_remap(void)
 		map.type = MT_MEMORY_DMA_READY;
 
 		/*
-		 * Clear previous low-memory mapping
+		 * Clear previous low-memory mapping to ensure that the
+		 * TLB does not see any conflicting entries, then flush
+		 * the TLB of the old entries before creating new mappings.
+		 *
+		 * This ensures that any speculatively loaded TLB entries
+		 * (even though they may be rare) can not cause any problems,
+		 * and ensures that this code is architecturally compliant.
 		 */
 		for (addr = __phys_to_virt(start); addr < __phys_to_virt(end);
 		     addr += PMD_SIZE)
 			pmd_clear(pmd_off_k(addr));
 
+		flush_tlb_kernel_range(__phys_to_virt(start),
+				       __phys_to_virt(end));
+
 		iotable_init(&map, 1);
 	}
 }
author maintainer
#29 lore raw

PATCH 29/51 ARM: 8108/1: mm: Introduce {pte,pmd}_isset and {pte,pmd}_isclear

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Steven Capper <[email protected]>

commit f2950706871c4b6e8c0f0d7c3f62d35930b8de63 upstream.

Long descriptors on ARM are 64 bits, and some pte functions such as pte_dirty return a bitwise-and of a flag with the pte value. If the flag to be tested resides in the upper 32 bits of the pte, then we run into the danger of the result being dropped if downcast.

For example:

	gather_stats(page, md, pte_dirty(*pte), 1);

where pte_dirty(*pte) is downcast to an int.

This patch introduces a new macro pte_isset which performs the bitwise and, then performs a double logical invert (where needed) to ensure predictable downcasting. The logical inverse pte_isclear is also introduced.

Equivalent pmd functions for Transparent HugePages have also been added.

[hpy: Backported to 3.10:
- adjust the context
- ignore change to pmd, because 3.10 does not support HugePage.]

---
 arch/arm/include/asm/pgtable.h |   14 +++++++++-----
 1 file changed, 9 insertions(+), 5 deletions(-)
arch/arm/include/asm/pgtable.h +9 -5
@@ -211,12 +211,16 @@ static inline pte_t *pmd_page_vaddr(pmd_
 
 #define pte_clear(mm,addr,ptep)	set_pte_ext(ptep, __pte(0), 0)
 
+#define pte_isset(pte, val)	((u32)(val) == (val) ? pte_val(pte) & (val) \
+						: !!(pte_val(pte) & (val)))
+#define pte_isclear(pte, val)	(!(pte_val(pte) & (val)))
+
 #define pte_none(pte)		(!pte_val(pte))
-#define pte_present(pte)	(pte_val(pte) & L_PTE_PRESENT)
-#define pte_write(pte)		(!(pte_val(pte) & L_PTE_RDONLY))
-#define pte_dirty(pte)		(pte_val(pte) & L_PTE_DIRTY)
-#define pte_young(pte)		(pte_val(pte) & L_PTE_YOUNG)
-#define pte_exec(pte)		(!(pte_val(pte) & L_PTE_XN))
+#define pte_present(pte)	(pte_isset((pte), L_PTE_PRESENT))
+#define pte_write(pte)		(pte_isclear((pte), L_PTE_RDONLY))
+#define pte_dirty(pte)		(pte_isset((pte), L_PTE_DIRTY))
+#define pte_young(pte)		(pte_isset((pte), L_PTE_YOUNG))
+#define pte_exec(pte)		(pte_isclear((pte), L_PTE_XN))
 #define pte_special(pte)	(0)
 
 #define pte_present_user(pte)  (pte_present(pte) && (pte_val(pte) & L_PTE_USER))
author maintainer
#30 lore raw

PATCH 30/51 ARM: 8109/1: mm: Modify pte_write and pmd_write logic for LPAE

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Steven Capper <[email protected]>

commit ded9477984690d026e46dd75e8157392cea3f13f upstream.

For LPAE, we have the following means for encoding writable or dirty ptes:

                              L_PTE_DIRTY       L_PTE_RDONLY
    !pte_dirty && !pte_write        0               1
    !pte_dirty && pte_write         0               1
    pte_dirty && !pte_write         1               1
    pte_dirty && pte_write          1               0

So we can't distinguish between writeable clean ptes and read only ptes. This can cause problems with ptes being incorrectly flagged as read only when they are writeable but not dirty.

This patch renumbers L_PTE_RDONLY from AP[2] to a software bit #58, and adds additional logic to set AP[2] whenever the pte is read only or not dirty. That way we can distinguish between clean writeable ptes and read only ptes.

HugeTLB pages will use this new logic automatically.

We need to add some logic to Transparent HugePages to ensure that they correctly interpret the revised pgprot permissions (L_PTE_RDONLY has moved and no longer matches PMD_SECT_AP2). In the process of revising THP, the names of the PMD software bits have been prefixed with L_ to make them easier to distinguish from their hardware bit counterparts.

[hpy: Backported to 3.10
- adjust the context
- ignore change related to pmd, because 3.10 does not support HugePage ]

---
 arch/arm/include/asm/pgtable-3level-hwdef.h |    1 +
 arch/arm/include/asm/pgtable-3level.h       |    6 +++---
 arch/arm/mm/proc-v7-3level.S                |    9 +++++++--
 3 files changed, 11 insertions(+), 5 deletions(-)
arch/arm/include/asm/pgtable-3level-hwdef.h +1 -0
@@ -68,6 +68,7 @@
 #define PTE_TYPE_PAGE		(_AT(pteval_t, 3) << 0)
 #define PTE_BUFFERABLE		(_AT(pteval_t, 1) << 2)		/* AttrIndx[0] */
 #define PTE_CACHEABLE		(_AT(pteval_t, 1) << 3)		/* AttrIndx[1] */
+#define PTE_AP2			(_AT(pteval_t, 1) << 7)		/* AP[2] */
 #define PTE_EXT_SHARED		(_AT(pteval_t, 3) << 8)		/* SH[1:0], inner shareable */
 #define PTE_EXT_AF		(_AT(pteval_t, 1) << 10)	/* Access Flag */
 #define PTE_EXT_NG		(_AT(pteval_t, 1) << 11)	/* nG */
arch/arm/include/asm/pgtable-3level.h +3 -3
@@ -71,13 +71,13 @@
 #define L_PTE_PRESENT		(_AT(pteval_t, 3) << 0)		/* Present */
 #define L_PTE_FILE		(_AT(pteval_t, 1) << 2)		/* only when !PRESENT */
 #define L_PTE_USER		(_AT(pteval_t, 1) << 6)		/* AP[1] */
-#define L_PTE_RDONLY		(_AT(pteval_t, 1) << 7)		/* AP[2] */
 #define L_PTE_SHARED		(_AT(pteval_t, 3) << 8)		/* SH[1:0], inner shareable */
 #define L_PTE_YOUNG		(_AT(pteval_t, 1) << 10)	/* AF */
 #define L_PTE_XN		(_AT(pteval_t, 1) << 54)	/* XN */
-#define L_PTE_DIRTY		(_AT(pteval_t, 1) << 55)	/* unused */
-#define L_PTE_SPECIAL		(_AT(pteval_t, 1) << 56)	/* unused */
+#define L_PTE_DIRTY		(_AT(pteval_t, 1) << 55)
+#define L_PTE_SPECIAL		(_AT(pteval_t, 1) << 56)
 #define L_PTE_NONE		(_AT(pteval_t, 1) << 57)	/* PROT_NONE */
+#define L_PTE_RDONLY		(_AT(pteval_t, 1) << 58)	/* READ ONLY */
 
 /*
  * To be used in assembly code with the upper page attributes.
arch/arm/mm/proc-v7-3level.S +7 -2
@@ -78,8 +78,13 @@ ENTRY(cpu_v7_set_pte_ext)
 	tst	rh, #1 << (57 - 32)		@ L_PTE_NONE
 	bicne	rl, #L_PTE_VALID
 	bne	1f
-	tst	rh, #1 << (55 - 32)		@ L_PTE_DIRTY
-	orreq	rl, #L_PTE_RDONLY
+
+	eor	ip, rh, #1 << (55 - 32)	@ toggle L_PTE_DIRTY in temp reg to
+					@ test for !L_PTE_DIRTY || L_PTE_RDONLY
+	tst	ip, #1 << (55 - 32) | 1 << (58 - 32)
+	orrne	rl, #PTE_AP2
+	biceq	rl, #PTE_AP2
+
 1:	strd	r2, r3, [r0]
 	ALT_SMP(W(nop))
 	ALT_UP (mcr	p15, 0, r0, c7, c10, 1)		@ flush_pte
author maintainer
#31 lore raw

PATCH 31/51 pstore: Fail to unlink if a driver has not defined pstore_erase

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Aruna Balakrishnaiah <[email protected]>

commit bf2883339a33b7544b92ea465b90c3de55082032 upstream.

pstore_erase is used to erase the record from the persistent store. So if a driver has not defined pstore_erase callback return -EPERM instead of unlinking a file as deleting the file without erasing its record in persistent store will give a wrong impression to customers.

---
 fs/pstore/inode.c |    2 ++
 1 file changed, 2 insertions(+)
fs/pstore/inode.c +2 -0
@@ -178,6 +178,8 @@ static int pstore_unlink(struct inode *d
 	if (p->psi->erase)
 		p->psi->erase(p->type, p->id, p->count,
 			      dentry->d_inode->i_ctime, p->psi);
+	else
+		return -EPERM;
 
 	return simple_unlink(dir, dentry);
 }
author maintainer
#32 lore raw

PATCH 32/51 pstore: d_alloc_name() doesnt return an ERR_PTR

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Dan Carpenter <[email protected]>

commit c39524e6744284452ef45480d3153bec28960c32 upstream.

d_alloc_name() returns NULL on error. Also I changed the error code from -ENOSPC to -ENOMEM to reflect that we were short on RAM not disk space.

---
 fs/pstore/inode.c |    3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
fs/pstore/inode.c +1 -2
@@ -336,9 +336,8 @@ int pstore_mkfile(enum pstore_type_id ty
 
 	mutex_lock(&root->d_inode->i_mutex);
 
-	rc = -ENOSPC;
 	dentry = d_alloc_name(root, name);
-	if (IS_ERR(dentry))
+	if (!dentry)
 		goto fail_lockedalloc;
 
 	memcpy(private->data, data, size);
author maintainer
#33 lore raw

PATCH 33/51 pstore: clarify clearing of _read_cnt in ramoops_context

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Liu ShuoX <[email protected]>

commit 57fd835385a043577457a385f28c08be693991bf upstream.

*_read_cnt in ramoops_context need to be cleared during pstore ->open to support mutli times getting the records. The patch added missed ftrace_read_cnt clearing and removed duplicate clearing in ramoops_probe.

---
 fs/pstore/ram.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
fs/pstore/ram.c +2 -1
@@ -92,6 +92,7 @@ struct ramoops_context {
 	struct persistent_ram_ecc_info ecc_info;
 	unsigned int max_dump_cnt;
 	unsigned int dump_write_cnt;
+	/* _read_cnt need clear on ramoops_pstore_open */
 	unsigned int dump_read_cnt;
 	unsigned int console_read_cnt;
 	unsigned int ftrace_read_cnt;
@@ -107,6 +108,7 @@ static int ramoops_pstore_open(struct ps
 
 	cxt->dump_read_cnt = 0;
 	cxt->console_read_cnt = 0;
+	cxt->ftrace_read_cnt = 0;
 	return 0;
 }
 
@@ -415,7 +417,6 @@ static int ramoops_probe(struct platform
 	if (!is_power_of_2(pdata->ftrace_size))
 		pdata->ftrace_size = rounddown_pow_of_two(pdata->ftrace_size);
 
-	cxt->dump_read_cnt = 0;
 	cxt->size = pdata->mem_size;
 	cxt->phys_addr = pdata->mem_address;
 	cxt->memtype = pdata->mem_type;
author maintainer
#34 lore raw

PATCH 34/51 pstore: skip zero size persistent ram buffer in traverse

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Liu ShuoX <[email protected]>

commit aa9a4a1edfbd3d223af01db833da2f07850bc655 upstream.

In ramoops_pstore_read, a valid prz pointer with zero size buffer will break traverse of all persistent ram buffers. The latter buffer might be lost.

---
 fs/pstore/ram.c |   10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)
fs/pstore/ram.c +5 -5
@@ -126,12 +126,12 @@ ramoops_get_next_prz(struct persistent_r
 
 	prz = przs[i];
 
-	if (update) {
-		/* Update old/shadowed buffer. */
+	/* Update old/shadowed buffer. */
+	if (update)
 		persistent_ram_save_old(prz);
-		if (!persistent_ram_old_size(prz))
-			return NULL;
-	}
+
+	if (!persistent_ram_old_size(prz))
+		return NULL;
 
 	*typep = type;
 	*id = i;
author maintainer
#35 lore raw

PATCH 35/51 pstore: Fix NULL pointer fault if get NULL prz in ramoops_get_next_prz

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Liu ShuoX <[email protected]>

commit b0aa931fb84431394d995472d0af2a6c2b61064d upstream.

ramoops_get_next_prz get the prz according the paramters. If it get a uninitialized prz, access its members by following persistent_ram_old_size(prz) will cause a NULL pointer crash.
Ex: if ftrace_size is 0, fprz will be NULL.

Fix it by return NULL in advance.

---
 fs/pstore/ram.c |    2 ++
 1 file changed, 2 insertions(+)
fs/pstore/ram.c +2 -0
@@ -125,6 +125,8 @@ ramoops_get_next_prz(struct persistent_r
 		return NULL;
 
 	prz = przs[i];
+	if (!prz)
+		return NULL;
 
 	/* Update old/shadowed buffer. */
 	if (update)
author maintainer
#36 lore raw

PATCH 36/51 pstore/ram: avoid atomic accesses for ioremapped regions

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Rob Herring <[email protected]>

commit 0405a5cec3406f19e69da07c8111a6bf1088ac29 upstream.

For persistent RAM outside of main memory, the memory may have limitations on supported accesses. For internal RAM on highbank platform exclusive accesses are not supported and will hang the system. So atomic_cmpxchg cannot be used. This commit uses spinlock protection for buffer size and start updates on ioremapped regions instead.

[hkp: Backported to 3.10: adjust context]
---
 fs/pstore/ram_core.c |   54 +++++++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 52 insertions(+), 2 deletions(-)
fs/pstore/ram_core.c +52 -2
@@ -46,7 +46,7 @@ static inline size_t buffer_start(struct
 }
 
 /* increase and wrap the start pointer, returning the old value */
-static inline size_t buffer_start_add(struct persistent_ram_zone *prz, size_t a)
+static size_t buffer_start_add_atomic(struct persistent_ram_zone *prz, size_t a)
 {
 	int old;
 	int new;
@@ -62,7 +62,7 @@ static inline size_t buffer_start_add(st
 }
 
 /* increase the size counter until it hits the max size */
-static inline void buffer_size_add(struct persistent_ram_zone *prz, size_t a)
+static void buffer_size_add_atomic(struct persistent_ram_zone *prz, size_t a)
 {
 	size_t old;
 	size_t new;
@@ -78,6 +78,53 @@ static inline void buffer_size_add(struc
 	} while (atomic_cmpxchg(&prz->buffer->size, old, new) != old);
 }
 
+static DEFINE_RAW_SPINLOCK(buffer_lock);
+
+/* increase and wrap the start pointer, returning the old value */
+static size_t buffer_start_add_locked(struct persistent_ram_zone *prz, size_t a)
+{
+	int old;
+	int new;
+	unsigned long flags;
+
+	raw_spin_lock_irqsave(&buffer_lock, flags);
+
+	old = atomic_read(&prz->buffer->start);
+	new = old + a;
+	while (unlikely(new > prz->buffer_size))
+		new -= prz->buffer_size;
+	atomic_set(&prz->buffer->start, new);
+
+	raw_spin_unlock_irqrestore(&buffer_lock, flags);
+
+	return old;
+}
+
+/* increase the size counter until it hits the max size */
+static void buffer_size_add_locked(struct persistent_ram_zone *prz, size_t a)
+{
+	size_t old;
+	size_t new;
+	unsigned long flags;
+
+	raw_spin_lock_irqsave(&buffer_lock, flags);
+
+	old = atomic_read(&prz->buffer->size);
+	if (old == prz->buffer_size)
+		goto exit;
+
+	new = old + a;
+	if (new > prz->buffer_size)
+		new = prz->buffer_size;
+	atomic_set(&prz->buffer->size, new);
+
+exit:
+	raw_spin_unlock_irqrestore(&buffer_lock, flags);
+}
+
+static size_t (*buffer_start_add)(struct persistent_ram_zone *, size_t) = buffer_start_add_atomic;
+static void (*buffer_size_add)(struct persistent_ram_zone *, size_t) = buffer_size_add_atomic;
+
 static void notrace persistent_ram_encode_rs8(struct persistent_ram_zone *prz,
 	uint8_t *data, size_t len, uint8_t *ecc)
 {
@@ -379,6 +426,9 @@ static void *persistent_ram_iomap(phys_a
 		return NULL;
 	}
 
+	buffer_start_add = buffer_start_add_locked;
+	buffer_size_add = buffer_size_add_locked;
+
 	if (memtype)
 		va = ioremap(start, size);
 	else
author maintainer
#37 lore raw

PATCH 37/51 efi-pstore: Make efi-pstore return a unique id

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Madper Xie <[email protected]>

commit fdeadb43fdf1e7d5698c027b555c389174548e5a upstream.

Pstore fs expects that backends provide a unique id which could avoid pstore making entries as duplication or denominating entries the same name. So I combine the timestamp, part and count into id.

Cc Seiji Aguchi <[email protected]>
[hkp: Backported to 3.10: adjust context]
---
 drivers/firmware/efi/efi-pstore.c |   18 +++++++++++++-----
 1 file changed, 13 insertions(+), 5 deletions(-)
drivers/firmware/efi/efi-pstore.c +13 -5
@@ -38,6 +38,12 @@ struct pstore_read_data {
 	char **buf;
 };
 
+static inline u64 generic_id(unsigned long timestamp,
+			     unsigned int part, int count)
+{
+	return (timestamp * 100 + part) * 1000 + count;
+}
+
 static int efi_pstore_read_func(struct efivar_entry *entry, void *data)
 {
 	efi_guid_t vendor = LINUX_EFI_CRASH_GUID;
@@ -56,7 +62,7 @@ static int efi_pstore_read_func(struct e
 
 	if (sscanf(name, "dump-type%u-%u-%d-%lu",
 		   cb_data->type, &part, &cnt, &time) == 4) {
-		*cb_data->id = part;
+		*cb_data->id = generic_id(time, part, cnt);
 		*cb_data->count = cnt;
 		cb_data->timespec->tv_sec = time;
 		cb_data->timespec->tv_nsec = 0;
@@ -67,7 +73,7 @@ static int efi_pstore_read_func(struct e
 		 * which doesn't support holding
 		 * multiple logs, remains.
 		 */
-		*cb_data->id = part;
+		*cb_data->id = generic_id(time, part, 0);
 		*cb_data->count = 0;
 		cb_data->timespec->tv_sec = time;
 		cb_data->timespec->tv_nsec = 0;
@@ -185,14 +191,16 @@ static int efi_pstore_erase(enum pstore_
 	char name[DUMP_NAME_LEN];
 	efi_char16_t efi_name[DUMP_NAME_LEN];
 	int found, i;
+	unsigned int part;
 
-	sprintf(name, "dump-type%u-%u-%d-%lu", type, (unsigned int)id, count,
-		time.tv_sec);
+	do_div(id, 1000);
+	part = do_div(id, 100);
+	sprintf(name, "dump-type%u-%u-%d-%lu", type, part, count, time.tv_sec);
 
 	for (i = 0; i < DUMP_NAME_LEN; i++)
 		efi_name[i] = name[i];
 
-	edata.id = id;
+	edata.id = part;
 	edata.type = type;
 	edata.count = count;
 	edata.time = time;
author maintainer
#38 lore raw

PATCH 38/51 gpio: squelch a compiler warning

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Martin Kaiser <[email protected]>

drivers/gpio/gpiolib-of.c: In function 'of_gpiochip_find_and_xlate':
drivers/gpio/gpiolib-of.c:51:21: warning: assignment makes integer from pointer without a cast [enabled by default]

   gg_data->out_gpio = ERR_PTR(ret);
                     ^

this was introduced in d1c3449160df60fac4abb56f0ba0a3784305e43e

the upstream kernel changed the type of out_gpio from int to struct gpio_desc * as part of a larger refactoring that wasn't backported

---
 drivers/gpio/gpiolib-of.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpio/gpiolib-of.c +1 -1
@@ -48,7 +48,7 @@ static int of_gpiochip_find_and_xlate(st
 		 * Return true to stop looking and return the translation
 		 * error via out_gpio
 		 */
-		gg_data->out_gpio = ERR_PTR(ret);
+		gg_data->out_gpio = ret;
 		return true;
 	 }
 
author maintainer
#39 lore raw

PATCH 39/51 workqueue: fix subtle pool management issue which can stall whole worker_pool

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Tejun Heo <[email protected]>

commit 29187a9eeaf362d8422e62e17a22a6e115277a49 upstream.

A worker_pool's forward progress is guaranteed by the fact that the last idle worker assumes the manager role to create more workers and summon the rescuers if creating workers doesn't succeed in timely manner before proceeding to execute work items.

This manager role is implemented in manage_workers(), which indicates whether the worker may proceed to work item execution with its return value. This is necessary because multiple workers may contend for the manager role, and, if there already is a manager, others should proceed to work item execution.

Unfortunately, the function also indicates that the worker may proceed to work item execution if need_to_create_worker() is false at the head of the function. need_to_create_worker() tests the following conditions.

	pending work items && !nr_running && !nr_idle

The first and third conditions are protected by pool->lock and thus won't change while holding pool->lock; however, nr_running can change asynchronously as other workers block and resume and while it's likely to be zero, as someone woke this worker up in the first place, some other workers could have become runnable inbetween making it non-zero.

If this happens, manage_worker() could return false even with zero nr_idle making the worker, the last idle one, proceed to execute work items. If then all workers of the pool end up blocking on a resource which can only be released by a work item which is pending on that pool, the whole pool can deadlock as there's no one to create more workers or summon the rescuers.

This patch fixes the problem by removing the early exit condition from maybe_create_worker() and making manage_workers() return false iff there's already another manager, which ensures that the last worker doesn't start executing work items.

We can leave the early exit condition alone and just ignore the return value but the only reason it was put there is because the manage_workers() used to perform both creations and destructions of workers and thus the function may be invoked while the pool is trying to reduce the number of workers. Now that manage_workers() is called only when more workers are needed, the only case this early exit condition is triggered is rare race conditions rendering it pointless.

Tested with simulated workload and modified workqueue code which trigger the pool deadlock reliably without this patch.

---
 kernel/workqueue.c |   39 +++++++++++++--------------------------
 1 file changed, 13 insertions(+), 26 deletions(-)
kernel/workqueue.c +13 -26
@@ -1934,17 +1934,13 @@ static void pool_mayday_timeout(unsigned
  * spin_lock_irq(pool->lock) which may be released and regrabbed
  * multiple times.  Does GFP_KERNEL allocations.  Called only from
  * manager.
- *
- * RETURNS:
- * %false if no action was taken and pool->lock stayed locked, %true
- * otherwise.
  */
-static bool maybe_create_worker(struct worker_pool *pool)
+static void maybe_create_worker(struct worker_pool *pool)
 __releases(&pool->lock)
 __acquires(&pool->lock)
 {
 	if (!need_to_create_worker(pool))
-		return false;
+		return;
 restart:
 	spin_unlock_irq(&pool->lock);
 
@@ -1961,7 +1957,7 @@ restart:
 			start_worker(worker);
 			if (WARN_ON_ONCE(need_to_create_worker(pool)))
 				goto restart;
-			return true;
+			return;
 		}
 
 		if (!need_to_create_worker(pool))
@@ -1978,7 +1974,7 @@ restart:
 	spin_lock_irq(&pool->lock);
 	if (need_to_create_worker(pool))
 		goto restart;
-	return true;
+	return;
 }
 
 /**
@@ -1991,15 +1987,9 @@ restart:
  * LOCKING:
  * spin_lock_irq(pool->lock) which may be released and regrabbed
  * multiple times.  Called only from manager.
- *
- * RETURNS:
- * %false if no action was taken and pool->lock stayed locked, %true
- * otherwise.
  */
-static bool maybe_destroy_workers(struct worker_pool *pool)
+static void maybe_destroy_workers(struct worker_pool *pool)
 {
-	bool ret = false;
-
 	while (too_many_workers(pool)) {
 		struct worker *worker;
 		unsigned long expires;
@@ -2013,10 +2003,7 @@ static bool maybe_destroy_workers(struct
 		}
 
 		destroy_worker(worker);
-		ret = true;
 	}
-
-	return ret;
 }
 
 /**
@@ -2036,13 +2023,14 @@ static bool maybe_destroy_workers(struct
  * multiple times.  Does GFP_KERNEL allocations.
  *
  * RETURNS:
- * spin_lock_irq(pool->lock) which may be released and regrabbed
- * multiple times.  Does GFP_KERNEL allocations.
+ * %false if the pool doesn't need management and the caller can safely
+ * start processing works, %true if management function was performed and
+ * the conditions that the caller verified before calling the function may
+ * no longer be true.
  */
 static bool manage_workers(struct worker *worker)
 {
 	struct worker_pool *pool = worker->pool;
-	bool ret = false;
 
 	/*
 	 * Managership is governed by two mutexes - manager_arb and
@@ -2066,7 +2054,7 @@ static bool manage_workers(struct worker
 	 * manager_mutex.
 	 */
 	if (!mutex_trylock(&pool->manager_arb))
-		return ret;
+		return false;
 
 	/*
 	 * With manager arbitration won, manager_mutex would be free in
@@ -2076,7 +2064,6 @@ static bool manage_workers(struct worker
 		spin_unlock_irq(&pool->lock);
 		mutex_lock(&pool->manager_mutex);
 		spin_lock_irq(&pool->lock);
-		ret = true;
 	}
 
 	pool->flags &= ~POOL_MANAGE_WORKERS;
@@ -2085,12 +2072,12 @@ static bool manage_workers(struct worker
 	 * Destroy and then create so that may_start_working() is true
 	 * on return.
 	 */
-	ret |= maybe_destroy_workers(pool);
-	ret |= maybe_create_worker(pool);
+	maybe_destroy_workers(pool);
+	maybe_create_worker(pool);
 
 	mutex_unlock(&pool->manager_mutex);
 	mutex_unlock(&pool->manager_arb);
-	return ret;
+	return true;
 }
 
 /**
author maintainer
#40 lore raw

PATCH 40/51 IB/isert: Adjust CQ size to HW limits

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Chris Moore <[email protected]>

commit b1a5ad006b34ded9dc7ec64988deba1b3ecad367 upstream.

isert has an issue of trying to create a CQ with more CQEs than are supported by the hardware, that currently results in failures during isert_device creation during first session login.

This is the isert version of the patch that Minh Tran submitted for iser, and is simple a workaround required to function with existing ocrdma hardware.

---
 drivers/infiniband/ulp/isert/ib_isert.c |   14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)
drivers/infiniband/ulp/isert/ib_isert.c +12 -2
@@ -212,6 +212,13 @@ isert_create_device_ib_res(struct isert_
 	struct ib_device *ib_dev = device->ib_device;
 	struct isert_cq_desc *cq_desc;
 	int ret = 0, i, j;
+	int max_rx_cqe, max_tx_cqe;
+	struct ib_device_attr dev_attr;
+
+	memset(&dev_attr, 0, sizeof(struct ib_device_attr));
+	ret = isert_query_device(device->ib_device, &dev_attr);
+	if (ret)
+		return ret;
 
 	device->cqs_used = min_t(int, num_online_cpus(),
 				 device->ib_device->num_comp_vectors);
@@ -234,6 +241,9 @@ isert_create_device_ib_res(struct isert_
 		goto out_cq_desc;
 	}
 
+	max_rx_cqe = min(ISER_MAX_RX_CQ_LEN, dev_attr.max_cqe);
+	max_tx_cqe = min(ISER_MAX_TX_CQ_LEN, dev_attr.max_cqe);
+
 	for (i = 0; i < device->cqs_used; i++) {
 		cq_desc[i].device = device;
 		cq_desc[i].cq_index = i;
@@ -242,7 +252,7 @@ isert_create_device_ib_res(struct isert_
 						isert_cq_rx_callback,
 						isert_cq_event_callback,
 						(void *)&cq_desc[i],
-						ISER_MAX_RX_CQ_LEN, i);
+						max_rx_cqe, i);
 		if (IS_ERR(device->dev_rx_cq[i])) {
 			ret = PTR_ERR(device->dev_rx_cq[i]);
 			device->dev_rx_cq[i] = NULL;
@@ -253,7 +263,7 @@ isert_create_device_ib_res(struct isert_
 						isert_cq_tx_callback,
 						isert_cq_event_callback,
 						(void *)&cq_desc[i],
-						ISER_MAX_TX_CQ_LEN, i);
+						max_tx_cqe, i);
 		if (IS_ERR(device->dev_tx_cq[i])) {
 			ret = PTR_ERR(device->dev_tx_cq[i]);
 			device->dev_tx_cq[i] = NULL;
author maintainer
#41 lore raw

PATCH 41/51 ib_isert: Add max_send_sge=2 minimum for control PDU responses

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Or Gerlitz <[email protected]>

commit f57915cfa5b2b14c1cffa2e83c034f55e3f0e70d upstream.

This patch adds a max_send_sge=2 minimum in isert_conn_setup_qp() to ensure outgoing control PDU responses with tx_desc->num_sge=2 are able to function correctly.

This addresses a bug with RDMA hardware using dev_attr.max_sge=3, that in the original code with the ConnectX-2 work-around would result in isert_conn->max_sge=1 being negotiated.

Originally reported by Chris with ocrdma driver.

---
 drivers/infiniband/ulp/isert/ib_isert.c |    7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
drivers/infiniband/ulp/isert/ib_isert.c +5 -2
@@ -107,9 +107,12 @@ isert_conn_setup_qp(struct isert_conn *i
 	attr.cap.max_recv_wr = ISERT_QP_MAX_RECV_DTOS;
 	/*
 	 * FIXME: Use devattr.max_sge - 2 for max_send_sge as
-	 * work-around for RDMA_READ..
+	 * work-around for RDMA_READs with ConnectX-2.
+	 *
+	 * Also, still make sure to have at least two SGEs for
+	 * outgoing control PDU responses.
 	 */
-	attr.cap.max_send_sge = devattr.max_sge - 2;
+	attr.cap.max_send_sge = max(2, devattr.max_sge - 2);
 	isert_conn->max_sge = attr.cap.max_send_sge;
 
 	attr.cap.max_recv_sge = 1;
author maintainer
#42 lore raw

PATCH 42/51 vhost-scsi: Take configfs group dependency during VHOST_SCSI_SET_ENDPOINT

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Nicholas Bellinger <[email protected]>

commit ab8edab132829b26dd13db6caca3c242cce35dc1 upstream.

This patch addresses a bug where individual vhost-scsi configfs endpoint groups can be removed from below while active exports to QEMU userspace still exist, resulting in an OOPs.

It adds a configfs_depend_item() in vhost_scsi_set_endpoint() to obtain an explicit dependency on se_tpg->tpg_group in order to prevent individual vhost-scsi WWPN endpoints from being released via normal configfs methods while an QEMU ioctl reference still exists.

Also, add matching configfs_undepend_item() in vhost_scsi_clear_endpoint() to release the dependency, once QEMU's reference to the individual group at /sys/kernel/config/target/vhost/$WWPN/$TPGT is released.

(Fix up vhost_scsi_clear_endpoint() error path - DanC)

Cc Michael S. Tsirkin <[email protected]>
Cc Paolo Bonzini <[email protected]>
Cc Stefan Hajnoczi <[email protected]>
---
 drivers/vhost/scsi.c |   24 ++++++++++++++++++++++++
 1 file changed, 24 insertions(+)
drivers/vhost/scsi.c +24 -0
@@ -1150,6 +1150,7 @@ static int vhost_scsi_set_endpoint(
 	struct vhost_scsi *vs,
 	struct vhost_scsi_target *t)
 {
+	struct se_portal_group *se_tpg;
 	struct tcm_vhost_tport *tv_tport;
 	struct tcm_vhost_tpg *tv_tpg;
 	struct tcm_vhost_tpg **vs_tpg;
@@ -1197,6 +1198,21 @@ static int vhost_scsi_set_endpoint(
 				ret = -EEXIST;
 				goto out;
 			}
+			/*
+			 * In order to ensure individual vhost-scsi configfs
+			 * groups cannot be removed while in use by vhost ioctl,
+			 * go ahead and take an explicit se_tpg->tpg_group.cg_item
+			 * dependency now.
+			 */
+			se_tpg = &tv_tpg->se_tpg;
+			ret = configfs_depend_item(se_tpg->se_tpg_tfo->tf_subsys,
+						   &se_tpg->tpg_group.cg_item);
+			if (ret) {
+				pr_warn("configfs_depend_item() failed: %d\n", ret);
+				kfree(vs_tpg);
+				mutex_unlock(&tv_tpg->tv_tpg_mutex);
+				goto out;
+			}
 			tv_tpg->tv_tpg_vhost_count++;
 			tv_tpg->vhost_scsi = vs;
 			vs_tpg[tv_tpg->tport_tpgt] = tv_tpg;
@@ -1240,6 +1256,7 @@ static int vhost_scsi_clear_endpoint(
 	struct vhost_scsi *vs,
 	struct vhost_scsi_target *t)
 {
+	struct se_portal_group *se_tpg;
 	struct tcm_vhost_tport *tv_tport;
 	struct tcm_vhost_tpg *tv_tpg;
 	struct vhost_virtqueue *vq;
@@ -1288,6 +1305,13 @@ static int vhost_scsi_clear_endpoint(
 		vs->vs_tpg[target] = NULL;
 		match = true;
 		mutex_unlock(&tv_tpg->tv_tpg_mutex);
+		/*
+		 * Release se_tpg->tpg_group.cg_item configfs dependency now
+		 * to allow vhost-scsi WWPN se_tpg->tpg_group shutdown to occur.
+		 */
+		se_tpg = &tv_tpg->se_tpg;
+		configfs_undepend_item(se_tpg->se_tpg_tfo->tf_subsys,
+				       &se_tpg->tpg_group.cg_item);
 	}
 	if (match) {
 		for (i = 0; i < VHOST_SCSI_MAX_VQ; i++) {
author maintainer
#43 lore raw

PATCH 43/51 tcm_loop: Fix wrong I_T nexus association

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Hannes Reinecke <[email protected]>

commit 506787a2c7daed45f0a213674ca706cbc83a9089 upstream.

tcm_loop has the I_T nexus associated with the HBA. This causes commands to become misdirected if the HBA has more than one target portal group; any command is then being sent to the first target portal group instead of the correct one.

The nexus needs to be associated with the target portal group instead.

---
 drivers/target/loopback/tcm_loop.c |   29 ++++++++++++++---------------
 drivers/target/loopback/tcm_loop.h |    7 +------
 2 files changed, 15 insertions(+), 21 deletions(-)
drivers/target/loopback/tcm_loop.c +14 -15
@@ -179,7 +179,7 @@ static void tcm_loop_submission_work(str
 		goto out_done;
 	}
 
-	tl_nexus = tl_hba->tl_nexus;
+	tl_nexus = tl_tpg->tl_nexus;
 	if (!tl_nexus) {
 		scmd_printk(KERN_ERR, sc, "TCM_Loop I_T Nexus"
 				" does not exist\n");
@@ -258,20 +258,20 @@ static int tcm_loop_device_reset(struct
 	 */
 	tl_hba = *(struct tcm_loop_hba **)shost_priv(sc->device->host);
 	/*
+	 * Locate the tl_tpg and se_tpg pointers from TargetID in sc->device->id
+	 */
+	tl_tpg = &tl_hba->tl_hba_tpgs[sc->device->id];
+	se_tpg = &tl_tpg->tl_se_tpg;
+	/*
 	 * Locate the tl_nexus and se_sess pointers
 	 */
-	tl_nexus = tl_hba->tl_nexus;
+	tl_nexus = tl_tpg->tl_nexus;
 	if (!tl_nexus) {
 		pr_err("Unable to perform device reset without"
 				" active I_T Nexus\n");
 		return FAILED;
 	}
 	se_sess = tl_nexus->se_sess;
-	/*
-	 * Locate the tl_tpg and se_tpg pointers from TargetID in sc->device->id
-	 */
-	tl_tpg = &tl_hba->tl_hba_tpgs[sc->device->id];
-	se_tpg = &tl_tpg->tl_se_tpg;
 
 	tl_cmd = kmem_cache_zalloc(tcm_loop_cmd_cache, GFP_KERNEL);
 	if (!tl_cmd) {
@@ -879,8 +879,8 @@ static int tcm_loop_make_nexus(
 	struct tcm_loop_nexus *tl_nexus;
 	int ret = -ENOMEM;
 
-	if (tl_tpg->tl_hba->tl_nexus) {
-		pr_debug("tl_tpg->tl_hba->tl_nexus already exists\n");
+	if (tl_tpg->tl_nexus) {
+		pr_debug("tl_tpg->tl_nexus already exists\n");
 		return -EEXIST;
 	}
 	se_tpg = &tl_tpg->tl_se_tpg;
@@ -915,7 +915,7 @@ static int tcm_loop_make_nexus(
 	 */
 	__transport_register_session(se_tpg, tl_nexus->se_sess->se_node_acl,
 			tl_nexus->se_sess, tl_nexus);
-	tl_tpg->tl_hba->tl_nexus = tl_nexus;
+	tl_tpg->tl_nexus = tl_nexus;
 	pr_debug("TCM_Loop_ConfigFS: Established I_T Nexus to emulated"
 		" %s Initiator Port: %s\n", tcm_loop_dump_proto_id(tl_hba),
 		name);
@@ -931,9 +931,8 @@ static int tcm_loop_drop_nexus(
 {
 	struct se_session *se_sess;
 	struct tcm_loop_nexus *tl_nexus;
-	struct tcm_loop_hba *tl_hba = tpg->tl_hba;
 
-	tl_nexus = tpg->tl_hba->tl_nexus;
+	tl_nexus = tpg->tl_nexus;
 	if (!tl_nexus)
 		return -ENODEV;
 
@@ -949,13 +948,13 @@ static int tcm_loop_drop_nexus(
 	}
 
 	pr_debug("TCM_Loop_ConfigFS: Removing I_T Nexus to emulated"
-		" %s Initiator Port: %s\n", tcm_loop_dump_proto_id(tl_hba),
+		" %s Initiator Port: %s\n", tcm_loop_dump_proto_id(tpg->tl_hba),
 		tl_nexus->se_sess->se_node_acl->initiatorname);
 	/*
 	 * Release the SCSI I_T Nexus to the emulated SAS Target Port
 	 */
 	transport_deregister_session(tl_nexus->se_sess);
-	tpg->tl_hba->tl_nexus = NULL;
+	tpg->tl_nexus = NULL;
 	kfree(tl_nexus);
 	return 0;
 }
@@ -971,7 +970,7 @@ static ssize_t tcm_loop_tpg_show_nexus(
 	struct tcm_loop_nexus *tl_nexus;
 	ssize_t ret;
 
-	tl_nexus = tl_tpg->tl_hba->tl_nexus;
+	tl_nexus = tl_tpg->tl_nexus;
 	if (!tl_nexus)
 		return -ENODEV;
 
drivers/target/loopback/tcm_loop.h +1 -6
@@ -25,11 +25,6 @@ struct tcm_loop_tmr {
 };
 
 struct tcm_loop_nexus {
-	int it_nexus_active;
-	/*
-	 * Pointer to Linux/SCSI HBA from linux/include/scsi_host.h
-	 */
-	struct scsi_host *sh;
 	/*
 	 * Pointer to TCM session for I_T Nexus
 	 */
@@ -45,6 +40,7 @@ struct tcm_loop_tpg {
 	atomic_t tl_tpg_port_count;
 	struct se_portal_group tl_se_tpg;
 	struct tcm_loop_hba *tl_hba;
+	struct tcm_loop_nexus *tl_nexus;
 };
 
 struct tcm_loop_hba {
@@ -53,7 +49,6 @@ struct tcm_loop_hba {
 	struct se_hba_s *se_hba;
 	struct se_lun *tl_hba_lun;
 	struct se_port *tl_hba_lun_sep;
-	struct tcm_loop_nexus *tl_nexus;
 	struct device dev;
 	struct Scsi_Host *sh;
 	struct tcm_loop_tpg tl_hba_tpgs[TL_TPGS_PER_HBA];
author maintainer
#44 lore raw

PATCH 44/51 vhost-scsi: Add missing virtio-scsi -> TCM attribute conversion

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Nicholas Bellinger <[email protected]>

commit 46243860806bdc2756f3ce8ac86b4d7c616bcd6c upstream.

While looking at hch's recent conversion to drop the MSG_*_TAG definitions, I noticed a long standing bug in vhost-scsi where the VIRTIO_SCSI_S_* attribute definitions where incorrectly being passed directly into target_submit_cmd_map_sgls().

This patch adds the missing virtio-scsi to TCM/SAM task attribute conversion.

Cc Christoph Hellwig <[email protected]>
Cc Michael S. Tsirkin <[email protected]>
Cc Paolo Bonzini <[email protected]>
---
 drivers/vhost/scsi.c |   23 ++++++++++++++++++++---
 1 file changed, 20 insertions(+), 3 deletions(-)
drivers/vhost/scsi.c +20 -3
@@ -820,6 +820,23 @@ static int vhost_scsi_map_iov_to_sgl(str
 	return 0;
 }
 
+static int vhost_scsi_to_tcm_attr(int attr)
+{
+	switch (attr) {
+	case VIRTIO_SCSI_S_SIMPLE:
+		return MSG_SIMPLE_TAG;
+	case VIRTIO_SCSI_S_ORDERED:
+		return MSG_ORDERED_TAG;
+	case VIRTIO_SCSI_S_HEAD:
+		return MSG_HEAD_TAG;
+	case VIRTIO_SCSI_S_ACA:
+		return MSG_ACA_TAG;
+	default:
+		break;
+	}
+	return MSG_SIMPLE_TAG;
+}
+
 static void tcm_vhost_submission_work(struct work_struct *work)
 {
 	struct tcm_vhost_cmd *tv_cmd =
@@ -846,9 +863,9 @@ static void tcm_vhost_submission_work(st
 	rc = target_submit_cmd_map_sgls(se_cmd, tv_nexus->tvn_se_sess,
 			tv_cmd->tvc_cdb, &tv_cmd->tvc_sense_buf[0],
 			tv_cmd->tvc_lun, tv_cmd->tvc_exp_data_len,
-			tv_cmd->tvc_task_attr, tv_cmd->tvc_data_direction,
-			0, sg_ptr, tv_cmd->tvc_sgl_count,
-			sg_bidi_ptr, sg_no_bidi);
+			vhost_scsi_to_tcm_attr(tv_cmd->tvc_task_attr),
+			tv_cmd->tvc_data_direction, 0, sg_ptr,
+			tv_cmd->tvc_sgl_count, sg_bidi_ptr, sg_no_bidi);
 	if (rc < 0) {
 		transport_send_check_condition_and_sense(se_cmd,
 				TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE, 0);
author maintainer
#45 lore raw

PATCH 45/51 iscsi,iser-target: Initiate termination only once

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Sagi Grimberg <[email protected]>

commit 954f23722b5753305be490330cf2680b7a25f4a3 upstream.

Since commit 0fc4ea701fcf ("Target/iser: Don't put isert_conn inside disconnected handler") we put the conn kref in isert_wait_conn, so we need .wait_conn to be invoked also in the error path.

Introduce call to isert_conn_terminate (called under lock) which transitions the connection state to TERMINATING and calls rdma_disconnect. If the state is already teminating, just bail out back (temination started).

Also, make sure to destroy the connection when getting a connect error event if didn't get to connected (state UP). Same for the handling of REJECTED and UNREACHABLE cma events.

Squashed:

iscsi-target: Add call to wait_conn in establishment error flow

---
 drivers/infiniband/ulp/isert/ib_isert.c   |   84 ++++++++++++++++++------------
 drivers/infiniband/ulp/isert/ib_isert.h   |    1 
 drivers/target/iscsi/iscsi_target_login.c |    3 +
 3 files changed, 54 insertions(+), 34 deletions(-)
drivers/infiniband/ulp/isert/ib_isert.c +51 -33
@@ -565,6 +565,33 @@ isert_put_conn(struct isert_conn *isert_
 	kref_put(&isert_conn->conn_kref, isert_release_conn_kref);
 }
 
+/**
+ * isert_conn_terminate() - Initiate connection termination
+ * @isert_conn: isert connection struct
+ *
+ * Notes:
+ * In case the connection state is UP, move state
+ * to TEMINATING and start teardown sequence (rdma_disconnect).
+ *
+ * This routine must be called with conn_mutex held. Thus it is
+ * safe to call multiple times.
+ */
+static void
+isert_conn_terminate(struct isert_conn *isert_conn)
+{
+	int err;
+
+	if (isert_conn->state == ISER_CONN_UP) {
+		isert_conn->state = ISER_CONN_TERMINATING;
+		pr_info("Terminating conn %p state %d\n",
+			   isert_conn, isert_conn->state);
+		err = rdma_disconnect(isert_conn->conn_cm_id);
+		if (err)
+			pr_warn("Failed rdma_disconnect isert_conn %p\n",
+				   isert_conn);
+	}
+}
+
 static void
 isert_disconnect_work(struct work_struct *work)
 {
@@ -573,33 +600,15 @@ isert_disconnect_work(struct work_struct
 
 	pr_debug("isert_disconnect_work(): >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>\n");
 	mutex_lock(&isert_conn->conn_mutex);
-	if (isert_conn->state == ISER_CONN_UP)
-		isert_conn->state = ISER_CONN_TERMINATING;
-
-	if (isert_conn->post_recv_buf_count == 0 &&
-	    atomic_read(&isert_conn->post_send_buf_count) == 0) {
-		mutex_unlock(&isert_conn->conn_mutex);
-		goto wake_up;
-	}
-	if (!isert_conn->conn_cm_id) {
-		mutex_unlock(&isert_conn->conn_mutex);
-		isert_put_conn(isert_conn);
-		return;
-	}
-
-	if (isert_conn->disconnect) {
-		/* Send DREQ/DREP towards our initiator */
-		rdma_disconnect(isert_conn->conn_cm_id);
-	}
-
+	isert_conn_terminate(isert_conn);
 	mutex_unlock(&isert_conn->conn_mutex);
 
-wake_up:
+	pr_info("conn %p completing conn_wait\n", isert_conn);
 	complete(&isert_conn->conn_wait);
 }
 
 static int
-isert_disconnected_handler(struct rdma_cm_id *cma_id, bool disconnect)
+isert_disconnected_handler(struct rdma_cm_id *cma_id)
 {
 	struct isert_conn *isert_conn;
 
@@ -612,18 +621,24 @@ isert_disconnected_handler(struct rdma_c
 
 	isert_conn = (struct isert_conn *)cma_id->context;
 
-	isert_conn->disconnect = disconnect;
 	INIT_WORK(&isert_conn->conn_logout_work, isert_disconnect_work);
 	schedule_work(&isert_conn->conn_logout_work);
 
 	return 0;
 }
 
+static void
+isert_connect_error(struct rdma_cm_id *cma_id)
+{
+	struct isert_conn *isert_conn = (struct isert_conn *)cma_id->context;
+
+	isert_put_conn(isert_conn);
+}
+
 static int
 isert_cma_handler(struct rdma_cm_id *cma_id, struct rdma_cm_event *event)
 {
 	int ret = 0;
-	bool disconnect = false;
 
 	pr_debug("isert_cma_handler: event %d status %d conn %p id %p\n",
 		 event->event, event->status, cma_id->context, cma_id);
@@ -641,11 +656,14 @@ isert_cma_handler(struct rdma_cm_id *cma
 	case RDMA_CM_EVENT_ADDR_CHANGE:    /* FALLTHRU */
 	case RDMA_CM_EVENT_DISCONNECTED:   /* FALLTHRU */
 	case RDMA_CM_EVENT_DEVICE_REMOVAL: /* FALLTHRU */
-		disconnect = true;
 	case RDMA_CM_EVENT_TIMEWAIT_EXIT:  /* FALLTHRU */
-		ret = isert_disconnected_handler(cma_id, disconnect);
+		ret = isert_disconnected_handler(cma_id);
 		break;
+	case RDMA_CM_EVENT_REJECTED:       /* FALLTHRU */
+	case RDMA_CM_EVENT_UNREACHABLE:    /* FALLTHRU */
 	case RDMA_CM_EVENT_CONNECT_ERROR:
+		isert_connect_error(cma_id);
+		break;
 	default:
 		pr_err("Unhandled RDMA CMA event: %d\n", event->event);
 		break;
@@ -1496,7 +1514,7 @@ isert_cq_rx_comp_err(struct isert_conn *
 		msleep(3000);
 
 	mutex_lock(&isert_conn->conn_mutex);
-	isert_conn->state = ISER_CONN_DOWN;
+	isert_conn_terminate(isert_conn);
 	mutex_unlock(&isert_conn->conn_mutex);
 
 	iscsit_cause_connection_reinstatement(isert_conn->conn, 0);
@@ -2268,10 +2286,6 @@ static void isert_wait_conn(struct iscsi
 	pr_debug("isert_wait_conn: Starting \n");
 
 	mutex_lock(&isert_conn->conn_mutex);
-	if (isert_conn->conn_cm_id) {
-		pr_debug("Calling rdma_disconnect from isert_wait_conn\n");
-		rdma_disconnect(isert_conn->conn_cm_id);
-	}
 	/*
 	 * Only wait for conn_wait_comp_err if the isert_conn made it
 	 * into full feature phase..
@@ -2280,13 +2294,17 @@ static void isert_wait_conn(struct iscsi
 		mutex_unlock(&isert_conn->conn_mutex);
 		return;
 	}
-	if (isert_conn->state == ISER_CONN_UP)
-		isert_conn->state = ISER_CONN_TERMINATING;
+	isert_conn_terminate(isert_conn);
 	mutex_unlock(&isert_conn->conn_mutex);
 
 	wait_for_completion(&isert_conn->conn_wait_comp_err);
-
 	wait_for_completion(&isert_conn->conn_wait);
+
+	mutex_lock(&isert_conn->conn_mutex);
+	isert_conn->state = ISER_CONN_DOWN;
+	mutex_unlock(&isert_conn->conn_mutex);
+
+	pr_info("Destroying conn %p\n", isert_conn);
 	isert_put_conn(isert_conn);
 }
 
drivers/infiniband/ulp/isert/ib_isert.h +0 -1
@@ -105,7 +105,6 @@ struct isert_conn {
 	struct completion	conn_wait;
 	struct completion	conn_wait_comp_err;
 	struct kref		conn_kref;
-	bool			disconnect;
 };
 
 #define ISERT_MAX_CQ 64
drivers/target/iscsi/iscsi_target_login.c +3 -0
@@ -1360,6 +1360,9 @@ old_sess_out:
 		conn->sock = NULL;
 	}
 
+	if (conn->conn_transport->iscsit_wait_conn)
+		conn->conn_transport->iscsit_wait_conn(conn);
+
 	if (conn->conn_transport->iscsit_free_conn)
 		conn->conn_transport->iscsit_free_conn(conn);
 
author maintainer
#46 lore raw

PATCH 46/51 iser-target: Fix flush + disconnect completion handling

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Sagi Grimberg <[email protected]>

commit 128e9cc84566a84146baea2335b3824288eed817 upstream.

ISER_CONN_UP state is not sufficient to know if we should wait for completion of flush errors and disconnected_handler event.

Instead, split it to 2 states:
- ISER_CONN_UP: Got to CM connected phase, This state indicates that we need to wait for a CM disconnect event before going to teardown.

- ISER_CONN_FULL_FEATURE: Got to full feature phase after we posted login response, This state indicates that we posted recv buffers and we need to wait for flush completions before going to teardown.

Also avoid deffering disconnected handler to a work, and handle it within disconnected handler. More work here is needed to handle DEVICE_REMOVAL event correctly (cleanup all resources).

Squashed:

iser-target: Don't deffer disconnected handler to a work

---
 drivers/infiniband/ulp/isert/ib_isert.c |   50 ++++++++++++++++++--------------
 drivers/infiniband/ulp/isert/ib_isert.h |    2 -
 2 files changed, 30 insertions(+), 22 deletions(-)
drivers/infiniband/ulp/isert/ib_isert.c +29 -21
@@ -544,6 +544,9 @@ isert_connected_handler(struct rdma_cm_i
 {
 	struct isert_conn *isert_conn = cma_id->context;
 
+	pr_info("conn %p\n", isert_conn);
+
+	isert_conn->state = ISER_CONN_UP;
 	kref_get(&isert_conn->conn_kref);
 }
 
@@ -570,8 +573,9 @@ isert_put_conn(struct isert_conn *isert_
  * @isert_conn: isert connection struct
  *
  * Notes:
- * In case the connection state is UP, move state
+ * In case the connection state is FULL_FEATURE, move state
  * to TEMINATING and start teardown sequence (rdma_disconnect).
+ * In case the connection state is UP, complete flush as well.
  *
  * This routine must be called with conn_mutex held. Thus it is
  * safe to call multiple times.
@@ -581,32 +585,31 @@ isert_conn_terminate(struct isert_conn *
 {
 	int err;
 
-	if (isert_conn->state == ISER_CONN_UP) {
-		isert_conn->state = ISER_CONN_TERMINATING;
+	switch (isert_conn->state) {
+	case ISER_CONN_TERMINATING:
+		break;
+	case ISER_CONN_UP:
+		/*
+		 * No flush completions will occur as we didn't
+		 * get to ISER_CONN_FULL_FEATURE yet, complete
+		 * to allow teardown progress.
+		 */
+		complete(&isert_conn->conn_wait_comp_err);
+	case ISER_CONN_FULL_FEATURE: /* FALLTHRU */
 		pr_info("Terminating conn %p state %d\n",
 			   isert_conn, isert_conn->state);
+		isert_conn->state = ISER_CONN_TERMINATING;
 		err = rdma_disconnect(isert_conn->conn_cm_id);
 		if (err)
 			pr_warn("Failed rdma_disconnect isert_conn %p\n",
 				   isert_conn);
+		break;
+	default:
+		pr_warn("conn %p teminating in state %d\n",
+			   isert_conn, isert_conn->state);
 	}
 }
 
-static void
-isert_disconnect_work(struct work_struct *work)
-{
-	struct isert_conn *isert_conn = container_of(work,
-				struct isert_conn, conn_logout_work);
-
-	pr_debug("isert_disconnect_work(): >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>\n");
-	mutex_lock(&isert_conn->conn_mutex);
-	isert_conn_terminate(isert_conn);
-	mutex_unlock(&isert_conn->conn_mutex);
-
-	pr_info("conn %p completing conn_wait\n", isert_conn);
-	complete(&isert_conn->conn_wait);
-}
-
 static int
 isert_disconnected_handler(struct rdma_cm_id *cma_id)
 {
@@ -621,8 +624,12 @@ isert_disconnected_handler(struct rdma_c
 
 	isert_conn = (struct isert_conn *)cma_id->context;
 
-	INIT_WORK(&isert_conn->conn_logout_work, isert_disconnect_work);
-	schedule_work(&isert_conn->conn_logout_work);
+	mutex_lock(&isert_conn->conn_mutex);
+	isert_conn_terminate(isert_conn);
+	mutex_unlock(&isert_conn->conn_mutex);
+
+	pr_info("conn %p completing conn_wait\n", isert_conn);
+	complete(&isert_conn->conn_wait);
 
 	return 0;
 }
@@ -865,7 +872,8 @@ isert_put_login_tx(struct iscsi_conn *co
 			if (ret)
 				return ret;
 
-			isert_conn->state = ISER_CONN_UP;
+			/* Now we are in FULL_FEATURE phase */
+			isert_conn->state = ISER_CONN_FULL_FEATURE;
 			goto post_send;
 		}
 
drivers/infiniband/ulp/isert/ib_isert.h +1 -1
@@ -21,6 +21,7 @@ enum iser_ib_op_code {
 enum iser_conn_state {
 	ISER_CONN_INIT,
 	ISER_CONN_UP,
+	ISER_CONN_FULL_FEATURE,
 	ISER_CONN_TERMINATING,
 	ISER_CONN_DOWN,
 };
@@ -100,7 +101,6 @@ struct isert_conn {
 	struct ib_mr		*conn_mr;
 	struct ib_qp		*conn_qp;
 	struct isert_device	*conn_device;
-	struct work_struct	conn_logout_work;
 	struct mutex		conn_mutex;
 	struct completion	conn_wait;
 	struct completion	conn_wait_comp_err;
author maintainer
#47 lore raw

PATCH 47/51 iser-target: Parallelize CM connection establishment

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Sagi Grimberg <[email protected]>

commit 2371e5da8cfe91443339b54444dec6254fdd6dfc upstream.

There is no point in accepting a new CM request only when we are completely done with the last iscsi login. Instead we accept immediately, this will also cause the CM connection to reach connected state and the initiator is allowed to send the first login. We mark that we got the initial login and let iscsi layer pick it up when it gets there.

This reduces the parallel login sequence by a factor of more then 4 (and more for multi-login) and also prevents the initiator (who does all logins in parallel) from giving up on login timeout expiration.

In order to support multiple login requests sequence (CHAP) we call isert_rx_login_req from isert_rx_completion insead of letting isert_get_login_rx call it.

Squashed:

iser-target: Use kref_get_unless_zero in connected_handler
iser-target: Acquire conn_mutex when changing connection state
iser-target: Reject connect request in failure path

---
 drivers/infiniband/ulp/isert/ib_isert.c |   90 +++++++++++++++++++++++---------
 drivers/infiniband/ulp/isert/ib_isert.h |    2 
 2 files changed, 67 insertions(+), 25 deletions(-)
drivers/infiniband/ulp/isert/ib_isert.c +65 -25
@@ -42,6 +42,11 @@ static struct workqueue_struct *isert_rx
 static struct workqueue_struct *isert_comp_wq;
 static struct kmem_cache *isert_cmd_cache;
 
+static int
+isert_rdma_post_recvl(struct isert_conn *isert_conn);
+static int
+isert_rdma_accept(struct isert_conn *isert_conn);
+
 static void
 isert_qp_event_callback(struct ib_event *e, void *context)
 {
@@ -414,6 +419,7 @@ isert_connect_request(struct rdma_cm_id
 	isert_conn->state = ISER_CONN_INIT;
 	INIT_LIST_HEAD(&isert_conn->conn_accept_node);
 	init_completion(&isert_conn->conn_login_comp);
+	init_completion(&isert_conn->login_req_comp);
 	init_completion(&isert_conn->conn_wait);
 	init_completion(&isert_conn->conn_wait_comp_err);
 	kref_init(&isert_conn->conn_kref);
@@ -479,6 +485,14 @@ isert_connect_request(struct rdma_cm_id
 	if (ret)
 		goto out_conn_dev;
 
+	ret = isert_rdma_post_recvl(isert_conn);
+	if (ret)
+		goto out_conn_dev;
+
+	ret = isert_rdma_accept(isert_conn);
+	if (ret)
+		goto out_conn_dev;
+
 	mutex_lock(&isert_np->np_accept_mutex);
 	list_add_tail(&isert_conn->conn_accept_node, &isert_np->np_accept_list);
 	mutex_unlock(&isert_np->np_accept_mutex);
@@ -499,6 +513,7 @@ out_login_buf:
 	kfree(isert_conn->login_buf);
 out:
 	kfree(isert_conn);
+	rdma_reject(cma_id, NULL, 0);
 	return ret;
 }
 
@@ -546,8 +561,15 @@ isert_connected_handler(struct rdma_cm_i
 
 	pr_info("conn %p\n", isert_conn);
 
-	isert_conn->state = ISER_CONN_UP;
-	kref_get(&isert_conn->conn_kref);
+	if (!kref_get_unless_zero(&isert_conn->conn_kref)) {
+		pr_warn("conn %p connect_release is running\n", isert_conn);
+		return;
+	}
+
+	mutex_lock(&isert_conn->conn_mutex);
+	if (isert_conn->state != ISER_CONN_FULL_FEATURE)
+		isert_conn->state = ISER_CONN_UP;
+	mutex_unlock(&isert_conn->conn_mutex);
 }
 
 static void
@@ -873,7 +895,9 @@ isert_put_login_tx(struct iscsi_conn *co
 				return ret;
 
 			/* Now we are in FULL_FEATURE phase */
+			mutex_lock(&isert_conn->conn_mutex);
 			isert_conn->state = ISER_CONN_FULL_FEATURE;
+			mutex_unlock(&isert_conn->conn_mutex);
 			goto post_send;
 		}
 
@@ -890,18 +914,17 @@ post_send:
 }
 
 static void
-isert_rx_login_req(struct iser_rx_desc *rx_desc, int rx_buflen,
-		   struct isert_conn *isert_conn)
+isert_rx_login_req(struct isert_conn *isert_conn)
 {
+	struct iser_rx_desc *rx_desc = (void *)isert_conn->login_req_buf;
+	int rx_buflen = isert_conn->login_req_len;
 	struct iscsi_conn *conn = isert_conn->conn;
 	struct iscsi_login *login = conn->conn_login;
 	int size;
 
-	if (!login) {
-		pr_err("conn->conn_login is NULL\n");
-		dump_stack();
-		return;
-	}
+	pr_info("conn %p\n", isert_conn);
+
+	WARN_ON_ONCE(!login);
 
 	if (login->first_request) {
 		struct iscsi_login_req *login_req =
@@ -931,7 +954,8 @@ isert_rx_login_req(struct iser_rx_desc *
 		 size, rx_buflen, MAX_KEY_VALUE_PAIRS);
 	memcpy(login->req_buf, &rx_desc->data[0], size);
 
-	complete(&isert_conn->conn_login_comp);
+	if (login->first_request)
+		complete(&isert_conn->conn_login_comp);
 }
 
 static void
@@ -1208,11 +1232,20 @@ isert_rx_completion(struct iser_rx_desc
 		 hdr->opcode, hdr->itt, hdr->flags,
 		 (int)(xfer_len - ISER_HEADERS_LEN));
 
-	if ((char *)desc == isert_conn->login_req_buf)
-		isert_rx_login_req(desc, xfer_len - ISER_HEADERS_LEN,
-				   isert_conn);
-	else
+	if ((char *)desc == isert_conn->login_req_buf) {
+		isert_conn->login_req_len = xfer_len - ISER_HEADERS_LEN;
+		if (isert_conn->conn) {
+			struct iscsi_login *login = isert_conn->conn->conn_login;
+
+			if (login && !login->first_request)
+				isert_rx_login_req(isert_conn);
+		}
+		mutex_lock(&isert_conn->conn_mutex);
+		complete(&isert_conn->login_req_comp);
+		mutex_unlock(&isert_conn->conn_mutex);
+	} else {
 		isert_rx_do_work(desc, isert_conn);
+	}
 
 	ib_dma_sync_single_for_device(ib_dev, rx_dma, rx_buflen,
 				      DMA_FROM_DEVICE);
@@ -2177,13 +2210,27 @@ isert_get_login_rx(struct iscsi_conn *co
 	struct isert_conn *isert_conn = (struct isert_conn *)conn->context;
 	int ret;
 
-	pr_debug("isert_get_login_rx before conn_login_comp conn: %p\n", conn);
+	pr_info("before login_req comp conn: %p\n", isert_conn);
+	ret = wait_for_completion_interruptible(&isert_conn->login_req_comp);
+	if (ret) {
+		pr_err("isert_conn %p interrupted before got login req\n",
+			isert_conn);
+		return ret;
+	}
+	isert_conn->login_req_comp.done = 0;
+
+	if (!login->first_request)
+		return 0;
+
+	isert_rx_login_req(isert_conn);
+
+	pr_info("before conn_login_comp conn: %p\n", conn);
 
 	ret = wait_for_completion_interruptible(&isert_conn->conn_login_comp);
 	if (ret)
 		return ret;
 
-	pr_debug("isert_get_login_rx processing login->req: %p\n", login->req);
+	pr_info("processing login->req: %p\n", login->req);
 	return 0;
 }
 
@@ -2261,17 +2308,10 @@ accept_wait:
 	isert_conn->conn = conn;
 	max_accept = 0;
 
-	ret = isert_rdma_post_recvl(isert_conn);
-	if (ret)
-		return ret;
-
-	ret = isert_rdma_accept(isert_conn);
-	if (ret)
-		return ret;
-
 	isert_set_conn_info(np, conn, isert_conn);
 
-	pr_debug("Processing isert_accept_np: isert_conn: %p\n", isert_conn);
+	pr_debug("Processing isert_conn: %p\n", isert_conn);
+
 	return 0;
 }
 
drivers/infiniband/ulp/isert/ib_isert.h +2 -0
@@ -88,6 +88,7 @@ struct isert_conn {
 	char			*login_req_buf;
 	char			*login_rsp_buf;
 	u64			login_req_dma;
+	int			login_req_len;
 	u64			login_rsp_dma;
 	unsigned int		conn_rx_desc_head;
 	struct iser_rx_desc	*conn_rx_descs;
@@ -95,6 +96,7 @@ struct isert_conn {
 	struct iscsi_conn	*conn;
 	struct list_head	conn_accept_node;
 	struct completion	conn_login_comp;
+	struct completion	login_req_comp;
 	struct iser_tx_desc	conn_login_tx_desc;
 	struct rdma_cm_id	*conn_cm_id;
 	struct ib_pd		*conn_pd;
author maintainer
#48 lore raw

PATCH 48/51 iser-target: Fix connected_handler + teardown flow race

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Sagi Grimberg <[email protected]>

commit 19e2090fb246ca21b3e569ead51a6a7a1748eadd upstream.

Take isert_conn pointer from cm_id->qp->qp_context. This will allow us to know that the cm_id context is always the network portal. This will make the cm_id event check (connection or network portal) more reliable.

In order to avoid a NULL dereference in cma_id->qp->qp_context we destroy the qp after we destroy the cm_id (and make the dereference safe). session stablishment/teardown sequences can happen in parallel, we should take into account that connected_handler might race with connection teardown flow.

Also, protect isert_conn->conn_device->active_qps decrement within the error patch during QP creation failure and the normal teardown path in isert_connect_release().

Squashed:

iser-target: Decrement completion context active_qps in error flow

---
 drivers/infiniband/ulp/isert/ib_isert.c |   31 +++++++++++++++++++------------
 1 file changed, 19 insertions(+), 12 deletions(-)
drivers/infiniband/ulp/isert/ib_isert.c +19 -12
@@ -132,12 +132,18 @@ isert_conn_setup_qp(struct isert_conn *i
 	ret = rdma_create_qp(cma_id, isert_conn->conn_pd, &attr);
 	if (ret) {
 		pr_err("rdma_create_qp failed for cma_id %d\n", ret);
-		return ret;
+		goto err;
 	}
 	isert_conn->conn_qp = cma_id->qp;
 	pr_debug("rdma_create_qp() returned success >>>>>>>>>>>>>>>>>>>>>>>>>.\n");
 
 	return 0;
+err:
+	mutex_lock(&device_list_mutex);
+	device->cq_active_qps[min_index]--;
+	mutex_unlock(&device_list_mutex);
+
+	return ret;
 }
 
 static void
@@ -425,7 +431,6 @@ isert_connect_request(struct rdma_cm_id
 	kref_init(&isert_conn->conn_kref);
 	mutex_init(&isert_conn->conn_mutex);
 
-	cma_id->context = isert_conn;
 	isert_conn->conn_cm_id = cma_id;
 	isert_conn->responder_resources = event->param.conn.responder_resources;
 	isert_conn->initiator_depth = event->param.conn.initiator_depth;
@@ -526,18 +531,20 @@ isert_connect_release(struct isert_conn
 
 	pr_debug("Entering isert_connect_release(): >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>\n");
 
+	isert_free_rx_descriptors(isert_conn);
+	rdma_destroy_id(isert_conn->conn_cm_id);
+
 	if (isert_conn->conn_qp) {
 		cq_index = ((struct isert_cq_desc *)
 			isert_conn->conn_qp->recv_cq->cq_context)->cq_index;
 		pr_debug("isert_connect_release: cq_index: %d\n", cq_index);
+		mutex_lock(&device_list_mutex);
 		isert_conn->conn_device->cq_active_qps[cq_index]--;
+		mutex_unlock(&device_list_mutex);
 
-		rdma_destroy_qp(isert_conn->conn_cm_id);
+		ib_destroy_qp(isert_conn->conn_qp);
 	}
 
-	isert_free_rx_descriptors(isert_conn);
-	rdma_destroy_id(isert_conn->conn_cm_id);
-
 	if (isert_conn->login_buf) {
 		ib_dma_unmap_single(ib_dev, isert_conn->login_rsp_dma,
 				    ISER_RX_LOGIN_SIZE, DMA_TO_DEVICE);
@@ -557,7 +564,7 @@ isert_connect_release(struct isert_conn
 static void
 isert_connected_handler(struct rdma_cm_id *cma_id)
 {
-	struct isert_conn *isert_conn = cma_id->context;
+	struct isert_conn *isert_conn = cma_id->qp->qp_context;
 
 	pr_info("conn %p\n", isert_conn);
 
@@ -635,16 +642,16 @@ isert_conn_terminate(struct isert_conn *
 static int
 isert_disconnected_handler(struct rdma_cm_id *cma_id)
 {
+	struct iscsi_np *np = cma_id->context;
+	struct isert_np *isert_np = np->np_context;
 	struct isert_conn *isert_conn;
 
-	if (!cma_id->qp) {
-		struct isert_np *isert_np = cma_id->context;
-
+	if (isert_np->np_cm_id == cma_id) {
 		isert_np->np_cm_id = NULL;
 		return -1;
 	}
 
-	isert_conn = (struct isert_conn *)cma_id->context;
+	isert_conn = cma_id->qp->qp_context;
 
 	mutex_lock(&isert_conn->conn_mutex);
 	isert_conn_terminate(isert_conn);
@@ -659,7 +666,7 @@ isert_disconnected_handler(struct rdma_c
 static void
 isert_connect_error(struct rdma_cm_id *cma_id)
 {
-	struct isert_conn *isert_conn = (struct isert_conn *)cma_id->context;
+	struct isert_conn *isert_conn = cma_id->qp->qp_context;
 
 	isert_put_conn(isert_conn);
 }
author maintainer
#49 lore raw

PATCH 49/51 iser-target: Handle ADDR_CHANGE event for listener cm_id

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Sagi Grimberg <[email protected]>

commit ca6c1d82d12d8013fb75ce015900d62b9754623c upstream.

The np listener cm_id will also get ADDR_CHANGE event upcall (in case it is bound to a specific IP). Handle it correctly by creating a new cm_id and implicitly destroy the old one.

Since this is the second event a listener np cm_id may encounter, we move the np cm_id event handling to a routine.

Squashed:

iser-target: Move cma_id setup to a function

---
 drivers/infiniband/ulp/isert/ib_isert.c |  107 ++++++++++++++++++++++----------
 drivers/infiniband/ulp/isert/ib_isert.h |    1 
 2 files changed, 77 insertions(+), 31 deletions(-)
drivers/infiniband/ulp/isert/ib_isert.c +76 -31
@@ -46,6 +46,7 @@ static int
 isert_rdma_post_recvl(struct isert_conn *isert_conn);
 static int
 isert_rdma_accept(struct isert_conn *isert_conn);
+struct rdma_cm_id *isert_setup_id(struct isert_np *isert_np);
 
 static void
 isert_qp_event_callback(struct ib_event *e, void *context)
@@ -399,8 +400,8 @@ isert_device_find_by_ib_dev(struct rdma_
 static int
 isert_connect_request(struct rdma_cm_id *cma_id, struct rdma_cm_event *event)
 {
-	struct iscsi_np *np = cma_id->context;
-	struct isert_np *isert_np = np->np_context;
+	struct isert_np *isert_np = cma_id->context;
+	struct iscsi_np *np = isert_np->np;
 	struct isert_conn *isert_conn;
 	struct isert_device *device;
 	struct ib_device *ib_dev = cma_id->device;
@@ -640,17 +641,41 @@ isert_conn_terminate(struct isert_conn *
 }
 
 static int
-isert_disconnected_handler(struct rdma_cm_id *cma_id)
+isert_np_cma_handler(struct isert_np *isert_np,
+		     enum rdma_cm_event_type event)
 {
-	struct iscsi_np *np = cma_id->context;
-	struct isert_np *isert_np = np->np_context;
-	struct isert_conn *isert_conn;
+	pr_debug("isert np %p, handling event %d\n", isert_np, event);
 
-	if (isert_np->np_cm_id == cma_id) {
+	switch (event) {
+	case RDMA_CM_EVENT_DEVICE_REMOVAL:
 		isert_np->np_cm_id = NULL;
-		return -1;
+		break;
+	case RDMA_CM_EVENT_ADDR_CHANGE:
+		isert_np->np_cm_id = isert_setup_id(isert_np);
+		if (IS_ERR(isert_np->np_cm_id)) {
+			pr_err("isert np %p setup id failed: %ld\n",
+				 isert_np, PTR_ERR(isert_np->np_cm_id));
+			isert_np->np_cm_id = NULL;
+		}
+		break;
+	default:
+		pr_err("isert np %p Unexpected event %d\n",
+			  isert_np, event);
 	}
 
+	return -1;
+}
+
+static int
+isert_disconnected_handler(struct rdma_cm_id *cma_id,
+			   enum rdma_cm_event_type event)
+{
+	struct isert_np *isert_np = cma_id->context;
+	struct isert_conn *isert_conn;
+
+	if (isert_np->np_cm_id == cma_id)
+		return isert_np_cma_handler(cma_id->context, event);
+
 	isert_conn = cma_id->qp->qp_context;
 
 	mutex_lock(&isert_conn->conn_mutex);
@@ -693,7 +718,7 @@ isert_cma_handler(struct rdma_cm_id *cma
 	case RDMA_CM_EVENT_DISCONNECTED:   /* FALLTHRU */
 	case RDMA_CM_EVENT_DEVICE_REMOVAL: /* FALLTHRU */
 	case RDMA_CM_EVENT_TIMEWAIT_EXIT:  /* FALLTHRU */
-		ret = isert_disconnected_handler(cma_id);
+		ret = isert_disconnected_handler(cma_id, event->event);
 		break;
 	case RDMA_CM_EVENT_REJECTED:       /* FALLTHRU */
 	case RDMA_CM_EVENT_UNREACHABLE:    /* FALLTHRU */
@@ -2123,13 +2148,51 @@ isert_response_queue(struct iscsi_conn *
 	return ret;
 }
 
+struct rdma_cm_id *
+isert_setup_id(struct isert_np *isert_np)
+{
+	struct iscsi_np *np = isert_np->np;
+	struct rdma_cm_id *id;
+	struct sockaddr *sa;
+	int ret;
+
+	sa = (struct sockaddr *)&np->np_sockaddr;
+	pr_debug("ksockaddr: %p, sa: %p\n", &np->np_sockaddr, sa);
+
+	id = rdma_create_id(isert_cma_handler, isert_np,
+			    RDMA_PS_TCP, IB_QPT_RC);
+	if (IS_ERR(id)) {
+		pr_err("rdma_create_id() failed: %ld\n", PTR_ERR(id));
+		ret = PTR_ERR(id);
+		goto out;
+	}
+	pr_debug("id %p context %p\n", id, id->context);
+
+	ret = rdma_bind_addr(id, sa);
+	if (ret) {
+		pr_err("rdma_bind_addr() failed: %d\n", ret);
+		goto out_id;
+	}
+
+	ret = rdma_listen(id, ISERT_RDMA_LISTEN_BACKLOG);
+	if (ret) {
+		pr_err("rdma_listen() failed: %d\n", ret);
+		goto out_id;
+	}
+
+	return id;
+out_id:
+	rdma_destroy_id(id);
+out:
+	return ERR_PTR(ret);
+}
+
 static int
 isert_setup_np(struct iscsi_np *np,
 	       struct __kernel_sockaddr_storage *ksockaddr)
 {
 	struct isert_np *isert_np;
 	struct rdma_cm_id *isert_lid;
-	struct sockaddr *sa;
 	int ret;
 
 	isert_np = kzalloc(sizeof(struct isert_np), GFP_KERNEL);
@@ -2141,9 +2204,8 @@ isert_setup_np(struct iscsi_np *np,
 	mutex_init(&isert_np->np_accept_mutex);
 	INIT_LIST_HEAD(&isert_np->np_accept_list);
 	init_completion(&isert_np->np_login_comp);
+	isert_np->np = np;
 
-	sa = (struct sockaddr *)ksockaddr;
-	pr_debug("ksockaddr: %p, sa: %p\n", ksockaddr, sa);
 	/*
 	 * Setup the np->np_sockaddr from the passed sockaddr setup
 	 * in iscsi_target_configfs.c code..
@@ -2151,37 +2213,20 @@ isert_setup_np(struct iscsi_np *np,
 	memcpy(&np->np_sockaddr, ksockaddr,
 	       sizeof(struct __kernel_sockaddr_storage));
 
-	isert_lid = rdma_create_id(isert_cma_handler, np, RDMA_PS_TCP,
-				IB_QPT_RC);
+	isert_lid = isert_setup_id(isert_np);
 	if (IS_ERR(isert_lid)) {
-		pr_err("rdma_create_id() for isert_listen_handler failed: %ld\n",
-		       PTR_ERR(isert_lid));
 		ret = PTR_ERR(isert_lid);
 		goto out;
 	}
 
-	ret = rdma_bind_addr(isert_lid, sa);
-	if (ret) {
-		pr_err("rdma_bind_addr() for isert_lid failed: %d\n", ret);
-		goto out_lid;
-	}
-
-	ret = rdma_listen(isert_lid, ISERT_RDMA_LISTEN_BACKLOG);
-	if (ret) {
-		pr_err("rdma_listen() for isert_lid failed: %d\n", ret);
-		goto out_lid;
-	}
-
 	isert_np->np_cm_id = isert_lid;
 	np->np_context = isert_np;
-	pr_debug("Setup isert_lid->context: %p\n", isert_lid->context);
 
 	return 0;
 
-out_lid:
-	rdma_destroy_id(isert_lid);
 out:
 	kfree(isert_np);
+
 	return ret;
 }
 
drivers/infiniband/ulp/isert/ib_isert.h +1 -0
@@ -132,6 +132,7 @@ struct isert_device {
 };
 
 struct isert_np {
+	struct iscsi_np         *np;
 	struct semaphore	np_sem;
 	struct rdma_cm_id	*np_cm_id;
 	struct mutex		np_accept_mutex;
author maintainer
#50 lore raw

PATCH 50/51 iser-target: Fix implicit termination of connections

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Sagi Grimberg <[email protected]>

commit b02efbfc9a051b41e71fe8f94ddf967260e024a6 upstream.

In situations such as bond failover, The new session establishment implicitly invokes the termination of the old connection.

So, we don't want to wait for the old connection wait_conn to completely terminate before we accept the new connection and post a login response.

The solution is to deffer the comp_wait completion and the conn_put to a work so wait_conn will effectively be non-blocking (flush errors are assumed to come very fast).

We allocate isert_release_wq with WQ_UNBOUND and WQ_UNBOUND_MAX_ACTIVE to spread the concurrency of release works.

---
 drivers/infiniband/ulp/isert/ib_isert.c |   45 ++++++++++++++++++++++++--------
 drivers/infiniband/ulp/isert/ib_isert.h |    1 
 2 files changed, 36 insertions(+), 10 deletions(-)
drivers/infiniband/ulp/isert/ib_isert.c +35 -10
@@ -40,6 +40,7 @@ static DEFINE_MUTEX(device_list_mutex);
 static LIST_HEAD(device_list);
 static struct workqueue_struct *isert_rx_wq;
 static struct workqueue_struct *isert_comp_wq;
+static struct workqueue_struct *isert_release_wq;
 static struct kmem_cache *isert_cmd_cache;
 
 static int
@@ -2379,6 +2380,24 @@ isert_free_np(struct iscsi_np *np)
 	kfree(isert_np);
 }
 
+static void isert_release_work(struct work_struct *work)
+{
+	struct isert_conn *isert_conn = container_of(work,
+						     struct isert_conn,
+						     release_work);
+
+	pr_info("Starting release conn %p\n", isert_conn);
+
+	wait_for_completion(&isert_conn->conn_wait);
+
+	mutex_lock(&isert_conn->conn_mutex);
+	isert_conn->state = ISER_CONN_DOWN;
+	mutex_unlock(&isert_conn->conn_mutex);
+
+	pr_info("Destroying conn %p\n", isert_conn);
+	isert_put_conn(isert_conn);
+}
+
 static void isert_wait_conn(struct iscsi_conn *conn)
 {
 	struct isert_conn *isert_conn = conn->context;
@@ -2398,14 +2417,9 @@ static void isert_wait_conn(struct iscsi
 	mutex_unlock(&isert_conn->conn_mutex);
 
 	wait_for_completion(&isert_conn->conn_wait_comp_err);
-	wait_for_completion(&isert_conn->conn_wait);
-
-	mutex_lock(&isert_conn->conn_mutex);
-	isert_conn->state = ISER_CONN_DOWN;
-	mutex_unlock(&isert_conn->conn_mutex);
 
-	pr_info("Destroying conn %p\n", isert_conn);
-	isert_put_conn(isert_conn);
+	INIT_WORK(&isert_conn->release_work, isert_release_work);
+	queue_work(isert_release_wq, &isert_conn->release_work);
 }
 
 static void isert_free_conn(struct iscsi_conn *conn)
@@ -2451,20 +2465,30 @@ static int __init isert_init(void)
 		goto destroy_rx_wq;
 	}
 
+	isert_release_wq = alloc_workqueue("isert_release_wq", WQ_UNBOUND,
+					WQ_UNBOUND_MAX_ACTIVE);
+	if (!isert_release_wq) {
+		pr_err("Unable to allocate isert_release_wq\n");
+		ret = -ENOMEM;
+		goto destroy_comp_wq;
+	}
+
 	isert_cmd_cache = kmem_cache_create("isert_cmd_cache",
 			sizeof(struct isert_cmd), __alignof__(struct isert_cmd),
 			0, NULL);
 	if (!isert_cmd_cache) {
 		pr_err("Unable to create isert_cmd_cache\n");
 		ret = -ENOMEM;
-		goto destroy_tx_cq;
+		goto destroy_release_wq;
 	}
 
 	iscsit_register_transport(&iser_target_transport);
-	pr_debug("iSER_TARGET[0] - Loaded iser_target_transport\n");
+	pr_info("iSER_TARGET[0] - Loaded iser_target_transport\n");
 	return 0;
 
-destroy_tx_cq:
+destroy_release_wq:
+	destroy_workqueue(isert_release_wq);
+destroy_comp_wq:
 	destroy_workqueue(isert_comp_wq);
 destroy_rx_wq:
 	destroy_workqueue(isert_rx_wq);
@@ -2475,6 +2499,7 @@ static void __exit isert_exit(void)
 {
 	flush_scheduled_work();
 	kmem_cache_destroy(isert_cmd_cache);
+	destroy_workqueue(isert_release_wq);
 	destroy_workqueue(isert_comp_wq);
 	destroy_workqueue(isert_rx_wq);
 	iscsit_unregister_transport(&iser_target_transport);
drivers/infiniband/ulp/isert/ib_isert.h +1 -0
@@ -107,6 +107,7 @@ struct isert_conn {
 	struct completion	conn_wait;
 	struct completion	conn_wait_comp_err;
 	struct kref		conn_kref;
+	struct work_struct	release_work;
 };
 
 #define ISERT_MAX_CQ 64
author maintainer
#51 lore raw

PATCH 51/51 target: Drop arbitrary maximum I/O size limit

In reply to Greg Kroah-Hartman

3.10-stable review patch. If anyone has any objections, please let me know.

------------------

From: Nicholas Bellinger <[email protected]>

commit 046ba64285a4389ae5e9a7dfa253c6bff3d7c341 upstream.

This patch drops the arbitrary maximum I/O size limit in sbc_parse_cdb(), which currently for fabric_max_sectors is hardcoded to 8192 (4 MB for 512 byte sector devices), and for hw_max_sectors is a backend driver dependent value.

This limit is problematic because Linux initiators have only recently started to honor block limits MAXIMUM TRANSFER LENGTH, and other non-Linux based initiators (eg: MSFT Fibre Channel) can also generate I/Os larger than 4 MB in size.

Currently when this happens, the following message will appear on the target resulting in I/Os being returned with non recoverable status:

SCSI OP 28h with too big sectors 16384 exceeds fabric_max_sectors: 8192

Instead, drop both [fabric,hw]_max_sector checks in sbc_parse_cdb(), and convert the existing hw_max_sectors into a purely informational attribute used to represent the granuality that backend driver and/or subsystem code is splitting I/Os upon.

Also, update FILEIO with an explicit FD_MAX_BYTES check in fd_execute_rw() to deal with the one special iovec limitiation case.

v2 changes:
- Drop hw_max_sectors check in sbc_parse_cdb()

---
 drivers/target/target_core_device.c |    8 ++++----
 drivers/target/target_core_file.c   |   11 ++++++++++-
 drivers/target/target_core_iblock.c |    2 +-
 drivers/target/target_core_sbc.c    |   15 ---------------
 drivers/target/target_core_spc.c    |    5 +----
 5 files changed, 16 insertions(+), 25 deletions(-)
drivers/target/target_core_device.c +4 -4
@@ -1037,10 +1037,10 @@ int se_dev_set_optimal_sectors(struct se
 				" changed for TCM/pSCSI\n", dev);
 		return -EINVAL;
 	}
-	if (optimal_sectors > dev->dev_attrib.fabric_max_sectors) {
+	if (optimal_sectors > dev->dev_attrib.hw_max_sectors) {
 		pr_err("dev[%p]: Passed optimal_sectors %u cannot be"
-			" greater than fabric_max_sectors: %u\n", dev,
-			optimal_sectors, dev->dev_attrib.fabric_max_sectors);
+			" greater than hw_max_sectors: %u\n", dev,
+			optimal_sectors, dev->dev_attrib.hw_max_sectors);
 		return -EINVAL;
 	}
 
@@ -1442,7 +1442,6 @@ struct se_device *target_alloc_device(st
 				DA_UNMAP_GRANULARITY_ALIGNMENT_DEFAULT;
 	dev->dev_attrib.max_write_same_len = DA_MAX_WRITE_SAME_LEN;
 	dev->dev_attrib.fabric_max_sectors = DA_FABRIC_MAX_SECTORS;
-	dev->dev_attrib.optimal_sectors = DA_FABRIC_MAX_SECTORS;
 
 	return dev;
 }
@@ -1475,6 +1474,7 @@ int target_configure_device(struct se_de
 	dev->dev_attrib.hw_max_sectors =
 		se_dev_align_max_sectors(dev->dev_attrib.hw_max_sectors,
 					 dev->dev_attrib.hw_block_size);
+	dev->dev_attrib.optimal_sectors = dev->dev_attrib.hw_max_sectors;
 
 	dev->dev_index = scsi_get_new_index(SCSI_DEVICE_INDEX);
 	dev->creation_time = get_jiffies_64();
drivers/target/target_core_file.c +10 -1
@@ -554,7 +554,16 @@ fd_execute_rw(struct se_cmd *cmd)
 	enum dma_data_direction data_direction = cmd->data_direction;
 	struct se_device *dev = cmd->se_dev;
 	int ret = 0;
-
+	/*
+	 * We are currently limited by the number of iovecs (2048) per
+	 * single vfs_[writev,readv] call.
+	 */
+	if (cmd->data_length > FD_MAX_BYTES) {
+		pr_err("FILEIO: Not able to process I/O of %u bytes due to"
+		       "FD_MAX_BYTES: %u iovec count limitiation\n",
+			cmd->data_length, FD_MAX_BYTES);
+		return TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE;
+	}
 	/*
 	 * Call vectorized fileio functions to map struct scatterlist
 	 * physical memory addresses to struct iovec virtual memory.
drivers/target/target_core_iblock.c +1 -1
@@ -122,7 +122,7 @@ static int iblock_configure_device(struc
 	q = bdev_get_queue(bd);
 
 	dev->dev_attrib.hw_block_size = bdev_logical_block_size(bd);
-	dev->dev_attrib.hw_max_sectors = UINT_MAX;
+	dev->dev_attrib.hw_max_sectors = queue_max_hw_sectors(q);
 	dev->dev_attrib.hw_queue_depth = q->nr_requests;
 
 	/*
drivers/target/target_core_sbc.c +0 -15
@@ -561,21 +561,6 @@ sbc_parse_cdb(struct se_cmd *cmd, struct
 	if (cmd->se_cmd_flags & SCF_SCSI_DATA_CDB) {
 		unsigned long long end_lba;
 
-		if (sectors > dev->dev_attrib.fabric_max_sectors) {
-			printk_ratelimited(KERN_ERR "SCSI OP %02xh with too"
-				" big sectors %u exceeds fabric_max_sectors:"
-				" %u\n", cdb[0], sectors,
-				dev->dev_attrib.fabric_max_sectors);
-			return TCM_INVALID_CDB_FIELD;
-		}
-		if (sectors > dev->dev_attrib.hw_max_sectors) {
-			printk_ratelimited(KERN_ERR "SCSI OP %02xh with too"
-				" big sectors %u exceeds backend hw_max_sectors:"
-				" %u\n", cdb[0], sectors,
-				dev->dev_attrib.hw_max_sectors);
-			return TCM_INVALID_CDB_FIELD;
-		}
-
 		end_lba = dev->transport->get_blocks(dev) + 1;
 		if (cmd->t_task_lba + sectors > end_lba) {
 			pr_err("cmd exceeds last lba %llu "
drivers/target/target_core_spc.c +1 -4
@@ -444,7 +444,6 @@ static sense_reason_t
 spc_emulate_evpd_b0(struct se_cmd *cmd, unsigned char *buf)
 {
 	struct se_device *dev = cmd->se_dev;
-	u32 max_sectors;
 	int have_tp = 0;
 
 	/*
@@ -469,9 +468,7 @@ spc_emulate_evpd_b0(struct se_cmd *cmd,
 	/*
 	 * Set MAXIMUM TRANSFER LENGTH
 	 */
-	max_sectors = min(dev->dev_attrib.fabric_max_sectors,
-			  dev->dev_attrib.hw_max_sectors);
-	put_unaligned_be32(max_sectors, &buf[8]);
+	put_unaligned_be32(dev->dev_attrib.hw_max_sectors, &buf[8]);
 
 	/*
 	 * Set OPTIMAL TRANSFER LENGTH
#52 lore raw
In reply to Greg Kroah-Hartman
Quoted 8 lines · Greg Kroah-Hartman

This is the start of the stable review cycle for the 3.10.68 release. There are 51 patches in this series, all will be posted as a response to this one. If anyone has any issues with these being applied, please let me know.

Responses should be made by Thu Feb 5 23:17:14 UTC 2015. Anything received after that time might be too late.

Build results:

	total: 137 pass: 137 fail: 0
Qemu tests:
	total: 27 pass: 27 fail: 0

Details are available at http://server.roeck-us.net:8010/builders.

Guenter

#53 lore raw
In reply to Greg Kroah-Hartman
Quoted 8 lines · Greg Kroah-Hartman

This is the start of the stable review cycle for the 3.10.68 release. There are 51 patches in this series, all will be posted as a response to this one. If anyone has any issues with these being applied, please let me know.

Responses should be made by Thu Feb 5 23:17:14 UTC 2015. Anything received after that time might be too late.

Quoted 8 lines

The whole patch series can be found in one patch at:

	kernel.org/pub/linux/kernel/v3.0/stable-review/patch-3.10.68-rc1.gz

and the diffstat can be found below.

thanks,

greg k-h

Compiled and booted on my test system. No dmesg regressions.

thanks,
-- Shuah

Signature · 5 lines
Shuah Khan
Sr. Linux Kernel Developer
Open Source Innovation Group
Samsung Research America (Silicon Valley)
[email protected] | (970) 217-8978