Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > linux.kernel > #1582156
| From | Ben Hutchings <ben@decadent.org.uk> |
|---|---|
| Newsgroups | linux.kernel |
| Subject | [PATCH 3.2 026/126] net/mlx4_core: Fix deadlock when switching between polling and event fw commands |
| Date | 2017-02-16 01:40 +0100 |
| Message-ID | <tbih3-768-5@gated-at.bofh.it> (permalink) |
| References | <tbgIi-5XZ-3@gated-at.bofh.it> |
| Organization | linux.* mail to news gateway |
3.2.85-rc1 review patch. If anyone has any objections, please let me know.
------------------
From: Jack Morgenstein <jackm@dev.mellanox.co.il>
commit a7e1f04905e5b2b90251974dddde781301b6be37 upstream.
When switching from polling-based fw commands to event-based fw
commands, there is a race condition which could cause a fw command
in another task to hang: that task will keep waiting for the polling
sempahore, but may never be able to acquire it. This is due to
mlx4_cmd_use_events, which "down"s the sempahore back to 0.
During driver initialization, this is not a problem, since no other
tasks which invoke FW commands are active.
However, there is a problem if the driver switches to polling mode
and then back to event mode during normal operation.
The "test_interrupts" feature does exactly that.
Running "ethtool -t <eth device> offline" causes the PF driver to
temporarily switch to polling mode, and then back to event mode.
(Note that for VF drivers, such switching is not performed).
Fix this by adding a read-write semaphore for protection when
switching between modes.
Fixes: 225c7b1feef1 ("IB/mlx4: Add a driver Mellanox ConnectX InfiniBand adapters")
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
[bwh: Backported to 3.2: adjust context, indentation]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
---
--- a/drivers/net/ethernet/mellanox/mlx4/cmd.c
+++ b/drivers/net/ethernet/mellanox/mlx4/cmd.c
@@ -313,12 +313,18 @@ int __mlx4_cmd(struct mlx4_dev *dev, u64
int out_is_imm, u32 in_modifier, u8 op_modifier,
u16 op, unsigned long timeout)
{
+ int ret;
+
+ down_read(&mlx4_priv(dev)->cmd.switch_sem);
if (mlx4_priv(dev)->cmd.use_events)
- return mlx4_cmd_wait(dev, in_param, out_param, out_is_imm,
- in_modifier, op_modifier, op, timeout);
+ ret = mlx4_cmd_wait(dev, in_param, out_param, out_is_imm,
+ in_modifier, op_modifier, op, timeout);
else
- return mlx4_cmd_poll(dev, in_param, out_param, out_is_imm,
- in_modifier, op_modifier, op, timeout);
+ ret = mlx4_cmd_poll(dev, in_param, out_param, out_is_imm,
+ in_modifier, op_modifier, op, timeout);
+
+ up_read(&mlx4_priv(dev)->cmd.switch_sem);
+ return ret;
}
EXPORT_SYMBOL_GPL(__mlx4_cmd);
@@ -326,6 +332,7 @@ int mlx4_cmd_init(struct mlx4_dev *dev)
{
struct mlx4_priv *priv = mlx4_priv(dev);
+ init_rwsem(&priv->cmd.switch_sem);
mutex_init(&priv->cmd.hcr_mutex);
sema_init(&priv->cmd.poll_sem, 1);
priv->cmd.use_events = 0;
@@ -372,6 +379,7 @@ int mlx4_cmd_use_events(struct mlx4_dev
if (!priv->cmd.context)
return -ENOMEM;
+ down_write(&priv->cmd.switch_sem);
for (i = 0; i < priv->cmd.max_cmds; ++i) {
priv->cmd.context[i].token = i;
priv->cmd.context[i].next = i + 1;
@@ -390,6 +398,7 @@ int mlx4_cmd_use_events(struct mlx4_dev
--priv->cmd.token_mask;
priv->cmd.use_events = 1;
+ up_write(&priv->cmd.switch_sem);
down(&priv->cmd.poll_sem);
@@ -404,6 +413,7 @@ void mlx4_cmd_use_polling(struct mlx4_de
struct mlx4_priv *priv = mlx4_priv(dev);
int i;
+ down_write(&priv->cmd.switch_sem);
priv->cmd.use_events = 0;
for (i = 0; i < priv->cmd.max_cmds; ++i)
@@ -412,6 +422,7 @@ void mlx4_cmd_use_polling(struct mlx4_de
kfree(priv->cmd.context);
up(&priv->cmd.poll_sem);
+ up_write(&priv->cmd.switch_sem);
}
struct mlx4_cmd_mailbox *mlx4_alloc_cmd_mailbox(struct mlx4_dev *dev)
--- a/drivers/net/ethernet/mellanox/mlx4/mlx4.h
+++ b/drivers/net/ethernet/mellanox/mlx4/mlx4.h
@@ -42,6 +42,7 @@
#include <linux/timer.h>
#include <linux/semaphore.h>
#include <linux/workqueue.h>
+#include <linux/rwsem.h>
#include <linux/mlx4/device.h>
#include <linux/mlx4/driver.h>
@@ -190,6 +191,7 @@ struct mlx4_cmd {
struct mutex hcr_mutex;
struct semaphore poll_sem;
struct semaphore event_sem;
+ struct rw_semaphore switch_sem;
int max_cmds;
spinlock_t context_lock;
int free_head;
Back to linux.kernel | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread
[PATCH 3.2 000/126] 3.2.85-rc1 review Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 00:40 +0100
[PATCH 3.2 111/126] lockdep: Silence warning if CONFIG_LOCKDEP isn't set Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 00:40 +0100
[PATCH 3.2 095/126] locking/rtmutex: Prevent dequeue vs. unlock race Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 00:40 +0100
[PATCH 3.2 057/126] scsi: arcmsr: Send SYNCHRONIZE_CACHE command to firmware Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 00:40 +0100
[PATCH 3.2 110/126] perf: Fix perf_event_for_each() to use sibling Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 00:40 +0100
[PATCH 3.2 098/126] net: ping: check minimum size on ICMP header length Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 00:40 +0100
[PATCH 3.2 114/126] perf/core: Fix concurrent sys_perf_event_open() vs. 'move_group' race Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 00:40 +0100
Re: [PATCH 3.2 114/126] perf/core: Fix concurrent sys_perf_event_open() vs. 'move_group' race Ben Hutchings <ben@decadent.org.uk> - 2017-02-21 01:50 +0100
[PATCH 3.2 085/126] Fix USB CB/CBI storage devices with CONFIG_VMAP_STACK=y Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 00:50 +0100
[PATCH 3.2 124/126] sg_write()/bsg_write() is not fit to be called under KERNEL_DS Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 00:50 +0100
[PATCH 3.2 053/126] scsi: megaraid_sas: Fix data integrity failure for JBOD (passthrough) devices Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 00:50 +0100
[PATCH 3.2 112/126] perf: Fix event->ctx locking Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 00:50 +0100
[PATCH 3.2 089/126] IB/uverbs: Fix leak of XRC target QPs Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 00:50 +0100
[PATCH 3.2 037/126] fuse: fix killing s[ug]id in setattr Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:40 +0100
[PATCH 3.2 026/126] net/mlx4_core: Fix deadlock when switching between polling and event fw commands Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:40 +0100
[PATCH 3.2 047/126] ubifs: Abort readdir upon error Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:40 +0100
[PATCH 3.2 003/126] zfcp: fix ELS/GS request&response length for hardware data router Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:40 +0100
[PATCH 3.2 065/126] ubifs: Fix regression in ubifs_readdir() Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:40 +0100
[PATCH 3.2 046/126] ubifs: Fix xattr_names length in exit paths Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:40 +0100
[PATCH 3.2 039/126] crypto: gcm - Fix IV buffer size in crypto_gcm_setkey Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:40 +0100
[PATCH 3.2 015/126] [media] mb86a20s: fix the locking logic Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:40 +0100
[PATCH 3.2 049/126] batman-adv: fix splat on disabling an interface Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:40 +0100
Re: [PATCH 3.2 049/126] batman-adv: fix splat on disabling an interface Linus Lüssing <linus.luessing@c0d3.blue> - 2017-02-16 08:00 +0100
Re: [PATCH 3.2 049/126] batman-adv: fix splat on disabling an interface Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 17:10 +0100
[PATCH 3.2 005/126] zfcp: retain trace level for SCSI and HBA FSF response records Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:40 +0100
[PATCH 3.2 032/126] powerpc/vdso64: Use double word compare on pointers Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:40 +0100
[PATCH 3.2 027/126] ALSA: usb-audio: Extend DragonFly dB scale quirk to cover other variants Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:40 +0100
[PATCH 3.2 061/126] vt: clear selection before resizing Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:40 +0100
[PATCH 3.2 040/126] powerpc/64: Fix incorrect return value from __copy_tofrom_user Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:40 +0100
[PATCH 3.2 044/126] isofs: Do not return EACCES for unknown filesystems Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:40 +0100
[PATCH 3.2 030/126] USB: serial: cp210x: Add ID for a Juniper console Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:40 +0100
[PATCH 3.2 034/126] s390/con3270: fix use of uninitialised data Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:40 +0100
[PATCH 3.2 016/126] [media] cx231xx: don't return error on success Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:40 +0100
[PATCH 3.2 033/126] ext4: release bh in make_indexed_dir Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:40 +0100
[PATCH 3.2 035/126] s390/con3270: fix insufficient space padding Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:40 +0100
[PATCH 3.2 024/126] ALSA: ali5451: Fix out-of-bound position reporting Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:40 +0100
[PATCH 3.2 010/126] zfcp: fix payload trace length for SAN request&response Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:50 +0100
[PATCH 3.2 028/126] regulator: tps65910: Work around silicon erratum SWCZ010 Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:50 +0100
[PATCH 3.2 008/126] zfcp: restore tracing of handle for port and LUN with HBA records Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:50 +0100
[PATCH 3.2 022/126] scsi: ibmvfc: Fix I/O hang when port is not mapped Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:50 +0100
[PATCH 3.2 004/126] zfcp: close window with unblocked rport during rport gone Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:50 +0100
[PATCH 3.2 017/126] [media] cx231xx: fix GPIOs for Pixelview SBTVD hybrid Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:50 +0100
[PATCH 3.2 029/126] mmc: block: don't use CMD23 with very old MMC cards Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:50 +0100
[PATCH 3.2 014/126] rtlwifi: Fix missing country code for Great Britain Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:50 +0100
[PATCH 3.2 011/126] zfcp: trace full payload of all SAN records (req,resp,iels) Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:50 +0100
[PATCH 3.2 002/126] zfcp: fix fc_host port_type with NPIV Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:50 +0100
[PATCH 3.2 013/126] rtlwifi: Update regulatory database Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:50 +0100
[PATCH 3.2 006/126] zfcp: restore: Dont use 0 to indicate invalid LUN in rec trace Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:50 +0100
[PATCH 3.2 009/126] zfcp: fix D_ID field with actual value on tracing SAN responses Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:50 +0100
[PATCH 3.2 025/126] usb: misc: legousbtower: Fix NULL pointer deference Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:50 +0100
[PATCH 3.2 038/126] fuse: listxattr: verify xattr list Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:50 +0100
[PATCH 3.2 043/126] scsi: zfcp: spin_lock_irqsave() is not nestable Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:50 +0100
[PATCH 3.2 036/126] fuse: invalidate dir dentry after chmod Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:50 +0100
[PATCH 3.2 019/126] reiserfs: Unlock superblock before calling reiserfs_quota_on_mount() Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 01:50 +0100
Re: [PATCH 3.2 000/126] 3.2.85-rc1 review Guenter Roeck <linux@roeck-us.net> - 2017-02-16 07:10 +0100
Re: [PATCH 3.2 000/126] 3.2.85-rc1 review Ben Hutchings <ben@decadent.org.uk> - 2017-02-16 17:00 +0100
csiph-web