Re: [PATCH] platform/x86: ISST: do not hold lock punit_misc_dev_lock when register misc
From: Srinivas Pandruvada
Date: Tue Jan 11 2022 - 16:20:45 EST
Hi Song,
Please check the attached and see if you can reproduce this.
Thanks,
Srinivas
On Tue, 2022-01-11 at 18:31 +0800, Liwei Song wrote:
> exist the below call sequences may cause deadlock:
>
> isst_if_probe()
> --> isst_if_cdev_register()
> --> mutex_lock(&punit_misc_dev_lock)
> --> misc_register()
> --> mutex_lock(&misc_mtx)
>
> misc_open()
> --> mutex_lock(&misc_mtx)
> --> isst_if_open()
> --> mutex_lock(&punit_misc_dev_lock)
>
> to fix this do not hold punit_misc_dev_lock when call misc_register
> since it has misc_mtx lock for sync.
>
> [ 256.104522] ======================================================
> [ 256.113783] WARNING: possible circular locking dependency detected
> [ 256.120093] 5.16.0-rc6-yocto-standard+ #99 Not tainted
> [ 256.125362] ------------------------------------------------------
> [ 256.131673] intel-speed-sel/844 is trying to acquire lock:
> [ 256.137290] ffffffffc036f0d0 (punit_misc_dev_lock){+.+.}-{3:3},
> at: isst_if_open+0x18/0x90 [isst_if_common]
> [ 256.147171]
> [ 256.147171] but task is already holding lock:
> [ 256.153135] ffffffff8ee7cb50 (misc_mtx){+.+.}-{3:3}, at:
> misc_open+0x2a/0x170
> [ 256.160407]
> [ 256.160407] which lock already depends on the new lock.
> [ 256.160407]
> [ 256.168712]
> [ 256.168712] the existing dependency chain (in reverse order) is:
> [ 256.176327]
> [ 256.176327] -> #1 (misc_mtx){+.+.}-{3:3}:
> [ 256.181946] lock_acquire+0x1e6/0x330
> [ 256.186265] __mutex_lock+0x9b/0x9b0
> [ 256.190497] mutex_lock_nested+0x1b/0x20
> [ 256.195075] misc_register+0x32/0x1a0
> [ 256.199390] isst_if_cdev_register+0x65/0x180
> [isst_if_common]
> [ 256.205878] isst_if_probe+0x144/0x16e [isst_if_mmio]
> [ 256.209991] hrtimer: interrupt took 10370 ns
> [ 256.211582] local_pci_probe+0x47/0xa0
> [ 256.220384] work_for_cpu_fn+0x17/0x30
> [ 256.224790] process_one_work+0x26a/0x650
> [ 256.229456] worker_thread+0x1dd/0x3b0
> [ 256.233861] kthread+0x191/0x1c0
> [ 256.237745] ret_from_fork+0x1f/0x30
> [ 256.241976]
> [ 256.241976] -> #0 (punit_misc_dev_lock){+.+.}-{3:3}:
> [ 256.248552] validate_chain+0xbc6/0x1750
> [ 256.253131] __lock_acquire+0x88c/0xc10
> [ 256.257618] lock_acquire+0x1e6/0x330
> [ 256.261933] __mutex_lock+0x9b/0x9b0
> [ 256.266165] mutex_lock_nested+0x1b/0x20
> [ 256.270739] isst_if_open+0x18/0x90 [isst_if_common]
> [ 256.276356] misc_open+0x100/0x170
> [ 256.280409] chrdev_open+0xa5/0x1e0
> [ 256.284550] do_dentry_open+0x23d/0x3c0
> [ 256.289039] vfs_open+0x2f/0x40
> [ 256.292836] path_openat+0x87a/0x940
> [ 256.297064] do_filp_open+0xc5/0x140
> [ 256.301295] do_sys_openat2+0x23d/0x320
> [ 256.305782] do_sys_open+0x59/0x80
> [ 256.309836] __x64_sys_openat+0x20/0x30
> [ 256.314324] do_syscall_64+0x3f/0x90
> [ 256.318552] entry_SYSCALL_64_after_hwframe+0x44/0xae
> [ 256.324259]
> [ 256.324259] other info that might help us debug this:
> [ 256.324259]
> [ 256.332394] Possible unsafe locking scenario:
> [ 256.332394]
> [ 256.338444] CPU0 CPU1
> [ 256.343105] ---- ----
> [ 256.347768] lock(misc_mtx);
> [ 256.350870] lock(punit_misc_dev_loc
> k);
> [ 256.357441] lock(misc_mtx);
> [ 256.363058] lock(punit_misc_dev_lock);
> [ 256.367110]
> [ 256.367110] *** DEADLOCK ***
> [ 256.367110]
> [ 256.373162] 1 lock held by intel-speed-sel/844:
> [ 256.377824] #0: ffffffff8ee7cb50 (misc_mtx){+.+.}-{3:3}, at:
> misc_open+0x2a/0x170
> [ 256.385531]
> [ 256.385531] stack backtrace:
> [ 256.390021] CPU: 12 PID: 844 Comm: intel-speed-sel Not tainted
> 5.16.0-rc6-yocto-standard+ #99
> [ 256.398678] Hardware name: ACCTON MOROCITY/MOROCITY, BIOS
> IDVLCRB1.86B.0021.D09.2111010103 11/01/2021
> [ 256.408028] Call Trace:
> [ 256.410605] <TASK>
> [ 256.412837] dump_stack_lvl+0x5b/0x82
> [ 256.416635] dump_stack+0x10/0x12
> [ 256.420085] print_circular_bug.isra.43+0x261/0x2c0
> [ 256.425095] check_noncircular+0x126/0x140
> [ 256.429326] ? __this_cpu_preempt_check+0x13/0x20
> [ 256.434167] validate_chain+0xbc6/0x1750
> [ 256.438223] ? validate_chain+0xbc6/0x1750
> [ 256.442451] ? validate_chain+0x236/0x1750
> [ 256.446687] __lock_acquire+0x88c/0xc10
> [ 256.450658] lock_acquire+0x1e6/0x330
> [ 256.454452] ? isst_if_open+0x18/0x90 [isst_if_common]
> [ 256.459726] ? __mutex_lock+0x79/0x9b0
> [ 256.463610] ? __mutex_lock+0x79/0x9b0
> [ 256.467493] ? isst_if_open+0x18/0x90 [isst_if_common]
> [ 256.472764] ? isst_if_open+0x18/0x90 [isst_if_common]
> [ 256.478038] __mutex_lock+0x9b/0x9b0
> [ 256.481748] ? isst_if_open+0x18/0x90 [isst_if_common]
> [ 256.487021] ? __mutex_lock+0x102/0x9b0
> [ 256.490993] ? __this_cpu_preempt_check+0x13/0x20
> [ 256.495837] mutex_lock_nested+0x1b/0x20
> [ 256.499893] ? mutex_lock_nested+0x1b/0x20
> [ 256.504121] isst_if_open+0x18/0x90 [isst_if_common]
> [ 256.509222] misc_open+0x100/0x170
> [ 256.512759] chrdev_open+0xa5/0x1e0
> [ 256.516386] ? cdev_put.part.1+0x20/0x20
> [ 256.520441] do_dentry_open+0x23d/0x3c0
> [ 256.524414] vfs_open+0x2f/0x40
> [ 256.527689] path_openat+0x87a/0x940
> [ 256.531399] do_filp_open+0xc5/0x140
> [ 256.535112] ? trace_preempt_on+0x28/0xd0
> [ 256.539255] ? alloc_fd+0x152/0x230
> [ 256.542880] ? preempt_count_sub+0x9b/0x100
> [ 256.547200] ? _raw_spin_unlock+0x2c/0x50
> [ 256.551348] do_sys_openat2+0x23d/0x320
> [ 256.555320] ? do_sys_openat2+0x23d/0x320
> [ 256.559467] do_sys_open+0x59/0x80
> [ 256.563003] __x64_sys_openat+0x20/0x30
> [ 256.566972] do_syscall_64+0x3f/0x90
> [ 256.570680] entry_SYSCALL_64_after_hwframe+0x44/0xae
> [ 256.575866] RIP: 0033:0x7f9be4b97c27
> [ 256.579576] Code: 25 00 00 41 00 3d 00 00 41 00 74 37 64 8b 04 25
> 18 00 00 00 85 c0 75 5b 44 89 e2 48 89 ee bf 9c ff ff ff b8 01 01 00
> 00 0f 05 <48> 3d 00 f0 ff ff 0f 87 85 00 00 00 48 83 c4 68 5d 41 5c
> c3 0f 1f
> [ 256.598474] RSP: 002b:00007ffd8fc01b70 EFLAGS: 00000246 ORIG_RAX:
> 0000000000000101
> [ 256.606177] RAX: ffffffffffffffda RBX: 00005572f20332b0 RCX:
> 00007f9be4b97c27
> [ 256.613443] RDX: 0000000000000000 RSI: 00005572f202936a RDI:
> 00000000ffffff9c
> [ 256.620709] RBP: 00005572f202936a R08: 0000000000000008 R09:
> 0000000000000001
> [ 256.627974] R10: 0000000000000000 R11: 0000000000000246 R12:
> 0000000000000000
> [ 256.635241] R13: 00005572f20332b0 R14: 0000000000000001 R15:
> 0000000000000000
> [ 256.642513] </TASK>
>
> Signed-off-by: Liwei Song <liwei.song@xxxxxxxxxxxxx>
> ---
> drivers/platform/x86/intel/speed_select_if/isst_if_common.c | 6
> +++++-
> 1 file changed, 5 insertions(+), 1 deletion(-)
>
> diff --git
> a/drivers/platform/x86/intel/speed_select_if/isst_if_common.c
> b/drivers/platform/x86/intel/speed_select_if/isst_if_common.c
> index c9a85eb2e860..bcbc0d508ec4 100644
> --- a/drivers/platform/x86/intel/speed_select_if/isst_if_common.c
> +++ b/drivers/platform/x86/intel/speed_select_if/isst_if_common.c
> @@ -693,10 +693,12 @@ int isst_if_cdev_register(int device_type,
> struct isst_if_cmd_cb *cb)
> if (!misc_usage_count) {
> int ret;
>
> + mutex_unlock(&punit_misc_dev_lock);
> misc_device_ret = misc_register(&isst_if_char_driver);
> if (misc_device_ret)
> - goto unlock_exit;
> + return misc_device_ret;
>
> + mutex_lock(&punit_misc_dev_lock);
> ret = isst_if_cpu_info_init();
> if (ret) {
> misc_deregister(&isst_if_char_driver);
> @@ -731,7 +733,9 @@ void isst_if_cdev_unregister(int device_type)
> if (device_type == ISST_IF_DEV_MBOX)
> isst_delete_hash();
> if (!misc_usage_count && !misc_device_ret) {
> + mutex_unlock(&punit_misc_dev_lock);
> misc_deregister(&isst_if_char_driver);
> + mutex_lock(&punit_misc_dev_lock);
> isst_if_cpu_info_exit();
> }
> mutex_unlock(&punit_misc_dev_lock);
From e92c9c429dbd259778b94c45c4723ddb0d1670fb Mon Sep 17 00:00:00 2001
From: Srinivas Pandruvada <srinivas.pandruvada@xxxxxxxxxxxxxxx>
Date: Tue, 11 Jan 2022 12:04:28 -0800
Subject: [PATCH] platform/x86: ISST: Fix possible circular locking dependency
detected
As reported:
[ 256.104522] ======================================================
[ 256.113783] WARNING: possible circular locking dependency detected
[ 256.120093] 5.16.0-rc6-yocto-standard+ #99 Not tainted
[ 256.125362] ------------------------------------------------------
[ 256.131673] intel-speed-sel/844 is trying to acquire lock:
[ 256.137290] ffffffffc036f0d0 (punit_misc_dev_lock){+.+.}-{3:3}, at: isst_if_open+0x18/0x90 [isst_if_common]
[ 256.147171]
[ 256.147171] but task is already holding lock:
[ 256.153135] ffffffff8ee7cb50 (misc_mtx){+.+.}-{3:3}, at: misc_open+0x2a/0x170
[ 256.160407]
[ 256.160407] which lock already depends on the new lock.
[ 256.160407]
[ 256.168712]
[ 256.168712] the existing dependency chain (in reverse order) is:
[ 256.176327]
[ 256.176327] -> #1 (misc_mtx){+.+.}-{3:3}:
[ 256.181946] lock_acquire+0x1e6/0x330
[ 256.186265] __mutex_lock+0x9b/0x9b0
[ 256.190497] mutex_lock_nested+0x1b/0x20
[ 256.195075] misc_register+0x32/0x1a0
[ 256.199390] isst_if_cdev_register+0x65/0x180 [isst_if_common]
[ 256.205878] isst_if_probe+0x144/0x16e [isst_if_mmio]
...
[ 256.241976]
[ 256.241976] -> #0 (punit_misc_dev_lock){+.+.}-{3:3}:
[ 256.248552] validate_chain+0xbc6/0x1750
[ 256.253131] __lock_acquire+0x88c/0xc10
[ 256.257618] lock_acquire+0x1e6/0x330
[ 256.261933] __mutex_lock+0x9b/0x9b0
[ 256.266165] mutex_lock_nested+0x1b/0x20
[ 256.270739] isst_if_open+0x18/0x90 [isst_if_common]
[ 256.276356] misc_open+0x100/0x170
[ 256.280409] chrdev_open+0xa5/0x1e0
...
The call sequence suggested that misc_device /dev file can be opened
before misc device is yet to be registered, which is done only once.
Here punit_misc_dev_lock was used as common lock, to protect the
registration by ISST HW drivers, one time setup, prevent duplicate
registry of misc device and prevent load/unload when device is open.
We can split into locks:
- One which just prevent duplicate call to misc_register() and one
time setup. Also never call again if the misc_register() failed or
required one time setup is failed. This lock is not shared with
any misc device callbacks.
- The other lock protects registry, load and unload of HW drivers.
Sequence in isst_if_cdev_register()
- Register callbacks under punit_misc_dev_open_lock
- Call isst_misc_reg() which registers misc_device on the first
registry which is under punit_misc_dev_reg_lock, which is not
shared with callbacks.
Sequence in isst_if_cdev_unregister
Just opposite of isst_if_cdev_register
Reported-by: Liwei Song <liwei.song@xxxxxxxxxxxxx>
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@xxxxxxxxxxxxxxx>
---
.../intel/speed_select_if/isst_if_common.c | 97 ++++++++++++-------
1 file changed, 63 insertions(+), 34 deletions(-)
diff --git a/drivers/platform/x86/intel/speed_select_if/isst_if_common.c b/drivers/platform/x86/intel/speed_select_if/isst_if_common.c
index c9a85eb2e860..e8424e70d81d 100644
--- a/drivers/platform/x86/intel/speed_select_if/isst_if_common.c
+++ b/drivers/platform/x86/intel/speed_select_if/isst_if_common.c
@@ -596,7 +596,10 @@ static long isst_if_def_ioctl(struct file *file, unsigned int cmd,
return ret;
}
-static DEFINE_MUTEX(punit_misc_dev_lock);
+/* Lock to prevent module registration when already opened by user space */
+static DEFINE_MUTEX(punit_misc_dev_open_lock);
+/* Lock to allow one share misc device for all ISST interace */
+static DEFINE_MUTEX(punit_misc_dev_reg_lock);
static int misc_usage_count;
static int misc_device_ret;
static int misc_device_open;
@@ -606,7 +609,7 @@ static int isst_if_open(struct inode *inode, struct file *file)
int i, ret = 0;
/* Fail open, if a module is going away */
- mutex_lock(&punit_misc_dev_lock);
+ mutex_lock(&punit_misc_dev_open_lock);
for (i = 0; i < ISST_IF_DEV_MAX; ++i) {
struct isst_if_cmd_cb *cb = &punit_callbacks[i];
@@ -628,7 +631,7 @@ static int isst_if_open(struct inode *inode, struct file *file)
} else {
misc_device_open++;
}
- mutex_unlock(&punit_misc_dev_lock);
+ mutex_unlock(&punit_misc_dev_open_lock);
return ret;
}
@@ -637,7 +640,7 @@ static int isst_if_relase(struct inode *inode, struct file *f)
{
int i;
- mutex_lock(&punit_misc_dev_lock);
+ mutex_lock(&punit_misc_dev_open_lock);
misc_device_open--;
for (i = 0; i < ISST_IF_DEV_MAX; ++i) {
struct isst_if_cmd_cb *cb = &punit_callbacks[i];
@@ -645,7 +648,7 @@ static int isst_if_relase(struct inode *inode, struct file *f)
if (cb->registered)
module_put(cb->owner);
}
- mutex_unlock(&punit_misc_dev_lock);
+ mutex_unlock(&punit_misc_dev_open_lock);
return 0;
}
@@ -662,6 +665,43 @@ static struct miscdevice isst_if_char_driver = {
.fops = &isst_if_char_driver_ops,
};
+static int isst_misc_reg(void)
+{
+ mutex_lock(&punit_misc_dev_reg_lock);
+ if (misc_device_ret)
+ goto unlock_exit;
+
+ if (!misc_usage_count) {
+ misc_device_ret = isst_if_cpu_info_init();
+ if (misc_device_ret)
+ goto unlock_exit;
+
+ misc_device_ret = misc_register(&isst_if_char_driver);
+ if (misc_device_ret) {
+ isst_if_cpu_info_exit();
+ goto unlock_exit;
+ }
+ }
+ misc_usage_count++;
+
+unlock_exit:
+ mutex_unlock(&punit_misc_dev_reg_lock);
+
+ return misc_device_ret;
+}
+
+static void isst_misc_unreg(void)
+{
+ mutex_lock(&punit_misc_dev_reg_lock);
+ if (misc_usage_count)
+ misc_usage_count--;
+ if (!misc_usage_count && !misc_device_ret) {
+ misc_deregister(&isst_if_char_driver);
+ isst_if_cpu_info_exit();
+ }
+ mutex_unlock(&punit_misc_dev_reg_lock);
+}
+
/**
* isst_if_cdev_register() - Register callback for IOCTL
* @device_type: The device type this callback handling.
@@ -679,38 +719,31 @@ static struct miscdevice isst_if_char_driver = {
*/
int isst_if_cdev_register(int device_type, struct isst_if_cmd_cb *cb)
{
- if (misc_device_ret)
- return misc_device_ret;
+ int ret;
if (device_type >= ISST_IF_DEV_MAX)
return -EINVAL;
- mutex_lock(&punit_misc_dev_lock);
+ mutex_lock(&punit_misc_dev_open_lock);
+ /* Device is already open, we don't want to add new callbacks */
if (misc_device_open) {
- mutex_unlock(&punit_misc_dev_lock);
+ mutex_unlock(&punit_misc_dev_open_lock);
return -EAGAIN;
}
- if (!misc_usage_count) {
- int ret;
-
- misc_device_ret = misc_register(&isst_if_char_driver);
- if (misc_device_ret)
- goto unlock_exit;
-
- ret = isst_if_cpu_info_init();
- if (ret) {
- misc_deregister(&isst_if_char_driver);
- misc_device_ret = ret;
- goto unlock_exit;
- }
- }
memcpy(&punit_callbacks[device_type], cb, sizeof(*cb));
punit_callbacks[device_type].registered = 1;
- misc_usage_count++;
-unlock_exit:
- mutex_unlock(&punit_misc_dev_lock);
+ mutex_unlock(&punit_misc_dev_open_lock);
- return misc_device_ret;
+ ret = isst_misc_reg();
+ if (ret) {
+ /*
+ * No need of mutex as the misc device register failed
+ * as no one can open device yet. Hence no contention.
+ */
+ punit_callbacks[device_type].registered = 0;
+ return ret;
+ }
+ return 0;
}
EXPORT_SYMBOL_GPL(isst_if_cdev_register);
@@ -725,16 +758,12 @@ EXPORT_SYMBOL_GPL(isst_if_cdev_register);
*/
void isst_if_cdev_unregister(int device_type)
{
- mutex_lock(&punit_misc_dev_lock);
- misc_usage_count--;
+ isst_misc_unreg();
+ mutex_lock(&punit_misc_dev_open_lock);
punit_callbacks[device_type].registered = 0;
if (device_type == ISST_IF_DEV_MBOX)
isst_delete_hash();
- if (!misc_usage_count && !misc_device_ret) {
- misc_deregister(&isst_if_char_driver);
- isst_if_cpu_info_exit();
- }
- mutex_unlock(&punit_misc_dev_lock);
+ mutex_unlock(&punit_misc_dev_open_lock);
}
EXPORT_SYMBOL_GPL(isst_if_cdev_unregister);
--
2.25.1