* [PATCH v2 1/2] tty: Register device *after* creating the cdev for a tty
@ 2025-05-28 13:28 Max Staudt
2025-05-28 13:28 ` [PATCH v2 2/2] tty: Fix race against tty_open() in tty_register_device_attr() Max Staudt
0 siblings, 1 reply; 6+ messages in thread
From: Max Staudt @ 2025-05-28 13:28 UTC (permalink / raw)
To: Greg Kroah-Hartman, Jiri Slaby
Cc: Johan Hovold, linux-serial, linux-kernel, Max Staudt, stable
This change makes the tty device file available only after the tty's
backing character device is ready.
Since 6a7e6f78c235975cc14d4e141fa088afffe7062c, the class device is
registered before the cdev is created, and userspace may pick it up,
yet open() will fail because the backing cdev doesn't exist yet.
Userspace is racing the bottom half of tty_register_device_attr() here,
specifically the call to tty_cdev_add().
dev_set_uevent_suppress() was used to work around this, but this fails
on embedded systems that rely on bare devtmpfs rather than udev.
On such systems, the device file is created as part of device_add(),
and userspace can pick it up via inotify, irrespective of uevent
suppression.
So let's undo the existing patch, and create the cdev first, and only
afterwards register the class device in the kernel's device tree.
However, this restores the original race of the cdev existing before the
class device is registered, and an attempt to tty_[k]open() the chardev
between these two steps will lead to tty->dev being assigned NULL by
alloc_tty_struct().
This will be addressed in a second patch.
Fixes: 6a7e6f78c235 ("tty: close race between device register and open")
Signed-off-by: Max Staudt <max@enpas.org>
Cc: <stable@vger.kernel.org>
---
drivers/tty/tty_io.c | 54 +++++++++++++++++++++++++-------------------
1 file changed, 31 insertions(+), 23 deletions(-)
diff --git a/drivers/tty/tty_io.c b/drivers/tty/tty_io.c
index ca9b7d7bad2b..e922b84524d2 100644
--- a/drivers/tty/tty_io.c
+++ b/drivers/tty/tty_io.c
@@ -3245,6 +3245,7 @@ struct device *tty_register_device_attr(struct tty_driver *driver,
struct ktermios *tp;
struct device *dev;
int retval;
+ bool cdev_added = false;
if (index >= driver->num) {
pr_err("%s: Attempt to register invalid tty line number (%d)\n",
@@ -3257,24 +3258,6 @@ struct device *tty_register_device_attr(struct tty_driver *driver,
else
tty_line_name(driver, index, name);
- dev = kzalloc(sizeof(*dev), GFP_KERNEL);
- if (!dev)
- return ERR_PTR(-ENOMEM);
-
- dev->devt = devt;
- dev->class = &tty_class;
- dev->parent = device;
- dev->release = tty_device_create_release;
- dev_set_name(dev, "%s", name);
- dev->groups = attr_grp;
- dev_set_drvdata(dev, drvdata);
-
- dev_set_uevent_suppress(dev, 1);
-
- retval = device_register(dev);
- if (retval)
- goto err_put;
-
if (!(driver->flags & TTY_DRIVER_DYNAMIC_ALLOC)) {
/*
* Free any saved termios data so that the termios state is
@@ -3288,19 +3271,44 @@ struct device *tty_register_device_attr(struct tty_driver *driver,
retval = tty_cdev_add(driver, devt, index, 1);
if (retval)
- goto err_del;
+ return ERR_PTR(retval);
+
+ cdev_added = true;
+ }
+
+ dev = kzalloc(sizeof(*dev), GFP_KERNEL);
+ if (!dev) {
+ retval = -ENOMEM;
+ goto err_del_cdev;
}
- dev_set_uevent_suppress(dev, 0);
- kobject_uevent(&dev->kobj, KOBJ_ADD);
+ dev->devt = devt;
+ dev->class = &tty_class;
+ dev->parent = device;
+ dev->release = tty_device_create_release;
+ dev_set_name(dev, "%s", name);
+ dev->groups = attr_grp;
+ dev_set_drvdata(dev, drvdata);
+
+ retval = device_register(dev);
+ if (retval)
+ goto err_put;
return dev;
-err_del:
- device_del(dev);
err_put:
+ /*
+ * device_register() calls device_add(), after which
+ * we must use put_device() instead of kfree().
+ */
put_device(dev);
+err_del_cdev:
+ if (cdev_added) {
+ cdev_del(driver->cdevs[index]);
+ driver->cdevs[index] = NULL;
+ }
+
return ERR_PTR(retval);
}
EXPORT_SYMBOL_GPL(tty_register_device_attr);
--
2.39.5
^ permalink raw reply related [flat|nested] 6+ messages in thread
* [PATCH v2 2/2] tty: Fix race against tty_open() in tty_register_device_attr()
2025-05-28 13:28 [PATCH v2 1/2] tty: Register device *after* creating the cdev for a tty Max Staudt
@ 2025-05-28 13:28 ` Max Staudt
2025-06-02 10:31 ` Ilpo Järvinen
2025-06-03 8:49 ` kernel test robot
0 siblings, 2 replies; 6+ messages in thread
From: Max Staudt @ 2025-05-28 13:28 UTC (permalink / raw)
To: Greg Kroah-Hartman, Jiri Slaby
Cc: Johan Hovold, linux-serial, linux-kernel, Max Staudt, stable
Since the chardev is now created before the class device is registered,
an attempt to tty_[k]open() the chardev between these two steps will
lead to tty->dev being assigned NULL by alloc_tty_struct().
alloc_tty_struct() is called via tty_init_dev() when the tty is firstly
opened, and is entered with tty_mutex held, so let's lock the critical
section in tty_register_device_attr() with the same global mutex.
This guarantees that tty->dev can be assigned a sane value.
Fixes: 6a7e6f78c235 ("tty: close race between device register and open")
Signed-off-by: Max Staudt <max@enpas.org>
Cc: <stable@vger.kernel.org>
---
drivers/tty/tty_io.c | 9 ++++++++-
1 file changed, 8 insertions(+), 1 deletion(-)
diff --git a/drivers/tty/tty_io.c b/drivers/tty/tty_io.c
index e922b84524d2..94768509e2d2 100644
--- a/drivers/tty/tty_io.c
+++ b/drivers/tty/tty_io.c
@@ -3258,6 +3258,8 @@ struct device *tty_register_device_attr(struct tty_driver *driver,
else
tty_line_name(driver, index, name);
+ mutex_lock(&tty_mutex);
+
if (!(driver->flags & TTY_DRIVER_DYNAMIC_ALLOC)) {
/*
* Free any saved termios data so that the termios state is
@@ -3271,7 +3273,7 @@ struct device *tty_register_device_attr(struct tty_driver *driver,
retval = tty_cdev_add(driver, devt, index, 1);
if (retval)
- return ERR_PTR(retval);
+ goto err_unlock;
cdev_added = true;
}
@@ -3294,6 +3296,8 @@ struct device *tty_register_device_attr(struct tty_driver *driver,
if (retval)
goto err_put;
+ mutex_unlock(&tty_mutex);
+
return dev;
err_put:
@@ -3309,6 +3313,9 @@ struct device *tty_register_device_attr(struct tty_driver *driver,
driver->cdevs[index] = NULL;
}
+err_unlock:
+ mutex_unlock(&tty_mutex);
+
return ERR_PTR(retval);
}
EXPORT_SYMBOL_GPL(tty_register_device_attr);
--
2.39.5
^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [PATCH v2 2/2] tty: Fix race against tty_open() in tty_register_device_attr()
2025-05-28 13:28 ` [PATCH v2 2/2] tty: Fix race against tty_open() in tty_register_device_attr() Max Staudt
@ 2025-06-02 10:31 ` Ilpo Järvinen
2025-06-02 13:40 ` Max Staudt
2025-06-03 8:49 ` kernel test robot
1 sibling, 1 reply; 6+ messages in thread
From: Ilpo Järvinen @ 2025-06-02 10:31 UTC (permalink / raw)
To: Max Staudt
Cc: Greg Kroah-Hartman, Jiri Slaby, Johan Hovold, linux-serial, LKML,
stable
On Wed, 28 May 2025, Max Staudt wrote:
> Since the chardev is now created before the class device is registered,
> an attempt to tty_[k]open() the chardev between these two steps will
> lead to tty->dev being assigned NULL by alloc_tty_struct().
>
> alloc_tty_struct() is called via tty_init_dev() when the tty is firstly
> opened, and is entered with tty_mutex held, so let's lock the critical
> section in tty_register_device_attr() with the same global mutex.
> This guarantees that tty->dev can be assigned a sane value.
>
> Fixes: 6a7e6f78c235 ("tty: close race between device register and open")
> Signed-off-by: Max Staudt <max@enpas.org>
> Cc: <stable@vger.kernel.org>
> ---
> drivers/tty/tty_io.c | 9 ++++++++-
> 1 file changed, 8 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/tty/tty_io.c b/drivers/tty/tty_io.c
> index e922b84524d2..94768509e2d2 100644
> --- a/drivers/tty/tty_io.c
> +++ b/drivers/tty/tty_io.c
> @@ -3258,6 +3258,8 @@ struct device *tty_register_device_attr(struct tty_driver *driver,
> else
> tty_line_name(driver, index, name);
>
> + mutex_lock(&tty_mutex);
Use guard() so you don't need to change the returns and rollback path.
> +
> if (!(driver->flags & TTY_DRIVER_DYNAMIC_ALLOC)) {
> /*
> * Free any saved termios data so that the termios state is
> @@ -3271,7 +3273,7 @@ struct device *tty_register_device_attr(struct tty_driver *driver,
>
> retval = tty_cdev_add(driver, devt, index, 1);
> if (retval)
> - return ERR_PTR(retval);
> + goto err_unlock;
>
> cdev_added = true;
> }
> @@ -3294,6 +3296,8 @@ struct device *tty_register_device_attr(struct tty_driver *driver,
> if (retval)
> goto err_put;
>
> + mutex_unlock(&tty_mutex);
> +
> return dev;
>
> err_put:
> @@ -3309,6 +3313,9 @@ struct device *tty_register_device_attr(struct tty_driver *driver,
> driver->cdevs[index] = NULL;
> }
>
> +err_unlock:
> + mutex_unlock(&tty_mutex);
> +
> return ERR_PTR(retval);
> }
> EXPORT_SYMBOL_GPL(tty_register_device_attr);
>
--
i.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH v2 2/2] tty: Fix race against tty_open() in tty_register_device_attr()
2025-06-02 10:31 ` Ilpo Järvinen
@ 2025-06-02 13:40 ` Max Staudt
2025-06-03 13:43 ` Jiri Slaby
0 siblings, 1 reply; 6+ messages in thread
From: Max Staudt @ 2025-06-02 13:40 UTC (permalink / raw)
To: Ilpo Järvinen
Cc: Greg Kroah-Hartman, Jiri Slaby, Johan Hovold, linux-serial, LKML,
stable
On 6/2/25 19:31, Ilpo Järvinen wrote:
>> + mutex_lock(&tty_mutex);
>
> Use guard() so you don't need to change the returns and rollback path.
Thanks, I didn't know about this new kind of helper.
I'll leave it up to the TTY maintainers - if they don't express a
preference for guard(), then I deem this code simple enough to leave it
as-is, because I don't have any experience with guard(), and in fact,
until 5 minutes ago, I didn't know at all that GCC cleanup attributes
even exist.
Interestingly, Documentation/process/maintainer-netdev.rst documents a
preference against guard(). I wonder why, but that's for another day.
Do you have an idea on how to solve the circular lock that the kernel
test robot found for v1 of this patch?
https://lore.kernel.org/linux-serial/202505281412.8c836cb7-lkp@intel.com/
Max
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH v2 2/2] tty: Fix race against tty_open() in tty_register_device_attr()
2025-05-28 13:28 ` [PATCH v2 2/2] tty: Fix race against tty_open() in tty_register_device_attr() Max Staudt
2025-06-02 10:31 ` Ilpo Järvinen
@ 2025-06-03 8:49 ` kernel test robot
1 sibling, 0 replies; 6+ messages in thread
From: kernel test robot @ 2025-06-03 8:49 UTC (permalink / raw)
To: Max Staudt
Cc: oe-lkp, lkp, linux-kernel, linux-serial, Greg Kroah-Hartman,
Jiri Slaby, Johan Hovold, Max Staudt, stable, oliver.sang
Hello,
kernel test robot noticed "WARNING:possible_circular_locking_dependency_detected" on:
commit: c3184e841ee2a1585b89e4bbd90cacb41063871f ("[PATCH v2 2/2] tty: Fix race against tty_open() in tty_register_device_attr()")
url: https://github.com/intel-lab-lkp/linux/commits/Max-Staudt/tty-Fix-race-against-tty_open-in-tty_register_device_attr/20250528-213024
base: https://git.kernel.org/cgit/linux/kernel/git/gregkh/tty.git tty-testing
patch link: https://lore.kernel.org/all/20250528132816.11433-2-max@enpas.org/
patch subject: [PATCH v2 2/2] tty: Fix race against tty_open() in tty_register_device_attr()
in testcase: boot
config: x86_64-randconfig-001-20250530
compiler: gcc-12
test machine: qemu-system-x86_64 -enable-kvm -cpu SandyBridge -smp 2 -m 16G
(please refer to attached dmesg/kmsg for entire log/backtrace)
+-------------------------------------------------------------------------------------------------------------------------------------------------------+------------+------------+
| | 44db64e370 | c3184e841e |
+-------------------------------------------------------------------------------------------------------------------------------------------------------+------------+------------+
| WARNING:possible_circular_locking_dependency_detected | 0 | 6 |
| WARNING:possible_circular_locking_dependency_detected_swapper_is_trying_to_acquire_lock:at:tty_port_open_but_task_is_already_holding_lock:at:tty_lock | 0 | 6 |
+-------------------------------------------------------------------------------------------------------------------------------------------------------+------------+------------+
If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <oliver.sang@intel.com>
| Closes: https://lore.kernel.org/oe-lkp/202506031638.a5c78088-lkp@intel.com
[ 22.678060][ T1] WARNING: possible circular locking dependency detected
[ 22.678860][ T1] 6.15.0-rc4-00108-gc3184e841ee2 #1 Tainted: G T
[ 22.679596][ T1] ------------------------------------------------------
[ 22.680280][ T1] swapper/1 is trying to acquire lock:
[ 22.680750][ T1] ffff888101cf02d0 (&port->mutex){+.+.}-{4:4}, at: tty_port_open (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/tty/tty_port.c:761)
[ 22.681597][ T1]
[ 22.681597][ T1] but task is already holding lock:
[ 22.682232][ T1] ffff8881084f91e0 (&tty->legacy_mutex){+.+.}-{4:4}, at: tty_lock (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/tty/tty_mutex.c:19)
[ 22.683067][ T1]
[ 22.683067][ T1] which lock already depends on the new lock.
[ 22.683067][ T1]
[ 22.683981][ T1]
[ 22.683981][ T1] the existing dependency chain (in reverse order) is:
[ 22.684732][ T1]
[ 22.684732][ T1] -> #2 (&tty->legacy_mutex){+.+.}-{4:4}:
[ 22.685465][ T1] validate_chain (kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/lockdep.c:3286 kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/lockdep.c:3909)
[ 22.685990][ T1] __lock_acquire (kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/lockdep.c:5235)
[ 22.688240][ T1] lock_acquire (kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/lockdep.c:472 kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/lockdep.c:5868 kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/lockdep.c:5823)
[ 22.688677][ T1] __mutex_lock (kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/mutex.c:603 kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/mutex.c:746)
[ 22.689128][ T1] mutex_lock_nested (kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/mutex.c:799)
[ 22.689636][ T1] tty_lock (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/tty/tty_mutex.c:19)
[ 22.690065][ T1] tty_init_dev (kbuild/obj/consumer/x86_64-randconfig-001-20250530/include/linux/module.h:730)
[ 22.690553][ T1] tty_init_dev (kbuild/obj/consumer/x86_64-randconfig-001-20250530/include/linux/module.h:730)
[ 22.691065][ T1] tty_open (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/tty/tty_io.c:2073 kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/tty/tty_io.c:2120)
[ 22.691582][ T1] chrdev_open (kbuild/obj/consumer/x86_64-randconfig-001-20250530/fs/char_dev.c:414)
[ 22.692010][ T1] do_dentry_open (kbuild/obj/consumer/x86_64-randconfig-001-20250530/fs/open.c:956)
[ 22.692506][ T1] vfs_open (kbuild/obj/consumer/x86_64-randconfig-001-20250530/fs/open.c:1087)
[ 22.692910][ T1] do_open (kbuild/obj/consumer/x86_64-randconfig-001-20250530/fs/namei.c:3881)
[ 22.693344][ T1] path_openat (kbuild/obj/consumer/x86_64-randconfig-001-20250530/fs/namei.c:4039)
[ 22.693796][ T1] do_filp_open (kbuild/obj/consumer/x86_64-randconfig-001-20250530/fs/namei.c:4067)
[ 22.694218][ T1] file_open_name (kbuild/obj/consumer/x86_64-randconfig-001-20250530/fs/open.c:1374)
[ 22.694669][ T1] filp_open (kbuild/obj/consumer/x86_64-randconfig-001-20250530/fs/open.c:1394)
[ 22.695063][ T1] console_on_rootfs (kbuild/obj/consumer/x86_64-randconfig-001-20250530/init/main.c:1528)
[ 22.695624][ T1] kernel_init_freeable (kbuild/obj/consumer/x86_64-randconfig-001-20250530/init/main.c:1578)
[ 22.696135][ T1] kernel_init (kbuild/obj/consumer/x86_64-randconfig-001-20250530/init/main.c:1459)
[ 22.696554][ T1] ret_from_fork (kbuild/obj/consumer/x86_64-randconfig-001-20250530/arch/x86/kernel/process.c:159)
[ 22.697060][ T1] ret_from_fork_asm (kbuild/obj/consumer/x86_64-randconfig-001-20250530/arch/x86/entry/entry_64.S:258)
[ 22.697516][ T1]
[ 22.697516][ T1] -> #1 (tty_mutex){+.+.}-{4:4}:
[ 22.698166][ T1] validate_chain (kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/lockdep.c:3286 kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/lockdep.c:3909)
[ 22.698652][ T1] __lock_acquire (kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/lockdep.c:5235)
[ 22.699158][ T1] lock_acquire (kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/lockdep.c:472 kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/lockdep.c:5868 kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/lockdep.c:5823)
[ 22.699577][ T1] __mutex_lock (kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/mutex.c:603 kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/mutex.c:746)
[ 22.700113][ T1] mutex_lock_nested (kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/mutex.c:799)
[ 22.700564][ T1] tty_register_device_attr (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/tty/tty_io.c:3250)
[ 22.701135][ T1] tty_port_register_device_attr_serdev (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/tty/tty_port.c:199)
[ 22.701722][ T1] serial_core_add_one_port (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/tty/serial/serial_core.c:3184)
[ 22.702255][ T1] serial_core_register_port (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/tty/serial/serial_core.c:3388)
[ 22.702868][ T1] serial_ctrl_register_port (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/tty/serial/serial_ctrl.c:42)
[ 22.703443][ T1] uart_add_one_port (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/tty/serial/serial_port.c:144)
[ 22.703888][ T1] serial8250_register_8250_port (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/tty/serial/8250/8250_core.c:822)
[ 22.704473][ T1] serial_pnp_probe (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/tty/serial/8250/8250_pnp.c:480)
[ 22.704935][ T1] pnp_device_probe (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/pnp/driver.c:113)
[ 22.705438][ T1] really_probe (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/base/dd.c:579 kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/base/dd.c:657)
[ 22.705882][ T1] __driver_probe_device (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/base/dd.c:799)
[ 22.706388][ T1] driver_probe_device (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/base/dd.c:829)
[ 22.706886][ T1] __driver_attach (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/base/dd.c:1216 kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/base/dd.c:1155)
[ 22.707345][ T1] bus_for_each_dev (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/base/bus.c:369)
[ 22.707854][ T1] driver_attach (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/base/dd.c:1234)
[ 22.708343][ T1] bus_add_driver (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/base/bus.c:678)
[ 22.708788][ T1] driver_register (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/base/driver.c:249)
[ 22.709313][ T1] pnp_register_driver (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/pnp/driver.c:281)
[ 22.709779][ T1] serial8250_pnp_init (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/tty/serial/8250/8250_pnp.c:531)
[ 22.710314][ T1] serial8250_init (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/tty/serial/8250/8250_platform.c:315)
[ 22.710850][ T1] do_one_initcall (kbuild/obj/consumer/x86_64-randconfig-001-20250530/init/main.c:1257)
[ 22.711331][ T1] do_initcalls (kbuild/obj/consumer/x86_64-randconfig-001-20250530/init/main.c:1318 kbuild/obj/consumer/x86_64-randconfig-001-20250530/init/main.c:1335)
[ 22.711819][ T1] kernel_init_freeable (kbuild/obj/consumer/x86_64-randconfig-001-20250530/init/main.c:1569)
[ 22.712320][ T1] kernel_init (kbuild/obj/consumer/x86_64-randconfig-001-20250530/init/main.c:1459)
[ 22.712739][ T1] ret_from_fork (kbuild/obj/consumer/x86_64-randconfig-001-20250530/arch/x86/kernel/process.c:159)
[ 22.713196][ T1] ret_from_fork_asm (kbuild/obj/consumer/x86_64-randconfig-001-20250530/arch/x86/entry/entry_64.S:258)
[ 22.713650][ T1]
[ 22.713650][ T1] -> #0 (&port->mutex){+.+.}-{4:4}:
[ 22.714356][ T1] check_noncircular (kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/lockdep.c:2215)
[ 22.714826][ T1] check_prev_add (kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/lockdep.c:3167)
[ 22.715274][ T1] validate_chain (kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/lockdep.c:3286 kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/lockdep.c:3909)
[ 22.715793][ T1] __lock_acquire (kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/lockdep.c:5235)
[ 22.716237][ T1] lock_acquire (kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/lockdep.c:472 kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/lockdep.c:5868 kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/lockdep.c:5823)
[ 22.716741][ T1] __mutex_lock (kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/mutex.c:603 kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/mutex.c:746)
[ 22.717172][ T1] mutex_lock_nested (kbuild/obj/consumer/x86_64-randconfig-001-20250530/kernel/locking/mutex.c:799)
[ 22.717631][ T1] tty_port_open (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/tty/tty_port.c:761)
[ 22.718106][ T1] uart_open (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/tty/serial/serial_core.c:1974)
[ 22.718642][ T1] tty_open (kbuild/obj/consumer/x86_64-randconfig-001-20250530/drivers/tty/tty_io.c:2140)
[ 22.719094][ T1] chrdev_open (kbuild/obj/consumer/x86_64-randconfig-001-20250530/fs/char_dev.c:414)
[ 22.719514][ T1] do_dentry_open (kbuild/obj/consumer/x86_64-randconfig-001-20250530/fs/open.c:956)
[ 22.719966][ T1] vfs_open (kbuild/obj/consumer/x86_64-randconfig-001-20250530/fs/open.c:1087)
[ 22.720411][ T1] do_open (kbuild/obj/consumer/x86_64-randconfig-001-20250530/fs/namei.c:3881)
[ 22.720873][ T1] path_openat (kbuild/obj/consumer/x86_64-randconfig-001-20250530/fs/namei.c:4039)
[ 22.721298][ T1] do_filp_open (kbuild/obj/consumer/x86_64-randconfig-001-20250530/fs/namei.c:4067)
[ 22.721848][ T1] file_open_name (kbuild/obj/consumer/x86_64-randconfig-001-20250530/fs/open.c:1374)
[ 22.722280][ T1] filp_open (kbuild/obj/consumer/x86_64-randconfig-001-20250530/fs/open.c:1394)
[ 22.722739][ T1] console_on_rootfs (kbuild/obj/consumer/x86_64-randconfig-001-20250530/init/main.c:1528)
[ 22.723194][ T1] kernel_init_freeable (kbuild/obj/consumer/x86_64-randconfig-001-20250530/init/main.c:1578)
[ 22.723695][ T1] kernel_init (kbuild/obj/consumer/x86_64-randconfig-001-20250530/init/main.c:1459)
[ 22.724200][ T1] ret_from_fork (kbuild/obj/consumer/x86_64-randconfig-001-20250530/arch/x86/kernel/process.c:159)
[ 22.724709][ T1] ret_from_fork_asm (kbuild/obj/consumer/x86_64-randconfig-001-20250530/arch/x86/entry/entry_64.S:258)
[ 22.725164][ T1]
[ 22.725164][ T1] other info that might help us debug this:
[ 22.725164][ T1]
[ 22.726026][ T1] Chain exists of:
[ 22.726026][ T1] &port->mutex --> tty_mutex --> &tty->legacy_mutex
[ 22.726026][ T1]
The kernel config and materials to reproduce are available at:
https://download.01.org/0day-ci/archive/20250603/202506031638.a5c78088-lkp@intel.com
--
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH v2 2/2] tty: Fix race against tty_open() in tty_register_device_attr()
2025-06-02 13:40 ` Max Staudt
@ 2025-06-03 13:43 ` Jiri Slaby
0 siblings, 0 replies; 6+ messages in thread
From: Jiri Slaby @ 2025-06-03 13:43 UTC (permalink / raw)
To: Max Staudt, Ilpo Järvinen
Cc: Greg Kroah-Hartman, Johan Hovold, linux-serial, LKML, stable
On 02. 06. 25, 15:40, Max Staudt wrote:
> On 6/2/25 19:31, Ilpo Järvinen wrote:
>>> + mutex_lock(&tty_mutex);
>>
>> Use guard() so you don't need to change the returns and rollback path.
>
> Thanks, I didn't know about this new kind of helper.
>
> I'll leave it up to the TTY maintainers - if they don't express a
> preference for guard(),
I prefer guard(). Actually, I have a patchset to add a support for
guard() for uart_lock and console_lock too and use it all over (incl.
__free). They untangle the code on many places and get rid of much
unneeded churn.
But in this very case, I see there is a label, I am not sure if it works
right here. Try compiling with clang -- it will tell you. You likely
won't cross the label with the guard().
thanks,
--
js
suse labs
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2025-06-03 13:43 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2025-05-28 13:28 [PATCH v2 1/2] tty: Register device *after* creating the cdev for a tty Max Staudt
2025-05-28 13:28 ` [PATCH v2 2/2] tty: Fix race against tty_open() in tty_register_device_attr() Max Staudt
2025-06-02 10:31 ` Ilpo Järvinen
2025-06-02 13:40 ` Max Staudt
2025-06-03 13:43 ` Jiri Slaby
2025-06-03 8:49 ` kernel test robot
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox