Re: [PATCH] pps: fix warning in pps_register_cdev when register device fail

[Wang Liang <wangliang74@huawei.com>]

在 2025/8/28 21:24, Calvin Owens 写道:
> On Wednesday 08/27 at 08:32 -0700, Calvin Owens wrote:
>> On Wednesday 08/27 at 14:50 +0800, Wang Liang wrote:
>>> Similar to previous commit 2a934fdb01db ("media: v4l2-dev: fix error
>>> handling in __video_register_device()"), the release hook should be set
>>> before device_register(). Otherwise, when device_register() return error
>>> and put_device() try to callback the release function, the below warning
>>> may happen.
>>>
>>>    ------------[ cut here ]------------
>>>    WARNING: CPU: 1 PID: 4760 at drivers/base/core.c:2567 device_release+0x1bd/0x240 drivers/base/core.c:2567
>>>    Modules linked in:
>>>    CPU: 1 UID: 0 PID: 4760 Comm: syz.4.914 Not tainted 6.17.0-rc3+ #1 NONE
>>>    RIP: 0010:device_release+0x1bd/0x240 drivers/base/core.c:2567
>>>    Call Trace:
>>>     <TASK>
>>>     kobject_cleanup+0x136/0x410 lib/kobject.c:689
>>>     kobject_release lib/kobject.c:720 [inline]
>>>     kref_put include/linux/kref.h:65 [inline]
>>>     kobject_put+0xe9/0x130 lib/kobject.c:737
>>>     put_device+0x24/0x30 drivers/base/core.c:3797
>>>     pps_register_cdev+0x2da/0x370 drivers/pps/pps.c:402
>>>     pps_register_source+0x2f6/0x480 drivers/pps/kapi.c:108
>>>     pps_tty_open+0x190/0x310 drivers/pps/clients/pps-ldisc.c:57
>>>     tty_ldisc_open+0xa7/0x120 drivers/tty/tty_ldisc.c:432
>>>     tty_set_ldisc+0x333/0x780 drivers/tty/tty_ldisc.c:563
>>>     tiocsetd drivers/tty/tty_io.c:2429 [inline]
>>>     tty_ioctl+0x5d1/0x1700 drivers/tty/tty_io.c:2728
>>>     vfs_ioctl fs/ioctl.c:51 [inline]
>>>     __do_sys_ioctl fs/ioctl.c:598 [inline]
>>>     __se_sys_ioctl fs/ioctl.c:584 [inline]
>>>     __x64_sys_ioctl+0x194/0x210 fs/ioctl.c:584
>>>     do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
>>>     do_syscall_64+0x5f/0x2a0 arch/x86/entry/syscall_64.c:94
>>>     entry_SYSCALL_64_after_hwframe+0x76/0x7e
>>>     </TASK>
>>>
>>> Before commit c79a39dc8d06 ("pps: Fix a use-after-free"),
>>> pps_register_cdev() call device_create() to create pps->dev, which will
>>> init dev->release to device_create_release(). Now the comment is outdated,
>>> just remove it.
>> Hi Wang,
>>
>> I'm curious why pps_register_cdev() is failing, is there possibly a
>> second issue to investigate there? Or was it fault injection?
>>
>> Otherwise, makes perfect sense to me. I'm new to this code, so grain of
>> salt, but since I exposed it:
>>
>> Reviewed-by: Calvin Owens <calvin@wbinvd.org>
> I apologize Wang, I missed something when I looked at this yesterday:
>
> I think your patch introduces a double-free, because the pps object is
> already freed by pps_register_source() when pps_register_cdev() returns
> an error. Because put_device() was a nop with the missing release_fn, it
> wasn't actually doing anything. That's very confusing, and my fault.


You are right! Thanks for the reminder. I reproduce the double-free issue:

==================================================================
BUG: KASAN: double-free in pps_register_source+0x2c1/0x3a0
Free of addr ffff8881078fd800 by task insmod/323
Call Trace:
  <TASK>
  dump_stack_lvl+0x55/0x70
  print_report+0xcb/0x610
  kasan_report_invalid_free+0x94/0xc0
  check_slab_allocation+0x105/0x130
  kfree+0xba/0x360
  pps_register_source+0x2c1/0x3a0

It is triggered by kzalloc fail in device_private_init().

> Your patch is clearly an improvement. But I think you additionally need
> to remove the 'kfree_pps' error logic at the bottom of
> pps_register_source() to avoid a double free in the failure case:


Thanks for your suggestions, it is helpful!

I will send a v2 patch later. Thanks!

------
Best regards
Wang Liang

>
> diff --git a/drivers/pps/kapi.c b/drivers/pps/kapi.c
> index 92d1b62ea239..4d0ba35590a3 100644
> --- a/drivers/pps/kapi.c
> +++ b/drivers/pps/kapi.c
> @@ -116,9 +116,6 @@ struct pps_device *pps_register_source(struct pps_source_info *info,
>   
>   	return pps;
>   
> -kfree_pps:
> -	kfree(pps);
> -
>   pps_register_source_exit:
>   	pr_err("%s: unable to register source\n", info->name);
>   
>
> The only reason I put that kfree() there is because I noticed the memory
> was being leaked in the failure case. Doh. Your patch makes this right.
>
>> Thanks,
>> Calvin
>>
>>> Fixes: c79a39dc8d06 ("pps: Fix a use-after-free")
>>> Signed-off-by: Wang Liang <wangliang74@huawei.com>
>>> ---
>>>   drivers/pps/pps.c | 4 +---
>>>   1 file changed, 1 insertion(+), 3 deletions(-)
>>>
>>> diff --git a/drivers/pps/pps.c b/drivers/pps/pps.c
>>> index 9463232af8d2..0d2d57250575 100644
>>> --- a/drivers/pps/pps.c
>>> +++ b/drivers/pps/pps.c
>>> @@ -383,13 +383,11 @@ int pps_register_cdev(struct pps_device *pps)
>>>   	pps->dev.devt = MKDEV(pps_major, pps->id);
>>>   	dev_set_drvdata(&pps->dev, pps);
>>>   	dev_set_name(&pps->dev, "pps%d", pps->id);
>>> +	pps->dev.release = pps_device_destruct;
>>>   	err = device_register(&pps->dev);
>>>   	if (err)
>>>   		goto free_idr;
>>>   
>>> -	/* Override the release function with our own */
>>> -	pps->dev.release = pps_device_destruct;
>>> -
>>>   	pr_debug("source %s got cdev (%d:%d)\n", pps->info.name, pps_major,
>>>   		 pps->id);
>>>   
>>> -- 
>>> 2.33.0
>>>