Re: [BUG] general protection fault when connecting an old mp3/usb device

["David Wang" <00107082@163.com>]

At 2025-08-20 03:20:51, "Bart Van Assche" <bvanassche@acm.org> wrote:
>On 8/18/25 2:50 AM, David Wang wrote:
>> [Sat Aug 23 03:56:09 2025] Oops: general protection fault, probably for non-canonical address 0x2e2e2f2e2e2f308e: 0000 [#1] SMP NOPTI
>> [Sat Aug 23 03:56:09 2025] Call Trace:
>> [Sat Aug 23 03:56:09 2025]  <TASK>
>> [Sat Aug 23 03:56:09 2025]  sr_do_ioctl+0x5b/0x1c0 [sr_mod]
>> [Sat Aug 23 03:56:09 2025]  sr_packet+0x2c/0x50 [sr_mod]
>> [Sat Aug 23 03:56:09 2025]  cdrom_get_disc_info+0x60/0xe0 [cdrom]
>> [Sat Aug 23 03:56:09 2025]  cdrom_mrw_exit+0x29/0xb0 [cdrom]
>> [Sat Aug 23 03:56:09 2025]  ? xa_destroy+0xaa/0x120
>> [Sat Aug 23 03:56:09 2025]  unregister_cdrom+0x76/0xc0 [cdrom]
>> [Sat Aug 23 03:56:09 2025]  sr_free_disk+0x44/0x50 [sr_mod]
>> [Sat Aug 23 03:56:09 2025]  disk_release+0xb0/0xe0
>> [Sat Aug 23 03:56:09 2025]  device_release+0x37/0x90
>> [Sat Aug 23 03:56:09 2025]  kobject_put+0x8e/0x1d0
>> [Sat Aug 23 03:56:09 2025]  blkdev_release+0x11/0x20
>> [Sat Aug 23 03:56:09 2025]  __fput+0xe3/0x2a0
>> [Sat Aug 23 03:56:09 2025]  task_work_run+0x59/0x90
>> [Sat Aug 23 03:56:09 2025]  exit_to_user_mode_loop+0xd6/0xe0
>> [Sat Aug 23 03:56:09 2025]  do_syscall_64+0x1c1/0x1e0
>> [Sat Aug 23 03:56:09 2025]  entry_SYSCALL_64_after_hwframe+0x76/0x7e
>
>Phillip, is this behavior perhaps introduced by commit 5ec9d26b78c4
>("cdrom: Call cdrom_mrw_exit from cdrom_release function")? Please do

I caught this on 6.16.0 which does not have this commit, and on the contrary, base
on the commit message, this commit may help in the positive direction.

I will try to figure out a procedure to reproduce this on my laptop, and if that possible,
I will upgrade my laptop to 6.17-rc1 to give it a trial.
And update later.

Thanks
David 

>not call code that invokes ioctls from the disk_release() callback.
>
>Thanks,
>
>Bart.