On 2020/05/25 21:06, Oliver Neukum wrote: > Am Freitag, den 22.05.2020, 17:26 +0900 schrieb Tetsuo Handa: >> On 2020/05/22 17:04, Oliver Neukum wrote: >>> May I ask you to redo the patch with comments added stating >>> that the wake up is done for the sake of wdm_flush(), change >>> the description and add the link to syzkaller? >> >> You can take over this patch. syzbot tried this patch on 2020/02/11 01:23 at >> https://syzkaller.appspot.com/bug?id=e7b761593b23eb50855b9ea31e3be5472b711186 , >> but this patch did not solve the problem syzbot has found. Thus, I don't add >> a link to syzkaller... > > Hi, > > this is odd. I looked at the driver and it looks to me like > pre/post_reset() is the only other place that touches WDM_IN_USE. > And it does so correctly. Any idea what could be wrong? I haven't understood the root cause of this problem. You were in the thread at https://groups.google.com/forum/#!msg/syzkaller-bugs/q3QMjt_t83c/GqP58GaTAQAJ . My patch was made from that thread (which simply makes sure to call wake_up_all() when clearing WDM_IN_USE bit) apart from the root cause of this problem.
