I was running a modified syzkaller locally and the crash came from that setup. I've since lost the original log, and looking back, I may have confused an issue in my tooling with a kernel bug. I should have verified more carefully before sending this. Sorry for the confusion.
Soowan Park 2026년 5월 26일 (화) 오후 11:18, Sven Eckelmann <[email protected]>님이 작성: > > On Tuesday, 26 May 2026 15:56:58 CEST 박수완 / 학생 / 컴퓨터공학부 wrote: > > Hi Sven, > > > > You're right. I re-examined __cancel_work_sync() and confirmed that it > > internally disables the work before flushing, so the re-queue race I > > described does not exist. I'll withdraw this patch. > > > > Thanks for the review. > > But you said that this was reported by syzkaller. I can't find the report in > the official list [1]. I am guessing you have a reproducer in a private > instance and tested with it. And if you tested your patch with it, you have > found that your patch fixes it, right? > > Can you give us more information about this to let us figure out what is > actually going on? > > Regards, > Sven > > [1] https://syzkaller.appspot.com/upstream/s/batman
