Bug 218902
| Summary: | Kernel crash/reboot on cifs copy | ||
|---|---|---|---|
| Product: | File System | Reporter: | dufgrinder |
| Component: | Samba/SMB | Assignee: | fs_samba-smb |
| Status: | NEW --- | ||
| Severity: | blocking | CC: | kernelbugzilla.jim, pc, regressions, smfrench, toracat |
| Priority: | P3 | ||
| Hardware: | Intel | ||
| OS: | Linux | ||
| Kernel Version: | 6.8.4 | Subsystem: | |
| Regression: | Yes | Bisected commit-id: | |
|
Description
dufgrinder
2024-05-27 20:24:38 UTC
(In reply to dufgrinder from comment #0) > > This bug exists since kernel 6.8.4 So 6.8.3 works fine? Or did you mean 6.7.y worked fine? Then a bisection would be good (and might be required): https://docs.kernel.org/admin-guide/verify-bugs-and-bisect-regressions.html According to : https://elrepo.org/bugs/view.php?id=1454 There is one other person who reported what looks like the same issue. kernel 6.7.9-1 works fine. Sure, elrepo reporter is me. The bug appears on 2 of my last updated computers to AlmaLinux 9.4. I'll try to do the bisect, but it'll be my first on a kernel ! KR, Olivier Just for info, the other reporter on the ELRepo's bug tracker runs CentOS Stream 9. FWIW, there is a patch [1] that has been sent to the mailing list that is a good candidate for fixing this deadlock. [1] https://lore.kernel.org/r/20240606161313.25521-1-ematsumiya@suse.de Using the patch provided in Comment 5, I have built a kernel-ml package set, kernel-ml-6.9.3-1.1.el9.elrepo. https://toracat.org/test/kernel/bug218902/ Can you test and see if the issue is fixed? (In reply to Akemi Yagi from comment #6) > Using the patch provided in Comment 5, I have built a kernel-ml package set, > kernel-ml-6.9.3-1.1.el9.elrepo. > > https://toracat.org/test/kernel/bug218902/ > > Can you test and see if the issue is fixed? I get the same crash with this kernel with CentOS Stream 9 > I get the same crash with this kernel with CentOS Stream 9
Apparently, the candidate patch did not fix the issue.
(dufgrinder also reported the same result.)
Sure, Crash confirmed with candidate fix (kernel-ml-6.9.3-1.1.el9.elrepo) on AlmaLinux 9.4 Anyone try it on 6.10-rc3 yet? We (elrepo) just built kernel-ml-6.10.0-0.rc3.el9.elrepo. We'll make it available for testing. kernel-ml-6.10.0-0.rc3.el9.elrepo.x86_64 is available here: https://elrepo.org/people/akemi/testing/el9/kernel/218902/x86_64/ kernel-ml-6.10.0-0.rc3.el9.elrepo.x86_64 worked for me with CentOS Stream 9. Writing to a CIFS mount no longer crashes my system. That's good news. kernel-ml-6.10 GA version is expected to be out in mid July. Steve, you need to figure out which commit fixed the copy regression so it can be backported to v6.8.y and v6.9.y -- if it hasn't been backported yet. Note that they reported two different oopses. The one from description seems to be fixed by [1]. [1] https://lore.kernel.org/r/20240606161313.25521-1-ematsumiya@suse.de Hi kernel-ml-6.10.0-0.rc3.el9.elrepo.x86_64 worked for me using AlmaLinux 9.4 on two laptops ( Intel i5-8250U and an old Intel Pentium B970) Tested with the same kind of copy to the CIFS mounted directory I think this can be closed. CIFS writes still work with kernel-ml-6.10.6-1.el9.elrepo.x86_64 (CentOS Stream 9). I confirm, CIFS copy works correctly now This bug can be closed |