Bug 218658
| Summary: | Hibernate stuck after recent kernel/workqueue.c changes in Stable 6.6.23 | ||
|---|---|---|---|
| Product: | Linux | Reporter: | Petri Kaukasoina (petri.kaukasoina) |
| Component: | Kernel | Assignee: | Virtual assignee for kernel bugs (linux-kernel) |
| Status: | NEW --- | ||
| Severity: | normal | CC: | padulodeveloper, petri.kaukasoina, philippolson886, regressions |
| Priority: | P3 | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Kernel Version: | 6.6.23 | Subsystem: | |
| Regression: | Yes | Bisected commit-id: | 5a70baec2294e8a7d0fcc4558741c23e752dad5c |
| Attachments: |
dmesg of the broken 6.6.23
bisection log |
||
Could you please bisect which of the changes that modified that file broke things? And check if mainline is affected as well? This guide explains the steps: https://docs.kernel.org/admin-guide/verify-bugs-and-bisect-regressions.html When actually doing the regression, you can limit the range to changes to that modified said file like this: git bisect start -- kernel/workqueue.c Thank you for your answer. 5a70baec2294e8a7d0fcc4558741c23e752dad5c is the first bad commit commit 5a70baec2294e8a7d0fcc4558741c23e752dad5c Author: Tejun Heo <tj@kernel.org> Date: Mon Jan 29 08:11:25 2024 -1000 workqueue: Implement system-wide nr_active enforcement for unbound workqueues I also tried mainline (6.9.0-rc2) but it couldn't even mount the root fs for some reason. Probably .config would have needed some work. Created attachment 306076 [details]
bisection log
Could you please give 6.6.24-rc1 a quick try; there was a workqueue change missing that might or might not be related: https://lore.kernel.org/all/20240401152547.867452742@linuxfoundation.org/ 6.6.24-rc1 did not fix this problem. When I revert commit 5a70baec2294e8a7d0fcc4558741c23e752dad5c from 6.6.24-rc1, hibernating and waking up work again. I got the same problem on Debian 11 bullseye updating kernel from version linux-image-5.10.0-28-amd to version linux-image-5.10.0-29-amd64. From an hibernated system: - When in GRUB I select the new kernel (5.10.0-29) the system waked up and the previous display comes visible, but it is stuck there; - When in GRUB I select the previous kernel (5.10.0-28-amd), the waked up works. As explained in comment #8, the system "wake up" doesn't work after hibernation. I have updated Debian 11 bullseye to the kernel version "linux-image-5.10.0-30-amd64", but the bug persists. From an hibernated system: - When in GRUB I select the new kernel (5.10.0-30) the system waked up and the previous display comes visible, but it is stuck there; - When in GRUB I select the working kernel (5.10.0-28-amd), the system correctly wake up. padulodeveloper those versions mean nothing here, those are not upstream kernel versions; reporting problems in an existing and likely fixed bug is also bad. Both things are listed here: https://linux-regtracking.leemhuis.info/post/frequent-reasons-why-linux-kernel-bug-reports-are-ignored/ |
Created attachment 306057 [details] dmesg of the broken 6.6.23 With kernel 6.6.23 hibernating usually hangs here: the display stays on but the mouse pointer does not move and the keyboard does not work. But SysRq REISUB does reboot. Sometimes it seems to hibernate: the computer powers down and can be waked up and the previous display comes visible, but it is stuck there. With 6.6.22 and earlier, hibernate works ok. When I rebuilt 6.6.23 with the previous linux-6.6.22/kernel/workqueue.c, hibernating works again.