VM_WARN_ON in switch_mm_irqs_off() when offlining a CPU

Open

#469 opened on Sep 21, 2023 3 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

bug

Dominant language: No language data
Stars: 14
Forks: 0
PR merge metrics: No merged PRs in 30d

Description

The `VM_WARN_ON()` in `switch_mm_irqs_off()` added by commit torvalds/linux@177255afb40548fdf504384b361d18d6cbe35d1e
some times fires when offling a CPU:

```
# echo
[ 3.278821][
[ 3.278823][
[ 3.278834][
[ 3.278837][
[ 3.278841][
[ 3.278844][
[ 3.278847][
[ 3.278850][
[ 3.278866][
[ 3.278866][
[ 3.278866][
[ 3.278866][
[ 3.278866][
[ 3.278866][
[ 3.278866][
[ 3.278866][
[ 3.278866][
[ 3.278920][
[ 3.278926][
[ 3.278931][
[ 3.278932][
[ 3.278939][
[ 3.278946][
[ 3.278952][
[ 3.278958][
[ 3.278963][
[ 3.278969][
[ 3.278975][
[ 3.278980][
[ 3.279001][
``` 0 > /sys/devices/system/cpu/cpu1/online^Mecho 0 > /sys/devices/system/cpu/cpu1/online T0] ------------[ cut here ]------------ T0] WARNING: CPU: 1 PID: 0 at arch/powerpc/mm/mmu_context.c:106 switch_mm_irqs_off+0x120/0x150 T0] Modules linked in: T0] CPU: 1 PID: 0 Comm: swapper/1 Not tainted 6.6.0-rc2 #410 T0] Hardware name: IBM pSeries (emulated by qemu) POWER9 (raw) 0x4e1202 0xf000004 of:SLOF,HEAD hv:linux,kvm pSeries T0] NIP: c00000000008c6d0 LR: c00000000008c6b8 CTR: 0000000000000000 T0] REGS: c0000000049bfb50 TRAP: 0700 Not tainted (6.6.0-rc2) T0] MSR: 8000000002823033 CR: 2800020a XER: 00000000 T0] CFAR: c00000000008c690 IRQMASK: 3 T0] GPR00: c00000000008c6b8 c0000000049bfdf0 c000000001578a00 c0000000fffff408 T0] GPR04: c000000002a10b22 000000000000000a 0000000000000000 0000000000000000 T0] GPR08: 0000000000000000 0000000000000000 0000000000000003 0000000000000000 T0] GPR12: c0000000000fbed0 c0000000fffff300 0000000000000000 0000000000000000 T0] GPR16: 0000000000000000 0000000000000000 0000000000000000 0000000000000000 T0] GPR20: 0000000000000000 0000000000000000 0000000000000000 0000000000000001 T0] GPR24: 0000000000000001 0000000000000000 0000000000000000 c0000000029a13f4 T0] GPR28: 0000000000000001 0000000000000000 0000000000000001 c000000006fb6e00 T0] NIP [c00000000008c6d0] switch_mm_irqs_off+0x120/0x150 T0] LR [c00000000008c6b8] switch_mm_irqs_off+0x108/0x150 T0] Call Trace: T0] [c0000000049bfdf0] [c00000000008c6b8] switch_mm_irqs_off+0x108/0x150 (unreliable) T0] [c0000000049bfe30] [c0000000001a7dc0] idle_task_exit+0x90/0xc0 T0] [c0000000049bfe60] [c0000000000fbf00] pseries_cpu_offline_self+0x30/0xf0 T0] [c0000000049bfed0] [c00000000005c7b4] arch_cpu_idle_dead+0x44/0x50 T0] [c0000000049bfef0] [c0000000001cb114] do_idle+0x284/0x390 T0] [c0000000049bff60] [c0000000001cb4c0] cpu_startup_entry+0x40/0x50 T0] [c0000000049bff90] [c00000000005c21c] start_secondary+0x29c/0x2b0 T0] [c0000000049bffe0] [c00000000000e258] start_secondary_prolog+0x10/0x14 T0] Code: eba1ffe8 ebc1fff0 ebe1fff8 4e800020 7ca32b78 48009425 60000000 4bffffc4 0fe00000 4bffff44 60000000 60420000 <0fe00000> e8010050 7c0803a6 4bffffc0 T0] ---[ end trace 0000000000000000 ]---

Contributor guide

No contributing guide indexed for this repository

First steps

Read the whole issue, then the project's contributing guide.
Comment on the issue to say you are picking it up — it saves two people doing the same work.
Fork the repository and make your change on a branch.
Open a pull request that references the issue number.

Research direction

Start in arch/powerpc/mm/mmu_context.c at switch_mm_irqs_off(), then trace the idle_task_exit and pseries_cpu_offline_self path shown in the report. Reproduce CPU offlining with the provided /sys/devices/system/cpu/cpu1/online command and investigate why the VM_WARN_ON fires; done means offlining no longer emits this warning.

Written by the indexing model from the issue text.

Assessment

Tech stack: c, linux
Domain: operating-systems
Issue type: Bug
Difficulty: 4/5
Estimated time: 3-5 days
Activity status: Stale
Clarity: Mostly clear
Newbie friendliness: 35/100