blob: ffef3e78274165f0ac00056948913275cf698959 (
plain) (
blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
|
ppc: 0.903
peripherals: 0.849
kernel: 0.766
graphic: 0.761
device: 0.736
debug: 0.667
semantic: 0.654
architecture: 0.569
performance: 0.503
files: 0.502
vnc: 0.490
risc-v: 0.443
register: 0.432
boot: 0.395
permissions: 0.369
PID: 0.329
TCG: 0.317
socket: 0.284
network: 0.276
user-level: 0.206
VMM: 0.192
x86: 0.172
virtual: 0.163
i386: 0.146
arm: 0.132
assembly: 0.130
mistranslation: 0.124
hypervisor: 0.121
KVM: 0.036
ppc64: hard lockup / hang in Linux kernel v5.17-rc1
Description of problem:
The kernel deterministically triggers a hard lockup / hang under QEMU since v5.17-rc1 (upgrading from v5.16).
Bisecting points to the kernel's [0faf20a1ad16](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=0faf20a1ad1647c0fc0f5a367c71e5e84deaf899) ("powerpc/64s/interrupt: Don't enable MSR[EE] in irq handlers unless perf is in use"). Reverting it on top of v5.17-rc1 fixes the issue.
Reported to [linuxppc-dev](https://lore.kernel.org/linuxppc-dev/CANiq72n_FmDx=r-o9J8gYc6LpwRL5EGmhM6Xzwv27Xc7h1TNDw@mail.gmail.com/). Confirmed. Suspected QEMU modeling issue by Cédric Le Goater.
Steps to reproduce:
1. Build kernel v5.17-rc1 or commit 0faf20a1ad16 for ppc64le with the attached config (either GCC or Clang).
2. Run it under QEMU with at least `-smp 2`.
Additional information:
[qemu-and-dmesg.log](/uploads/7cb5ce1cb70e06262800c16f4c800157/qemu-and-dmesg.log)
[kernel.config](/uploads/327e9cec48a731abc9e44cb40db67de9/kernel.config)
|
