Skip to content

Suppress sysrq Kernel panics for Kdump tests #4473

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
SRIKKANTH wants to merge 7 commits into
base: main
Choose a base branch
from
Open

Suppress sysrq Kernel panics for Kdump tests #4473

Changes from 1 commit
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
3df958b
suppress sysrq
SRIKKANTH May 8, 2026
d7c960b
print log about sysrq panic suppresssion
SRIKKANTH May 11, 2026
23714ee
Added Sample sysrq generated crash from Ubuntu 24.04
SRIKKANTH May 11, 2026
eb15160
flake8 fixes
SRIKKANTH May 11, 2026
8848678
address comments
SRIKKANTH May 12, 2026
0e9db01
Revert "address comments"
SRIKKANTH May 13, 2026
9bcb0cf
Update kdump.py
SRIKKANTH May 13, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
26 changes: 26 additions & 0 deletions lisa/tools/kdump.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -898,6 +898,32 @@ def kdump_test(
# We should clean up the vmcore file since the test is passed
self.node.execute(f"rm -rf {kdump.dump_path}/*", shell=True, sudo=True)

# The sysrq-triggered panic is intentional for this test. Suppress the
# post-case panic check so it doesn't flag the expected crash as a
# failure when SerialConsole.check_panic re-scans the boot diagnostics.
self._suppress_expected_sysrq_panic()
Comment thread
SRIKKANTH marked this conversation as resolved.

Comment thread
SRIKKANTH marked this conversation as resolved.
def _suppress_expected_sysrq_panic(self) -> None:
from lisa.features import SerialConsole

if not self.node.features.is_supported(SerialConsole):
return
serial_console = self.node.features[SerialConsole]
# Patterns matching the panic this test intentionally triggered via sysrq.
expected_patterns: List[re.Pattern[str]] = [
re.compile(
r"^(.*Kernel panic - not syncing: sysrq triggered crash.*)$",
re.MULTILINE,
),
re.compile(r"^(.*sysrq: SysRq : Trigger a crash.*)$", re.MULTILINE),
Comment thread
SRIKKANTH marked this conversation as resolved.
# The RIP line accompanies the sysrq-triggered panic.
re.compile(r"^(.*RIP:.*)$", re.MULTILINE),
Copy link
Copy Markdown
Collaborator

@LiliDeng LiliDeng May 11, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@SRIKKANTH RIP is too broad

Copy link
Copy Markdown
Collaborator Author

@SRIKKANTH SRIKKANTH May 11, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There is not much to compare after RIP.
[ 17.532453] RIP: 0033:0x75f94711c5a4

Copy link
Copy Markdown
Collaborator

@vyadavmsft vyadavmsft May 11, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

RIP is too broad , kdump is being run by many teams and with this change we will start getting false panic detection . i had discussion with Vijay/Tyle some months ago regarding kernel panic detection and from meeting it was decided we should not add overly broad detectiom and every kernel panic detection change should go through kernel team for review .

Copy link
Copy Markdown
Collaborator

@vyadavmsft vyadavmsft May 11, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You can do like this :
re.compile(
r"(?ms)^.Kernel panic - not syncing: sysrq triggered crash.\n"
r"(?:^.\n){0,80}?"
r"^.sysrq_handle_crash.\n"
r"(?:^.\n){0,80}?"
r"^(.RIP: 0033:.)$"
)

It does not ignore every RIP: line.
It does not require sysrq_handle_crash to be on the RIP: line.
It scopes the ignored RIP: 0033: to a panic block that is clearly the intentional sysrq-triggered kdump crash.
It still returns only the RIP: line as the ignored candidate, so it fits the current behavior.

Copy link
Copy Markdown
Collaborator Author

@SRIKKANTH SRIKKANTH May 13, 2026 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@vyadavmsft ,
Given regex is not catching the sysrq panic.

Added '0033:' which will restrict the regex only to user space generated kernel panics like 'echo c > /proc/sysrq-trigger'
re.compile(r"^(.*RIP: 0033:.*)$", re.MULTILINE),

Copy link
Copy Markdown
Collaborator Author

@SRIKKANTH SRIKKANTH May 20, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@LiliDeng , @johnsongeorge-w please approve this PR if there are no more questions.

Copy link
Copy Markdown
Collaborator

@LiliDeng LiliDeng May 21, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@SRIKKANTH RIP: 0033:.* is not only for the sysrq kernel crash.

Copy link
Copy Markdown
Collaborator Author

@SRIKKANTH SRIKKANTH May 21, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

RIP: 0033:.* says the crash was generated by user space application. Any kernel or hardware generated crashes are not covered by it. I think its safe to have this one.

Copy link
Copy Markdown
Collaborator

@kanchansenlaskar kanchansenlaskar May 11, 2026 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

re.compile(r"^(.*RIP:.sysrq_handle_crash.)$", re.MULTILINE) re.compile(r"^(.*RIP:.*)$", re.MULTILINE) This pattern matches any RIP: line in the entire console log, not just the one from the sysrq crash. If a genuine, unrelated kernel panic also produces a RIP: line (which they almost always do), it would be silently ignored. Please scope this to the sysrq crash handler:
re.compile(r"^(.*RIP:.*sysrq_handle_crash.*)$", re.MULTILINE)

fixed the typo

Copy link
Copy Markdown
Collaborator Author

@SRIKKANTH SRIKKANTH May 11, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Sample sysrq generated kernel panic. There is not much to compare after RIP.

[   17.519224] sysrq: Trigger a crash
[   17.519617] Kernel panic - not syncing: sysrq triggered crash
[   17.519964] CPU: 18 UID: 0 PID: 8948 Comm: echo Kdump: loaded Not tainted 6.17.0-1008-azure #8~24.04.1-Ubuntu VOLUNTARY 
[   17.520541] Hardware name: Microsoft Corporation Virtual Machine/Virtual Machine, BIOS Hyper-V UEFI Release v4.1 02/25/2026
[   17.521245] Call Trace:
[   17.521366]  <TASK>
[   17.521510]  dump_stack_lvl+0x27/0x70
[   17.521758]  dump_stack+0x10/0x20
[   17.521996]  vpanic+0x31f/0x3b0
[   17.522236]  panic+0x5f/0x60
[   17.522468]  sysrq_handle_crash+0x15/0x20
[   17.522726]  __handle_sysrq+0xe0/0x250
[   17.522931]  write_sysrq_trigger+0x5c/0x80
[   17.523182]  proc_reg_write+0x5e/0xa0
[   17.523443]  ? __cond_resched+0x1a/0x50
[   17.523654]  vfs_write+0xf9/0x440
[   17.523868]  ? srso_return_thunk+0x5/0x5f
[   17.524077]  ? xas_load+0x17/0x100
[   17.524290]  ? get_page_from_freelist+0x471/0x6c0
[   17.524601]  ? srso_return_thunk+0x5/0x5f
[   17.524808]  ? __cond_resched+0x1a/0x50
[   17.525025]  ? srso_return_thunk+0x5/0x5f
[   17.525234]  ? mutex_lock+0x12/0x40
[   17.525444]  ksys_write+0x71/0xf0
[   17.525625]  __x64_sys_write+0x19/0x20
[   17.525791]  x64_sys_call+0x79/0x20d0
[   17.526007]  do_syscall_64+0x7b/0xb70
[   17.526173]  ? srso_return_thunk+0x5/0x5f
[   17.526487]  ? cp_new_stat+0x141/0x170
[   17.526720]  ? srso_return_thunk+0x5/0x5f
[   17.526958]  ? __do_sys_newfstat+0x4c/0x80
[   17.527207]  ? srso_return_thunk+0x5/0x5f
[   17.527499]  ? srso_return_thunk+0x5/0x5f
[   17.527781]  ? arch_exit_to_user_mode_prepare.isra.0+0xd/0xc0
[   17.528261]  ? srso_return_thunk+0x5/0x5f
[   17.528528]  ? do_syscall_64+0xad/0xb70
[   17.528740]  ? srso_return_thunk+0x5/0x5f
[   17.528996]  ? count_memcg_events+0xba/0x1a0
[   17.529252]  ? srso_return_thunk+0x5/0x5f
[   17.529472]  ? handle_mm_fault+0x1d3/0x2d0
[   17.529677]  ? srso_return_thunk+0x5/0x5f
[   17.529939]  ? do_user_addr_fault+0x1b9/0x860
[   17.530191]  ? srso_return_thunk+0x5/0x5f
[   17.530460]  ? arch_exit_to_user_mode_prepare.isra.0+0xd/0xe0
[   17.530811]  ? srso_return_thunk+0x5/0x5f
[   17.531025]  ? irqentry_exit_to_user_mode+0x2d/0x1b0
[   17.531283]  ? srso_return_thunk+0x5/0x5f
[   17.531443]  ? irqentry_exit+0x1d/0x30
[   17.531705]  ? srso_return_thunk+0x5/0x5f
[   17.531957]  ? exc_page_fault+0x84/0x150
[   17.532196]  entry_SYSCALL_64_after_hwframe+0x76/0x7e
[   17.532453] RIP: 0033:0x75f94711c5a4
[   17.532661] Code: c7 00 16 00 00 00 b8 ff ff ff ff c3 66 2e 0f 1f 84 00 00 00 00 00 f3 0f 1e fa 80 3d a5 ea 0e 00 00 74 13 b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 54 c3 0f 1f 00 55 48 89 e5 48 83 ec 20 48 89
[   17.533822] RSP: 002b:00007ffe3b627158 EFLAGS: 00000202 ORIG_RAX: 0000000000000001
[   17.534173] RAX: ffffffffffffffda RBX: 0000000000000002 RCX: 000075f94711c5a4
[   17.534609] RDX: 0000000000000002 RSI: 000063539a6b10b0 RDI: 0000000000000001
[   17.535042] RBP: 00007ffe3b627180 R08: 0000000000000000 R09: 0000000000000410
[   17.535587] R10: 0000000000000001 R11: 0000000000000202 R12: 0000000000000002
[   17.536028] R13: 000063539a6b10b0 R14: 000075f9472045c0 R15: 000075f947201ee0
[   17.536469]  </TASK>
[   17.542975] Kernel Offset: 0x28200000 from 0xffffffff81000000 (relocation range: 0xffffffff80000000-0xffffffffbfffffff)

Copy link
Copy Markdown
Collaborator

@kanchansenlaskar kanchansenlaskar May 11, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There was a typo in my comment, corrected it, now check

Copy link
Copy Markdown
Collaborator

@kanchansenlaskar kanchansenlaskar May 11, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Please scope this to the sysrq crash handler:
re.compile(r"^(.*RIP:.*sysrq_handle_crash.*)$", re.MULTILINE)

Copy link
Copy Markdown
Collaborator

@kanchansenlaskar kanchansenlaskar May 11, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

can you try changing it
from:
re.compile(r"^(.*RIP:.*)$", re.MULTILINE)
to:
re.compile(r"^(.*RIP: 0033:.*)$", re.MULTILINE)

]
# Shadow the class attribute on the instance so other nodes are unaffected.
existing = list(serial_console.panic_ignorable_patterns)
existing.extend(expected_patterns)
serial_console.panic_ignorable_patterns = existing
Comment thread
kanchansenlaskar marked this conversation as resolved.

def trigger_kdump_on_specified_cpu(self, cpu_num: int, log_path: Path) -> None:
lscpu = self.node.tools[Lscpu]
thread_count = lscpu.get_thread_count()
Expand Down
Loading