Bug 457245

Summary: kernel oops + lockup in utrace at gdm startup
Product: [Fedora] Fedora Reporter: Will Woods <wwoods>
Component: kernelAssignee: Roland McGrath <roland>
Status: CLOSED RAWHIDE QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: low Docs Contact:
Priority: low    
Version: rawhideCC: dcantrell, kernel-maint
Target Milestone: ---   
Target Release: ---   
Hardware: All   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2008-10-05 17:51:51 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Bug Depends On:    
Bug Blocks: 438943    

Description Will Woods 2008-07-30 14:25:38 UTC
Current rawhide - kernel 2.6.27-0.186.rc0.git15.fc10

At gdm startup, something goes wrong on my system. gdb gets started - presumably
to trace the problem - and then the kernel emits an oops in
get_utrace_lock_attached. 

The system is then soft-locked. 60s later I get a BUG: soft lockup from gdm-binary. 

Unfortunately, since the system is locked, it's hard to get full oops logs.
Here's a summary of the soft lockup trace:

__delay
_raw_spin_lock
_spin_lock
finish_callback
? ptrace_report_signal
utrace_get_signal
? check_object
? ptrace_report_signal
get_signal_to_deliver
? trace_hardirqs_on
do_notify_resume
? remove_wait_queue
? trace_hardirqs_on_caller
? trace_hardirqs_on
? do_wait
? default_wake_function
? path_put
work_notifysig

I'll try to get info on the original Oops, but it's tricky, since I only get 60
seconds to copy info down before the soft lockup BUG is triggered.

Comment 1 Will Woods 2008-07-30 14:29:58 UTC
Here's what I could copy down in those 60s:

utrace_control
ptrace_detach_utrace
ptrace_attach
sys_ptrace
syscall_call
__mutex_unlock_slowpath

EIP: [<c045fa72>] get_utrace_lock_attached+0x5e/0x91 SS:ESP 0068:f2b89f44

Comment 2 Will Woods 2008-07-30 14:46:57 UTC
kernel 2.6.27-0.166.rc0.git8.fc10 (the kernel currently tagged for F10Alpha)
does not have this bug. 

The gdm crash that triggers this seems to have been brought on by an SELinux
denial. Booting with 'enforcing=0' works fine.

Comment 3 Jesse Keating 2008-10-03 23:36:25 UTC
So can this be closed, or is it still happening?

Comment 4 Will Woods 2008-10-04 02:51:21 UTC
Well, the kernel bug that was triggering this is fixed, so it's hard to tell if the gdb + utrace lockup is still around.

Still, I tried killing Xorg off while gdm was up (killall -SEGV Xorg) and nothing bad happens, and there have been plenty of other X crashes that didn't cause anything like this. So unless someone came up with a better way to reproduce the old trigger, I'd consider it fixed.