Bug 42300 - RRGetScreenInfo possibly takes a long time to complete on i915
Summary: RRGetScreenInfo possibly takes a long time to complete on i915
Status: RESOLVED INVALID
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/intel (show other bugs)
Version: git
Hardware: x86 (IA32) Linux (All)
: medium normal
Assignee: Eugeni Dodonov
QA Contact: Xorg Project Team
URL:
Whiteboard: 2011BRB_Reviewed
Keywords:
Depends on:
Blocks:
 
Reported: 2011-10-26 23:58 UTC by Knut Petersen
Modified: 2012-10-21 14:30 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:


Attachments
xorg log with errmsg and backtrace (28.54 KB, text/plain)
2011-10-26 23:58 UTC, Knut Petersen
no flags Details

Description Knut Petersen 2011-10-26 23:58:44 UTC
Created attachment 52813 [details]
xorg log with errmsg and backtrace

Got this problem once, tried to reproduce it but failed.

Hardware: i915GM on AOpen i915GMm-hfs motherboard, Pentium M (Dothan)
Software: linux kernel 3.1, xorg git 10-26-2011

[...]
[  4758.759] [mi] EQ overflowing.  Additional events will be discarded until existing events are processed.
[  4758.759] 
Backtrace:
[  4758.759] 0: /usr/bin/Xorg (xorg_backtrace+0x2c) [0x81d5b5c]
[  4758.759] 1: /usr/bin/Xorg (mieqEnqueue+0xdb) [0x81b5552]
[  4758.759] 2: /usr/bin/Xorg (0x8048000+0x4bcac) [0x8093cac]
[  4758.759] 3: /usr/bin/Xorg (QueuePointerEvents+0x54) [0x80942be]
[  4758.759] 4: /usr/bin/Xorg (xf86PostMotionEventM+0x190) [0x80cda1a]
[  4758.759] 5: /usr/lib/xorg/modules/input/evdev_drv.so (0xb6ed4000+0x3e37) [0xb6ed7e37]
[  4758.759] 6: /usr/lib/xorg/modules/input/evdev_drv.so (0xb6ed4000+0x41c9) [0xb6ed81c9]
[  4758.759] 7: /usr/lib/xorg/modules/input/evdev_drv.so (0xb6ed4000+0x4350) [0xb6ed8350]
[  4758.759] 8: /usr/lib/xorg/modules/input/evdev_drv.so (0xb6ed4000+0x449a) [0xb6ed849a]
[  4758.760] 9: /usr/bin/Xorg (0x8048000+0x74bda) [0x80bcbda]
[  4758.760] 10: /usr/bin/Xorg (0x8048000+0x998f5) [0x80e18f5]
[  4758.760] 11: (vdso) (__kernel_sigreturn+0x0) [0xb77dd400]
[  4758.760] 12: /usr/lib/libdrm.so.2 (drmModeGetConnector+0x14c) [0xb7064d1a]
[  4758.760] 13: /usr/lib/xorg/modules/drivers/intel_drv.so (0xb6ee2000+0xa908) [0xb6eec908]
[  4758.760] 14: /usr/bin/Xorg (xf86ProbeOutputModes+0xd9) [0x80f4f79]
[  4758.760] 15: /usr/bin/Xorg (0x8048000+0xb9a29) [0x8101a29]
[  4758.760] 16: /usr/bin/Xorg (RRGetInfo+0xd2) [0x8135fb7]
[  4758.760] 17: /usr/bin/Xorg (ProcRRGetScreenInfo+0x9c) [0x813b745]
[  4758.760] 18: /usr/bin/Xorg (0x8048000+0xe9f8b) [0x8131f8b]
[  4758.760] 19: /usr/bin/Xorg (0x8048000+0x2af82) [0x8072f82]
[  4758.760] 20: /usr/bin/Xorg (0x8048000+0x1e46f) [0x806646f]
[  4758.760] 21: /lib/libc.so.6 (__libc_start_main+0xfe) [0xb7105c2e]
[  4758.760] [mi] These backtraces from mieqEnqueue may point to a culprit higher up the stack.
[  4758.760] [mi] mieq is *NOT* the cause.  It is a victim.
[  4759.107] [mi] Increasing EQ size to 512 to prevent dropped events.
[  4759.107] [mi] EQ processing has resumed after 25 dropped events.
[  4759.107] [mi] This may be caused my a misbehaving driver monopolizing the server's resources.
[...]

Knut
Comment 1 Jeremy Huddleston Sequoia 2011-10-31 16:22:42 UTC
Comment on attachment 52813 [details]
xorg log with errmsg and backtrace

Please remember to set the correct mime type for attachments.
Comment 2 Jeremy Huddleston Sequoia 2011-10-31 16:29:35 UTC
The mieq code is behaving as intended there.  The server dropped 25 events on the floor because the buffer wasn't large enough to hold them.  It then resized the buffed to hold more events when it became active again.

Over to intel to investigate what was possibly taking so long in handling RRGetScreenInfo
Comment 3 Paulo Zanoni 2011-11-10 03:45:24 UTC
Dupe of bug #41059? It has patches floating on intel-gfx@ mailing list.
Comment 4 Chris Wilson 2011-11-10 03:52:03 UTC
(In reply to comment #3)
> Dupe of bug #41059? It has patches floating on intel-gfx@ mailing list.

I believe so, once 41059 is resolved it will be useful to see what remains. 915GM has other problems such as requiring long load detection on CRT outputs which is also slow.
Comment 5 Chris Wilson 2012-04-03 13:36:28 UTC
Bumping to needinfo as the slow EDID read fixes have landed upstream and it is now useful to retest.
Comment 6 Chris Wilson 2012-10-21 14:30:33 UTC
Timeout. Please do reopen if you can still reproduce the issue and help us diagnose the problem, thanks.


Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.