Bug 5535 - x11perf crashes the x server
Summary: x11perf crashes the x server
Status: RESOLVED DUPLICATE of bug 6325
Alias: None
Product: xorg
Classification: Unclassified
Component: Server/General (show other bugs)
Version: 6.9.0
Hardware: x86 (IA32) Linux (All)
: high major
Assignee: Adam Jackson
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2006-01-07 10:07 UTC by Florin Iucha
Modified: 2006-04-03 14:31 UTC (History)
0 users

See Also:
i915 platform:
i915 features:


Attachments
The config file. (15.93 KB, text/plain)
2006-01-07 10:09 UTC, Florin Iucha
no flags Details
The log file. (27.62 KB, text/plain)
2006-01-07 10:10 UTC, Florin Iucha
no flags Details
XFree86 4.2 x11perf output (156.20 KB, text/plain)
2006-01-07 10:10 UTC, Florin Iucha
no flags Details
XOrg 6.9.0 x11perf output (83.59 KB, text/plain)
2006-01-07 10:11 UTC, Florin Iucha
no flags Details
Xorg log from Xorg 6.9 compiled with gcc 4.0.2 (27.47 KB, text/plain)
2006-01-11 09:51 UTC, Florin Iucha
no flags Details
x11perf 4.2 running on xorg 6.9 (156.24 KB, text/plain)
2006-01-12 02:06 UTC, Florin Iucha
no flags Details

Description Florin Iucha 2006-01-07 10:07:18 UTC
Steps to reproduce:

   Start X using "xinit"
   Run "twm &"
   Move the initial xterm to the lower-right corner
   Run "x11perf -all > logfile"

The video hardware is intel 815 on-board graphics. The OS is redhat 7.3.

x11perf works fine with the "native" XFree86 4.2

XOrg 6.9 was compiled using the default compiler:
gcc version 2.96 20000731 (Red Hat Linux 7.3 2.96-113)

I will attach the xorg.conf, Xorg.0.log and the perf output.

BTM, many performance regression vs. 4.2...
Comment 1 Florin Iucha 2006-01-07 10:09:57 UTC
Created attachment 4270 [details]
The config file.
Comment 2 Florin Iucha 2006-01-07 10:10:18 UTC
Created attachment 4271 [details]
The log file.
Comment 3 Florin Iucha 2006-01-07 10:10:53 UTC
Created attachment 4272 [details]
XFree86 4.2 x11perf output
Comment 4 Florin Iucha 2006-01-07 10:11:17 UTC
Created attachment 4273 [details]
XOrg 6.9.0 x11perf output
Comment 5 Florin Iucha 2006-01-11 09:51:53 UTC
Created attachment 4322 [details]
Xorg log from Xorg 6.9 compiled with gcc 4.0.2
Comment 6 Florin Iucha 2006-01-12 02:05:15 UTC
I have run x11perf from XFree86 4.2 using the Xorg 6.9 distribution and it 
passed!

It seems that the x11perf distributed with Xorg 6.9 has a bug or tickles some 
new area in Xorg that has a bug.

On a side note, it seems that the x11perf.69 obtains scores 2-3% lower than x11
perf.42 when running on the same xorg.69 server!
Comment 7 Florin Iucha 2006-01-12 02:06:40 UTC
Created attachment 4329 [details]
x11perf 4.2 running on xorg 6.9
Comment 8 Ruben Nunez Francisco 2006-03-01 22:37:46 UTC
more info here:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=354721
Comment 9 Adam Jackson 2006-03-21 05:49:57 UTC
i can't see any reason why fbFetch_a4 would be undefined, but i'll take a look.
Comment 10 Ruben Nunez Francisco 2006-03-21 21:15:51 UTC
(In reply to comment #9)
> i can't see any reason why fbFetch_a4 would be undefined, but i'll take a look.
it's not fbFetch_a4, its some 8bpp format
from xc/programs/Xserver/render/picture.h:
#define PICT_FORMAT(bpp,type,a,r,g,b)   (((bpp) << 24) |  \
                                         ((type) << 16) | \
                                         ((a) << 12) | \
                                         ((r) << 8) | \
                                         ((g) << 4) | \
                                         ((b)))

#define PICT_a4         PICT_FORMAT(4,PICT_TYPE_A,4,0,0,0)

the default case is taken in fetchProcForPicture(...) and NULL is returned

the format before the crash:
(gdb) p *pMask
$1 = {pDrawable = 0xb16f8008, pFormat = 0x824b968, format = 134299648,
refcnt = 1, id = 2097161, pNext = 0x0, repeat = 0,
  graphicsExposures = 0, subWindowMode = 0, polyEdge = 0, polyMode = 0,
freeCompClip = 1, clientClipType = 0,
  componentAlpha = 0, repeatType = 0, unused = 0, alphaMap = 0x0,
alphaOrigin = {x = 0, y = 0}, clipOrigin = {x = 0, y = 0},
  clientClip = 0x0, dither = 0, stateChanges = 0, serialNumber = 46,
pCompositeClip = 0x83e5118, devPrivates = 0x84d711c,
  transform = 0x0, filter = 0, filter_params = 0x0, filter_nparams = 0,
pSourcePict = 0x0}
(gdb) p/x pMask->format
$2 = 0x8014000
Comment 11 Adam Jackson 2006-04-04 07:31:25 UTC

*** This bug has been marked as a duplicate of 6325 ***


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.