On sobota, 31 maja 2008, Michel Dänzer wrote:
> On Sat, 2008-05-31 at 09:07 +0200, [EMAIL PROTECTED] wrote:
> > On wtorek, 27 maja 2008, Michel Dänzer wrote:
> > > On Mon, 2008-05-26 at 19:18 +0200, [EMAIL PROTECTED] wrote:
> > > > On poniedziałek, 26 maja 2008 you wrote:
> > > > > The EXA offscreen area may be too small. Does Option
> > > > > "FBTexPercent" "0" or reducing the maximum desktop size with
> > > > > the Virtual directive help?
> > > >
> > > > I forgot to mention it, but xcompmgr spits out some misterious
> > > > error messages, maybe they can help:
> > > >
> > > > [EMAIL PROTECTED] ~ $ xcompmgr -cf
> > > > error 3 request 20 minor 0 serial 845855
> > >
> > > I think these are harmless.
> > >
> > > It would be interesting to know where the X server is spending
> > > 40% of the CPU cycles.
> >
> > It sure would. Is there a way for a mere mortal like me to compile
> > xorg with some sort of profiling (preferably under Gentoo) or are
> > there some other ways ?
>
> Something like sysprof or oprofile is preferable.

I`ve run oprofile and gathered some information when the xcompmgr -cfF was 
running. 
However, I wasn`t able to reproduce the situation from before with xorg taking 
40 % CPU cycles. Still, here is the top of the report:

[EMAIL PROTECTED] /usr/share/doc/oprofile $ opreport -p 
/lib/modules/2.6.25-idaho/kernel -l | head -40
warning: /truecrypt could not be found.
warning: [vdso] (tgid:3959 range:0xb8046000-0xb8047000) could not be found.
warning: [vdso] (tgid:4043 range:0xb8053000-0xb8054000) could not be found.
warning: [vdso] (tgid:4060 range:0xb802d000-0xb802e000) could not be found.
warning: [vdso] (tgid:4664 range:0xb7fd8000-0xb7fd9000) could not be found.
warning: [vdso] (tgid:5077 range:0xb80c3000-0xb80c4000) could not be found.
CPU: CPU with timer interrupt, speed 0 MHz (estimated)
Profiling through timer interrupt
warning: could not check that the binary file 
/lib/modules/2.6.25-idaho/kernel/fs/ext3/ext3.ko has not been modified 
since the profile was taken. Results may be inaccurate.
warning: could not check that the binary file 
/lib/modules/2.6.25-idaho/kernel/fs/jbd/jbd.ko has not been modified since 
the profile was taken. Results may be inaccurate.
warning: could not check that the binary file 
/lib/modules/2.6.25-idaho/kernel/drivers/acpi/processor.ko has not been 
modified since the profile was taken. Results may be inaccurate.
warning: could not check that the binary file 
/lib/modules/2.6.25-idaho/kernel/drivers/char/drm/radeon.ko has not been 
modified since the profile was taken. Results may be inaccurate.
samples  %        image name               app name                 symbol name
44822    71.3272  processor.ko             processor                
acpi_processor_idle
3671      5.8418  libc-2.7.so              libc-2.7.so              memcpy
3590      5.7129  libc-2.7.so              libc-2.7.so              memmove
844       1.3431  vmlinux                  vmlinux                  
native_read_tsc
538       0.8561  libexa.so                libexa.so                
ExaOffscreenMarkUsed
485       0.7718  opreport                 opreport                 calc_crc32
370       0.5888  radeon.ko                radeon                   
radeon_do_wait_for_idle
218       0.3469  libc-2.7.so              libc-2.7.so              _int_malloc
154       0.2451  vmlinux                  vmlinux                  
__copy_to_user_ll
154       0.2451  vmlinux                  vmlinux                  
_spin_unlock_irqrestore
151       0.2403  libpixman-1.so.0.10.0    libpixman-1.so.0.10.0    
pixman_blt_mmx
145       0.2307  libpixman-1.so.0.10.0    libpixman-1.so.0.10.0    pixman_op
141       0.2244  eagle                    eagle                    (no symbols)
139       0.2212  vmlinux                  vmlinux                  
_spin_unlock_irq
134       0.2132  libpixman-1.so.0.10.0    libpixman-1.so.0.10.0    expand8888
129       0.2053  vmlinux                  vmlinux                  delay_tsc
121       0.1926  vmlinux                  vmlinux                  
get_page_from_freelist
114       0.1814  radeon_drv.so            radeon_drv.so            
RadeonCompositeCP
109       0.1735  libpixman-1.so.0.10.0    libpixman-1.so.0.10.0    in_over
93        0.1480  libc-2.7.so              libc-2.7.so              _int_free
93        0.1480  truecrypt                truecrypt                (no symbols)
91        0.1448  libXfont.so.1.4.1        libXfont.so.1.4.1        PatternMatch
86        0.1369  libexa.so                libexa.so                
exaTryDriverComposite
84        0.1337  libexa.so                libexa.so                exaComposite
80        0.1273  libexa.so                libexa.so                
exaPixmapIsOffscreen
75        0.1194  radeon.ko                radeon                   
radeon_cp_indirect
69        0.1098  libexa.so                libexa.so                exaGlyphs
69        0.1098  radeon_drv.so            radeon_drv.so            
R200TextureSetupCP
66        0.1050  libc-2.7.so              libc-2.7.so              free
66        0.1050  libc-2.7.so              libc-2.7.so              malloc
64        0.1018  libpixman-1.so.0.10.0    libpixman-1.so.0.10.0    over
54        0.0859  libc-2.7.so              libc-2.7.so              
malloc_consolidate
53        0.0843  vmlinux                  vmlinux                  
search_by_key
50        0.0796  vmlinux                  vmlinux                  
__link_path_walk
49        0.0780  radeon.ko                radeon                   
radeon_freelist_get
49        0.0780  radeon_drv.so            radeon_drv.so            
R200PrepareCompositeCP
46        0.0732  libexa.so                libexa.so                
exaGetPixmapOffset

Is there anything, that can be deduced from this dump ? I see, that libexa.a is
the biggest hog but I leave the analysis to the experts ;).

Best regards,
-- 
Maciej Grela

Attachment: oprofile-profile.gz
Description: GNU Zip compressed data

_______________________________________________
xorg-driver-ati mailing list
[email protected]
http://lists.x.org/mailman/listinfo/xorg-driver-ati

Reply via email to