Description of problem: when I startup firefox 1.5.0.9 in windowmaker, Xorg needs 5s CPU time (Pentium D) to redraw or whatever it's doing till I can start using firefox. also, when I change workspace and go back to workspsace where firefox is located, 5s CPU time is needed for something again. earlier version 1.1.1-57 needed around 0.01s CPU time. Version-Release number of selected component (if applicable): 1.2.0-1 How reproducible: always Steps to Reproduce: 1. startup firefox 2. 3. Actual results: laggy as ****, when I move window it refreshes 5 times a second. Expected results: working like earlier (last 10 years?). Additional info: Xorg.0.log attached. other programs do not cause this kind of CPU usage...
Created attachment 146432 [details] Xorg.0.log
Totally screwed. Starting up command "alsamixer -V all" in gnome-terminal: 15s CPU time in Xorg. Starting up command "alsamixer -V all" in xterm: ~0s CPU time in Xorg.
Thanks for the bug report. We have reviewed the information you have provided above, and there is some additional information we require that will be helpful in our diagnosis of this issue. Please attach your X server config file (/etc/X11/xorg.conf) and X server log file (/var/log/Xorg.*.log) to the bug report as individual uncompressed file attachments using the bugzilla file attachment link below. Could you please also try to run without any /etc/X11/xorg.conf whatsoever and let X11 autodetect your display and video card? Attach to this bug /var/log/Xorg.0.log from this attempt as well, please. We will review this issue again once you've had a chance to attach this information. Thanks in advance.
Created attachment 146906 [details] xorg.conf don't you have bunzip2?
I can't reproduce the slowdown since now when I try to startup Xorg it stalls after starting wmaker and I have to press sysrq+k. It works when I disable EXA: # Option "AccelMethod" "EXA" But now with 1.2.0 redrawing isn't that great, but I don't get the 5s stalls now. screencap of firefox, after I opened Tools menu and closed it. xrefresh does not make any diff.
Created attachment 146920 [details] corrupted garbage in screencap
Created attachment 146922 [details] startup log with EXA
Created attachment 146923 [details] using XAA - works but some corruption
Funny that when xephem is running, firefox skin is drawn just okay.
I wasted some hours playing with different broken versions of software, here the matrix: +-----+-----+ | XAA | EXA | -------------+-----+-----+ xorg 1.1.1 a | OK | OK | xorg 1.2.0 a | OK x| 2) | xorg 1.1.1 b | ?) | 1) | xorg 1.2.0 b | 3) | 4) | -------------+-----------+ x) currently using ?) haven't tried a) i810 modesetting from 20070103 b) i810 modesetting from 20070130 1) https://bugs.freedesktop.org/show_bug.cgi?id=9637 2) this is the 5s-delay-when-switching-to-firefox 3) the firefox redraw problem in screencap 4) crash, sysrq+k needed
Wow. 20070202 modesetting works in XAA, but EXA is a bit broken. Fonts do not show up in e.g. wmaker menus or firefox. (II) intel(0): Successful allocation of EXA offscreen memory at 0x660000, size 22560 KB (II) intel(0): Allocated 64 kB for the G965 exa state buffer at 0xbff0000 - 0xc000000. (EE) EXA(0): driver's EXA version requirements (2.1) are incompatible with EXA version (2.0) (II) intel(0): EXA initialization failed; trying older version
Now with server 1.2.99.902 Xorg waster 0.5s of CPU time when switching workspace in wmaker. using 1.9.92 intel_drv.so with EXA. later versions do not work (undefined reference to xf86_cursors_init). XAA segfaults: --- snip --- (II) intel(0): [dri] visual configs initialized (II) intel(0): Page Flipping disabled (==) intel(0): Write-combining range (0x40000000,0x10000000) (II) intel(0): vgaHWGetIOBase: hwp->IOBase is 0x03d0, hwp->PIOOffset is 0x0000 Backtrace: 0: X(xf86SigHandler+0x85) [0x80be837] 1: X(AddScreen+0x1ff) [0x807024d] Fatal server error: Caught signal 11. Server aborting --- snip --- profile when I switched workspace 20 times. CPU: P4 / Xeon, speed 2797.42 MHz (estimated) Counted GLOBAL_POWER_EVENTS events (time during which processor is not stopped) with a unit mask of 0x01 (mandatory) count 45000 Counted FSB_DATA_ACTIVITY events (DRDY or DBSY events on the front side bus) with a unit mask of 0x03 (multiple flags) count 45000 Counted BRANCH_RETIRED events (retired branches) with a unit mask of 0x05 (multiple flags) count 45000 Counted BRANCH_RETIRED events (retired branches) with a unit mask of 0x0a (multiple flags) count 45000 samples % samples % samples % samples % linenr info image name app name symbol name 722165 58.9155 5632 53.4497 1670 7.9904 19 2.9688 (no location information) libc-2.5.so libc-2.5.so memcpy 154260 12.5848 24 0.2278 1341 6.4163 7 1.0938 delay.c:42 vmlinux vmlinux delay_tsc 21432 1.7485 182 1.7272 74 0.3541 1 0.1562 (no location information) intel_drv.so intel_drv.so i965_prepare_composite 10988 0.8964 1014 9.6232 43 0.2057 1 0.1562 page_alloc.c:578 vmlinux vmlinux prep_new_page 7390 0.6029 22 0.2088 18 0.0861 0 0 (no location information) vmlinux vmlinux page_fault 6230 0.5083 29 0.2752 82 0.3923 5 0.7812 (no location information) intel_drv.so intel_drv.so I830WaitLpRing 5360 0.4373 17 0.1613 34 0.1627 1 0.1562 time.c:165 vmlinux vmlinux timer_interrupt 5202 0.4244 3 0.0285 23 0.1100 1 0.1562 (no location information) anon (tgid:12410 range:0xeb2000-0xeb3000) Xorg (no symbols) 5117 0.4175 95 0.9016 105 0.5024 0 0 usercopy.c:717 vmlinux vmlinux __copy_to_user_ll 4682 0.3820 187 1.7747 14 0.0670 0 0 highmem.c:89 vmlinux vmlinux kunmap_atomic 4321 0.3525 74 0.7023 231 1.1053 3 0.4688 (no location information) oprofiled oprofiled odb_update_node 3845 0.3137 73 0.6928 19 0.0909 0 0 process.c:252 vmlinux vmlinux mwait_idle_with_hints 3685 0.3006 17 0.1613 197 0.9426 30 4.6875 (no location information) libc-2.5.so libc-2.5.so _int_malloc 3431 0.2799 9 0.0854 310 1.4833 21 3.2812 (no location information) Xorg Xorg miRegionOp that was profile for all... now for Xorg. CPU: P4 / Xeon, speed 2797.42 MHz (estimated) Counted GLOBAL_POWER_EVENTS events (time during which processor is not stopped) with a unit mask of 0x01 (mandatory) count 45000 Counted FSB_DATA_ACTIVITY events (DRDY or DBSY events on the front side bus) with a unit mask of 0x03 (multiple flags) count 45000 Counted BRANCH_RETIRED events (retired branches) with a unit mask of 0x05 (multiple flags) count 45000 Counted BRANCH_RETIRED events (retired branches) with a unit mask of 0x0a (multiple flags) count 45000 samples % samples % samples % samples % linenr info image name symbol name 5202 20.7714 3 1.6575 23 1.7385 1 1.4085 (no location information) anon (tgid:12410 range:0xeb2000-0xeb3000) (no symbols) 3431 13.6999 9 4.9724 310 23.4316 21 29.5775 (no location information) Xorg miRegionOp 1089 4.3483 12 6.6298 69 5.2154 2 2.8169 (no location information) Xorg miUnion 806 3.2183 3 1.6575 83 6.2736 6 8.4507 (no location information) Xorg miUnionO 572 2.2840 6 3.3149 51 3.8549 4 5.6338 (no location information) Xorg ReadRequestFromClient 511 2.0404 7 3.8674 18 1.3605 1 1.4085 (no location information) Xorg Dispatch 473 1.8887 4 2.2099 43 3.2502 2 2.8169 (no location information) Xorg XaceHook 461 1.8408 3 1.6575 19 1.4361 1 1.4085 (no location information) Xorg WaitForSomething 460 1.8368 9 4.9724 23 1.7385 2 2.8169 (no location information) Xorg SecurityLookupIDByType 455 1.8168 2 1.1050 26 1.9652 1 1.4085 (no location information) Xorg miComputeCompositeRegion 419 1.6731 0 0 32 2.4187 0 0 (no location information) Xorg __i686.get_pc_thunk.bx 404 1.6132 8 4.4199 16 1.2094 0 0 (no location information) Xorg FreeResource 397 1.5852 0 0 3 0.2268 0 0 (no location information) Xorg GetTimeInMillis 331 1.3217 5 2.7624 58 4.3840 0 0 (no location information) Xorg miRectIn
Created attachment 150586 [details] server 1.2.99.902 + intel_drv 1.9.92 + EXA
Having DRI on/off or GLX on/off makes not difference. Also, looks like only gtk apps are causing this burning of CPU cycles. When I cycle workspace between xterms, qt apps or motif apps, not much CPU power is needed.
currently EXA is not enabled by default, there are still some issues to be resolved with EXA before this can be done in Fedora, so can you quantify what the problem is with XAA? We are working on getting EXA to a more acceptable level.
*** This bug has been marked as a duplicate of 431340 ***