Bug 504510
Summary: | xorg intel driver 2.7.0.7.fc11 generates strange Xorg log messages | ||||||
---|---|---|---|---|---|---|---|
Product: | [Fedora] Fedora | Reporter: | Simone Deponti <simone.deponti> | ||||
Component: | xorg-x11-drv-intel | Assignee: | Kristian Høgsberg <krh> | ||||
Status: | CLOSED DUPLICATE | QA Contact: | Fedora Extras Quality Assurance <extras-qa> | ||||
Severity: | low | Docs Contact: | |||||
Priority: | low | ||||||
Version: | 11 | CC: | ajax, fabio.canepa, giulio.martinat, ismael, mcepl, toze, tselovan, twscofi, xgl-maint | ||||
Target Milestone: | --- | ||||||
Target Release: | --- | ||||||
Hardware: | i686 | ||||||
OS: | Linux | ||||||
Whiteboard: | |||||||
Fixed In Version: | Doc Type: | Bug Fix | |||||
Doc Text: | Story Points: | --- | |||||
Clone Of: | Environment: | ||||||
Last Closed: | 2009-10-27 17:44:08 UTC | Type: | --- | ||||
Regression: | --- | Mount Type: | --- | ||||
Documentation: | --- | CRM: | |||||
Verified Versions: | Category: | --- | |||||
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||
Cloudforms Team: | --- | Target Upstream Version: | |||||
Embargoed: | |||||||
Attachments: |
|
Okay, this http://bugs.freedesktop.org/show_bug.cgi?id=21246#c3 might be related. The mentioned patch went upstream though, and should be in the current kernel I have installed (went upstream into 2.6.29.3, I'm on 2.6.29.4). However, many other people report performance problems and they "disappeared" when switching to kernel 2.6.30 (http://beranger.org/v3/wordpress/2009/05/04/jaunty-kernel-2630-fixes-the-intel-video/). And in fact, 2.6.30 prepatch changelog does contain an awful lot of i915 fixes (http://kernel.org/pub/linux/kernel/v2.6/testing/ChangeLog-2.6.30-rc8), although my nonexistant knowledge doesn't allow me to say they solve the problem. An update to 2.6.30 might solve this. This bug appears to have been reported against 'rawhide' during the Fedora 11 development cycle. Changing version to '11'. More information and reason for this action is here: http://fedoraproject.org/wiki/BugZappers/HouseKeeping I'm running fedora 11 with the latest kernel from koji (2.6.29.6-211.fc11.x86_64). Inside /var/log/Xorg.0.log I see the error "(EE) intel(0): max_gtt_map_size: 60457kb" but now the performance is much more better!! Before this kernel I'd to disable tiling (Option "Tiling" "false" inside /etc/X11/xorg.conf) to get a decent performance with desktop effect enabled... (In reply to comment #3) > I'm running fedora 11 with the latest kernel from koji > (2.6.29.6-211.fc11.x86_64). > Inside /var/log/Xorg.0.log I see the error "(EE) intel(0): max_gtt_map_size: > 60457kb" but now the performance is much more better!! > Before this kernel I'd to disable tiling (Option "Tiling" "false" inside > /etc/X11/xorg.conf) to get a decent performance with desktop effect enabled... That error message is mostly harmless and red herring for most of the real problems. So, can we say that this bug has been fixed by the kernel package 2.6.29.6-211.fc11.x86_64? Much faster but generate a kernel oops: http://www.kerneloops.org/submitresult.php?number=509120 WARNING: at drivers/gpu/drm/i915/i915_gem_tiling.c:473 i915_gem_set_tiling+0x4a0/0x4f2 [i915]() (Not tainted) Hardware name: 900 failed to unbind object for tiling switchModules linked in: bridge stp llc bnep sco l2cap bluetooth sunrpc ip6t_REJECT nf_conntrack_ipv6 ip6table_filter ip6_tables ipv6 p4_clockmod ext2 dm_multipath uinput pcspkr joydev i2c_i801 iTCO_wdt iTCO_vendor_support uvcvideo videodev v4l1_compat usb_storage arc4 ecb ath5k snd_hda_codec_realtek mac80211 cfg80211 snd_hda_intel atl2 snd_hda_codec snd_hwdep snd_pcm snd_timer snd soundcore snd_page_alloc eeepc_laptop hwmon ata_generic pata_acpi i915 drm i2c_algo_bit i2c_core video output [last unloaded: scsi_wait_scan] Pid: 1799, comm: Xorg Not tainted 2.6.29.6-211.fc11.i586 #1 Call Trace: [<c042ebc6>] warn_slowpath+0x7c/0xa4 [<c0535db3>] ? selinux_capable+0xb4/0xc1 [<c0705a10>] ? printk+0x14/0x1c [<f7f42193>] ? i915_gem_object_unbind+0x43/0x1f9 [i915] [<f7dbb405>] ? drm_gem_object_reference+0xd/0xf [drm] [<f7f46441>] i915_gem_set_tiling+0x4a0/0x4f2 [i915] [<f7dba602>] drm_ioctl+0x202/0x292 [drm] [<f7f45fa1>] ? i915_gem_set_tiling+0x0/0x4f2 [i915] [<c0532919>] ? inode_has_perm+0x60/0x6a [<c04461db>] ? getnstimeofday+0x59/0xea [<c04aa494>] vfs_ioctl+0x5a/0x74 [<c04aad3d>] do_vfs_ioctl+0x480/0x4ba [<c0532ccf>] ? selinux_file_ioctl+0x3f/0x42 [<c04aadbd>] sys_ioctl+0x46/0x66 [<c0403f72>] syscall_call+0x7/0xb ---[ end trace 6f2ea718258f111d ]--- I didn't get the oops but I've booted only for some hours and I don't have selinux enabled (I see a selinux reference in the trace). I found kernel 2.6.29.6-211.fc11.x86_64 faster on my card (i945GM) yes but my setup is different: for once you are using xorg.conf and I'm not, second you are on x86_64 and I'm on i586 (no PAE), and third, it's roughly 50% more fast with the new kernel, but gave me the oops (although the system did not crash). Now I've delete the xorg.conf because it was created only to disable tiling and I still have a very good performance.But you are righ my setup it's different... I'm having a similar problem but in my case the X freezes and it freezes the keyboard, also. I am not able to do anything unless i reboot the machine or connect through ssh. Trace X: page allocation failure. order:1, mode:0x50d0 Pid: 5129, comm: X Not tainted 2.6.29.5-191.fc11.i686.PAE #1 Call Trace: [<c0714776>] ? printk+0x14/0x16 [<c04840ff>] __alloc_pages_internal+0x36b/0x386 [<c048415e>] __get_free_pages+0x19/0x29 [<c04a2f87>] __kmalloc+0x2d/0x11d [<c05df34e>] agp_alloc_page_array+0x2a/0x45 [<c05df3a7>] agp_generic_alloc_user+0x3e/0xa1 [<c05df4bb>] agp_allocate_memory+0x3c/0xc1 [<f81bf37d>] drm_agp_allocate_memory+0xd/0xf [drm] [<f81bf4bc>] drm_agp_bind_pages+0x3e/0xb0 [drm] [<f82b63ea>] i915_gem_object_bind_to_gtt+0x1a7/0x20a [i915] [<f82b646e>] i915_gem_object_pin+0x21/0x139 [i915] [<f81bb895>] ? drm_gem_object_lookup+0x35/0x3d [drm] [<f82b7941>] i915_gem_execbuffer+0x327/0xb37 [i915] [<f81ba73f>] drm_ioctl+0x202/0x292 [drm] [<f82b761a>] ? i915_gem_execbuffer+0x0/0xb37 [i915] [<c042e832>] ? default_wake_function+0x0/0x12 [<c0426b5d>] ? __wake_up_common+0x39/0x61 [<c04b2652>] vfs_ioctl+0x5c/0x76 [<c04b2f06>] do_vfs_ioctl+0x480/0x4ba [<c04306cd>] ? finish_task_switch+0x34/0x9f [<c0439fca>] ? __do_softirq+0x12a/0x139 [<c046bd02>] ? audit_syscall_entry+0x163/0x185 [<c04b2f86>] sys_ioctl+0x46/0x66 [<c040945f>] sysenter_do_call+0x12/0x34 *** This bug has been marked as a duplicate of bug 489907 *** |
Created attachment 346788 [details] The X log Description of problem: X, when compiz is enabled on GNOME, feels consistently slower than with Fedora 9 on the same machine. In addition, strange errors appears in Xorg.0.log: (**) intel(0): Kernel mode setting active, disabling FBC. (**) intel(0): Framebuffer compression disabled (**) intel(0): Tiling enabled (==) intel(0): VideoRam: 4194303 KB (II) intel(0): Attempting memory allocation with tiled buffers. (EE) intel(0): Failed to set tiling on front buffer: rejected by kernel (EE) intel(0): max_gtt_map_size: 61727kb. (II) intel(0): Tiled allocation successful. And, subsequently through the whole file, seemingly random repetitions of: (EE) intel(0): max_gtt_map_size: 61727kb. This might be unrelated to the slowness, but that is what pèrompted me to investigate Version-Release number of selected component (if applicable): 2.7.0.7.fc11 How reproducible: Steps to Reproduce: 1. Log into X and activate desktop effects, checking that no xorg.conf is present 2. grep "(EE) " /var/log/Xorg.0.log 3. Actual results: (WW) warning, (EE) error, (NI) not implemented, (??) unknown. (EE) intel(0): Failed to set tiling on front buffer: rejected by kernel (EE) intel(0): max_gtt_map_size: 61727kb. (EE) intel(0): max_gtt_map_size: 61727kb. (EE) intel(0): max_gtt_map_size: 61727kb. (EE) intel(0): max_gtt_map_size: 61727kb. (EE) intel(0): Failed to set tiling on front buffer: rejected by kernel (EE) intel(0): max_gtt_map_size: 59679kb. (EE) intel(0): max_gtt_map_size: 59679kb. (EE) intel(0): max_gtt_map_size: 59679kb. Expected results: (WW) warning, (EE) error, (NI) not implemented, (??) unknown. Additional info: This is not a fatal bug, and the slowness is not enough to make the system unusable, however the error messages are somewhat odd (tiling fails on the front buffer: is it because the FBC was disabled?) and also because there is no trace of said errors launch in the vanilla 2.7.1 driver code from freedesktop's git: especially, all references to max_gtt_map_size are not there.