Compiz Running With Mesa On R600/700 GPUs

[Perry3D]

Yes, I do.
Actually, there is one because the 3D functions does not exist in the mainline DRM. So, the driver does not use it and so I get close to no-corruption on screen.

I will soon (this week end?) set up a nightly-build system on my computer to compile all the packages from GIT and upload them on my repository (accessible from here http://mupuf.org/packages/).
I will build the 64bit packages, if someone wants to build the 32bits edition, this could be great

PS: My server is down for the moment because of an OVH problem. So, if it doesn't work, wait a bit and try to refresh again

You don't have to recompile the kernel or remove the original drm modules. drm-radeon-module-git-r6xx-r7xx-3d installs the drm modules from agd5f's branch in /lib/modules/2.6.30-ARCH/updates/. So the old modules are replaced with the modules from this package.

At least this works for me.

[MuPuF]

You don't have to recompile the kernel or remove the original drm modules. drm-radeon-module-git-r6xx-r7xx-3d installs the drm modules from agd5f's branch in /lib/modules/2.6.30-ARCH/updates/. So the old modules are replaced with the modules from this package.

At least this works for me.

Sure, I don't need them. In fact, WHEN I want to get a stable driver, I simply remove the package drm-radeon-module-git-r6xx-r7xx-3d. This is quite convenient as I don't need to reinstall every single package whenever I want to try out the latest development version

Also, I have written an article on my blog and made a video of what I get using the driver on my HD4770.
Here is the link http://mupuf.org/blog/article/9/ .
You'll see a lot of screen corruption but otherwise, windows are already quite snappy

[javum]

With latest radeon-git drivers, Gnome 2.26.3 and latest compiz-git I have problem with refresh of input text in gnome-terminal and sometimes in other other apps (f.е. Opera textareas and seach fields).




after switching to another window and returning back it redraws:

Force synchronization between X and GLX option in Workarounds doesn't help.
Any ideas to fix this?

[nanonyme]

Wait? It looks like an issue that quite a few users are getting.

[javum]

Ok. I will be waiting fo the fix

[kernelOfTruth]

that's what we're talking about !

I'm getting the exactly / similar same text corruption during input

hopefully there soon will be a patch or a branch with the KMS Page-Flipping Ioctl and other related stuff which can be merged to 2.6.30 or 2.6.31 in order to get rid of the corruption

[ojab]

I'm getting a X hardlock (only reboot helps) on tex1d-2dborder piglit test.

xserver-git-master, mesa-git-master, xf86-video-ati-git-master, drm-git (r6xx-r7xx-3d branch), RS780 "ATI Radeon HD 3300 Graphics" (ChipID = 0x9614), linux-2.6.31-rc6+ (something after rc6 from git)

During shutdown before soft reboot (via ssh) i'm getting oops:

Aug 19 23:56:18 ojab kernel: [99522.866315] mtrr: no MTRR for d0000000,10000000 found
Aug 20 00:28:48 ojab kernel: [101473.033748] mtrr: no MTRR for d0000000,10000000 found
Aug 20 00:28:52 ojab kernel: [101476.722569] [drm] Setting GART location based on new memory map
Aug 20 00:28:52 ojab kernel: [101476.738082] [drm] Loading RS780/RS880 CP Microcode
Aug 20 00:28:52 ojab kernel: [101476.738792] [drm] Loading RS780/RS880 PFP Microcode
Aug 20 00:28:52 ojab kernel: [101476.753914] [drm] Resetting GPU
Aug 20 00:28:52 ojab kernel: [101476.753973] [drm] writeback test succeeded in 1 usecs
Aug 20 00:30:56 ojab kernel: [101601.082919] [drm] wait idle failed status : 0xA0003030 0x00000003
Aug 20 00:30:56 ojab kernel: [101601.213932] [drm] wait idle failed status : 0xA0003030 0x00000003
............many same records...................................
Aug 20 00:30:56 ojab kernel: [101601.365091] [drm] wait idle failed status : 0xA0003030 0x00000003
Aug 20 00:43:43 ojab kernel: [102367.446305] [drm] wait idle failed status : 0xA0003030 0x00000003
Aug 20 00:43:43 ojab kernel: [102367.578205] [drm] wait idle failed status : 0xA0003030 0x00000003
Aug 20 00:43:43 ojab kernel: [102367.578219] BUG: unable to handle kernel NULL pointer dereference at (null)
Aug 20 00:43:43 ojab kernel: [102367.578222] IP: [<ffffffffa01b9131>] radeon_read_ring_rptr+0x31/0x60 [radeon]
Aug 20 00:43:43 ojab kernel: [102367.578230] PGD 0 
Aug 20 00:43:43 ojab kernel: [102367.578232] Oops: 0000 [#1] PREEMPT SMP 
Aug 20 00:43:43 ojab kernel: [102367.578235] last sysfs file: /sys/devices/system/cpu/cpu1/topology/core_id
Aug 20 00:43:43 ojab kernel: [102367.578237] CPU 0 
Aug 20 00:43:43 ojab kernel: [102367.578238] Modules linked in: radeon drm isofs zlib_inflate sr_mod cdrom powernow_k8 i2c_dev k8temp cpufreq_stats af_packet kvm_amd kvm hwmon usbhid hid snd_hda_intel snd_hda_codec r8169 snd_hwdep snd_pcm ehci_hcd ohci_hcd bitrev snd_timer crc32 usbcore snd soundcore snd_page_alloc i2c_piix4 amd64_edac_mod i2c_core thermal edac_core sg rtc_cmos rtc_core evdev mii rtc_lib processor button unix [last unloaded: drm]
Aug 20 00:43:43 ojab kernel: [102367.578261] Pid: 32233, comm: X Not tainted 2.6.31-rc6-ojab #5 MS-7576
Aug 20 00:43:43 ojab kernel: [102367.578262] RIP: 0010:[<ffffffffa01b9131>]  [<ffffffffa01b9131>] radeon_read_ring_rptr+0x31/0x60 [radeon]
Aug 20 00:43:43 ojab kernel: [102367.578268] RSP: 0018:ffff8800460ffba8  EFLAGS: 00010246
Aug 20 00:43:43 ojab kernel: [102367.578270] RAX: ffff8800479f8540 RBX: ffff88006dca8000 RCX: 000000000000002c
Aug 20 00:43:43 ojab kernel: [102367.578271] RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffff88006dca8000
Aug 20 00:43:43 ojab kernel: [102367.578273] RBP: 0000000000000010 R08: 0000000000000000 R09: 00000000ffffffff
Aug 20 00:43:43 ojab kernel: [102367.578275] R10: 0000000000000001 R11: ffff88000157b318 R12: 0000000000000008
Aug 20 00:43:43 ojab kernel: [102367.578276] R13: ffff88005c8e5c60 R14: ffff88006fbb9700 R15: 0000000000000000
Aug 20 00:43:43 ojab kernel: [102367.578278] FS:  00007fa3ac0626f0(0000) GS:ffff880001569000(0000) knlGS:0000000000000000
Aug 20 00:43:43 ojab kernel: [102367.578280] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
Aug 20 00:43:43 ojab kernel: [102367.578281] CR2: 0000000000000000 CR3: 0000000001001000 CR4: 00000000000006f0
Aug 20 00:43:43 ojab kernel: [102367.578283] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
Aug 20 00:43:43 ojab kernel: [102367.578285] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Aug 20 00:43:43 ojab kernel: [102367.578287] Process X (pid: 32233, threadinfo ffff8800460fe000, task ffff8800019396c0)
Aug 20 00:43:43 ojab kernel: [102367.578288] Stack:
Aug 20 00:43:43 ojab kernel: [102367.578289]  0000000000000af8 00000000bb4b85b7 0000000000000003 ffffffffa01b9c12
Aug 20 00:43:43 ojab kernel: [102367.578292] <0> 0000000000000000 00000000bb4b85b7 ffff88006dca8000 ffffffffa01cdaab
Aug 20 00:43:43 ojab kernel: [102367.578295] <0> ffff88006dcaa000 00000000bb4b85b7 ffff88005c8e5c60 ffff88006dca8000
Aug 20 00:43:43 ojab kernel: [102367.578298] Call Trace:
Aug 20 00:43:43 ojab kernel: [102367.578303]  [<ffffffffa01b9c12>] ? radeon_commit_ring+0x62/0xe0 [radeon]
Aug 20 00:43:43 ojab kernel: [102367.578308]  [<ffffffffa01cdaab>] ? r600_do_cp_idle+0xcb/0x140 [radeon]
Aug 20 00:43:43 ojab kernel: [102367.578313]  [<ffffffffa01bd41e>] ? radeon_do_release+0x6e/0x230 [radeon]
Aug 20 00:43:43 ojab kernel: [102367.578326]  [<ffffffffa018df91>] ? drm_lastclose+0x51/0x320 [drm]
Aug 20 00:43:43 ojab kernel: [102367.578330]  [<ffffffff810dcea5>] ? __fput+0xe5/0x240
Aug 20 00:43:43 ojab kernel: [102367.578333]  [<ffffffff810d8d27>] ? filp_close+0x67/0xb0
Aug 20 00:43:43 ojab kernel: [102367.578336]  [<ffffffff8104e55d>] ? put_files_struct+0xed/0x130
Aug 20 00:43:43 ojab kernel: [102367.578338]  [<ffffffff81050568>] ? do_exit+0x6c8/0x730
Aug 20 00:43:43 ojab kernel: [102367.578341]  [<ffffffff8100cb0e>] ? common_interrupt+0xe/0x13
Aug 20 00:43:43 ojab kernel: [102367.578343]  [<ffffffff81050622>] ? do_group_exit+0x52/0xc0
Aug 20 00:43:43 ojab kernel: [102367.578346]  [<ffffffff8105d1e9>] ? get_signal_to_deliver+0x279/0x430
Aug 20 00:43:43 ojab kernel: [102367.578348]  [<ffffffff8100b640>] ? do_notify_resume+0x120/0x780
Aug 20 00:43:43 ojab kernel: [102367.578351]  [<ffffffff810ec5b9>] ? vfs_ioctl+0xa9/0xd0
Aug 20 00:43:43 ojab kernel: [102367.578353]  [<ffffffff810ec740>] ? do_vfs_ioctl+0xa0/0x5a0
Aug 20 00:43:43 ojab kernel: [102367.578355]  [<ffffffff81016fc8>] ? restore_i387_xstate+0x168/0x1e0
Aug 20 00:43:43 ojab kernel: [102367.578359]  [<ffffffff81314eda>] ? thread_return+0x4e/0x5b4
Aug 20 00:43:43 ojab kernel: [102367.578361]  [<ffffffff8100c271>] ? sysret_signal+0x74/0xc5
Aug 20 00:43:43 ojab kernel: [102367.578362] Code: 8b 04 25 28 00 00 00 48 89 44 24 08 31 c0 f6 87 d6 03 00 00 08 75 2b 48 8b 87 10 01 00 00 c1 ee 02 89 f6 48 c1 e6 02 48 03 70 18 <8b> 06 48 8b 54 24 08 65 48 33 14 25 28 00 00 00 75 16 48 83 c4 
Aug 20 00:43:43 ojab kernel: [102367.578381] RIP  [<ffffffffa01b9131>] radeon_read_ring_rptr+0x31/0x60 [radeon]
Aug 20 00:43:43 ojab kernel: [102367.578385]  RSP <ffff8800460ffba8>
Aug 20 00:43:43 ojab kernel: [102367.578386] CR2: 0000000000000000
Aug 20 00:43:43 ojab kernel: [102367.578388] ---[ end trace 7ad15c9b10ecbf30 ]---
Aug 20 00:43:43 ojab kernel: [102367.578390] Fixing recursive fault but reboot is needed!

Can anyone reproduce?

[kernelOfTruth]

nope but try re-compiling the stuff involved with the following C-/CXXFLAGS:

-fno-strict-overflow -fno-delete-null-pointer-checks

if that "fixes" it - it's a bug / problem caused by gcc-optimization

[nanonyme]

hopefully there soon will be a patch or a branch with the KMS Page-Flipping Ioctl and other related stuff which can be merged to 2.6.30 or 2.6.31 in order to get rid of the corruption

I think you mean 2.6.32.

[kernelOfTruth]

no - more specifically I should have written "kind of a backport" to make it clearer,

but because of its nature (deep changes) there's probably a must to upgrade to 2.6.32