AMD Catalyst 7.11 Linux Driver

[phoronix]

Phoronix: AMD Catalyst 7.11 Linux Driver

For the past two months, the AMD release train has been running full speed ahead as they introduced their new driver code-base and last month had dropped in the AIGLX support. However, this month the AMD train has taken an intermediate stop as this closed-source driver embarks on its next journey. As always, we have all of the details for you on this month's ATI Catalyst 7.11 Linux display driver release.

http://www.phoronix.com/vr.php?view=11462

[jasa68]

Aah new drivers finally, odd thing seems to me that release notes dont say anything about x-video problems (on AIGLX) as a known problem or fixed one. Guess i'll need to try those later on after i get some things done.

[Dottout]

https://a248.e.akamai.net/f/674/9206/0/www2.ati.com/drivers/linux/catalyst_711_linux.html <-- official relase notes. I read new features\resolved issues and I dont know if I want to laugh or cry. nomore nomore nomore ati cards in my life!

[Malikith]

https://a248.e.akamai.net/f/674/9206/0/www2.ati.com/drivers/linux/catalyst_711_linux.html <-- official relase notes. I read new features\resolved issues and I dont know if I want to laugh or cry. nomore nomore nomore ati cards in my life!

Yeah I don't like the looks of that. That reminds me of that other release earlier in the year that was the most useless release. I hope this one is not the case as well. But it sure looks like it.

[Dottout]

Yeah I don't like the looks of that. That reminds me of that other release earlier in the year that was the most useless release. I hope this one is not the case as well. But it sure looks like it.
do you have any spare hopes left? I dont, anymore :\

[carpex]

Resolved Issues
* An error message no longer appears during installation if dash is used as /bin/sh

Known Issues
* An error message appears during installation if dash is used as /bin/sh. A topic number is not available for this issue

Wow, that's pretty reassuring.

CP

[ZedDB]

And the memleak is still not fixed on my x1900xt...
700mb in 2min running glxgears...

[hmmm]

doesn't say anything about performance improvements either...

Michael: any chance you can interview the devs again??

[Nicolas]

Ok, I guess we must wait for 2008 to see if something works as it should...

[Dottout]

ebuild for gentoo here http://forums.gentoo.org/viewtopic-p-4529789.html#4529789

[Michael]

Michael: any chance you can interview the devs again??

If you guys come up with the questions, I will get the answers.

[chefkoch]

ok my observations:
- it builds without patching with the 2.6.24-rc3 kernel
- the control center works now
- some apps still hang (including the control center) when they are the first program that talks to the driver
- compiz still doesn't work with xserver 1.4 (but I don't need that)

Overall I'm happy with this release. It installed flawlessly even on the latest kernel, that hang thing doesn't bug me that much and I can live without compiz.

[lucien]

Could some say something about the AIGLX performance?

[Kano]

The person who renamed the driver from 8.433 to 7-11 AFTER packageing must be an idiot. Now I have to rename the version in my script from 7-11 to 8.433 after extracting. If someone wants to rename the driver fine, but then package it with the same name! Usually the script could use -v VER but in that case you need special handling, that makes the option useless - or you have to parse the real name which is used by the packages.

[hmmm]

Michael: should I start a new thread for it? btw the way how many devs are there meant to be working on the fglrxs now?

[Michael]

Michael: should I start a new thread for it? btw the way how many devs are there meant to be working on the fglrxs now?

New thread would be much better.

Regarding how many developers are working on the Linux driver now, that number is confidential. (FYI, they are hiring new developers though)

[macmus]

maybe that was finnaly corrected:

http://ubuntuforums.org/showpost.php?p=3804912&postcount=233

[Dottout]

Could some say something about the AIGLX performance?

here, with x1400 aiglx sux more than before. Ive lost 100fps with glxgears

[macmus]

here, with x1400 aiglx sux more than before. Ive lost 100fps with glxgears

what about rest of common bugs: flickering, compiz choking ?

[Dottout]

what about rest of common bugs: flickering, compiz choking ?
everything like 8.42. this is what I call shit

[konas]

everything like 8.42. this is what I call shit

Same here (X700 mobility, Gutsy 7.10 /64)

flickering, Xorg is more cpu consuming then 8.42.3, scrolling in firefox is not perfect, no improvement on speed ....

[macmus]

everything like 8.42. this is what I call shit
have u tried my link above doing that litte change.
Author says it does help with scrolling and chooking issues - this is fault in some wrong module insertion.

Nevertheless it seems cpu and flicering problems seems to lay somewhere deep is that NEW CODE shit :(.

Why they are correcting intaller issues .. soon there will be nobody uger even to touch the installer :(

[Kano]

The only visable change from 8.42.3 to 8.433/7-11 is that the kernel patch for 2.6.23 is not longer required. And you waited one month for that! Not even the Xserver 1.4.x bug is gone, that you need to start a second 3d app to make the first one go. That's what I call useless release. Btw. for current sid you need a trick to create debs, my script has a workaround included.

http://kanotix.com/files/install-fglrx-debian.sh

[Extreme Coder]

I have an ATI Radeon X1200;
I don't care much for Compiz, and such, and I have 8.42 working.

Do you think I should upgrade to this driver?

[Ole-Martin Broz]

seems like theyve worked on getting new distro installers up n running, well, thats not what i want !, however, its nice to have it though.

[Dottout]

I have an ATI Radeon X1200;
I don't care much for Compiz, and such, and I have 8.42 working.

Do you think I should upgrade to this driver? no gains, you would only waste time. as ati devs do

[d2kx]

Anyone thought about that AMD may be is preparing new features for 8.44? What do you think they have been working on for the last weeks.

I am sure they could have worked on some new features and 8.43 is a (very small!) bugfix release.

[macmus]

no gains, you would only waste time. as ati devs do

devs ?? i tought there is only one guy out there wating for christmas tree to arrive nice and shiny

[SledgeHammer_999]

I just installed 7.11 and thank god I have tv-out again on my X1800XL. But I have something to say: ATI Catalyst Control Center is absolutely USELESS. I used it to to change the position and size of the TV-OUT image and it worked like in Windows. But when I restarted X every change was lost. I had to use aticonfig(bliah).

[pedepy]

well i have myself the voice of optimism the past few releases but this time its getting pretty hard .. I mean its not so much about what the drivers can and cannot do, its about how its being made ... I havent tried the latest driver yet but if its true that its slower and as buggier than its predecessor, then someone at AMD is not doing their job right, period ... I think a good question to ask the devs, mr Micheal, is what the hell is going on behind those closed doors ?! ...

I just hope I can get to use my ati card to its full potential before it reaches obsolescence... :(

anyway, good things have happened in the past couple weeks, not just bad ones, lets not forget that .. and keep our hopes up that the next driver release could be named tagged the Christmas Miracle ... :P (even I though I think most of the rant is coming from our high expectations of just that, a miracle..)

[Dottout]

well i have myself the voice of optimism the past few releases but this time its getting pretty hard .. I mean its not so much about what the drivers can and cannot do, its about how its being made ... I havent tried the latest driver yet but if its true that its slower and as buggier than its predecessor, then someone at AMD is not doing their job right, period ... I think a good question to ask the devs, mr Micheal, is what the hell is going on behind those closed doors ?! ...

I just hope I can get to use my ati card to its full potential before it reaches obsolescence... :(

anyway, good things have happened in the past couple weeks, not just bad ones, lets not forget that .. and keep our hopes up that the next driver release could be named tagged the Christmas Miracle ... :P (even I though I think most of the rant is coming from our high expectations of just that, a miracle..)

save your hopes for radeonhd and open source world, amd will never provide anything good. how long have we been waiting? how can you still believe in a *better* tomorrow?

[Bigon]

amdccle seems to work with xcb now


(is it me or libstdc++5 is not needed anymore?)

Edit: only amdcccle needs libstdc++5 with the today release

[fettouhi]

Well this new driver hasn't changed much for me with my ATI HD 2400XT card. Scrolling in firefox seems remotely faster but my biggest problem is the screen going black for 2 secs (almost as if the x-server is crashing) every time I load a video in any player or in firefox via the mplayerplug-in. This problem started with the 8.42.3 driver.

Regards

André

[Ole-Martin Broz]

Uhm, gusy, i think they've planning to release their drivers at the same time as the windows driver, so to get rid of the number mess, with driver numbers everywhere :p

so, i think this is a just completed stuff released, who knows. ?

[koolmanoncampus]

Who else thinks AMD should give up on fglrx development; and just work on sanitizing specs for the guys at novell faster?

"No Video playback is allowed on second head". Pathetic. Heres for still chuggin on 8.40!

[Uchikoma]

Dottout: I have great patience. I've waited years for ATI (now ATI/AMD) to bring themselves to the top of their game with regards to their closed source. I'll keep waiting too. For now.

Now, the one thing I do note in the fairly unimpressive driver release docs is the inclusion of official Ubuntu 7.10 support. (early look it says). The question I then is:

Does it allow for Step 1 type installation? For those who don't know what I mean, I mean simply running the installer and installing it using the graphical user interface without doing the Step 2 commandline compile and install.

[Trek1701]

Fine. On 7.11 if I enable TV-out the max refresh rate I got is 60Hz. With the previous driver it was 85Hz.


Trek1701

[adamk]

Aah new drivers finally, odd thing seems to me that release notes dont say anything about x-video problems (on AIGLX) as a known problem or fixed one. Guess i'll need to try those later on after i get some things done.

The X-Video problem when running compiz exists for any driver that uses Xorg's direct rendering infrastructure, so I'm not inclined to think that this is a fglrx issue, which may be why you're not seeing mentioned in the release notes.

[Berniyh]

amdccle seems to work with xcb now


(is it me or libstdc++5 is not needed anymore?)

Edit: only amdcccle needs libstdc++5 with the today release
That was the case with the previous release, too.

@Michael, you write in your article, that further modification is needed for the driver to work with kernel 2.6.24. I'm running 2.6.24-rc3 (with a few more patches) and it build fine for me.
I'm also running 8.42.3 (patched for 2.6.23) for quite some time now with kernel 2.6.24, so I wouldn't expect any problems.

One question, do you know if they fixed the SLUB thing?


Edit: BTW, so I was right, it came today (and this week).:D

[Michael]

@Michael, you write in your article, that further modification is needed for the driver to work with kernel 2.6.24. I'm running 2.6.24-rc3 (with a few more patches) and it build fine for me.
I'm also running 8.42.3 (patched for 2.6.23) for quite some time now with kernel 2.6.24, so I wouldn't expect any problems.

Corrected, the support ended up landing late in the development cycle.

[ltmon]

One question, do you know if they fixed the SLUB thing?

I'd like to know this also. It seems to me that this would be about the only reason I'd risk any regressions by upgrading.

Anyone have any success with suspend/resume on Gutsy where they didn't previously?

Thanks,

L.

[chrisr]

Now that I have a working 2.6.23.x configuration, I'll be able to upgrade this laptop to Fedora 8 this weekend.
And I'm getting about 2670 fps out of glxgears, although World of Warcraft / Wine is still not working with pixel shaders or "Mark 2" (vertex?) shaders enabled.

[Berniyh]

I'd like to know this also. It seems to me that this would be about the only reason I'd risk any regressions by upgrading.

Anyone have any success with suspend/resume on Gutsy where they didn't previously?

Thanks,

L.
I'll recompile my kernel tomorrow, I'll also compile a SLUB version.
But I thought maybe he has got the info, that might spare time. ;)

[chrisr]

*sigh*. They still haven't fixed this issue with Xvideo playback:

Error in "atiddxMiscUpdateFile" -4

Backtrace:
0: /usr/bin/Xorg(xf86SigHandler+0x81) [0x80c2d21]
1: [0xb7fcd420]
2: /lib/libc.so.6(memcpy+0x1c) [0x46008bcc]
3: /usr/lib/xorg/modules//glesx.so [0xb5e57e02]
4: /usr/lib/xorg/modules//glesx.so [0xb5e5d408]
5: /usr/lib/xorg/modules//glesx.so [0xb5e38b02]
6: /usr/lib/xorg/modules//glesx.so [0xb5e20163]
7: /usr/lib/xorg/modules//glesx.so [0xb5ddab6d]
8: /usr/lib/xorg/modules//glesx.so [0xb5dc3c35]
9: /usr/lib/xorg/modules//glesx.so [0xb5dc2306]
10: /usr/lib/xorg/modules//glesx.so [0xb5dc07c0]
11: /usr/lib/xorg/modules//glesx.so [0xb5dbf1cf]
12: /usr/bin/Xorg [0x80d16cf]
13: /usr/lib/xorg/modules/extensions//libextmod.so(XvdiPutImage+0x174) [0xb7f3ecb4]
14: /usr/lib/xorg/modules/extensions//libextmod.so [0xb7f41c87]
15: /usr/bin/Xorg [0x815235e]
16: /usr/bin/Xorg(Dispatch+0x1aa) [0x80899fa]
17: /usr/bin/Xorg(main+0x495) [0x8071735]
18: /lib/libc.so.6(__libc_start_main+0xe0) [0x45faef70]
19: /usr/bin/Xorg(FontFileCompleteXLFD+0x1e9) [0x8070a21]

Fatal server error:
Caught signal 11. Server aborting

[chrisr]

I've passed this log through c++filt:
BUG: sleeping function called from invalid context at include/asm/semaphore.h:99
in_atomic():1, irqs_disabled():0
[<f9339e91>] __ke_down_struct_sem+0x14/0x2b [fglrx]
[<f93530fe>] firegl_cmmqs_unbind_pool+0x9e/0x170 [fglrx]
[<f939cd31>] NODElist_free::mergeNodeWithSuccessor(CMMNode*)+0x 51/0x80 [fglrx]
[<f934f169>] MCIL_UnmapFromGartSpace+0xa9/0x110 [fglrx]
[<f939cda4>] NODElist_free::doGarbageCollection(CMMNode*)+0x44/0x80 [fglrx]
[<f938cd6d>] OS::gart_Free(CMMPool*)+0xcd/0x110 [fglrx]
[<f9386eef>] SegmentMapManager::init(MemSegment const&)+0x18f/0x220 [fglrx]
[<f9391d88>] CMMHeap_GART::dropPoolFromHeap(CMMPool*)+0x38/0x70 [fglrx]
[<f9391eb2>] CMMHeap_GART::shrinkHeap(bool)+0x92/0xa0 [fglrx]
[<f938d904>] CMMPool::freeNode(CMMNode*)+0x24/0x40 [fglrx]
[<f93a0ae2>] CMMClient::~CMMClient()+0x32/0x50 [fglrx]
[<f9390064>] CMMHeapManager::shrinkAllHeaps()+0x34/0x60 [fglrx]
[<f9396c5d>] CMMClientManager_CORE::destroyClient(CMMDriver*, CMMClient*)+0x2d/0x50 [fglrx]
[<f9392845>] cmmUnregisterClientInternal(CMMDriver*, CMMClient*)+0x25/0x30 [fglrx]
[<f938ab64>] MSF::doGarbageCollection()+0x204/0x250 [fglrx]
[<f939a0a8>] QS_PRIVATE_CORE::PM4idle(Asic::idle_WaitMethod)+0x 48/0x70 [fglrx]
[<f939a0ec>] QS_PRIVATE_CORE::idleAll(Asic::idle_WaitMethod)+0x 1c/0x50 [fglrx]
[<f933b116>] drm_free+0x66/0x150 [fglrx]
[<f93887aa>] CMM::releaseChildren()+0x2a/0x2d0 [fglrx]
[<f938a93b>] MSF::processTerminated(unsigned int)+0x1b/0x40 [fglrx]
[<f933b116>] drm_free+0x66/0x150 [fglrx]
[<f938502b>] CMMQS_ProcessTerminate+0x1b/0x30 [fglrx]
[<f9384e54>] CMMQS_Uninitialize+0x14/0x30 [fglrx]
[<f93550d8>] firegl_cmmqs_deinit+0x48/0xf0 [fglrx]
[<f9341435>] firegl_release_helper+0x5f5/0xa20 [fglrx]
[<f9344b9b>] firegl_takedown+0x5b/0xc50 [fglrx]
[<f93443bf>] firegl_release+0x12f/0x190 [fglrx]
[<f933898f>] ip_firegl_release+0xf/0x12 [fglrx]
[<c016b744>] __fput+0xaf/0x157
[<c016920c>] filp_close+0x51/0x58
[<c0122e44>] put_files_struct+0x5f/0xa7
[<c0123f1d>] do_exit+0x220/0x6d0
[<c012443a>] sys_exit_group+0x0/0xd
[<c012b569>] get_signal_to_deliver+0x3f3/0x418
[<c011b6d2>] try_to_wake_up+0x276/0x281
[<c01036d1>] do_notify_resume+0x84/0x633
[<c0129e4c>] signal_wake_up+0x1e/0x2c
[<c0118951>] do_page_fault+0x308/0x729
[<c014961a>] audit_syscall_exit+0x2aa/0x2c6
[<c0149346>] audit_syscall_entry+0x10d/0x137
[<c0107190>] do_syscall_trace+0x13c/0x180
[<c01070c3>] do_syscall_trace+0x6f/0x180
[<c010405e>] work_notifysig+0x13/0x19
=======================
BUG: scheduling while atomic: Xorg/0x00000002/3110
[<c02ad05e>] __sched_text_start+0x86/0x5b7
[<c011b1ef>] __wake_up+0x32/0x43
[<f9351e31>] irqmgr_wrap_shutdown+0x101/0x170 [fglrx]
[<f9344ba3>] firegl_takedown+0x63/0xc50 [fglrx]
[<f93443bf>] firegl_release+0x12f/0x190 [fglrx]
[<f933898f>] ip_firegl_release+0xf/0x12 [fglrx]
[<c016b744>] __fput+0xaf/0x157
[<c016920c>] filp_close+0x51/0x58
[<c0122e44>] put_files_struct+0x5f/0xa7
[<c0123f1d>] do_exit+0x220/0x6d0
[<c012443a>] sys_exit_group+0x0/0xd
[<c012b569>] get_signal_to_deliver+0x3f3/0x418
[<c011b6d2>] try_to_wake_up+0x276/0x281
[<c01036d1>] do_notify_resume+0x84/0x633
[<c0129e4c>] signal_wake_up+0x1e/0x2c
[<c0118951>] do_page_fault+0x308/0x729
[<c014961a>] audit_syscall_exit+0x2aa/0x2c6
[<c0149346>] audit_syscall_entry+0x10d/0x137
[<c0107190>] do_syscall_trace+0x13c/0x180
[<c01070c3>] do_syscall_trace+0x6f/0x180
[<c010405e>] work_notifysig+0x13/0x19
=======================
BUG: sleeping function called from invalid context at include/asm/semaphore.h:99
in_atomic():1, irqs_disabled():0
[<f9339e91>] __ke_down_struct_sem+0x14/0x2b [fglrx]
[<f933ea6f>] firegl_remove_all_drawables+0x3f/0x100 [fglrx]
[<f93414f9>] firegl_release_helper+0x6b9/0xa20 [fglrx]
[<f9344bab>] firegl_takedown+0x6b/0xc50 [fglrx]
[<c0246400>] __cpufreq_remove_dev+0x195/0x197
[<f93443bf>] firegl_release+0x12f/0x190 [fglrx]
[<f933898f>] ip_firegl_release+0xf/0x12 [fglrx]
[<c016b744>] __fput+0xaf/0x157
[<c016920c>] filp_close+0x51/0x58
[<c0107190>] do_syscall_trace+0x13c/0x180
[<c016a289>] sys_close+0x67/0x9e
[<c0103fc2>] syscall_call+0x7/0xb
=======================
[fglrx] PCIe has already been initialized. Reinitializing ...
BUG: sleeping function called from invalid context at include/asm/semaphore.h:99
in_atomic():1, irqs_disabled():0
[<f9339e91>] __ke_down_struct_sem+0x14/0x2b [fglrx]
[<f933ea6f>] firegl_remove_all_drawables+0x3f/0x100 [fglrx]
[<f93414f9>] firegl_release_helper+0x6b9/0xa20 [fglrx]
[<f9344bab>] firegl_takedown+0x6b/0xc50 [fglrx]
[<c0118649>] do_page_fault+0x0/0x729
[<f93443bf>] firegl_release+0x12f/0x190 [fglrx]
[<f933898f>] ip_firegl_release+0xf/0x12 [fglrx]
[<c016b744>] __fput+0xaf/0x157
[<c016920c>] filp_close+0x51/0x58
[<c0107190>] do_syscall_trace+0x13c/0x180
[<c016a289>] sys_close+0x67/0x9e
[<c0103fc2>] syscall_call+0x7/0xb
[<c02a0000>] xfrm4_udp_encap_rcv+0x3c/0x14c
=======================
[fglrx] Reserve Block - 0 offset = 0Xfffb000 length = 0X5000
[fglrx] Reserve Block - 1 offset = 0X0 length = 0X1000000
[fglrx] Reserve Block - 2 offset = 0Xffbb000 length = 0X40000

[rolz]

yeah i'm still having all the same issues i had with .42 :(

ow GREAT! i was responsible for at least some of the problems. fixed them now it at least works, but yeah memory leak ..

[Ghnuberath]

- Installed fglrx 7.11/8.43 this afternoon (Mobility X300, Gutsy Gibbon)
- Pressed suspend button
- Watched the driver act like 8.42
- Died a little inside

But Compiz seems to be running more smoothly (a noticeable difference).

Someone else please confirm my SLUB/8.43 findings.
Or deny - that would be nice.

[Berniyh]

yeah i'm still having all the same issues i had with .42 :(

Yeah, what they did is cross out "8.42.3" and write 7.11 on it. :p

[oyvind]

Should I really be surprised ? None of the serious flaws and issues that plague fglrx have been fixed. This is just insane, and I'm getting ready to throw my ATI-X1400-based laptop out of the window (better than installing Windows on it).

:mad:

* Proper and stable XVideo support ? Hahahahaa !

* Good 2D-acceleration ? Hahahhaa !

* Working suspend-to-RAM-support, since this new version is said to officially support 2.6.23 kernel ? Hahahahaaha !

* Stability when running multiple X-sessions ? Hahahaha !

* Leaking memory like a coffee filter (OpenGL-apps).

* Composited Compiz-environment through AIGLX still slow, unstable and shows diagonal tearing. Forget playing movies using anything by plain X11-shm output.

* Firefox-scrolling in Compiz-environment horrible.

* Firefox-scrolling i standard 2D still sucks for some sites (blaming poor 2D-acceleration).

Damn, I hate being so negative.

Hoping for improvements in next release, but I have a feeling my hardware will be totally obsolete before fglrx ever gets any good. Tired of waiting *sigh* and actually spending so much energy focusing on a freakin' driver, instead of just using Linux to its fullest potential.

[Compxpert]

Most useless release ever. Is there even a point in updating? Does it fix anything important? Glad I wasn't waiting on this release like 8.42.3 I would have been in store for even more of a disappointment at least the last release presented AIGLX even though it was/still is somewhat broken.

[emms]

Just installed on Gusty. I am quite disappointed as well.

All that matters to me is at least to have working suspend and resume mode - it won't.. This makes this driver completely useless on a laptop. I guess, even if they committed to the 2.6.23 kernel, the SLUB feature isn't even support on Gusty's 2.6.22.

I'm heading back to the open radeon driver.

[pedepy]

save your hopes for radeonhd and open source world, amd will never provide anything good. how long have we been waiting? how can you still believe in a *better* tomorrow?


well ive only had this laptop for 3 months now so I can wait some more ;p

[Dandel]

I wonder about undocumented changes, so i'll try this out on my setup to see if it works, because version 8.42.3 had major issues with the detection of the screen frequency information.

[c0un7d0wn]

Another sucky release..

[fglrx:firegl_lock_free] *ERROR* lock was not held by 1! (*lock=0x80000004)
[fglrx:firegl_unlock] *ERROR* firegl_lock_free failed!
BUG: scheduling while atomic: X/0x00000002/4732

Call Trace:
[<ffffffff8121d4a2>] schedule+0xdd/0x69d
[<ffffffff8102c6a3>] __wake_up+0x38/0x4e
[<ffffffff88028541>] :fglrx:irqmgr_wrap_shutdown+0x101/0x170
[<ffffffff8801c1e5>] :fglrx:firegl_takedown+0x65/0xc10
[<ffffffff8801ba64>] :fglrx:firegl_release+0x104/0x150
[<ffffffff810886c9>] __fput+0xa1/0x16e
[<ffffffff81085e00>] filp_close+0x5d/0x65
[<ffffffff81035a8d>] put_files_struct+0x66/0xc4
[<ffffffff81036e03>] do_exit+0x2b5/0x84b
[<ffffffff81037415>] sys_exit_group+0x0/0xe
[<ffffffff8103ef19>] get_signal_to_deliver+0x422/0x451
[<ffffffff8100b205>] do_notify_resume+0xa8/0x716
[<ffffffff8103cd1e>] sigprocmask+0xbc/0xc3
[<ffffffff8103f44e>] sys_rt_sigprocmask+0x50/0xcf
[<ffffffff8100bda7>] sysret_signal+0x1c/0x27
[<ffffffff8100c037>] ptregscall_common+0x67/0xb0

[fglrx] PCIe has already been initialized. Reinitializing ...
[fglrx] Reserve Block - 0 offset = 0X1000000 length = 0X5000
[fglrx] Reserve Block - 1 offset = 0X0 length = 0X1000000
[fglrx] Reserve Block - 2 offset = 0Xffff000 length = 0X1000
[fglrx] Reserve Block - 3 offset = 0Xffbf000 length = 0X40000

[Skywalker]

Lots of flickering using Firefox with previous versions.

All gone with this new one. Much improved, i would say.

(This on Fedora 6, 64 bits. HD2600 video card.)

[Uchikoma]

Installed the Catalyst 7.11 driver a couple hours ago using the graphical installer and a "from-minimalist up" ubuntu minimal with xorg/xfce4.

Video overlay is Xv
Single head setup.

Video playback: good (480p and 720p res)
Scrolling: good
Display corruption? none
DRI: Yes
AIGLX: Yes (desktop effects/compiz not tested, nor do I intend to test it until...)
GL: Memory leak. Drops about 30 MB per 5 seconds until I come very close to running out of memory (in which I kill gears)

That's my take. From 8.42, I don't notice much different except I can do the graphical installer without a problem. Its completely usable for me with 2d only, so I'll keep it running.

[yoshi314]

wow, there are still ati linux users waiting for new fglrx. how strange ;-)

i don't care one bit about fglrx anymore. thanks go to the opensource ati driver developers for providing much better opensource solution, despite all legal and documentation problems they encountered on the way.

[FunkyRider]

Do you even need a video card if all you need is 2D? Go get an onboard or S3 or something should make you happier as it also saves you a good fortune... Today, right after this driver is released, I sold my AIW X1900 and got a MSI-8600GT-OC. Guess what, I couldn't be happier now. Quake 4 runs @ max settings, 1680x1050 like charm, good AIGLX, good video (no tearings!), no memory leak, I can now work on GLSL with 3D desktop enabled at the same time. What can I ask for more man!

[yoshi314]

Do you even need a video card if all you need is 2D?but of course. i don't want any integrated sub-par video card to eat my system ram.

besides i play darkplaces on opensource drivers now and then.

[Kano]

Onboard vga takes system ram, that's correct, but with current ram prices you could easyly add 1 or 2 gb ram, then a few mb are no problem usally. With dual channel ram the speed drop is so minimal that you would never feel it. For simple games Intel onboard for example would be really enough - but that D3/Q4 and those games would run fast you should not expect. It is of course fast enough for compiz. For your purpose you really don't need a dedicated gfx card...

[Ze..]

Does the GL memory leak bug affect X1600 mobility GPU's?

If it doesn't I'll install this driver and give it a go but if it does there is no point fucking around doing it. I'll stick with whatever the default for gutsy.

[yoshi314]

For your purpose you really don't need a dedicated gfx card...well no turning back now. i don't earn enough to waste my money on a new pc every couple of months.

[FunkyRider]

but unfortunately we are still wasting time and money waiting for ATI

[peper]

In this and previous release I have problem with cursor in my FC6. When I'm switching desktops sometimes cursor is changing to some random area of gray colors. Switching desktops back or just moving around screen so cursor can change its shape at the edges of windows corrects this problem. I wonder if its driver, configuration or KDE related problem. HWCursor option in confing is not solving problem. Any ideas?

[macmus]

i am pissed..
I bought laptop with ATI HD2600 after reading articles on this site about new driver HOPE.

I will never do same mistake again. I'm stuck with ATI now cause i cannot remove it from laptop.

What i can only say i was totaly screw .. :(

Thank you ATI developer.

[Berniyh]

In this and previous release I have problem with cursor in my FC6. When I'm switching desktops sometimes cursor is changing to some random area of gray colors. Switching desktops back or just moving around screen so cursor can change its shape at the edges of windows corrects this problem. I wonder if its driver, configuration or KDE related problem. HWCursor option in confing is not solving problem. Any ideas?

Hmpf, I thought they corrected that corruption thing.
Try: Option "XAANoOffscreenPixmaps" "true"
in your xorg.conf.

[garoth]

Acer Travelmate 4402WLMi laptop with ATI Radeon Mobility X700

With 8.42.3 I had the problem with video tearing/corruption when trying to use the hardware overlay (while Compiz was enabled)... and actually I recently was unable to play any video files at all even with Compiz disabled (not sure why).

Upgrading to this release
-Video playback works with Compiz enabled or disabled
-With compiz enabled, some flicker in video playback when NOT in full screen... when in full screen video plays fine with tolerable amount of tearing
-Scrolling in Firefox still a bit slow. In Firefox 3.0b1 it is tolerable, compared to Firefox 2

Overall I'm happier with this release - it has been an improvement, at least for me. I much prefer the look/feel of the desktop with being able to use Compiz - all of the fonts appear crisper to me and menus are more responsive, so this is important to me.

[peper]

Hmpf, I thought they corrected that corruption thing.
Try: Option "XAANoOffscreenPixmaps" "true"
in your xorg.conf.

Hi,
Thanks, this option solved problem, altho fglrx kernel module is crashing but working

BUG: sleeping function called from invalid context at include/asm/semaphore.h:105
in_atomic():1, irqs_disabled():0

Call Trace:
[<ffffffff8813907c>] :fglrx:__ke_down_struct_sem+0x26/0x3c
[<ffffffff8813de08>] :fglrx:firegl_remove_all_drawables+0x38/0xf0
[<ffffffff881438ed>] :fglrx:firegl_takedown+0x6d/0xc10
[<ffffffff88143164>] :fglrx:firegl_release+0x104/0x150
[<ffffffff80290b01>] __fput+0xbd/0x18d
[<ffffffff8028e35c>] filp_close+0x5d/0x65
[<ffffffff8028f41a>] sys_close+0x93/0xd7
[<ffffffff80209c4e>] system_call+0x7e/0x83

[fglrx] PCIe has already been initialized. Reinitializing ...
BUG: sleeping function called from invalid context at include/asm/semaphore.h:105
in_atomic():1, irqs_disabled():0

Call Trace:
[<ffffffff8813907c>] :fglrx:__ke_down_struct_sem+0x26/0x3c
[<ffffffff8813de08>] :fglrx:firegl_remove_all_drawables+0x38/0xf0
[<ffffffff881438ed>] :fglrx:firegl_takedown+0x6d/0xc10
[<ffffffff88143164>] :fglrx:firegl_release+0x104/0x150
[<ffffffff80290b01>] __fput+0xbd/0x18d
[<ffffffff8028e35c>] filp_close+0x5d/0x65
[<ffffffff8028f41a>] sys_close+0x93/0xd7
[<ffffffff80209c4e>] system_call+0x7e/0x83

[fglrx] GART Table is not in FRAME_BUFFER range
[fglrx] Reserve Block - 0 offset = 0X7ffb000 length = 0X5000
[fglrx] Reserve Block - 1 offset = 0X0 length = 0X1000000
hda-intel: Invalid position buffer, using LPIB read method instead.

[diverse_izzue]

Someone else please confirm my SLUB/8.43 findings.
Or deny - that would be nice.

Same story here. Was really eager to install this, then big disappointment.

I wonder if it'd work if Gutsy actually had a 2.6.23 kernel. Maybe fglrx just checks for the kernel version and not for SLUB/SLAB and thus gets it wrong (IIRC a 2.6.22 kernel is not "supposed" to use SLUB)

[karolbe]

Same story here. Was really eager to install this, then big disappointment.

I wonder if it'd work if Gutsy actually had a 2.6.23 kernel. Maybe fglrx just checks for the kernel version and not for SLUB/SLAB and thus gets it wrong (IIRC a 2.6.22 kernel is not "supposed" to use SLUB)

too bad, all I was waiting for was working suspend/hibernate...so good bye ati, thanks God I can replace video card in my laptop and I have spare nVidia 6400 :)

Dell 6400 users, if you have ati x1300 or x1400 then you can get used nVidia 6400 on eBay for 40 bucks and replace video card in 20 minutes, why suffer using ati? :)

[Nicolas]

too bad, all I was waiting for was working suspend/hibernate...so good bye ati, thanks God I can replace video card in my laptop and I have spare nVidia 6400 :)

Dell 6400 users, if you have ati x1300 or x1400 then you can get used nVidia 6400 on eBay for 40 bucks and replace video card in 20 minutes, why suffer using ati? :)

I wish I could do that on my Acer Travelmate 8103 :(.

[Dandel]

interestingly enough the issue related to CRT monitors that showed up in 8.42.3 is no longer an issue in this release.

[chikazuku]

What are they doing over there? I thought they would at least fix some bugs, but no: The "bus error" is back, and as such I can't start any OpenGL application with this driver..... :(

[gmplague]

I just installed 7.11 and thank god I have tv-out again on my X1800XL. But I have something to say: ATI Catalyst Control Center is absolutely USELESS. I used it to to change the position and size of the TV-OUT image and it worked like in Windows. But when I restarted X every change was lost. I had to use aticonfig(bliah).

Had the same issue in Ubuntu Gutsy 7.10/x64 ... had to go to System -> Screen Resolution and click the 'Make this default' box after the change. Don't know if this is some gnome tweak or what's going on under the covers, but it worked for me.

Also, aticonfig never worked right for me after login, so maybe it's a different issue.

[SledgeHammer_999]

Had the same issue in Ubuntu Gutsy 7.10/x64 ... had to go to System -> Screen Resolution and click the 'Make this default' box after the change. Don't know if this is some gnome tweak or what's going on under the covers, but it worked for me.

Also, aticonfig never worked right for me after login, so maybe it's a different issue.

I also have gutsy amd64 installed. But now I can't test your method(because I am bored and I don't want to have a nervous breakdown).

I think that Catalyst Center should write everything in x.org file like aticonfig. Otherwise what's the point of using Catalyst Center???

[mile]

one proof more that ati is making fun of us :(

[sabriah]

Does anyone know of Debian Sid specific installation instructions for 7.11?

[Kano]

Very easy, just execute my script:

http://kanotix.com/files/install-fglrx-debian.sh

[butdie]

they came up with the cheap idea, renaming fglrx.

AMD treats customers as brainless.

Please, AMD, we asked for firing of fglrx developers. Cheap tricks won't do it.

Anyway, I have to make the decision, when to open my ati based laptop, take out the ati card, and burn it.

May there be justice!

one proof more that ati is making fun of us :(

[Berniyh]

Anyway, I have to make the decision, when to open my ati based laptop, take out the ati card, and burn it.
Make sure you burn the whole thing, because unless you've got a mxm slot (which as far as I know maybe about 10 notebooks out there have :D) it will be useless without it. ;)

[sabriah]

Very easy, just execute my script:

http://kanotix.com/files/install-fglrx-debian.sh

Thanks!

Would it support uninstall too?

[d2kx]

Please, AMD, we asked for firing of fglrx developers.

AMD currently asks for new developers. Should be the better idea.

[Kano]

@sabriah

Uninstall is no option to the script,but:

cp /etc/X11/xorg.conf.1st /etc/X11/xorg.conf

apt-get remove --purge $(dpkg -l|awk '/fglrx/{print $2}')

will remove it.

[Berniyh]

AMD currently asks for new developers. Should be the better idea.

Did they recruit developers?
The recruited someone for the specifications, but that isn't the same.

[d2kx]

They are looking for fglrx developers at the moment. Look around on amd.com's job site to find the description.

[Michael]

Did they recruit developers?
The recruited someone for the specifications, but that isn't the same.


They are also hiring for blob work....

http://sh.webhire.com/servlet/av/jd?ai=405&ji=2103384&sn=I

[Kano]

"-Solid understanding of standard Software Development Life Cycles."

Thats the best requirement ;)

[sabriah]

@sabriah

Uninstall is no option to the script,but:

cp /etc/X11/xorg.conf.1st /etc/X11/xorg.conf

apt-get remove --purge $(dpkg -l|awk '/fglrx/{print $2}')

will remove it.


Excellent scripts!

I started my computer in single-user mode (non-X console) and purged the previous version. I ran the install script. Then, "exit" and the startup proceeded and 3D works!

glxgears:

39620 frames in 5.0 seconds = 7923.917 FPS
30950 frames in 5.0 seconds = 6189.878 FPS
30547 frames in 5.0 seconds = 6109.343 FPS
30761 frames in 5.0 seconds = 6152.027 FPS

chromium: 45-50 fps

Nexuiz etc run smooth!

etc

:)

Thanks for a WORKING script!!

Pushing my luck here... You don't happen to know if this setup would work with Compiz? :)

[piquadrat]

This is really getting ridiculous. There are several showstopper bugs in the driver right now. I mean, the memory leak is a nobrainer - you just don't release software with such a huge, known flaw. Than there's Xvideo which is broken since more than a year. Suspend, for crying out loud! I don't care about AIGLX (not anymore), just give me a driver capable of a bit of 3D, playing movies in acceptable quality and which is not butchering suspend/resume.

I've given up all hope on fglrx. I just hope that the radeon or radeonhd reach a usable state on R500 chipsets in the next few months.

[Kano]

@sabriah

The script does more than you might have thought, you do not need single user mode, it kills kdm/gdm or whatever, uninstalls previous driver, installs new one and patches xorg.conf. After that X is started again. The commands I told you are just to delete the driver if you do not want to use fglrx at all.

[sabriah]

@sabriah

The script does more than you might have thought, you do not need single user mode, it kills kdm/gdm or whatever, uninstalls previous driver, installs new one and patches xorg.conf. After that X is started again. The commands I told you are just to delete the driver if you do not want to use fglrx at all.

Possibly, but I couldn't install it from a Gnome-Terminal. There was a complaint that X interfered, or something, so I rebooted.

[aranarth]

and I am _very_ disappointed with this release.

I'll have to wait for the December release but I'm afraid ATI/AMD will continue to ignore the FireGL userbase...

[Kano]

It suggests to press ctrl+alt+f1, if that does not work:

chvt 1

as root.

[sabriah]

It suggests to press ctrl+alt+f1, if that does not work:

chvt 1

as root.

True, but I didn't get to chvt 1 step. I'm not sure what might be wrong with my setup, I often get messages about "hdb" and "hdc" are not ready yet (yes I do know two of my cdroms are broken, but...). Still, that stalled the installation, so I rebooted.

I run XFS, which might be extra sensitive to faulty hw?

[planetf1]

Only had a few mins to try, but with the pci_id hack, the 7.11 driver appears to work under RHEL5.1.
with a FireGL V5250

No Beryl/AIGLX tried yet, but suspend/resume works -- an improvement on 8.42. GLXgears looks ok.

Did have to add a DRI section with 0666 permissions into xorg.conf

Lost hotkeys.. but they were redhat specific.

Will try using "as normal" tomorrow to see if stable enough for the day job before experimenting with 3D etc.

[mintcoffee]

Hmm my Xorg Server still hard freezes / crashes (same with 8.42.3) upon resuming from suspend. I have a FireGL V5200, so I don't think there's a big difference.

All I really want is working suspend and resume! Arggg

[CPrgmSwR2]

The main two questions I have are.

1. Are they even aware of the bugs that exist in their drivers?
2. Can they make a special release to address bug fixes that doesn't require the two month additional testing?

Is AMD a good company to work for?, I would like to apply for a job and help develop their fglrx drivers.

[Trek1701]

Hi,

Have anyone noticed that when TV-out is enabled the refresh rate goes down to 60HZ only with this new driver or it's just happening with me?


Thanks,
Trek1701

[robitor]

does this fix the slow scrolling in firefox issue?

[lionheart]

Why Phoronix is all the time advertising these drivers with such emphasis? All the time I read words such as "great" "the drivers had a long run" etc etc followed by hate comments.

A bit comical, isn't it?

I decided to change my ATI-based laptop a few weeks ago with an X3100 based one. True, bugs are there due also to the newness of the 965G chipset, but at least I can pull out new releases from git almost every week seeing that bugs reported to freedesktop.org get *really* fixed.

There's no reason to keep an AMD/ATI card. One of the biggest hardware companies is behaving like they don't have enough money to spend on decent developers.. hard to believe but true.