Summary: | x11-drivers/ati-drivers-10.4_beta bump request | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Matt <jackdachef> |
Component: | New packages | Assignee: | Gentoo X packagers <x11> |
Status: | RESOLVED FIXED | ||
Severity: | enhancement | CC: | anm.mlist01, boltomli, bugzilla-gentoo, fcoiffie, lu_zero, mattemod, njdoyle+bugs, non7top, paluszak, shaun, strowi, tobias.pal, toto, tswatzke |
Priority: | High | Keywords: | Inclusion |
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: |
ati-drivers 8.721 (aka 10.4 beta) ebuild
A quick hack to rename it to 10.4_beta, like it should be. Output from failed patching, using ati-drivers-2.6.33.patch This one installs fine. ati-drivers-8.723.1 ebuild patch from 8.721 to 8.723.1 |
Description
Matt
2010-03-20 13:05:04 UTC
Please don't add arches yourself I had assumed that it was logical since it only includes support for x86 & amd64/x86_64 sorry for the annoyance & thanks Pacho ! Created attachment 224481 [details]
ati-drivers 8.721 (aka 10.4 beta) ebuild
I've tried to modify the 10.2 ebuild to make it work with this ubuntu package. It is a mess, but it compiles here. BUT i can't test it at the moment so it may fail you have been warned. But it may be a good start.
Oh i forgotten i used a newer package from here https://launchpad.net/ubuntu/lucid/+source/fglrx-installer/2:8.721-0ubuntu2 I put it in portage, it should support just xorg-server-1.7 only and it is already p.masked. Please test and report back =) i cannot see this package in portage! :( What could be wrong here? (In reply to comment #6) > i cannot see this package in portage! :( > What could be wrong here? > it's in the tree, but named ati-drivers-8.721 btw. whats the purpose of calling it 8.721 and not 10.4_pre/alpha/beta (In reply to comment #7) > it's in the tree, but named ati-drivers-8.721 > btw. whats the purpose of calling it 8.721 and not 10.4_pre/alpha/beta > It is just more handy for the ebuild structure and it is unsupported so devs want that only people that know what they are doing to use this :) (In reply to comment #5) > It is just more handy for the ebuild structure and it is unsupported so devs > want that only people that know what they are doing to use this :) Now you created a mess. Either all the packages use the 10.x-like scheme, or they all use the 8.xxx-like one. Now portage thinks that the newest ebuild is an old one. Hard-masking is already enough. Gentoo is not Windows. People are supposed to know what they are doing. Or else fall flat on their faces. ^^ Especially when unmasking a hard-masked package. Because that’s what it’s there for. I agree, that the 10.x scheme is a bit unclean. But this mess is even more unclean. Please rename all the ebuilds to their proper names. (8.xxx) Then everything is how it should be. :) Or you have to come up with something else that doesn’t break portage’s “this one is newer and this one is older” logic, when it becomes stable and tons of people will fall over it. ;) Oh, and good luck, wen ATi brings out a real 9.9 or 10.x driver, and you still got >=x11-drivers/ati-drivers-9.0 masked. (In reply to comment #9) > Now you created a mess. Either all the packages use the 10.x-like scheme, or > they all use the 8.xxx-like one. > Now portage thinks that the newest ebuild is an old one. Not me, i just followed the devs wishes. The previous beta of fglrx for ubuntu 9.10 was named with the 8.xxx version scheme for the reason i explained earlier. Ok, I told you it would go up in flames. Now I tried it, and it did: I updated xorg 1.7 not masked anymore, and this is what I get: !!! Multiple package instances within a single package slot have been pulled !!! into the dependency graph, resulting in a slot conflict: x11-drivers/ati-drivers:1 ('installed', '/', 'x11-drivers/ati-drivers-10.2', 'nomerge') pulled in by >=x11-drivers/ati-drivers-9.10 required by ('installed', '/', 'x11-libs/xvba-video-0.6.6', 'nomerge') (and 2 more) ('ebuild', '/', 'x11-drivers/ati-drivers-8.721', 'merge') pulled in by x11-drivers/ati-drivers required by ('ebuild', '/', 'x11-base/xorg-drivers-1.7', 'merge') x11-drivers/ati-drivers required by @world xvba-video is not a gentoo official package afaik. ad dep it has >=x11-drivers/ati-drivers-9.10 and of course this doesn't work with this ebuild since the version is 8.721 . But this issue should be reported to the mantainer of xvba-video. Anyway you can edit xvba-video ebuild and remove the version from the dep, this will solve your issue By the way: The ebuild internally acts, as if the driver would only support opengl 1.2. That’s certainly not true, is it? local ATI_ROOT=/usr/$(get_libdir)/opengl/ati # To make sure we do not miss a spot when these change. local libmajor=1 libminor=2 local libver=${libmajor}.${libminor} Created attachment 224549 [details] A quick hack to rename it to 10.4_beta, like it should be. (In reply to comment #13) Now it’s getting silly. ^^ As if that package were the only one. Missed the “(and 2 more)”? Please fix your stuff. I just modded the ebuild to be 10.4_beta, and am in the process of updating the system. We’ll see how it works out. Created attachment 224561 [details]
Output from failed patching, using ati-drivers-2.6.33.patch
The ebuild fails, since ati-drivers-2.6.33.patch can’t be applied.
Comment on attachment 224561 [details] Output from failed patching, using ati-drivers-2.6.33.patch (In reply to comment #16) Ok, forget it, I missed one of those "$majorversion -gt 8" checks. Created attachment 224567 [details]
This one installs fine.
Fixed.
Just have to start X with it and see if it works.
OK, I’m running the new driver on Xorg 1.7 right now!! Works as advertised. Yay :) Just to be sure, in case you run into trouble: I did not keep any of the files in /etc/ati/, did revdep-rebuild, and rebooted. Of course I had to reconfigure the displays via xrandr. But I prefer starting with a clean configuration. Also now the font sizes got much smaller, wich I have to fix too. But other than that, no problems yet. Beware: My ebuild is just a quick hack. I removed the check for the old slot of packages at two places, and used $REAL_PV (=8.721) instead of $PV. (A diff shows the changes.) So this ebuild is not meant to end up in portage just like that! Decide for yourself how you wanna do it. But the masking required with ati-drivers-8.721.ebuild WILL get you in trouble sooner or later. If unsure, wait until 10.4 is out of beta and officially in portage. :) thanks guys ! yeah, you don't need the other ati-drivers-2.6.33.patch any more - since ati-drivers-fix_compilation-bug-297322.patch provides the fix for the compilation-error with 2.6.33 ^^ I'll at earliest will be able to try that ebuild later this evening / night any experience yet with xvba and this driver ? does it work at all ? (I read in the thread over at phoronix.com/forums that it (xvba) seems to work with some cards of the 5xxx / Evergreen series is it stable enough for production ? in the past I always had made the experience, that the provided pre-release drivers for Ubuntu were rather unstable and poor in features (besides support for newer xorg-server or kernel versions) (In reply to comment #21) > yeah, you don't need the other ati-drivers-2.6.33.patch any more - since > ati-drivers-fix_compilation-bug-297322.patch provides the fix for the > compilation-error with 2.6.33 ^^ Ohoh… I still used that 2.6.33 patch. Could someone please work the new patch in, and remove the old one, if that is the case? > any experience yet with xvba and this driver ? does it work at all ? No idea how to test it. Would it accelerate H.264? Since I only have H.264 and XviD videos here. I heard it still does only support MPEG2… > (I read in the thread over at phoronix.com/forums that it (xvba) seems > to work with some cards of the 5xxx / Evergreen series I only have a HD 4860. > is it stable enough for production ? I’m running it for a couple of hours now. Switched to the console and back, between two different X sessions with different display setups, played a video, and am using a big desktop spread over two displays. The only thing I didn’t test is 3D acceleration (e.g. in games). But I think since KDE4 already uses that to render its compositing desktop, it should work. ;) (In reply to comment #22) > I only have a HD 4860. No, that card doesn’t exist. ^^ I have a HD 4850! Well, finally it is working. OpenGL is running fine and my fan is much more quiet. compiling fine against 2.6.34-rc2 Thanks for this contribution. @Lukas: it works also with 2.6.34-rc* ? that's amazing news ^^ this surely is a badass driver-release :) @Navid: I made a test-compile by manually unpacking, preparing and compiling the ebuild and it worked fine so ati-drivers-2.6.33.patch definitely shouldn't be needed anymore ... thanks for the info, btw ! afaik you need special builds of mplayer and other video players to be able to unleash or even access the UVD2 it should support several video formats, at least h264 (according to what I read at phoronix.com/forums) (In reply to comment #25) > this surely is a badass driver-release :) You mean just as badass as caring for your kids? ^^ (As Chris Rock would have said.) It’s supposed to work with xorg-1.7. That deserves no big tanks. :) > afaik you need special builds of mplayer and other video players to be able to > unleash or even access the UVD2 I know. I already tried that, a couple of month ago. I’ll find it. But I remember that it’s still very experimental. > it should support several video formats, at least h264 (according to what I > read at phoronix.com/forums) Well, I hope it’s not using the same buggy decoder that caused all my videos to have extreme whites and blacks (basically glowing and with nearly no contrast in-between) back then. We’ll see. assigning and renaming. old title: "[ebuild request] x11-drivers/ati-drivers-10.4pre (8.721) for Ubuntu released, support for xorg-server-1.7* !" By the way: While the drivers seems pretty stable, I have problems with KDE 4.4(.1) here. I can’t enable compositing, and while in the system settings for the desktop, I get crashes of X. No idea if this is because of the driver, or because of KDE4’s all-around crappiness, as this is the first time I try KDE4. Removed my old package.mask that prevented me from upgrading X.Org so far. Now running ~amd64 with ati-drivers-10.4_beta.ebuild with up to date X.Org. No issues whatsoever. Thank you so much for this ebuild! (In reply to comment #29) > Now running ~amd64 with ati-drivers-10.4_beta.ebuild with up to date X.Org. > No issues whatsoever. Do you have KDE 4.4.1 and compositing enabled? I can’t get it to work here with this driver. :/ Can someone confirm my issue, or show me wrong? My hardware is a Thinkpad W500 with Intel GM45 and Radeon Mobility HD3650 (FireGL) The internal card (intel) is switched off from the bios. I switched from 10.2 under xorg-server-1.6.5 to the new beta and xorg-server-1.7.6. With the old driver I could connect a external monitor via a DP-DVI-Adapter. Now xrandr doesn't work anymore. External Analog-VGA works, the displayport: zilch. No, yes I mean _no_ config-changes. Everything else works, the compiz lag is still visible, but at least we could switch to xorg-server-1.7. Is it my fault, or the drivers'? (In reply to comment #31) I think it’s way more complex than that. I can’t use compiz at all anymore, so I can’t directly confirm your compiz problem, but it shouldn’t be related. There are similar problems with kwm though. Plasma crashes often. And compositing fails to start, or starts and then when you go back to the configs it’s disabled again. Also configuring anything related to kwm makes the screen flicker twice and then often crash plasma or even X as a whole. Now the important part for your case: After such incidents, the xrandr settings are completely reset to a very strange setup (1024x764 on the first analog screen, and a 1024x768 desktop on a 1600x1200 resolution [yes, really, wtf?] on the second screen. You also have to change it *twice* in the settings, or the first screen will be cut off at the height of 786 or so. But it can just as well be a KDE error. (4.4.1 here) It’s just extremely annoying to have all those crashes. hi, thx for the updated ebuild, it seems to work fine here (kde3,even kaffeine doesn't crash anymore with "xv" and dvbt). I never used it, but the catalyst control center throws this error: $> amdcccle amdcccle: error while loading shared libraries: libQtGui.so.4: cannot open shared object file: No such file or directory (In reply to comment #33) I hope you’re doing your revdep-rebuilds (and perhaps emerge -a1tv @preserved_rebuild) after your updates. I’d do that right now, because it smells just like that. :) (In reply to comment #33) > $> amdcccle > amdcccle: error while loading shared libraries: libQtGui.so.4: cannot open > shared object file: No such file or directory > well you are right there is a missing RDEPEND x11-libs/qt-gui which is needed by amdcccle only. I might work on a patch for this (In reply to comment #35) > well you are right there is a missing RDEPEND x11-libs/qt-gui which is needed > by amdcccle only. I might work on a patch for this > That would be a good idea. Maybe just make the complete control-center optional? I am running a qt3-only system, and don't want to install the whole qt4-* stuff just because of this driver... (In reply to comment #36) > That would be a good idea. Maybe just make the complete control-center optional? I second that. It would be analog to Windows, where the catalyst control center also is optional. Also, it’s the Gentoo spirit of freedom trough choice. (In reply to comment #36) > That would be a good idea. Maybe just make the complete control-center > optional? I am running a qt3-only system, and don't want to install the whole > qt4-* stuff just because of this driver... take a look at bug 310993 :) (In reply to comment #37) > Also, it’s the Gentoo spirit of freedom trough choice. No no no no no no no, and for the last time, no. Gentoo isn't about choice, it's about power. We don't _have_ to give users all the freedom, but we do give users the _power_ to make their own changes easily (with overlays, for example). Say it with me: Gentoo is about empowering users. Sorry for the slightly off-topic rant, it had to be said. Thanks (In reply to comment #39) > Say it with me: Gentoo is about empowering users. Sorry to reply, but there is no power without choice. As power is defined as having the choice. :) (Don’t try logic on me. I’m a professional. ^^) But since freedom does not mean that you *have* to add options in there, but only that you won’t limit the user (intentionally), the spirit of “Gentoo is about empowering users.” is still correct. :) (In reply to comment #40) > But since freedom does not mean that you *have* to add options in there, but > only that you won’t limit the user (intentionally), the spirit of “Gentoo > is about empowering users.” is still correct. :) We _do_ intentionally limit choices when we think it makes sense. But the freedom you speak of comes out of us making it easy to bypass our choices. We _encourage_ users to do mini-forks of Gentoo by having their own ebuilds in their own overlays when we restrict choices intentionally. Cheers Thank you all, got a working 10.4_beta used for ATI Radeon HD 4890, kernel 2.6.33 and xorg-server 1.7. *thumbs-up* compiles fine against 2.6.34-rc3-git2, but errors against 2.6.34-rc3-git4 New beta release: https://lists.ubuntu.com/archives/lucid-changes/2010-April/010166.html Created attachment 227425 [details] ati-drivers-8.723.1 ebuild This is the ebuild for the new beta release for ubuntu. Tested with xorg-server-1.7.6 on amd64 no-multilib. see bug 308521 for the keyword file used to unmask xorg. (In reply to comment #45) Yay, I just noticed that I don’t need that anymore, since the open-source radeon drivers are good enough to free you from the chains of fglrx (tested with an R770 aka HD 4850, compiz, KDE4, big desktop, video, 3d)! Finally! :) If you got a HD 48xx card, and want to switch, ask me. But in return you have to fill some wiki somewhere with everything you get from me. ^^ (In reply to comment #46) > (In reply to comment #45) > Yay, I just noticed that I don’t need that anymore, since the open-source > radeon drivers are good enough to free you from the chains of fglrx (tested > with an R770 aka HD 4850, compiz, KDE4, big desktop, video, 3d)! Finally! :) > If you got a HD 48xx card, and want to switch, ask me. But in return you have > to fill some wiki somewhere with everything you get from me. ^^ Thank you very much for the offer. I already use the radeon driver from a long time. I just try to help to mantain the fglrx driver since i found it usefull for some needs Created attachment 227663 [details]
patch from 8.721 to 8.723.1
There is a newer beta out (version 8.723.1), here is a diff to the ebuild in the attachment (ati-drivers.patch). (In reply to comment #49) > There is a newer beta out (version 8.723.1), here is a diff to the ebuild in > the attachment (ati-drivers.patch). Hmm sorry I missed the ebuild for 8.723.1 :( Does it have xorg-server 1.8 support? Apparently Arch Linux has support for it: http://www.phoronix.com/forums/showthread.php?t=23082 (In reply to comment #51) > Does it have xorg-server 1.8 support? > > Apparently Arch Linux has support for it: > http://www.phoronix.com/forums/showthread.php?t=23082 > Officially they don't. Because the ati-driver detects the X-Version wrong, when running startx. On Arch they have patched the xorg-server-1.8 package with another version number so the ati-driver works. For gentoo i have made an ebuild including the patch, and it seems to work. sorry, forgot to mention: see this invalid bug: http://bugs.gentoo.org/show_bug.cgi?id=315415 09 Apr 2010; Christian Faulhammer <fauli@gentoo.org> xorg-server-1.7.6.ebuild: stable x86, bug 308521 16 Apr 2010; Pacho Ramos <pacho@gentoo.org> xorg-server-1.7.6.ebuild: amd64 stable, bug 308521 With xorg-server marked stable on two main arches, it would be a pleasure to see this in portage as ati-drivers-10.4_betaX, so that fglrx-users can easily unmask and use it. Using patch of comment #53, I can confirm a working ebuild, although I needed to fix both the patch and the upper ebuild: 1) xorg-server patch (removing line breaks after the version of the line to be changed) 2) applying patch of comment #48 3) Removing depency from ebuild: !>=x11-base/xorg-server-1.8.0 On my system, with a HD5750 card, both 8.721 and 8.723.1 freeze my system when starting X. uname -a: Linux jeroen 2.6.33-gentoo-r1 #1 SMP PREEMPT Fri Apr 23 20:57:52 CEST 2010 x86_64 AMD Phenom(tm) II X4 955 Processor AuthenticAMD GNU/Linux This is what shows up in slim.log (not in Xorg.0.log, for some weird reason) X.Org X Server 1.7.6 Release Date: 2010-03-17 X Protocol Version 11, Revision 0 Build Operating System: Linux 2.6.31-gentoo-r6 x86_64 Current Operating System: Linux jeroen 2.6.33-gentoo-r1 #1 SMP PREEMPT Fri Apr 23 20:57:52 CEST 2010 x86_64 Kernel command line: root=/dev/ram0 real_root=/dev/sda3 Build Date: 22 April 2010 09:37:42PM Current version of pixman: 0.17.2 Before reporting problems, check http://wiki.x.org to make sure that you have the latest version. Markers: (--) probed, (**) from config file, (==) default setting, (++) from command line, (!!) notice, (II) informational, (WW) warning, (EE) error, (NI) not implemented, (??) unknown. (==) Log file: "/var/log/Xorg.0.log", Time: Fri Apr 23 21:18:13 2010 (==) Using config file: "/etc/X11/xorg.conf" (EE) Failed to load module "xtrap" (module does not exist, 0) (WW) fglrx: No matching Device section for instance (BusID PCI:0@0:0:0) found (WW) fglrx: No matching Device section for instance (BusID PCI:0@0:2:0) found (WW) fglrx: No matching Device section for instance (BusID PCI:0@0:6:0) found (WW) fglrx: No matching Device section for instance (BusID PCI:0@0:17:0) found (WW) fglrx: No matching Device section for instance (BusID PCI:0@0:18:0) found (WW) fglrx: No matching Device section for instance (BusID PCI:0@0:18:1) found (WW) fglrx: No matching Device section for instance (BusID PCI:0@0:18:2) found (WW) fglrx: No matching Device section for instance (BusID PCI:0@0:19:0) found (WW) fglrx: No matching Device section for instance (BusID PCI:0@0:19:1) found (WW) fglrx: No matching Device section for instance (BusID PCI:0@0:19:2) found (WW) fglrx: No matching Device section for instance (BusID PCI:0@0:20:0) found (WW) fglrx: No matching Device section for instance (BusID PCI:0@0:20:1) found (WW) fglrx: No matching Device section for instance (BusID PCI:0@0:20:2) found (WW) fglrx: No matching Device section for instance (BusID PCI:0@0:20:3) found (WW) fglrx: No matching Device section for instance (BusID PCI:0@0:20:4) found (WW) fglrx: No matching Device section for instance (BusID PCI:0@0:20:5) found (WW) fglrx: N So, appearantly, it freezes while detecting the card... Update: I fixed the freeze with # eselect opengl set ati Hello. Can someone elaborate on how to: emerge ati-drivers-10.4_beta thanks! 10.4 is out: Release notes: https://a248.e.akamai.net/f/674/9206/0/www2.ati.com/drivers/linux/catalyst_104_linux.pdf Driver: https://a248.e.akamai.net/f/674/9206/0/www2.ati.com/drivers/linux/ati-driver-installer-10-4-x86.x86_64.run Haven't tried it yet. (In reply to comment #59) > 10.4 is out: Used slightly modified ebuild for 10.3. The driver works fine with xorg-server-1.7.6 (and backfill patch for smooth resizing). ati-drivers-10.4 builds and works wonderfully well with 2.6.34 kernel series but it needs an additionnal patch (slab.h inclusion in kcl_wait.c). Good news : some guys also reported that this version is functionnal with Xorg 7.6 (xserver 1.8) but the xserver version must be faked in its configuration files (rebuild required). :) (In reply to comment #54) > With xorg-server marked stable on two main arches, it would be a pleasure to > see this in portage as ati-drivers-10.4_betaX, so that fglrx-users can easily > unmask and use it. > It is a bit late since 10.4 is in portage now but ati-drivers-8.721 (aka the 10.4 beta) was added 21 Mar 2010 Since ati-drivers-10.4 is in the tree, I think ati-drivers-8.721 should be removed from tree to avoid more confusion and this bug should be closed. As per previous comment. |