Description of problem: Installation program hangs when booting into the graphical installation when using a ATI Radeon 9800 Pro using the auto-detected drivers. Version-Release number of selected component (if applicable): How reproducible: Place in boot CD and hit enter at the install prompt to use the graphical installation. Steps to Reproduce: 1. 2. 3. Actual results: System hangs. Expected results: Additional info: Installation continues if you choose 'linux text' at the install prompt. But will hang again at 'first boot' after install.
Work around so far is to use 'linux rescue', chroot, change /etc/inittab to reboot into text mode. Have the ATI made driver RPM for the 9800 and install. Run the 'fglrxconfig' utility that is included with the drivers to config the monitor system. After that, change back the /etc/inittab and reboot. Everything should be fine then. Will have 2D AND 3D support with FSAA etc etc.
It`s also easyier to do following instead using of using the boot rescue mode: Just hit a at the grub prompt and append "3 nogb" to the kernel parameters... 3 tells the system to only start into runlevel 3 (without xdm, so no xfree server is started) nogb tells the system that you do not want to use graphical boot (so x doesnt get started...) Afterwards you can do the changes you explained above.. but you dont need the cd to do it... I think this way is easier...
This is a duplicate of bug 109256
I have some questions regarding Scott Grants post. I need to change my card back the VESA as well so I can actually use Linux. But I'm quite new to Linux and need some help. 1. How do you "change /etc/inittab to reboot into text mode"? 2. I have the ATI driver on a floppy can I install from floppy? 3. How do I "Run the 'fglrxconfig' utility that is included with the drivers to config the monitor system" 4. How do I "After that, change back the /etc/inittab and reboot" (I know how to reboot). Thanks very much, I am not running a Radeon 9800 but a Radeon 9600 XT but I still need to start my Linux up.
This bug should probably be updated to Version: test1 as this still happens in the FC 1.90 release. lspci: 0000:01:00.0 VGA compatible controller: ATI Technologies Inc Radeon R350 [Radeon 9800] 0000:01:00.1 Display controller: ATI Technologies Inc Radeon R350 [Radeon 9800] (Secondary) Upon loading the graphical installer it freezes with a white screen with a mouse pointer on it. It can be installed using: linux text at the boot prompt, then after it finishes installing booting to init 3 and changing the radeon driver to vesa I have had success using the XFree86 4.4.0 pre-release radeon driver with this card. Any chance of a driver backport into FC's XFree86 4.3?
Yes, this problem is still present, as I haven't touched the Radeon driver in many months. I will update all Radeon related bug reports that are open in bugzilla when I have a chance to spend time on these issues. It wont be for a while at least. The 4.4.0 driver is drastically changed, it isn't just some small updates. Updating the driver would destabilize RHEL 3 potentially, and I'm not willing to do that. I need to keep one single src.rpm building between all OS releases we support, to minimize maintenance.
I've been having the same problem with all FC releases (All-In-Wonder 9800 Pro), but it works in RedHat 9 where the card is detected as generic. Couldn't you guys have a command-line option like "linux forcegeneric" so that we don't have to use the text install anymore?
I've found that if I enable the option: Section "Screen" Option "XaaNoScreenToScreenCopy" It solves all the lockups that I've had on my 9800 and 9600XT, plus I get the added benefit of using a refresh rate above 60 Hz. Sadly this Xaa option also seems to cause the greatest slowdown to this driver. I used the list of options I found here to try and determine the cause: http://www.xfree86.org/current/XF86Config.5.html#sect10 I also have a config that works most of the time, with the exception of opening Mozilla Mail, that causes it to freeze. Section "Screen" #Option "XaaNoScreenToScreenCopy" Option "XaaNoImageWriteRect" Option "XaaNoOffscreenPixmaps" Option "XaaNoSolidFillRect" Anyone else care to give it a try?
This is really disappointing that it can't be fixed. I'm currently using RedHat Professional Workstation and it has the same issue. In fact, there it's even worse, X won't even start. Fedora Core 1.90 obviously doesn't work either as stated above. I used to be able to use RedHat 9 with the 9800 by overriding the Chipset ID and forcing X to think it was on of the older radeon 9x00's. That allowed me to get fully working 2D without using ATi's shoddy drivers.
> This is really disappointing that it can't be fixed. Nobody's said it can't be fixed. It can be fixed, it just is not fixed right now currently. I'm currently using RedHat Professional Workstation and it has the same issue. In fact, there it's even worse, X won't even start. >Fedora Core 1.90 obviously doesn't work either as stated above. > >I used to be able to use RedHat 9 with the 9800 by overriding the >Chipset ID and forcing X to think it was on of the older radeon >9x00's. That allowed me to get fully working 2D without using ATi's >shoddy drivers. Radeon 9200, 9600, 9800 are not supported at all in any way by XFree86 4.3.0. XFree86 4.3.0 was released on February 27, 2003, a month or two before the ATI Radeon 9200/9600/9800 hardware was publically available, and as such, that hardware is not supported in 4.3.0 at all. ATI contributed patches to XFree86.org to add experimental support for the new hardware, which got committed to CVS in April or thereabouts. That code was just _preliminary_ support for the new hardware, and not complete in any way. I backported this code to 4.3.0 and made some additional fixes. I later backported further ATI contributed code as well, however ALL of this code is experimental only, and not considered "complete". Months later, ATI came out with another new generation of variants of the original 9200/9600/9800 with "SE" and "XT" suffixes. Later they contributed patches to provide initial support for these new cards also. All of that support is in XFree86 CVS as well, and again, I've backported some of it. Again, it is _EXPERIMENTAL_. So as it currently stands, neither XFree86.org's official XFree86 4.3.0 release, nor Red Hat "officially" support the Radeon 9200, 9600, or 9800 in any way. The driver support for these chips is expermental only, and provided "as-is" in hopes that it works for someone, and in order to have people test it and provide feedback so that problems can be fixed in a future "official" release which officially supports this hardware. The problem however, is that by adding "experimental" support in this fashion, many people _expect_ it to work, and to work flawlessly, and get very upset if it does not. Many people also demand that problems get fixed immediately, or as soon as possible, assuming that because the driver detects their card, that the card is "supported". I must admit, that this was a huge mistake on my part, because now there are many users out there who are upset that the driver does not work on their card, and are misled into believing that their card is supported, due to the config tools autodetecting and autoconfiguring the hardware. In order to correct this problem, I will no longer be including experimental driver support for new hardware in our XFree86 releases, until I can personally test each and every card, and either fix problems found myself, or obtain fixes from ATI or other vendors, or elsewhere. All unsupported hardware, will either be redirected to the generic unsupported "vesa" driver in the future, or if we know that vesa doesn't work, the hardware will remain unmapped to a driver. I sincerly apologize to all Radeon users affected by the current Radeon driver's instabilities on unsupported hardware, for the inconvenience that my personal efforts as a volunteer on the weekend to backport Radeon driver support to XFree86 4.3.0 have failed to result in a flawless user experience for many Radeon 9200/9600/9800 users. I have learned the hard way, that it is better to have no support at all, than to have something experimental that does not work for everyone, but conveys the incorrect message to them that it is a supported. I will make my best efforts very soon, to either fix the Radeon driver, or to remove all Radeon support patches for newer hardware in which I know it does not work properly due to the bug reports received. Again, my apologies.
Oops, the following comment in my last comment was supposed to be a quote from comment 9 above, which I forgot to add ">" to the beginning of. "I'm currently using RedHat Professional Workstation and it has the same issue. In fact, there it's even worse, X won't even start." Just clarifying here, that that was a quote, and not my own words.
It's not your fault the drivers aren't fully compatible. I do think that at least for the initial installation that experimental drivers should not be used, or at least some sort of warning message: "The drivers for your video card are experimental, would you like to use the experimental driver or a generic driver?"
*** Bug 116285 has been marked as a duplicate of this bug. ***
I once was also one of the Fedora/Redhat newbies that opened a bug about this issue and said there that this should be blocking the Core 1 Release and so on... Mike Harris: I am sorry that this is happening to you, but you should know that this is not a personal offence :/ People can`t know what you did to try to improve the XFree used in Fedora... They really just can't know that! You should not feel offended by bugs like this and like mine (months ago). I know you are a good guy and hope you won't take it too personal ;) Good luck to you and your coding ;)
I notice that Core 2, Test 2 seems to work with the Radeon 9800. Has this bug been fixed or is it just luck that things seem to work now?
Neither actually. Fedora Core 1 ships with XFree86 4.3.0. Stock 4.3.0 does not support the Radeon 9800 at all because the Radeon 9800 did not exist when 4.3.0 was released. Red Hat XFree86 4.3.0 has some patches added to it taken from XFree86 CVS after 4.3.0 was released, which were added as experimental code to try to get the 9800 working. It does work on some Radeon 9800 models, but not all. That is not really a bug per se, because the experimental patches are not officially supported by Red Hat. On the luck side of your question... it is not luck, it is because FC2t2 does not have XFree86 4.3.0 anymore. FC2t2 ships with an entirely brand new X server implementation - Xorg X11, which is based on XFree86 4.4.0 prior to the license change. Xorg X11 does support the various Radeon 9800 variants, so you'll probably find it works much better. Hope this helps. Thanks for testing.
Possible dupe of bug 121604?
That teaches me to do 10 things at once. Bug 121604 is a dupe of this bug. Sorry, I got mixed up.
Radeon 9800 and various variants should now work correctly in Fedora Core development, and should also autodetect with our config tools. Please test Fedora Core 3 test2 beta test release and if there are problems, please open a new bug report against Fedora Core 3 test2. Thanks in advance. Setting status to "RAWHIDE"