Fixed in next set of isos

Status: NEW => RESOLVED
CC: (none) => ennael1
Resolution: (none) => FIXED

sorry wrong bug updated...

Status: RESOLVED => REOPENED
Resolution: FIXED => (none)

I'd like to see the /var/log/Xorg.0.log of the first boot, and the /var/log/Xorg.0.log and dmesg after the second try after you reconfigured X.

Created attachment 2010 [details]
Xorg.0.log before config change

Created attachment 2011 [details]
prefdm.service show

Created attachment 2012 [details]
prefdm.service status

Created attachment 2013 [details]
Xorg.0.log after config change

Created attachment 2014 [details]
dmesg

OK, so it's not a problem I thought it could be (which is good). It seems things are properly starting. Just need to work out why VESA gets an IO error trying to read the vbios.

I don't know much about what would cause such an IO error. I doubt it's because e.g. plymouth is still running or similar such issues, but I'll see if I can test in a vbox.

The strange thing, when configuring the Sandy bridge integrated intel graphics on the same device, it is running.
So it must be connected to the ati card.

Magnus did report similar problems on pre release testing on his machines.

CC: (none) => magnus.mud

What does the xorg.conf look like initially? Does it have a BusID specified in it?

I suspect i915 gets loaded and blocks "vesa" then. Can you give your "lspci -nn | VGA", and if possible, also the following:

1. change the config back to vesa (and nokmsboot)
2. uncomment DEBUG_DISPLAY_DRIVER_HELPER=yes line in the beginning of /sbin/display_driver_helper
3. reboot
4. copy /dev/ddh_debug and dmesg somewhere (use e.g. ssh?), and then attach them here.

The "radeon" configuration you tried doesn't seem to work because the intel card is selected for display instead of radeon, so radeon can't display anything (PowerXpress). I think vga_switcheroo could probably be used to switch the active card, I'm not sure if we need some glue somewhere to make it work automatically. Hmm, or maybe we should just disable loading i915 if radeon is configured in xorg.conf, maybe it would work automatically then... Will probably have to test this, but lets investigate the "vesa" issue first with the above steps.

Created attachment 2017 [details]
dmesg using vesa

Created attachment 2018 [details]
ddh_debug

Created attachment 2019 [details]
xorg.conf using vesa

The requested output of lspci is:
00:02.0 VGA compatible controller [0300]: Intel Corporation 2nd Generation Core Processor Family Integrated Graphics Controller [8086:0116] (rev 09)
01:00.0 VGA compatible controller [0300]: ATI Technologies Inc Seymour [Radeon HD 6400M Series] [1002:6760]

To reconfigure to vesa, I had to uninstall radeon-firmware first. With installed radeon-firmware, the xorg.conf entry was "ati", no matter what I did.

When booting with vesa and nokmsboot, I didn't get a tty (as on firt boot of this installation), only when booting without nokmsboot, I get a tty.

Comment on attachment 2019 [details]
xorg.conf using vesa

(for future reference can you make sure to pick text/plain mime type when uploading text attachments without a .txt suffix - thanks :))

Attachment 2019 mime type: application/octet-stream => text/plain

(In reply to comment #13)
> Created attachment 2017 [details]
> dmesg using vesa

The log says the system was not booted with 'nokmsboot', which seems to be causing the issue (the current issue, at least). How did you configure the 'vesa' driver? What does your /boot/grub/menu.lst look like?

As said in comment 16, when booting with nokmsboot, the boot hangs and I don't even get to a tty, that's why I manually disabled nokmsboot on grub prompt the second time I booted with vesa.

When I booted with installed and configured fglrx and "rdloaddriver=radeon" as tmb asked on the dev ml the system did boot correctly, including X.

I can retry again using vesa and nokmsboot, and try to get the logs, but it will need some time.

What do you mean with "boot hangs"?
That you can't even SSH in?

If so, please look in /var/log/syslog for the log entries of the failed boot and attach that part here.

If you can still SSH in, I do need to see the dmesg and ddh_debug of the failed boot, in order to see where the issue is.

I've done a fresh network install of cauldron as of this morning (EST), on a system with an RS780M/RS780MN [Mobility Radeon HD 3200 Graphics], Vendor ID 0x1002 Device ID 0x9612 Subvendor ID 0x1179 Subdevice ID 0xffe2, and the boot does not hang.  plymouth hangs on tty1 and doesn't end, but getty is available on tty2/3/etc.

Going to one of those to examine Xorg.0.log shows that X is segfaulting.  I'll attach the log.

CC: (none) => ftg

Created attachment 2032 [details]
Xorg.0.log for segfaulting X.

(In reply to comment #21)
> I've done a fresh network install of cauldron as of this morning (EST), on a
> system with an RS780M/RS780MN [Mobility Radeon HD 3200 Graphics], Vendor ID
> 0x1002 Device ID 0x9612 Subvendor ID 0x1179 Subdevice ID 0xffe2, and the boot
> does not hang.  plymouth hangs on tty1 and doesn't end, but getty is available
> on tty2/3/etc.

For what's it work, although there is little we can do here, I have talked to upstream GDM folks about how to handle this case better. We agreed that when X fails GDM should exit with an error allowing all the subsequent OnFailure actions to kick in etc. Right now it's kinda lost inside GDM, KDM etc.

Anwyay, when we do this change, we'll propose it to KDE folks too so hopefully we can get better handling when things fail (e.g. show a nice Plymouth based graphic when things go wrong - kinda like how this page suggests for some failure conditions: https://live.gnome.org/GnomeOS/Design/Whiteboards/ProblemReporting)

clean beta3 install, runlevel 3, xserver will not start when doing a startx.
still fails after adding radeon.modeset=0 to kernel boot line.

Hardware: i586 => x86_64
CC: (none) => junk_no_spam

Created attachment 2033 [details]
xorg log

Well, as much as I like a nice error message, a fix for the problem would be better.
But of course such an error message is always a good idea.
Next to kdm and gdm we should not forget lxdm folks as well.
lxdm ist standard dm for lxde and xfce installations and for dual cd.

I have the same problem as Frank - a segfault in the X.org server. There is this highly suspicious line immediately before the crash:

[    28.424] (II) RADEON(0): Setting screen physical size to 444 x 277

and I see something very similar in Frank's log.
originally 
This seems to be a different bug to the one under discussion here - should we take this to a separate bug report?

CC: (none) => mageia

Further experiments show:

  - if the vesa driver is selected, all is well
  - if the radeon driver is selected AND the radeon-firmware package is installed
      - if kernel mode setting is enabled, all is well
      - if kernel mode setting is disabled, the X server crashes on startup

So it seems, at least with my hardware, that the radeon driver will now only work if kernel mode setting is enabled AND the firmware is installed (whereas in beta 2 it could be made to work either by disabling kernel mode setting OR by installing the firmware.

My hardware is reported as:

ATI Mobility Radeon HD 3430 (ChipID = 0x95c2)

(In reply to comment #27)
> I have the same problem as Frank - a segfault in the X.org server. There is
> this highly suspicious line immediately before the crash:
> 
> [    28.424] (II) RADEON(0): Setting screen physical size to 444 x 277
> 
> and I see something very similar in Frank's log.
> originally 
> This seems to be a different bug to the one under discussion here - should we
> take this to a separate bug report?

That's a normal message. Mine has
[   162.357] (II) RADEON(0): Setting screen physical size to 380 x 285
which is the physical size of the display in millimetres.

With my card, it works fine.
Card:ATI Radeon X1950 and earlier: ATI Technologies Inc|RV280 [Radeon 9200 SE] [DISPLAY_VGA] (rev: 01)

CC: (none) => davidwhodgins

(In reply to comment #29)
> (In reply to comment #27)
> > I have the same problem as Frank - a segfault in the X.org server. There is
> > this highly suspicious line immediately before the crash:
> > 
> > [    28.424] (II) RADEON(0): Setting screen physical size to 444 x 277

> That's a normal message. Mine has
> [   162.357] (II) RADEON(0): Setting screen physical size to 380 x 285
> which is the physical size of the display in millimetres.
> 
You're right, I didn't read that message carefully enough. However, the segfault isn't normal!

Fails with M2 beta3 live CD on my hardware.  I have not tried a real DVD installation.  

[root@live ~]# lspci -s 1:
01:00.0 VGA compatible controller: ATI Technologies Inc RV370 [Sapphire X550 Silent]
01:00.1 Display controller: ATI Technologies Inc RV370 secondary [Sapphire X550 Silent]

CC: (none) => 0123peter

I've now tested on a second machine, which has hybrid graphics:

  Intel Corporation 2nd Generation Core Processor Family
  Integrated Graphics Controller [8086:0116]

  ATI Technologies Inc Whistler [AMD Radeon HD 6600M Series]
  [1002:6741]

If you choose to just configure the IGP, the machine boots and starts X with no problems. If you then install radeon-firmware, and configure to use the free radeon driver, the machine boots with no problems (using the IGP), but fails to start X, due to a segfault.

If you boot with the rdloaddriver=radeon option, the boot hangs for a number of seconds whilst trying to load the radeon firmware, with the following messages (from dmesg):

[    2.694668] [drm] Loading TURKS Microcode
[   62.917595] ni_cp: Failed to load firmware "radeon/TURKS_pfp.bin"
[   62.926893] [drm:evergreen_startup] *ERROR* Failed to load firmware!
[   62.936103] radeon 0000:01:00.0: disabling GPU acceleration

then continues, but then fails again to start X, with the following messages in Xorg.0.log:

[    82.171] (II) Loading /usr/lib64/xorg/modules/drivers/radeon_drv.so
[    82.172] (II) [KMS] drm report modesetting isn't supported.
[    82.172] (II) GPU only supported with KMS, using vesa instead.
[    82.172] (II) UnloadModule: "radeon"
[    82.172] (II) Unloading radeon
[    82.172] (EE) Screen(s) found, but none have a usable configuration.

Note that from dmesg:

[    1.139809] [drm] Initialized drm 1.1.0 20060810
[    1.200624] [drm] radeon defaulting to kernel modesetting.
[    1.211195] [drm] radeon kernel modesetting enabled.

so it must be the subsequent failure to load the firmware that is causing the problem.

Also note that when booting without the rdloaddriver=radeon option, the firmware is loaded successfully later in the boot process.

Let me know if you want more logs.

Did you regenerate your initrd to include the firmware in it?

The radeon-firmware package does this during post-install. Is there a command to list the contents of the initrd, to check the firmware has really been included?

lsinitrd

# lsinitrd | grep TURKS
-rw-r--r-- 1 root root  5504 Apr 22 12:27 lib/firmware/radeon/TURKS_me.bin
-rw-r--r-- 1 root root 24096 Apr 22 12:27 lib/firmware/radeon/TURKS_mc.bin
-rw-r--r-- 1 root root  4480 Apr 22 12:27 lib/firmware/radeon/TURKS_pfp.bin

(In reply to comment #21)
> I've done a fresh network install of cauldron as of this morning (EST), on a
> system with an RS780M/RS780MN [Mobility Radeon HD 3200 Graphics], Vendor ID
> 0x1002 Device ID 0x9612 Subvendor ID 0x1179 Subdevice ID 0xffe2, and the boot
> does not hang.  plymouth hangs on tty1 and doesn't end, but getty is available
> on tty2/3/etc.
> 
> Going to one of those to examine Xorg.0.log shows that X is segfaulting.  I'll
> attach the log.

Just for completeness, I tested a few other scenarios here:

- nokmsboot / radeon driver / no firmware = segfault
- kmsboot + radeon driver / no firmware = kernel loop
- nokmsboot / vesa driver / no firmware = normal operation

Does this problem only occur when we have both Intel and AMD graphics hardware installed?  If that is true and since it's a release blocker, maybe we should change the bug title.  I can't tell if Frank Griffin and Peter D also have an Intel graphic hardware.

FWIW, the Beta 3 DVD install worked wonderfully on my machine with integrated HD6310.  It booted into KDE, I installed the radeon-firmware package and later reboot didn't give me a problem either.

CC: (none) => jeffrobinsSAE

(In reply to comment #38)
>I can't tell if Frank Griffin and Peter D also have an
> Intel graphic hardware.

I don't think so.  At least, harddrake only shows the ATI video chip.  It seems, from this and other related bugs, that some ATI chips are Ok and others not.  IIRC, there is no simple cutoff such that "this chip and earlier fail, but later chips are OK".

Agreed. I have one machine with ATI only (HD3430M) and one with Intel+ATI (HD6600M), and both have problems (as described above).

Just to make matters worse, the latest fglrx driver also has problems (bug #5559).

It seems that I actually got bitten by bug 5518.  

A real install from a DVD using the free drivers works.  There are not Intel graphics.

Today during 32 bit HTTP install on RV200 (7500 AGP device 0x5157) I saw this message about missing radeon-firmware. Detection must be broken, as there is no firmware required for this ancient chip. Plain old radeon is working fine with KMS @ 1600x1200x24bpp@85 on a CRT, with or without xorg.conf.

CC: (none) => mrmazda

(In reply to comment #42)

Even RV200 needs firmware (R200_cp.bin or R100_cp.bin) for full functionality (3D acceleration and such, I guess). It is true that KMS should work fine with such cards even without firmware, but there were reports of firmware loading failure causing KMS boot issues for even such cards, so firmware-less KMS was disabled at least for MGA2, just to be on the safe side.

I still see the same fault on the machine with ATI only (HD3430M) using the M2 RC DVD. Without radeon-firmware installed, the radeon driver segfaults when starting X, even with the nokmsboot option. As noted above, the nokmsboot workaround worked for this card prior to beta 3.

I will test on the machine with hybrid Intel/ATI tomorrow.

kde livecd rc fails to start X with ATI/intel card.
Tested boot with 
1) xdriver=radeon works, slow boot but X starts
2) xdriver=fglrx does not work, X does not start
3) xdriver=intel works, boot a little quicker

Created attachment 2241 [details]
boot without any additional parameters, X does not start

Created attachment 2242 [details]
boot with xdriver=fglrx X does not start

I still see the same fault on the machine with hybrid Intel/ATI (Radeon HD6600M):

1) Initial install using the intel driver, no radeon firmware, works OK.
2) Switching to the radeon driver, with radeon firmware installed, boots OK, but X fails to start (segfault in radeon driver).
3) Switching to fglrx driver, works OK.

Also, installing the radeon firmware but continuing to use the intel driver results in an increase in power consumption from ~27W to ~50W. Turning the ATI card off via vgaswitcheroo reduces this to ~44W. Using the fglrx driver with the ATI card enabled reduces this to ~19W.

Created attachment 2279 [details]
Xorg.log that ends up with a segfault on radeon HD 4570 only laptop (no firmware + nokmsboot + radeon.modeset=0)

X server segfault here on a fresh install of mga2rc from DVD on a ATI only (HD 4570) laptop. No firmware installed. noksmboot + radeon.modeset = 0.

Created attachment 2280 [details]
dmesg assocaited with previous Xorg.log (HD 4570, no firmware)

Associated dmesg

Do not know if this is the same: regression: it have become very slow.
Until a couple week ago ATI graphics was very fast on my thinkpad T40.
That system borked and is let alone for bug tracking, so i cloned another thinkpads cauldron disk, updated and configured graphics the same, but it is *very* slow.
Xorg log seem OK though.  Strange.

CC: (none) => fri

(In reply to comment #51)
> Do not know if this is the same: regression: ATI graphics has become very slow.

I saw that problem on beta3 with all updates and proprietary driver. 

teapot and glxgears would not even run.

Clean Mageia-2-rc-x86_64-DVD.iso install + updates + nonfree firmware + fglrx driver and urpme --auto-orphans y cleaned up the problem. Disabling Desktop Effects gets me the fastest frames per second of 186 according to teapot app.
Still not as good as Mandriva 2011.0's 220. :(

Disabling Desktop Effects also cleared up intermittent dropped escape/control xterm sequence characters.

CC: junk_no_spam => (none)

Clean install on a 64-bit system.

The motherboard is an Asus M4A785TD-V EVO with an Ati HD4200 video system on board, running under an AMD Phenom II X4 955 Black Edition (4 core) with 8GB of RAM. 

Set graphics to nonfree-vesa/fglrx. 

On booting, entered non-GUI login, was asked to authenticate myself (again) found myself in a dialog telling me my system doesn't support 3D desktop effects and inviting OK or cancel. Click OK, get told to log out so the changes can take effect 
and ...

Back to the log in!

So far no solution.21:03 UK time 13 May 2012

CC: (none) => alan

btw you can try a test-iso https://www.mageia.org/pipermail/mageia-dev/2012-May/015646.html

Trying the test ISO:

1) On my machine with ATI only graphics, it automatically selects the fglrx driver when booting the live CD. This of course avoids the bug. Doing an installation from the live CD and then manually selecting the radeon driver also gives a working system, but this is not surprising, as the radeon firmware is installed.

2) On my machine with hybrid Intel/ATI graphics, it automatically selects the radeon driver when booting the live CD. X fails to start due to a segfault, exactly as described before.

*** Bug 5859 has been marked as a duplicate of this bug. ***

CC: (none) => lebarhon

so according to the last changes/packages, this is will not be fixed with the release, any volunteer to write something exhaustive, complete and precise in the errata ?

I can write what I have done as a workaround. (bug 5859)
Is it correct ?

(In reply to comment #58)
> I can write what I have done as a workaround. (bug 5859)
> Is it correct ?

I don't think installing the radeon-firmware is necessary if you are going to install the proprietary driver. And your write-up is probably only suitable for experienced users.

I would just say:

"The Free DVD and CD are known to generate a non-working installation on many systems with Radeon graphics. If you are not an experienced Linux user, we recommend you use one of the Live CDs to install Mageia 2, and when asked, choose to use the proprietary graphics driver."

thanks to all propositions that I have seen in 3-4 bugs, I have beginning to write something: https://wiki.mageia.org/en/Mageia_2_Errata#Proprietary_card

feel free to complete, thanks too all ;)

Actually, think some warning like this should be right on the download page so inexperienced users select to download the Live CD and not the DVD (and later get angry about both wasted download and install time)

Hi,

This bug was filed against cauldron, but we do not have cauldron at the moment.

Please report whether this bug is still valid for Mageia 2.

Thanks :)

Cheers,
marja

Keywords: (none) => NEEDINFO

Yes, still valid.

Keywords: NEEDINFO => (none)

I wonder if the regression i experience in Bug 6438 may be related.

*** Bug 6779 has been marked as a duplicate of this bug. ***

CC: (none) => klaus.maerker

Is this bug valid for mageia 3 alpha 2? Or should it be closed?
I did not see any problems when I installed Mageia 3 alpha 2.

CC: (none) => micheelsen

Testing fresh install using Mageia 3 alpha 3 DVD.

System 1 - Desktop with Radeon HD7800

Installer selects vesa driver. Boots OK. Subsequent installation of fglrx driver gives fully working system.

System 2 - Notebook with Radeon HD3430M

Installer selects radeon driver. Boot fails (X segfaults when started, giving black screen). Enabling non-free repositories and installing radeon-firmware fixes this fault.

System 3 - Notebook with hybrid Intel + Radeon HD6600M

If "configure only Intel card" option selected, installer selects intel driver. Boots OK.

If "configure only Radeon card" option selected, installer selects vesa driver.
Boot fails (X exits with error "VESA(0): Cannot read V_BIOS (3) Input/output error"). Subsequent installation of radeon driver and radeon-firmware does not give working system (X segfaults when started, giving black screen). Subsequent installation of fglrx driver also does not give working system (X segfaults when started, giving black screen).

So this bug is still valid.

*** Bug 8214 has been marked as a duplicate of this bug. ***

CC: (none) => junk_no_spam

*** Bug 8147 has been marked as a duplicate of this bug. ***

CC: (none) => juan.baptiste

Update for fresh install using Mageia 3 beta 1 DVD and non-free repository enabled.

System 1 - Desktop with Radeon HD7800

Installer selects fglrx driver. Boots OK to fully working system.

System 2 - Notebook with Radeon HD3430M

Installer selects radeon driver and installs radeon-firmware. Boots OK to fully working system. Power consumption significantly higher than Mageia 2 + fglrx driver (see bug 7346).

System 3 - Notebook with hybrid Intel + Radeon HD6600M

If "configure only Intel card" option selected, installer selects intel driver.
Boots OK to fully working system. Power consumption extremely high (38W as compared to 18W with Mageia 2 + fglrx), because radeon driver is also initialised and powers up the Radeon graphics hardware.

If "configure only Radeon card" option selected, installer selects fglrx driver. Boots OK to text console. X startup fails due to segfault in intel driver. A bit of time with Google reveals this is an incompatibility between the fglrx driver and intel (> 2.20.2) driver. More info and a patch that reportedly fixes the issue can be found at

  https://code.launchpad.net/~andrikos/ubuntu/quantal/xserver-xorg-video-intel/fix-ati-hybrid/+merge/140578

So, some progress, but I'm still left with 2 out of 3 machines that are unusable with Mageia 3, due to power consumption issues and no option to use the high performance graphics on the hybrid graphics machine.

thierry, possible patch for fglrx see https://bugs.mageia.org/show_bug.cgi?id=5426#c70

(btw, maybe we should split this bug)

CC: (none) => thierry.vignaud

Beta1 live DVD KDE version boots fine on Dell Vostro 3750 (from duplicate bug #8147), I haven't tried installing it yet, I'll report back when done.

It mga 3 beta 1 live KDE DVD installs fine and it configures the Intel card. When using XFdrake to use the ATI card and rebooting I don't get video, the atieventsd service doesn't start. This is what journalctl says:

Dec 25 22:42:02 cauldron-laptop systemd[1]: Starting LSB: ATI Events Daemon...
Dec 25 22:42:02 cauldron-laptop systemd[1]: PID file /var/run/atieventsd.pid not readable (yet?) after start.
Dec 25 22:42:02 cauldron-laptop systemd[1]: Failed to start LSB: ATI Events Daemon.
Dec 25 22:42:02 cauldron-laptop systemd[1]: Unit atieventsd.service entered failed state

On a system with Core I5 3450 and Radeon HD7770, I can only install Mageia Beta 2 with the LiveDVD. 
Trying so with the InstallDVD give me an error (conflict with video driver it says) after first reboot when Xorg tries to start. I tried with and without Nonfree repo. 
Strange is that in manual configuration of the Graphic Card during installation process, hitting test works ( I see that rainbow background and the message to accept new graphic configuration ).
Installing the exact same driver (FGLRX for ATI HD 6400+)  with exact same configuration after installing from LiveDVD works perfectly... I don't understand.

CC: (none) => matali

Did you install updates as part of the installation which failed?

yes, updates repo where activated every time (I installed with netinstall iso). I installed 3 times, it failed 3 times. Once with all at default. Once with personalized config that was working properly when hiting test, once without nonfree repo, but it looks like the Fglrx driver was installed anyway...

It looks like the latest 13.2 beta will fix the problem with hanging boot - but introduces other interesting problems :-(   http://ubuntuforums.org/showthread.php?t=1930450&page=64

Any news from Mageia on this?

Is this bug related to this one? Or am I just confusing things?
https://bugs.mageia.org/show_bug.cgi?id=8852

Seems to be solved. I can boot with fglrx driver now.

Oliver can you confirm this also?

adding nokmsboot for boot linux if fglrx driver is in use, or modeset=1? else uninstall ati fglrx - then kms is working on dri (linus-kernel i've tested), or i've seen that probably beta ati driver can fix the problem with some catalyst-hook

CC: (none) => oeai

I have a couple of machines acquired in late 2009, and I have never been able to make the Xorg/ati driver work with either, under Mandriva 2010 onward and currently under Mageia 3B2 and 3B3.

One machine is a Gateway 4300-9, Athlon dual-core cpu, ATI Radeon HD3200 video chipset, and the other is a Microcenter Power Spec, Intel Core i7 870 cpu, ATI Radeon 4670 PCIe video card.  Mageia 3 alpha and beta have been installed only on the Gateway 4300-9.

I have routinely used boot option nokmsboot and would try nosetmode when having problems.  I have installed firmware packages when bug reports indicated that could be a problem, but that did nothing for the Xorg/ati driver.   Neither XFdrake nor the setup routines of mcc would successfully install the Xorg/ati driver (appears as Driver "ati" in /etc/X11/xorg.conf).

I was able to use the AMD proprietary driver fglrx on both machines until AMD discontinued support for cards earlier than HD5000.  Yes, XFdrake offers a legacy option for cards earlier than 5000, but at one point I dug around in the update package and found a README that declared cards prior to 5000 were no longer supported.  Perhaps that legacy version may work for some machines, but not for mine.

On the Intel cpu machine using Mandriva 2011 PLUS OS, I am still able to use the fglrx driver installed by  XFdrake (in xorg.conf the BoardName is listed as "ATI Radeon HD 2000 and later (radeon/fglrx)" but in Mageia 3 Alpha and Beta (and IIRC Mandriva 2012 Alpha installed last November) the proprietary fglrx driver would not work.

The Xorg/vesa driver will work on both machines, and it meets my needs, but if you think it would be useful I am willing to spend some time testing and reporting if requested.

CC: (none) => jdbeard

@Jim

Up to now, there is no fglrx driver for HD3200 & HD4670 that runs with Linux 3.8 & Xorg 1.13, the versions of mga3 (As you mention AMD drops support for HD < 5000 and the last fglrx-legacy v13.1 doesn't support Xorg 1.13, it is mainly a quick-fix for Team Fortress 2 players on Ubuntu 12.04 & Steam).

But it should be possible to run the open-source driver "radeon" on both machines. It runs quite nicely on my HD4570 laptop, except the power management which can be a real stopper for laptop users. The "radeon" driver is the default dirver installed by mga3, well, at least on my laptop.

CC: (none) => farfouille64

yeah i disabled fglrx too, but it is asking about using it just in the end of installation process, if i want to use proprietary drivers.
amdcccle - gives much more tweaks, and kms is used for boot? - it depends on which desktop you are using really, so for e17 i got now dri and it seems alright, and i got hd5xxx, but yes i got older machines with hd3870 and x1600, so for x1600 no support at all, but it would be nice to use it too.

@François,

Please identify EXACTLY what you mean by 'the open-source driver "radeon"'.

Using XFdrake, or the gui mcc, there is no "radeon"  other than under ATI or VENDOR/ATI.  The card identified as "Radeon HD 4870 and earlier" appears under ATI (not under Xorg which I would expect for open source), and if installed it is identified in /etc/X11/xorg.conf as 'Driver:  "ati"

That driver is the "ati" that has never worked on my machines.

With ati the driver in xorg.conf, the Xorg log file shows that ati_drv.so and radeon_drv.so are loaded but "drm report modesetting isn't supported" and radeon is unloaded due to no matching device.

I note that in times now long past the ATI proprietary fglrx driver did work, so I question whether modesetting is or is not available on my machine. (I do remember a time when under Mandriva the boot command to grub included a nomodeset option the first time a kernel was booted, but that was dropped many months ago.)

I further note that the list of ATI chipsets appearing in the Xorg log does include the HD 3200 and two versions of HD 4670, with one of the latter a laptop chipset and the other RV4670XT that is used in my deskside machine.  Confusion of laptop vs desktop for the 3200 may be why system settings thinks my display is a laptop, but that would seem to be a separate issue.

I'm not sure this is pertinent, but the complaint from drm that modesetting isn't supported promted me to try booting without the nokmsboot  option.

Odd behavior.  If I boot using a grub line that included nokmsboot but edit the line in grub to remove nokmsboot and then continue the boot, the machine boots and seems to work normally.  If I delete the nokmsboot from the grub stanza in menu.lst used to boot and reboot, I get an <OUT OF RANGE> error message on the monitor.

This did not help with driver selection, though.  When I edited the boot command during boot to delete nokmsboot, attempts to use XFdrake and mcc to install the Xorg/ati and the AMD/HD 4870 and earlier choices both result in no usable screen found.

i've installed linus-kernel and deinstalled fglrx
then it is working

but if you have no freeze but the error - you can try to rename Xorg.conf and start with default settings, or try to run drakxconf.

If a conditional was added to the boot up script that, upon detecting legacy Ati devices, substituted the Xorg-Vesa driver AND displayed a message to that effect, novice users might well continue with Mageia, rather than dump it with a disdainful sniff.

(In reply to Jim Beard from comment #85)
> @François,
> 
> Please identify EXACTLY what you mean by 'the open-source driver "radeon"'.

In fact it is the "ati" driver you mention as set in xorg.conf. It works pretty well for my HD 4570 laptop except the power management. Maybe you'll find some useful informations here : http://www.x.org/wiki/RadeonFeature

Francois,

Please see my Comments 85 and 86 above.

The "ati" radeon driver loads, but is then unloaded with the Xorg log comment that kernel modesetting is not supported.

The features of the "ati" radeon driver mean nothing if it will not load and remain loaded.  Have you suggestions on how I might get that to happen?

(In reply to Juan Luis Baptiste from comment #73)
> It mga 3 beta 1 live KDE DVD installs fine and it configures the Intel card.
> When using XFdrake to use the ATI card and rebooting I don't get video, the
> atieventsd service doesn't start. This is what journalctl says:
> 
> Dec 25 22:42:02 cauldron-laptop systemd[1]: Starting LSB: ATI Events
> Daemon...
> Dec 25 22:42:02 cauldron-laptop systemd[1]: PID file /var/run/atieventsd.pid
> not readable (yet?) after start.
> Dec 25 22:42:02 cauldron-laptop systemd[1]: Failed to start LSB: ATI Events
> Daemon.
> Dec 25 22:42:02 cauldron-laptop systemd[1]: Unit atieventsd.service entered
> failed state


For me this is still an issue, I still can't boot with fglrx on current cauldron.

(In reply to Martin Whitaker from comment #70)
> If "configure only Radeon card" option selected, installer selects fglrx
> driver. Boots OK to text console. X startup fails due to segfault in intel
> driver. A bit of time with Google reveals this is an incompatibility between
> the fglrx driver and intel (> 2.20.2) driver. More info and a patch that
> reportedly fixes the issue can be found at
> 
> https://code.launchpad.net/~andrikos/ubuntu/quantal/xserver-xorg-video-intel/
> fix-ati-hybrid/+merge/140578

@Thierry, do you think we could apply that patch (a revert of upstream intel driver commit)? As per the discussion at the link upstream has said reverting it would not cause issues.



(In reply to Juan Luis Baptiste from comment #73)
> It mga 3 beta 1 live KDE DVD installs fine and it configures the Intel card.
> When using XFdrake to use the ATI card and rebooting I don't get video, the
> atieventsd service doesn't start. This is what journalctl says:
> 
[...]
> Dec 25 22:42:02 cauldron-laptop systemd[1]: Unit atieventsd.service entered
> failed state

atieventsd failure is not fatal. What does your /var/log/Xorg.0.log contain after failure to get video?  (or /var/log/Xorg.0.log.old if restarted X since then).



(In reply to Jim Beard from comment #85)
> That driver is the "ati" that has never worked on my machines.
> 
> With ati the driver in xorg.conf, the Xorg log file shows that ati_drv.so
> and radeon_drv.so are loaded but "drm report modesetting isn't supported"
> and radeon is unloaded due to no matching device.

That seems strange, since we never disable kernel modesetting in mga3. Are you sure you are not booting with manually added "nomodeset" or something like that?
If so, please provide full /var/log/Xorg.0.log and output of "dmesg" after such an unsuccessful bootup. Thanks.

ping for answer...

@Anssi, @tv: are we gonna apply this patch?

if this is not solvable for Mga3 (allthough it seems this time, more ATI stuff is working than mga2), perhaps we need to set some errata?

Martin Whitaker's X startup issue should be fixed in fglrx 12.104 that was uploaded by Thomas 3 days ago, and there should be no need for the intel patch.

@Martin: please retest with the latest livecd (from 26th april) on your mirrors.

@anssi: is this the only issue left? can we then close this one as resolved, or is it more complicated?

Not sure, there were other issues but I did not get any replies in comment #93.

Though it could very well be that I missed something, this bug report is very hard to follow. Please, in the future, if you are not sure that the issue is the same, report a separate bug to make sure your issue is not missed.

@AL13N, sorry, I no longer have the notebook with ATI hybrid graphics, so you'll need to find someone else to test this.

@Anssi, the heat issue, as mentioned both here and in bug 7346, means Mageia 3 is unusable on my other notebook. But as I find Mageia 3 unusable for several other reasons, I've given up caring.

I have a GPT/EFI system (4Gb) with AMD Radeon HD7310 graphics.
Trying MAG3RC Gnome LiveDVD 64-bit dated 24 April, both 'run' & 'install' boot ends prematurely with on the main console (tty1):

[FAILED] Failed to start LSB: ATI Events Daemon
See 'syscntl status atieventsd.service' for details

Other Ctl/Alt/Fn consoles work.
dmesg shows near the end:

Several 'fglrx' lines complaining about
- Invalid AMD VBIOS
- Failed to read VBIOS
- Invalid image size requested [many such lines]
dmesg O/P ends:
systemd-readahead [261]: Out of memory

I wonder whether this is the same problem as this bug, or a different one. Please advise.

CC: (none) => lewyssmith

(In reply to Martin Whitaker from comment #99)
> @Anssi, the heat issue, as mentioned both here and in bug 7346, means Mageia
> 3 is unusable on my other notebook. But as I find Mageia 3 unusable for
> several other reasons, I've given up caring.

Well, for that one we can do nothing, AMD does not support the X server we have with fglrx.


(In reply to Lewis Smith from comment #100)
> I wonder whether this is the same problem as this bug, or a different one.
> Please advise.

Different, please open a separate bug report. As for the problem itself, any possibility of attaching the full dmesg?

>> I wonder whether this is the same problem as this bug, or a different  
>> one.
>> Please advise.
>
> Different, please open a separate bug report. As for the problem itself,  
> any
> possibility of attaching the full dmesg?
I will open a separate bug when I have a feel for what the problem actually is (Title). I will try vesa first, & upload all dmesg O/P. Not today! It is fiddly.

@anssi: perhaps we should close this one as "probably fixed" and let lewis file a different bug? cause i can't tell if this is all of it or not.

As per the description, this bug is about failure to boot with the "vesa" driver on ATI/Intel PowerXpress hardware as experienced by Oliver Burger when booting Mageia 2 Beta3. Additional information asked for in comment #20 was not provided, and I will therefore close this bug report as OLD as we have no confirmation on whether this has been fixed since then.

@Oliver, if you still experience the issue, please reopen so that we can work on resolving it.

All the other people who have commented on this bug are experiencing various different unrelated issues. Those issues have either been already fixed, or information asked for has not been provided, or the issue was not noticed among the over a hundred comments.

If you are still experiencing an issue, please open a new separate bug report so that it can be properly followed and eventually fixed. If I asked for any information from you in comment #93, please also provide that information in the new report.

I'm modifying the summary to better reflect the original issue and to avoid people incorrectly commenting on this bug report.

Status: REOPENED => RESOLVED
Summary: Beta3 doesn't boot on ati graphics hardware => Mageia 3 Beta3 doesn't boot on ati graphics hardware (vesa driver)
Resolution: (none) => OLD