ati and xorg 6.8.2

[Wedge_]

The HPT372 is definitely supported is 2.6, I have a board with the same controller. Try upgrading before you do anything else. You should just need to enable "Device drivers -> ATA/ATAPI/MFM/RLL support -> HPT36X/37X chipset support" in the kernel config.

First: Sorry for OT!
But Is it supported with Raid in Stripping and ad unique hardisk?Because, when I had try with kernel 2.6.x, the system find HPT372, but not the stripping. However, I try....

I'm not sure about that, since I'm just using it for an extra pair of IDE channels. I'd be a bit surprised if the 2.6 driver didn't support the same functionality as the 2.4 driver though.

Again... *something* is locking the box up hard. AND neither root nor users can get glxinfo to believe that opengl, direct rendering and glx are enabled. Attempting to use a glx app will lock the box up hard, but its far less consistent.

Is X locking up as soon as you start it? Are there any errors in the log? Any errors in dmesg? Does the card work under Windows?

I have to wonder where libGL.la in /usr/lib came from -- I believe that it comes out of opengl-update rebuilding the library or something. -- I have to ask -- why are we building libraries on the fly instead of simply moving softlinks around?

libGL.la is managed by openg-update, but it's not a binary file. It contains linking information for libGL.so.1 as plaintext, and so it can presumably be copied around without much difficulty.

[Sebell]

Wedge:

Thanks for responding:

Is X locking up as soon as you start it? Are there any errors in the log? Any errors in dmesg? Does the card work under Windows?

No the lockups do NOT occur immediately, and I haven't found any errors in the logs - there was *one* occurrance of vmalloc out of memory, however it was not timestamped as to indicate it was related to the lockup. And I have no MS products, except on my work laptop and my other half's pc. The lockups that occur have been irregular and inconsistent, ranging from:

1) a black screen in X and a working system underneath, that will continue to function if I leave it alone, but will require a reboot to a) get the screen back or b) get X cleared out.
to
2) Hard lock, not responsive to keyboard input, no networking, no magic keys, power cycle to get it back.


I am running on a dually Tyan board, AMD 761MP chipset, the XP processor I have is NOT an MP but I'm only running one CPU at the moment. The card is an ATI rv250 9000 card with 128Mb and dualhead/tvout caps.

libGL.la is managed by openg-update, but it's not a binary file. It contains linking information for libGL.so.1 as plaintext, and so it can presumably be copied around without much difficulty.

Coolness -- I didn't realize that -- I've been more interested in PARISC relocation lately so am rusty on libraries in x86 (Im an HPUX/Solaris admin)
Next question -- why is it that my dynamic library is in ati and the static is in xorg? -- there doesn't appear to be an ati static libGL (libGL.a)?

Major issue at this stage is that direct rendering is enabled, glx loads, permissions are set correctly but no screens have GLX.

Anyone else out there with this hardware combo? working?

I am now considering a complete rebuild with debugging switches and trying to figure out how to use my laptop as a serial console for the system.

[MrIch]

I had the same problem... a soon I started X the screen was black...

I solved the problem by editing the xorg file:

Code: Select all

    Option "UseInternalAGPGART"         "no"

[Wedge_]

No the lockups do NOT occur immediately, and I haven't found any errors in the logs - there was *one* occurrance of vmalloc out of memory, however it was not timestamped as to indicate it was related to the lockup. And I have no MS products, except on my work laptop and my other half's pc. The lockups that occur have been irregular and inconsistent, ranging from:

1) a black screen in X and a working system underneath, that will continue to function if I leave it alone, but will require a reboot to a) get the screen back or b) get X cleared out.
to
2) Hard lock, not responsive to keyboard input, no networking, no magic keys, power cycle to get it back.

And you get these with both the fglrx and radeon drivers? As MrIch suggested, it would probably be a good idea to set "UseInternalAGPGART" to "no" if you don't have it set to that already. Are you sure the lockups are being caused by the video drivers? Do they still occur when using the basic "vesa" driver?

I am running on a dually Tyan board, AMD 761MP chipset, the XP processor I have is NOT an MP but I'm only running one CPU at the moment. The card is an ATI rv250 9000 card with 128Mb and dualhead/tvout caps.

Some people in the past had problems with this board or similar models, but if I remember right (must be over a year ago) they didn't have lockups...

Coolness -- I didn't realize that -- I've been more interested in PARISC relocation lately so am rusty on libraries in x86 (Im an HPUX/Solaris admin) Next question -- why is it that my dynamic library is in ati and the static is in xorg? -- there doesn't appear to be an ati static libGL (libGL.a)?

That's right, ATI don't provide a libGL.a, so the xorg version gets used instead. This is all normal AFAIK, so I assume there aren't any problems doing things that way.

[Sebell]

Thanks guys --

And you get these with both the fglrx and radeon drivers? As MrIch suggested, it would probably be a good idea to set "UseInternalAGPGART" to "no" if you don't have it set to that already. Are you sure the lockups are being caused by the video drivers? Do they still occur when using the basic "vesa" driver?

Both drivers will precipitate the problem -- after a complete rebuild (emerge --unmerge -C xorg-x11 ati-drivers ati-drivers-extra opengl-update, and emerge xorg-x11 opengl-update ati-drivers ati-drivers-extra) I no longer get instant lockups starting X, but I still get lockups. I'm using the external agpgart in xorg.conf for fglrx driver. I suspect that I need to turn off kernel autoload to keep the kernel agp drivers out. I cannot rmmod the amd-k7-agp and agpgart modules once loaded (amd-k7-agp in use without radeon or fglrx loaded, or X started) -- fbsplash gets a nomtrr parameter in the kernel 'append' line.

What gets me here is that Xorg log shows direct rendering enabled, loads glx, dri, fglrxdrm etc. user has permissions on /dev/dri/card0 and I still get no glx caps per glxinfo and frglxinfo .. DRI screeninit succeeds against /dev/dri/card0 (result 5, ok), Video Overlay gets enabled etc -- everything looks great from there ..

The only relevant message that might indicate something is :

Code: Select all

Feb 26 17:26:42 ajftl1 [fglrx] AGP enabled,  AgpCommand = 0x0f000304 (selected caps)
Feb 26 17:26:42 ajftl1 [fglrx] free  AGP = 54800384
Feb 26 17:26:42 ajftl1 [fglrx] max   AGP = 54800384
Feb 26 17:26:42 ajftl1 [fglrx] free  LFB = 116387840
Feb 26 17:26:42 ajftl1 [fglrx] max   LFB = 116387840
Feb 26 17:26:42 ajftl1 [fglrx] free  Inv = 0
Feb 26 17:26:42 ajftl1 [fglrx] max   Inv = 0
Feb 26 17:26:42 ajftl1 [fglrx] total Inv = 0
Feb 26 17:26:42 ajftl1 [fglrx] total TIM = 0
Feb 26 17:26:42 ajftl1 [fglrx] total FB  = 0
Feb 26 17:26:42 ajftl1 [fglrx] total AGP = 16384

I'm not sure what the Inv TIM and FB are, but I can guess and have NO idea why I see 0 listed against them.
*sigh*

[geaaru]

I'm not sure about that, since I'm just using it for an extra pair of IDE channels. I'd be a bit surprised if the 2.6 driver didn't support the same functionality as the 2.4 driver though.

I try with 2.6, but raid doesn't see; system see hardisk but no raid.
In 2.4 there is a voice for Ata/atapi... -> Support for IDE raid Controllers -> Highpoint 370 software RAID that there isn't in 2.6 (beyond voice HPT36X/37x chipset support). I want, if I won't change motherboad, because as I say, it's down, to do the porting of this entry in kernel 2.6.
Thanks. See you.