Re: [radeonhd] SOLVED - 2e2e4d118e Kills RS690M (x1200) Cards [ was 0x791F:0x1179:0xFF68: ]
David C. Rankin writes:
Here's what's funny about the log files you've posted: The .old file (from commit e3f80bc6ec2a - which is apparently working) contains the line:
(EE) RADEONHD(0): RHDCSStop: Command Submission backend is not active!
This should not happen! The other log file doesn't - as it seems to lock up before. Still, the change made in commit 2e2e4d118e497 is very small and should be straight forward. I fail to see why this would cause any problems. Therefore I suspect a memory corruption somewhere else. This change should not even matter for you this code is not executed on RS690 - it is for R600 and later.
Cheers, Egbert.
Wow,
I get what you are saying, but regardless of whether the changes in that commit "should" matter, the "do" matter and "do" break the driver completely for the RS690M cards. The question is "What will it take to fix it?"
The card at issue is in a Toshiba 205D laptop. Here is all the card information I can find (I'm running fglrx at the moment):
It has nothing to do with the card. We've seen strange problems before which are impossible to explain. They seem to be related to memory corruptions. Then adding or removing totally unrelated code in the driver triggered or fixed those problems. I have no idea where to start chasing this as long as I don't see these issues locally. What you can try is to build the driver with -O0: 'make CFLAGS=-O0' should do and see if this makes a difference. Cheers, Egbert. -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Wednesday 26 August 2009 10:08:34 am Egbert Eich wrote:
David C. Rankin writes: <snip>
Wow,
I get what you are saying, but regardless of whether the changes in that commit "should" matter, the "do" matter and "do" break the driver completely for the RS690M cards. The question is "What will it take to fix it?"
The card at issue is in a Toshiba 205D laptop. Here is all the card information I can find (I'm running fglrx at the moment):
It has nothing to do with the card. We've seen strange problems before which are impossible to explain. They seem to be related to memory corruptions. Then adding or removing totally unrelated code in the driver triggered or fixed those problems. I have no idea where to start chasing this as long as I don't see these issues locally. What you can try is to build the driver with -O0: 'make CFLAGS=-O0' should do and see if this makes a difference.
Cheers, Egbert.
Egbert, I guess sometimes you do just have to hit me over the head with a big hammer ;-) I rebuilt the current driver with CFLAGS=-O0, installed, logged out and ... I was able to log back in on the current driver! Success! Of course, that doesn't explain why I need to set CFLAGS now, but at least I'm back in the saddle. Thank you for your help. Is there anything else I can run that would help you chase down where the problem is? If so, let me know. -- David C. Rankin, J.D.,P.E. Rankin Law Firm, PLLC 510 Ochiltree Street Nacogdoches, Texas 75961 Telephone: (936) 715-9333 Facsimile: (936) 715-9339 www.rankinlawfirm.com -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
participants (2)
-
David C. Rankin -
Egbert Eich