Hi!
I have the problem that when trying to use my TV Card my machine totally
freezes after a few seconds.
After some investigation and reporting on [email protected] it
became apparent, that the problem is with my AliMagik1 chipset. And all the
AliMagik1 users are affected with this problem.
After some searching on the relevant sites I have found this FAQ:
---------------------------------
Question :
When using PCI TV card with A7A266, the system will hang when TV program is
activated.
Answer :
This is a compatibility issue between Ali Chipset and PCI Card. Please
upgrade to ALi AGP Miniport Driver v1.74 and later or change the BIOS setting
from BIOS -> Advanced -> PCI Configuration -> PCI Latency Timer, set as 10
(Default is 32) to increase further compatibility
---------------------------------
Since the bios setting was not enough, I have contacted Ali support to ask
about what was that the Windows driver fixed. See the attached mail below.
It would be pretty good if somebody could implement the things described
below, and send me the patch, so I could test it out.
Does somebody volunteer for that?
It would be nice if the patch would be against a 2.4.x kernel. (like 2.4.19 or
2.4.20 prerelease).
Thanks
Zsolt
---------- Forwarded Message ----------
Subject: Re: agp driver update for linux
Date: Wednesday 20 November 2002 12:17
From: [email protected]
To:
Dear Sir,
I'm sorry we can't provide an AGP driver for Linux to fix this problem as in
Windows. But we are willing to provide the information to fix this problem.
If you can help us pass this information to Linux developer, please help us
find which version of driver or kernel will contains the fix. Hence we can
post this information in our website FAQ to let more users know where to get
the solution. My mail address is [email protected].
Below is the workaround we add in our driver:
1. scan all devices located at bus 0
2. if device class code (PCI config offset 0xb~0xa) is 0x400, do
2a. set latency timer (offset 0xd) of matched device to 0xa
2b. set offset 0x40 bit 1 of matched device to 1
Best Regards
ALi Support Team
-------------------------------------------------------
Like the question says, is there a way to invalidate the whole buffer
cache at once?
--
Mark Atwood | Well done is better than well said.
[email protected] |
http://www.pobox.com/~mra
On 20 Nov 2002, Mark Atwood wrote:
> Like the question says, is there a way to invalidate the whole buffer
> cache at once?
Except for unmounting all filesystems, no.
Rik
--
Bravely reimplemented by the knights who say "NIH".
http://www.surriel.com/ http://guru.conectiva.com/
Current spamtrap: <a href=mailto:"[email protected]">[email protected]</a>
Mark Atwood wrote:
>
> Like the question says, is there a way to invalidate the whole buffer
> cache at once?
>
"There is no buffer cache".
There's a pagecache, which caches the contents of files and block devices.
You can invalidate the block device pagecache with `blockdev --flushbufs' but
there is no interface for invalidating the pagecache of regular files.
Allocating and then freeing a ton of anonymous memory, then running a
swapoff/swapon cycle might suit.