Re: sis5513 fatal udma problems

[Steven Walter]

On Wed, Jan 03, 2001 at 03:10:39PM -0600, Jim Studt wrote:
> The sis5513 has for quite a while had fatal problems with udma on some
> machines.  (At least the thelinuxstore PIAs with the P6SET-ML
> motherboard, I have heard other people with the same problem).
> 
> The sis5513 driver deliberatly overrides the CONFIG_IDEDMA_AUTO and
> the BIOS settings that might disable UDMA and forces UDMA which works
> for a while then fails, attempts to fallback to a non-DMA mode and
> fails that as well.
> 
> The logs end up with something along the lines of...
>   hdc: timeout waiting for DMA
>   ide_dmaproc: chipset supported ide_dma_timeout func only: 14
>   hdc: irq timeout: status=0x58 { DriveReady SeekComplete DataRequest }
>   hdc: status timeout: status=0xd8 { Busy }
>   hdc: DMA disabled
>   hdc: drive not ready for command
>   ide1: reset: success
>   hdc: status error: status=0x58 { DriveReady SeekComplete DataRequest }
>   hdc: drive not ready for command
> ... but it nevers succeeds in the reset.  The drive never becomes ready.
> 
> I will make a patch, but I am uncertain what the `right' policy should
> be.  I am inclined to first and foremost respect the CONFIG_IDEDMA_AUTO
> flag, then to take the fastest mode which is both capable and enabled in
> the bios.
> 
> Thoughts?  Please? :-)

What kernel version is this?  If its a 2.2.x series kernel, try patching
it with Andre's IDE patches.  I've had good luck using UDMA/33 on a
SiS530 (which uses the 5513 ide chipset) using Andre's patches.  I've
also been able to use UDMA/33 and UDMA/66 with the latest 2.4.x kernels
and an 80-conductor cable (very important!).  I haven't tried 2.2.x+ide
with UDMA/66, though it should work as well.
-- 
-Steven
"Voters decide nothing.  Vote counters decide everything."
				-Joseph Stalin