Chipset is SIS 5571, ide chipset SIS 5513. I've some problems with
latest updates about SIS Ide driver update: When I sleep the drive with
hdparm -Y /dev/hda
then the drive sleeps. But it doesn't 'awake'. I can do 'Login: XXX \n Password: XXX' or write a command
in the shell. But when the unit has to read/write, it just does nothing.
Latest kernels without this patch (2.4.18-preX....) did something different:
I could sleep drive normally. But when the system had to read/write something:
Feb 15 18:13:08 localhost kernel: hda: timeout waiting for DMA
Feb 15 18:13:08 localhost kernel: ide_dmaproc: chipset supported ide_dma_timeout func only: 14
Feb 15 18:13:08 localhost kernel: hda: status timeout: status=0xd0 { Busy }
Feb 15 18:13:08 localhost kernel: hda: drive not ready for command
Feb 15 18:13:08 localhost kernel: ide0: reset: success
The system just stopped a few seconds, and then it started as always.
Drive was tuned with hdparm with the following options:
/sbin/hdparm -c3 -A1 -a8 -d1 -m16 -p4 -u1 -W1 -X34 /dev/hda
En Wed, 20 Feb 2002 15:52:30 -0500 (EST)
Mark Hahn <[email protected]> escribio...:
> > latest updates about SIS Ide driver update: When I sleep the drive with
> > hdparm -Y /dev/hda
> > then the drive sleeps.
>
> are you also doing APM/ACPI sleep?
I'm using APM && apmd, but it seems it does nothing. Now I'll test without apm
module && apmd unloaded.
Well, there's a strange thing: I have two drives:
/dev/hda on /boot,
/dev/hdc on /
If i do hdparm -Y /dev/hda, without apm loaded, hda stops, i can't awake it. hdc remains working.
Now, if i do hdparm -Y /dev/hdc i can sleep && awake hdc. But sometimes, when hda is working (or not)
and with/without apm loaded the system stops, and hdc doesn't awake. I'm getting crazy.
Now when I try to mount the ide cdrom, mount keeps hanged, too.
>
> > Feb 15 18:13:08 localhost kernel: hda: timeout waiting for DMA
>
> classic sign that the ide interface has forgotten its state.
>
> > Drive was tuned with hdparm with the following options:
> > /sbin/hdparm -c3 -A1 -a8 -d1 -m16 -p4 -u1 -W1 -X34 /dev/hda
>
> such tuning is unnecessary and indeed possibly bad on modern kernels.
> for instance, -m is irrelevant in any dma/udma mode, as is -u.
I didn't know, i just read the hdparm manual, which doesn't says that.
>