Re: DMA not permitted since 2.6.14
Le dimanche 11 décembre 2005 à 23:52 +0100, Hendrik Sattler a écrit :
> David LIMA wrote:
> >
> > I'm running Debian SID with linux-image-2.6-686, since few days i'm
> > facing DMA problems after upgrading kernel to 2.6.14 (from stock).
> > Hdparm reports "operation not permitted" when I try do activate DMA.
> >
> > I have read http://lists.debian.org/debian-user/2005/11/msg00410.html
> > and recompile my kernel but still same issue.
> [...]
> > /dev/hda:
> > UDMA modes: udma0 udma1 udma2 udma3 udma4 *udma5 udma6
> >
> > /dev/hdc:
> > UDMA modes: udma0 udma1 udma2 udma3 udma4 *udma5 udma6
> > * signifies the current active mode
> >
> > Has someone any clue about this problem that slow down my system ?
>
> Actually, no clue what you are complaining about: both drives run with UDMA5,
> there not much more what you can enable with respect to DMA transfers.
Hi, Henrik
For sure, DMA is not running, my system is very slow each time I access
my disks (CPU eating).
_______________________
hdparm -tT /dev/hd{a,c}
/dev/hda:
Timing cached reads: 768 MB in 2.01 seconds = 382.53 MB/sec
Timing buffered disk reads: 8 MB in 3.26 seconds = 2.45 MB/sec
/dev/hdc:
Timing cached reads: 144 MB in 2.00 seconds = 71.87 MB/sec
Timing buffered disk reads: 8 MB in 3.42 seconds = 2.34 MB/sec
______________________
hdparm -v /dev/hd{a,c}
/dev/hda:
multcount = 1 (on)
IO_support = 0 (default 16-bit)
unmaskirq = 0 (off)
using_dma = 0 (off)
keepsettings = 0 (off)
readonly = 0 (off)
readahead = 256 (on)
geometry = 65535/16/63, sectors = 160086528, start = 0
/dev/hdc:
multcount = 1 (on)
IO_support = 0 (default 16-bit)
unmaskirq = 0 (off)
using_dma = 0 (off)
keepsettings = 0 (off)
readonly = 0 (off)
readahead = 256 (on)
geometry = 19929/255/63, sectors = 320173056, start = 0
________________________
hdparm -d1 /dev/hd{a,c}
/dev/hda:
setting using_dma to 1 (on)
HDIO_SET_DMA failed: Operation not permitted
using_dma = 0 (off)
/dev/hdc:
setting using_dma to 1 (on)
HDIO_SET_DMA failed: Operation not permitted
using_dma = 0 (off)
TIA,
David
>
> HS
>
Reply to: