Bug#270199: Performance prob applies to 2.4.27-1-686 too

[Joshua Kwan <joshk@triplehelix.org>]

John wrote:
> Well, I've had a wee break and the computer came up okay. Here's the 
> kernel I booted:
> ns:~# ls /boot/vmlinuz-2.4*
> -rw-r--r--  1 root root 872499 Sep  3 14:44 /boot/vmlinuz-2.4.27-1-686
> ns:~#

firesong:/home/joshk# ls -l /boot/vmlinuz-2.4.27-1-686
-rw-r--r--  1 root root 872499 2004-09-02 23:44 /boot/vmlinuz-2.4.27-1-686

> ns:~# hdparm -t /dev/hda
>
> /dev/hda:
>  Timing buffered disk reads:   12 MB in  3.47 seconds =   3.46 MB/sec

firesong:/home/joshk# hdparm -t /dev/hda

/dev/hda:
 Timing buffered disk reads:   62 MB in  3.03 seconds =  20.46 MB/sec

> ns:~# hdparm  /dev/hda
>
> /dev/hda:
>  multcount    = 16 (on)
>  IO_support   =  0 (default 16-bit)
>  unmaskirq    =  0 (off)
>  using_dma    =  0 (off)
>  keepsettings =  0 (off)
>  readonly     =  0 (off)
>  readahead    =  8 (on)
>  geometry     = 14593/255/63, sectors = 234441648, start = 0
> ns:~# hdparm -d1  /dev/hda
>
> /dev/hda:
>  setting using_dma to 1 (on)
>  HDIO_SET_DMA failed: Operation not permitted
>  using_dma    =  0 (off)
> ns:~#

firesong:/home/joshk# hdparm /dev/hda

/dev/hda:
 multcount    = 16 (on)
 IO_support   =  0 (default 16-bit)
 unmaskirq    =  0 (off)
 using_dma    =  1 (on)
 keepsettings =  0 (off)
 readonly     =  0 (off)
 readahead    =  8 (on)
 geometry     = 58140/16/63, sectors = 58605120, start = 0
firesong:/home/joshk# hdparm -d1 /dev/hda

/dev/hda:
 setting using_dma to 1 (on)
 using_dma    =  1 (on)

So we've established that your controller doesn't seem to want to do 
DMA. You seem to be using the piix driver, correct? I do, too...
Could you post a dmesg from 2.4.27-1-686, and also try some older 
kernels to see if this is a regression and not a bug that has always 
been there? (Hint: kernel-image-2.4.26-1-686)

Behold piix.c, 2.4.27:

        /* UDMA 33 capable */
        case PCI_DEVICE_ID_INTEL_82371AB: <--- this is you
        case PCI_DEVICE_ID_INTEL_82443MX_1:
        case PCI_DEVICE_ID_INTEL_82451NX:
        case PCI_DEVICE_ID_INTEL_82801AB_1:
            return 1;

So it SHOULD be working.

Also attach the output from cat /proc/ide/piix.
Note, if it doesn't exist, piix refused to handle your device, which 
sounds like the older problem (except that it's fixed now!)

--
Joshua Kwan

Attachment: signature.asc
Description: OpenPGP digital signature