devb-eide not using DMA

QNX Newsgroups (Archives) qdn.public.qnxrtp.os
1

I was trying to figure out why disk access on a SuperMicro X5DPi is running
so slowly, and it appears that devb-eide may not be running in DMA mode.
Below are the relevant lines from slog. Another thing that may be of
interest is that during the initial boot, the dots come out relatively
slowly.

Thanks,

lew

12:38:37 3 Range check failed (IO) - Dev 248b - Vend 8086 - Class 1018a -
Addr 1f0 - Size 8
12:38:37 3 Range check failed (IO) - Dev 248b - Vend 8086 - Class 1018a -
Addr 170 - Size 8
12:38:37 2 eide_identify_devices: Generic IDE vid 8086, did 248b, class
1018a rev 2, busno 0, dfunc f9
12:38:37 2 eide_identify_devices: cmd_addr 1f0, cntl_addr 3f4, irq e, chnl
0, udma -1, mdma -1, sdma -1, pio 0
12:38:37 2 eide_identify_devices: probing dev 0 50
12:38:37 2 eide_identify_devices: probing dev 1 0
12:38:37 2 eide_display_devices: SAMSUNG SP4002H tid 0, cable 80, max udma
5, cur udma 5, max mdma 2, cur mdma 0, max sdma -1, cur sdma -1, pio 4, mblk
16
12:38:37 2 eide_identify_devices: Generic IDE vid 8086, did 248b, class
1018a rev 2, busno 0, dfunc f9
12:38:37 2 eide_identify_devices: cmd_addr 170, cntl_addr 374, irq f, chnl
1, udma -1, mdma -1, sdma -1, pio 0
12:38:37 2 eide_identify_devices: probing dev 0 0
12:38:37 2 eide_identify_devices: probing dev 1 0
12:38:37 2 eide_init_devices: SAMSUNG SP4002H path 0, tid 0, udma -1,
mdma -1, sdma -1, pio 4, mblk 16
12:38:37 2 scsi_interpret_sense: path=0, target=0, lun=0, cam_status=c4,
scsi_status=2
12:38:37 2 scsi_interpret_sense: error=70, sense=5, asc=24, ascq=0
12:38:37 2 scsi_interpret_sense: SK_ILLEGAL command 1a in databyte
12:38:37 2 scsi_interpret_sense: path=0, target=0, lun=0, cam_status=c4,
scsi_status=2
12:38:37 2 scsi_interpret_sense: error=70, sense=5, asc=24, ascq=0
12:38:37 2 scsi_interpret_sense: SK_ILLEGAL command 5a in databyte
12:38:37 2 scsi_interpret_sense: path=0, target=0, lun=0, cam_status=c4,
scsi_status=2
12:38:37 2 scsi_interpret_sense: error=70, sense=5, asc=24, ascq=0
12:38:37 2 scsi_interpret_sense: SK_ILLEGAL command 5a in databyte

2

Check to see if your system has the ICH4 (sometimes used with the 845 chipset).
It’s designation is 82801DB. DMA is not supported on this chip yet.

“Lewis Donzis” <lew@perftech.com> wrote in message news:b76u4n$bre$1@inn.qnx.com…

I was trying to figure out why disk access on a SuperMicro X5DPi is running
so slowly, and it appears that devb-eide may not be running in DMA mode.
Below are the relevant lines from slog. Another thing that may be of
interest is that during the initial boot, the dots come out relatively
slowly.

Thanks,

lew

12:38:37 3 Range check failed (IO) - Dev 248b - Vend 8086 - Class 1018a -
Addr 1f0 - Size 8
12:38:37 3 Range check failed (IO) - Dev 248b - Vend 8086 - Class 1018a -
Addr 170 - Size 8
12:38:37 2 eide_identify_devices: Generic IDE vid 8086, did 248b, class
1018a rev 2, busno 0, dfunc f9
12:38:37 2 eide_identify_devices: cmd_addr 1f0, cntl_addr 3f4, irq e, chnl
0, udma -1, mdma -1, sdma -1, pio 0
12:38:37 2 eide_identify_devices: probing dev 0 50
12:38:37 2 eide_identify_devices: probing dev 1 0
12:38:37 2 eide_display_devices: SAMSUNG SP4002H tid 0, cable 80, max udma
5, cur udma 5, max mdma 2, cur mdma 0, max sdma -1, cur sdma -1, pio 4, mblk
16
12:38:37 2 eide_identify_devices: Generic IDE vid 8086, did 248b, class
1018a rev 2, busno 0, dfunc f9
12:38:37 2 eide_identify_devices: cmd_addr 170, cntl_addr 374, irq f, chnl
1, udma -1, mdma -1, sdma -1, pio 0
12:38:37 2 eide_identify_devices: probing dev 0 0
12:38:37 2 eide_identify_devices: probing dev 1 0
12:38:37 2 eide_init_devices: SAMSUNG SP4002H path 0, tid 0, udma -1,
mdma -1, sdma -1, pio 4, mblk 16
12:38:37 2 scsi_interpret_sense: path=0, target=0, lun=0, cam_status=c4,
scsi_status=2
12:38:37 2 scsi_interpret_sense: error=70, sense=5, asc=24, ascq=0
12:38:37 2 scsi_interpret_sense: SK_ILLEGAL command 1a in databyte
12:38:37 2 scsi_interpret_sense: path=0, target=0, lun=0, cam_status=c4,
scsi_status=2
12:38:37 2 scsi_interpret_sense: error=70, sense=5, asc=24, ascq=0
12:38:37 2 scsi_interpret_sense: SK_ILLEGAL command 5a in databyte
12:38:37 2 scsi_interpret_sense: path=0, target=0, lun=0, cam_status=c4,
scsi_status=2
12:38:37 2 scsi_interpret_sense: error=70, sense=5, asc=24, ascq=0
12:38:37 2 scsi_interpret_sense: SK_ILLEGAL command 5a in databyte

3

“Art Hays” <avhays@nih.gov> wrote in message
news:b76vf8$ddr$1@inn.qnx.com…

Check to see if your system has the ICH4 (sometimes used with the 845
chipset).
It’s designation is 82801DB. DMA is not supported on this chip yet.

Yep, that’s the one. Any idea when it might be supported? It seems a shame
to have such a nifty system and watch the disk I/O not only run slowly, but
consume so much of the cpu.

Thanks,

lew

4

“Lewis Donzis” <lew@perftech.com> wrote in message news:b77d1o$rdu$1@inn.qnx.com…

“Art Hays” <> avhays@nih.gov> > wrote in message
news:b76vf8$ddr$> 1@inn.qnx.com> …
Check to see if your system has the ICH4 (sometimes used with the 845
chipset).
It’s designation is 82801DB. DMA is not supported on this chip yet.

Yep, that’s the one. Any idea when it might be supported? It seems a shame
to have such a nifty system and watch the disk I/O not only run slowly, but
consume so much of the cpu.

We were told last month that it would be at least 6 months. We are negotiating
to have it done thru custom engineering. I would assume the patch would then
be available to anyone.

5

Art Hays wrote:

We were told last month that it would be at least 6 months. We are negotiating
to have it done thru custom engineering. I would assume the patch would then
be available to anyone.

Sounds like we’d better take this up with our sales rep. Thanks very
much for your help.

lew