Yet another HDIO_SET_DMA failed problem
hdparm -d 1 -k 1 /dev/hda
HDIO_SET_DMA failed: operation not permitted
I went to Google and did some searching, finding what looks like a
fair amount of false information, like the person who said that you
just can't do it with a 2.4.21 kernel. Several people suggested a
command that I forget, but apparently the same info is in /proc/pci.
-- I should note: 2.4.21 kernel, AMD Athlon K7 processor, Iwill XP333
motherboard, new Maxtor DiamondMax Plus 9 200 GB IDE (6Y200P0).
00:04.0 IDE interface: ALi Corporation M5229 IDE (rev (4)
(prog-if & [Master SecP PriP])
ISA bridge: ALi Corporation [ALi] M1533 PCI to ISA bridge
[Alladin IV] (rev 0)
if my scrawled notes are to be trusted. One person had recommended
that the second line indicated that the alim15x3 module needed to be
loaded. I did, and bandwidth dropped from 4 MB/s to 1 MB/s!
The further frustrating thing is that Knoppix can handle it fine.
"hdparm -d 1" works fine, and bandwidth goes to the 45-55 MB/s range.
(I'd prefer higher, given than I'm doing a "badblocks -w" on a new 200
GB drive and will I hope be doing another, but this means it'll run
overnight and while I'm at work instead of weeks.)
(Other parameters (-c, -u, -A, -a, -m) have no noticable effect on the
speed reported by "hdparm -t", whether under my normal kernel or
under Knoppix.)
(This is the latest Knoppix, and uname -a returns 2.4.21, I think
-xfs.)
I did "lsmod" to see what modules were loaded in Knoppix that weren't
in my system, and they're these:
af_packet agpgart apm autofs4 cloop cmpci
efs hfs minix ntfs reiserfs serial
usb-ohci usbcore xfs
But Knoppix might perhaps have a proper driver built-in, and lsmod
would not then report it, right?
Any clues about what module I might need? Also, where do I put the
insmod line? (I'm afraid I haven't Googled for that info -- sorry for
not doig my research.)
--
Tim McDaniel, tmcd@panix.com; tmcd@us.ibm.com is my work address
Reply to: