Re: Request for help to identify some bugs (part two)
Benjamin Herrenschmidt wrote:
> The kernel should set the DMA speed of the CD to whatever is appropriate
> for the drive. If you have problem and slowing it _down_ helps, then
> it's indeed something interesting worth noting. If you are on the
> contrary pushing it up, then I can't guarantee anything. udma is faster
> than mdma so ...
Hi Benjamin,
What's the "appropriate" setting?
"hdparm -I /dev/hdc" tells me that the cdrom can cope with up to udma2:
/dev/hdc:
ATAPI CD-ROM, with removable media
Model Number: MATSHITACD-RW CW-8123
Serial Number:
Firmware Revision: CA10
Standards:
Likely used CD-ROM ATAPI-1
Configuration:
DRQ response: 50us.
Packet size: 12 bytes
Capabilities:
LBA, IORDY(can be disabled)
DMA: sdma0 sdma1 sdma2 mdma0 mdma1 *mdma2 udma0 udma1 udma2
Cycle time: min=120ns recommended=120ns
PIO: pio0 pio1 pio2 pio3 pio4
Cycle time: no flow control=120ns IORDY flow control=120ns
In fact, setting it to udma2 makes it no longer work, does the kernel is
"aware" of this limitation and sets mdma2 as a consequence?
> Now the problem is to diagnose kded eating CPU when you have the proper
> setting. When that happens, do you see anything in dmesg ? (like IDE
> errors etc...)
When inserting the cd, /var/log/message is flooded by:
hdc: tray open
end_request: I/O error, dev hdc, sector 5568072
hdc: tray open
end_request: I/O error, dev hdc, sector 5568076
hdc: tray open
end_request: I/O error, dev hdc, sector 5568072
hdc: tray open...
I'm not sure that this problem is related to hardware configuration since it
only happen with KDE "media discovery" (when turning off the "gestionnaire
de media kded", the cd insertion does not produce the kded madness but the
cdrom is not accessible via media:/) and not with de Gnome one.
Sincerely,
Yannick
Reply to: