Hello there,
accessing my CD-ROM drive causes my system to generate 94% "hi" in
top. I reproduced it using the following steps:
1. mount /dev/hda /cdrom
2. cat /cdrom/* > /dev/null
During this timeperiod, the system does almost stop to do any other
I/O, i.e. accessing my SATA drives on an LVM2 stripeset and doing "cat
largefile > /dev/null" does not read the usual 60 MB/s but only like 2
or 3. I am using the anticipatory scheduler, but tried out the
deadline elevator which doesn't change anything.
System configuration:
=====================
/proc/interrupts is at http://home.in.tum.de/foerstes/lkml/interrupts
/proc/slabinfo is at http://home.in.tum.de/foerstes/lkml/slabinfo
/proc/cpuinfo is at http://home.in.tum.de/foerstes/lkml/cpuinfo
lspci -vvv is at http://home.in.tum.de/foerstes/lkml/lspci
dmesg is at http://home.in.tum.de/foerstes/lkml/dmesg
.config is at http://home.in.tum.de/foerstes/lkml/config
I am using a patch for LIRC from:
http://flameeyes.web.ctonet.it/lirc/patch-lirc-2.6.1-rc1-20040106.diff.bz2
Nothing else is patched in.
Is this the expected behaviour of the 2.6 series kernel?
If you need any more informations, let me know.
ciao, Marc
On Fri, Jan 16, 2004 at 02:48:16PM +0100, Marc-Christian Petersen wrote:
> Hello there,
>
> accessing my CD-ROM drive causes my system to generate 94% "hi" in
> top. I reproduced it using the following steps:
>
> 1. mount /dev/hda /cdrom
> 2. cat /cdrom/* > /dev/null
>
> During this timeperiod, the system does almost stop to do any other
> I/O, i.e. accessing my SATA drives on an LVM2 stripeset and doing "cat
> largefile > /dev/null" does not read the usual 60 MB/s but only like 2
> or 3. I am using the anticipatory scheduler, but tried out the
> deadline elevator which doesn't change anything.
Try hdparm -d1 on your cdrom drive.