Re: 2.3.5_andrea2 IDE-CD Problem

synflood@endor.sick.cl
Sat, 5 Jun 1999 16:07:15 -0400

On Sat, Jun 05, 1999 at 02:45:36PM +0200, Andrea Arcangeli wrote:
> On Sat, 5 Jun 1999 synflood@endor.sick.cl wrote:
>
> >i was testing 2.3.5_andrea2 patch on 2 server and 1 workstation today,
>
> Do you use SMP? 2.3.5 is not SMP safe in the network code (warning, don't
> use it for production!).
No all the machines are single Celeron 300A Machine with 64Mb or 128Mb Ram and
IDE Disk, my workstation is a Celeron 300A overclocked to 450Mhz with 128Mb
ram and an ASUS P2B.

>
> >both servers work great, handling load of closer to 30 or 33 w/o problem
> >and almost with no swap on a 64Mb Ram Machine running cistron-radiusd and MySQL
> >(the load was produced by a test script)
>
> Good. I suppose 2.3.5 couldn't sustain a load of 30/33 without major lose
> of iterativeness. Also the global performances should be improved a lot.

well you were wrong :) but a second test today show that the machine freeze
w/o a message with a load of 24, but the machine normal load is 1 o 0.5 , so
is ok for me. and yeah swap usage is 0, and normally it is around 2 or 3 megs
of swal allocated just on startup.... IDE disk almost report no heavy work.
the machine is running MySQL and cistron-radiusd. in a 2.3.5 w/o patches there
is a lot of HD work, with your patches the work on the IDE disk is almost none.

> >Jun 5 00:09:38 endor kernel: end_request: I/O error, dev 16:40 (hdd), sector
> >108
>
> I have no idea about this. I really think to have changed nothing that may
> generate these errors. Is this reproducible? Are you sure this doesn't
> happen in 2.3.5 clean right now too?

no 2.3.5 vanilla works great. i check your patch and there is no modification
to the IDE Subsystem. also as i'm writing this mail i'm testing you patch again
and the ide cdrom is working great :( seems to be a random problem. right now
i'm burining the CDROM and the IDE CDROM is working w/o problems.
is strange when i write the e-mail i reboot the machine twice and both time
the problems appears.. and now it didn't show up...

>
> My first guess is that you had a real problem reading I/O from the CD, and
> then you triggered a buffer leakage (shown by my set_blocksize) in the
> isofs code.
>
maybe, but right now i'm unable to reproduce the problem... well i'll keep
trying to make the bug show again...

> Where is placed your CDWR?
is a SCSI CDR Attached to a NCR53c850 PCI Controller.

Thanks for your time and help.

-- 


SynFl00d


email : synflood@endor.sick.cl
Efax Number : (815) 366-3177
ICQ : 6994327


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/