Re: ide1 stalls while tar xvfz'ing

Anthony Barbachan (barbacha@Hinako.AMBusiness.com)
Tue, 13 Oct 1998 01:07:10 -0400

It could be the drive itself. I have always found Maxtors to not only have
a high risk of dying with a bunch of bad sectors but also to be slower than
other EIDE brands. It could be that you Maxtor is just one of those super
slow maxtors, the hiccup is probably caused when the disk cache becomes full
and the writes to the drive are suspended until some of the data in the
cache is written to the disk. Another possibility is your motherboard. I
read that some of the original PCI boards (both Pentiums and 486's) had the
primary EIDE on PCI and the secondary channel on ISA (probably expecting
that the second chain will only be used for a CDROM). This would cause any
drive on hdc and hdd to be much slower than it should be.

-----Original Message-----
From: Les Schaffer <godzilla@netmeg.net>
Newsgroups: sunsite.linux.kernel
To: linux-kernel@vger.rutgers.edu <linux-kernel@vger.rutgers.edu>
Date: Monday, October 12, 1998 5:51 PM
Subject: ide1 stalls while tar xvfz'ing

>
>Greetings:
>
>I installed a few months ago a new hard drive (Maxtor 4 Megish) on
>ide1 as /dev/hdc, and have noticed something queer which last nite i
>took the trouble to characterize clearly:
>
>lets say i go 'tar xvfz large-many-file-archive.tar.gz' onto
>/dev/hdc. you see the files start to spin out fast, then stop, then go
>a little bit, then stop and wait, then go again. And the stop and wait
>are not on large files, could be any size even small.
>
>when i go and tar xvfz the same archive on /dev/hda it speeds right
>through all the files.
>
>so i figure something is up with /dev/hdc trying to write lots of
>different files.
>
>originally i had MultiSect=0 and then set it to 8, but that didnt
>help.
>
>i am using irqtune to fix poor PPP download rates.
>
>but i dont think explains anything, since ide0 is just one higher
>priority than ide1 (and lower than the I03 serial, which carries PPP),
>but ide0 has no problem with stall.
>
>Any suggestions to fix this damn thing?
>
>thanks much.
>
>info follows
>============
>
>kernel = 2.0.35
>
>(gustav)~/: hdparm -I /dev/hdc
>
>/dev/hdc:
>
> Model=aMtxro9 40233D , FwRev=AW8S7293,
SerialNo=3K40K7AJ
> Config={ Fixed }
> RawCHS=8374/16/63, TrkSize=0, SectSize=0, ECCbytes=29
> BuffType=3(DualPortCache), BuffSize=256kB, MaxMultSect=16, MultSect=8
> DblWordIO=no, maxPIO=2(fast), DMA=yes, maxDMA=2(fast)
> CurCHS=8374/16/63, CurSects=8440992, LBA=yes, LBAsects=8440992
> tDMA={min:120,rec:120}, DMA modes: mword0 mword1 *mword2
> IORDY=on/off, tPIO={min:120,w/IORDY:120}, PIO modes: mode3 mode4
>
>
>[memo to Mark Lord: hdparm is swapping characters in the Model name,
>that should be Maxtor 94......]
>
>irqtune: trying command -- rmmod irqtune_mod
>I00/P05: 377686 timer
>I01/P06: 3596 keyboard
>I02/P07: 0 cascade
>I03/P00: 31164 + serial
>I04/P01: 31127 + serial
>I08/P09: 0 + rtc
>I11/P12: 529 + aha152x
>I13/P14: 1 math error
>I14/P07: 42916 + ide0
>I15/P08: 85861 + ide1
>
>
>--
>____ Les Schaffer godzilla@netmeg.net ___| --->> Engineering R&D <<---
>Theoretical & Applied Mechanics | Designspring, Inc.
>Center for Radiophysics & Space Research | Westport, CT USA
>Cornell Univ. schaffer@tam.cornell.edu | les@designspring.com
>.
>
>-
>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/
>

-
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/