Re: All filesystems hang under long periods of heavy load (read and write) on a filesystem

From: Ged Haywood (ged_at_www2.jubileegroup.co.uk)
Date: 12/29/03

  • Next message: Frank van Maarseveen: "Re: 2.4.23 can run with HZ==0!"
    Date:	Mon, 29 Dec 2003 15:32:54 +0000 (GMT)
    To: linux-kernel@vger.kernel.org
    
    

    Hello again all,

    On Tue, 23 Dec 2003, Ged Haywood wrote:
    [snip,snip]
    > On Tue, 04 Nov 2003 Randy Dunlap wrote:
    > >
    > > Can you try a recent kernel, like 2.4.23-pre8 or -pre9?
    >
    > Copying large files from the smaller IDE drive to the larger trashed
    > the filesystem on the partition being written to.

    It seems that this boils down to using PIO mode, after enabling DMA I am
    able to write 1.7G files with no trouble - of course it's early days yet,
    the fault might not be fixed, it might just be less obvious...

    Enabling DMA hadn't been top priority until I spoke to the 'hdparm'
    maintainer at Debian. The driver for the CMD Technology Inc CMD680 is
    cunningly hidden under "Silicon Image chipset support" in the 2.4.23
    config menus but was fairly easy to find once I started digging. :)

    Apparently the problem might still be present in 2.6.0 - see for example

    ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/must-fix/should-fix-7.txt

    Anyway I'm thinking I'll put together an identical machine in the next
    few weeks so I'll be able to reproduce the fault at will. Would anyone
    be interested in playing with it?

    73,
    Ged.

    PS: Not subscribed, please CC me if you're interested.

    PPS: Keeping the same subject line, but maybe this deserves a new one?

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


  • Next message: Frank van Maarseveen: "Re: 2.4.23 can run with HZ==0!"

    Relevant Pages

    • Re: [RFC] Changing COW detection to be memory hotplug friendly
      ... Keeping the swap slot ... is valuable if read fault, but once the page is dirtied, wouldn't ... I cannot imagine a good enough range of swap performance tests. ... send the line "unsubscribe linux-kernel" in ...
      (Linux-Kernel)
    • Re: Stop the linux kernel madness - SOLVED!
      ... > So simply removing this symlink and putting back a link to ... IF the fault here is SUSE's, then submit THEM a bug report and stop whining in lkml. ... could have between lkml readers. ... send the line "unsubscribe linux-kernel" in ...
      (Linux-Kernel)
    • Re: system-freeze: kprobe and do_gettimeofday
      ... "insmod kgettime.ko" from the console ... double fault, ... Then I removed all my modules I was able to load the module ... send the line "unsubscribe linux-kernel" in ...
      (Linux-Kernel)
    • Re: [PATCH] [Request for inclusion] Filesystem in Userspace
      ... > We actually had code that tracked the dirty bit in ... > software (ie make it unwritable by default, and take the write fault), but ... notes on kernel development methodology skipped ...] ... send the line "unsubscribe linux-kernel" in ...
      (Linux-Kernel)
    • Re: [RFT][PATCH] fix ServerWorks PIO auto-tuning
      ... We depend on BIOS hints for ServerWorks and BIOS is not enabling DMA. ... designs, ... send the line "unsubscribe linux-kernel" in ...
      (Linux-Kernel)