[lkml]   [2006]   [Mar]   [23]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
    SubjectRe: [PATCH] PM-Timer: doesn't use workaround if chipset is not buggy

    On Thu, Mar 23, 2006 at 03:49:18AM +0900, OGAWA Hirofumi wrote:
    > This patch adds blacklist of buggy chip, and if chip is not buggy,
    > this uses fast normal version instead of slow workaround version.

    Testing on ICH5 (P4HT) hasn't shown any problems so far.

    > If chip is buggy, warnings "pmtmr is slow". But sounds like there is
    > gray zone. I found the PIIX4 errata, but I couldn't find the ICH4
    > errata. But some motherboard seems to have problem.

    Same here, the ICH4 trace is nowhere to be found.

    Should I do a public request for chipset testing?
    (I wrote a small test app here that would hopefully identify a buggy

    Data that I have collected from internet snippets (mostly Intel errata
    Affected (PCI ID / rev):
    - ICH4???
    - PIIX4 A0 (0x7113 / 00?), A1 (0x7113 / 00?), B0 (0x7113 / 01?)
    - PIIX4E A0 (0x7113 / 02?)
    Probably fixed (PCI ID / rev):
    - PIIX4M A0 (0x7113 / 03?)

    My Toshiba Satellite 4280 seems to have non-buggy PIIX4M
    (since it's PCI rev. 03), haven't had time to test reliability yet, though.

    > So, if found a ICH4, also warnings, and uses a workaround version. If
    > user's ICH4 is good actually, user can specify the "pmtmr_good" boot
    > parameter to use fast version.

    Good, that's how it should be done as long as it's not entirely known which
    chipsets (both Intel and possibly also non-Intel!) are buggy.

    Also, I think that since triple-reading wastes 3% of system time, it would
    be very nice to be able to eventually write a software workaround for buggy
    systems, too.
    This could perhaps be done by keeping the *old* value in a global yet
    thread-safe variable and checking whether the new value is safely related
    to it if within short jiffies distance (but since jiffies are probably
    calculated from pmtmr readings... :-P).

    Andreas Mohr
    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to
    More majordomo info at
    Please read the FAQ at

     \ /
      Last update: 2006-03-23 18:06    [from the cache]
    ©2003-2014 Jasper Spaans. hosted at Digital Ocean