[patch] SMP alternatives, take #2

From: Gerd Knorr (kraxel_at_suse.de)
Date: 11/30/05

  • Next message: Folkert van Heusden: "Re: [RFC] un petite hack: /proc/*/ctl"
    Date:	Wed, 30 Nov 2005 11:19:24 +0100
    To: Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
    
    
    

       Hi,

    Here next version of the SMP alternatives patch.

    Had a more detailed look at the bits generated, the vast majority in the
    smp alternatives table where simply lock prefixes. So I decided to
    specical-case these to keep the table sizes small. There are three elf
    sections now:

       .smp_altinstructions
       .smp_altinstr_replacement
           These are pretty much the same like the existing non-smp
           alternative bits, with the exception that the replacement
           table also has some space to keep the original instruction, to
           allow patching both SMP->UP and back UP->SMP. It's used to patch
           the code for spinlocks with simpler UP versions.

       .smp_locks
           List of pointers to the lock prefixes to be NOP'ed for UP.

    For the core kernel these three sections are grouped together and are
    page-aligned, so they can be freed. Freeing happens automatically in
    some cases where they not needed any more after booting up (UP system
    found, kernel without CONFIG_HOTPLUG_CPU). There is also a kernel cmd
    line option to force the tables being released.

    For modules only the .smp_locks section is handled, the other ones
    simply don't appear in modules ;)

    cheers,

       Gerd

    
    

    -
    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: Folkert van Heusden: "Re: [RFC] un petite hack: /proc/*/ctl"

    Relevant Pages

    • Re: [PATCH 1/3] 2.6.8-rc4-mm1 - Fix UML build
      ... can access them all, and initialized data all before uninitialized, so ... SYMLINKS:= $,$/$f) ... semaphore.c-dir = kernel ... send the line "unsubscribe linux-kernel" in ...
      (Linux-Kernel)
    • Re: Real-Time Preemption, -RT-2.6.12-final-V0.7.50-24
      ... > rebooted to a kernel that doesn't have the RT-preempt patch but ... getting a very verbose running trail, almost like an strace output, ... Copyright 2005 by Maurice Eugene Heskett, ... send the line "unsubscribe linux-kernel" in ...
      (Linux-Kernel)
    • RE: Error mounting root fs on 72:01 using Promise FastTrak TX2000 (PDC20271)
      ... > Now I'm sucessfully booting my system with the 2.4.23 kernel using ... I think it's when the ATARAID driver is about to fire up. ... > the hard drives since this is the only thing that has changed. ... send the line "unsubscribe linux-kernel" in ...
      (Linux-Kernel)
    • Re: udev and devfs - The final word
      ... As each device driver loads and registers with the kernel, ... 'cookies' can be written out to keep them separately ID'd. ... e.g. increasing error rates and aging silicon will ... send the line "unsubscribe linux-kernel" in ...
      (Linux-Kernel)
    • Re: Linux 2.4.30-rc3 md/ext3 problems
      ... you have your root on ext3 on top of soft raid1 consisting in ... two IDE disks, which works in 2.4.21 but not on 2.4.30-rc3, that's ... top of scsi and with kernel 2.4.27, so there's not much to compare... ... send the line "unsubscribe linux-kernel" in ...
      (Linux-Kernel)