[PATCH] reorganized per-CPU data

From: Dimitris Michailidis (dimitris@cthulhu.engr.sgi.com)
Date: Fri Jun 16 2000 - 17:26:29 EDT

  • Next message: Jeff Epler: "User bug (was Re: 2.2 + LFS + -malign_double = BUG)"

    From: Dimitris Michailidis <dimitris@engr.sgi.com>
    Date: 16 Jun 2000 14:26:29 -0700
    In-Reply-To: Andi Kleen's message of "9 Jun 2000 08:48:04 -0700"
    Message-ID: <6yr66r9iagq.fsf_-_@darkside.engr.sgi.com>
    Lines: 27
    X-Mailer: Gnus v5.5/XEmacs 20.4 - "Emerald"

    A new version of the PDA patch for data maintained per-CPU is available from

      http://reality.sgi.com/dimitris_engr/pda_patch-2.4.0-1.

    New in this version:

    - ported to -test1
    - some more code has been converted to use the new scheme
    - on MP systems we allocate only as many PDAs as we need.

    Given that these structures are large (>2K each, larger with IPv6 enabled)
    the allocation change above can save substantial space on MP systems. On a
    4P box space savings amount to more than 79K, .text savings alone exceed 5600
    bytes (including small savings in the system call and IRQ paths in entry.S ;-)

    Andi Kleen wrote:
    > Just an ugly problem is the Config dependent layout of the pda data
    > and module binary compatibility. Maybe it would be possible to tag
    > some magic symbol with its size and export it, making modules reference
    > it ?

    These issues are still present. I don't think exporting the size of a pda is
    safe, you might be able to get the same size with different configurations.
    Maybe exporting the offsets of optional fields? 

    -- 
Dimitris Michailidis                    dimitris@engr.sgi.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/

    This archive was generated by hypermail 2b29 : Fri Jun 16 2000 - 17:45:07 EDT