[lkml]   [2002]   [Dec]   [17]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
    Available from:

    Notable events:

    (1) I switched to mainline releases since intermediate -bk snapshots
    vaporize after the following mainline release.
    (2) The driverfs oops fix got merged.
    (3) A new patch dynamically sizes the pidhash to save memory on tiny boxen.
    (4) A new patch has a small experiment with pte_chain pointer block sizes.

    My NUMA-Q has PCI-PCI bridges and I've not quite gotten
    them working yet. Here's the workaround. If someone
    would be kind enough to make PCI segments not require
    3 cleanup passes and an accessor that changes semantics
    between each pass and after the last one I'd be much obliged.

    Update __do_SAK() to use for_each_task_pid(). It was
    looking for all tasks in a given session.

    Keep count of real processes (not threads) and don't
    walk the tasklist in proc_fill_super().

    Use for_each_task_pid() in cap_set_pg(). It's trying
    to find all tasks in a given pgrp.

    05_vm86 Bugfix for not using release on exit() but instead
    scanning for a matching task pointer, which is oopsable.

    get_task() is attempting to do find_task_by_pid().
    Call find_task_by_pid() and remove tasklist scanning.

    Speed up NUMA-Q mem_map initialization by freeing
    higher-order pages during highpage init.

    Allocate pgdat's from node-local memory on NUMA-Q.

    Remove the unused has_stopped_jobs() function and
    rename __has_stopped_jobs() to has_stopped_jobs().

    Increase the size of the inode wait table.

    Use slab ctor's to reduce the initialization costs of
    PAE pgd's and pmd's.

    Dynamically size the pidhash hashtable at boot-time,
    proportional to memory. 128KB isn't used until 1GB RAM.
    Kernel BSS reduced by (almost) 128KB.

    Try forcibly increasing NRPTE to just slightly above the
    mean pte_chain length for theoretically less allocation
    overhead and/or L2 cache footprint (note: they're already
    >= L1_CACHE_BYTES). If this doesn't work, all you'll get
    is internal fragmentation (measurable by comparing the
    slab info to nr_reverse_maps).
    I'm not 100% sure I want to keep core VM stuff in here...
    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: 2005-03-22 13:31    [W:0.021 / U:49.140 seconds]
    ©2003-2016 Jasper Spaans. hosted at Digital OceanAdvertise on this site