Re: 2.4.20, .text.lock.swap cpu usage? (ibm x440)

From: William Lee Irwin III (wli@holomorphy.com)
Date: Thu Jan 09 2003 - 19:25:48 EST


On Thu, Jan 09, 2003 at 12:18:27PM -0800, Andrew Morton wrote:
> These numbers are a little odd. You seem to have only lost 200M of
> lowmem to buffer_heads. Bill, what's your take on this?

He's really low on lowmem. It's <= 16GB or so so mem_map is near-
irrelevant, say 60MB.

My interpretation of the numbers is as follows, where pae_pmd and
kernel_stack are both guessed from pae_pgd:

       buffer_head: 166994KB 183768KB 90.87
           pae_pmd: 21576KB 21576KB 100.0
      kernel_stack: 14384KB 14384KB 100.0
       inode_cache: 6904KB 10377KB 66.52
              filp: 6539KB 6547KB 99.87
    vm_area_struct: 4897KB 4897KB 100.0
         size-4096: 3924KB 4044KB 97.3
          size-256: 2137KB 2137KB 100.0
        signal_act: 1966KB 1966KB 100.0
      dentry_cache: 747KB 1758KB 42.47
              sock: 1368KB 1368KB 100.0
         size-1024: 1080KB 1080KB 100.0
       files_cache: 738KB 738KB 100.0
         size-2048: 624KB 684KB 91.22
           size-64: 323KB 525KB 61.69
           size-32: 158KB 445KB 35.54
          size-512: 416KB 416KB 100.0
         mm_struct: 405KB 405KB 100.0
          size-128: 356KB 356KB 100.0
 skbuff_head_cache: 171KB 255KB 67.15
      ip_dst_cache: 240KB 240KB 100.0
   file_lock_cache: 166KB 191KB 87.5
      journal_head: 77KB 176KB 43.99
          fs_cache: 113KB 123KB 92.3
           pae_pgd: 112KB 112KB 100.0
   blkdev_requests: 96KB 101KB 94.81
        size-16384: 80KB 80KB 100.0
        cdev_cache: 45KB 47KB 96.15
         arp_cache: 29KB 45KB 64.44
         size-8192: 40KB 40KB 100.0
       names_cache: 32KB 32KB 100.0
        size-32768: 32KB 32KB 100.0
   tcp_bind_bucket: 21KB 28KB 77.67
          sigqueue: 26KB 26KB 100.0
     tcp_tw_bucket: 18KB 18KB 100.0
         uid_cache: 15KB 17KB 89.46
  tcp_open_request: 15KB 15KB 100.0
        kmem_cache: 15KB 15KB 100.0
   inet_peer_cache: 6KB 14KB 46.12
     size-128(DMA): 0KB 7KB 3.33
      size-32(DMA): 2KB 7KB 29.31
       ip_fib_hash: 0KB 7KB 6.25
        bdev_cache: 0KB 7KB 7.75
         mnt_cache: 1KB 7KB 15.51
     dnotify_cache: 4KB 6KB 71.68
      fasync_cache: 4KB 6KB 68.25
      revoke_table: 0KB 2KB 2.80

== grand total of 253.015MB, fragmentation included.
        + 60MB mem_map
== grand total of 313MB or so

Either pollwait tables (invisible in 2.4 and 2.5), kernel stacks of
threads (which don't get pae_pgd's and are hence invisible in 2.4
and 2.5), or pagecache, with a much higher likelihood of pagecache.

Or there might be dark matter in the universe, and he's being bitten by
unaccounted !__GFP_HIGHMEM allocations, e.g. stock 2.4.x pagetables,
which aren't predictable from pae_pgd etc. highpte of any flavor (aa or
otherwise) should fix that. But there's no way to guess, as there's zero
2.4.x PTE accounting or even any hints from this report, like average
RSS and VSZ (which are still underestimates, as 2.4.x pagetables are
leaked over the lifetime of the process vs. 2.5.x's reap-on-munmap()).

Bill
-
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/



This archive was generated by hypermail 2b29 : Wed Jan 15 2003 - 22:00:30 EST