[PATCH] (1/2) reverse mapping VM for 2.5.23 (rmap-13b)

From: Craig Kulesa (ckulesa@as.arizona.edu)
Date: Wed Jun 19 2002 - 06:18:00 EST


Where: http://loke.as.arizona.edu/~ckulesa/kernel/rmap-vm/

This patch implements Rik van Riel's patches for a reverse mapping VM
atop the 2.5.23 kernel infrastructure. The principal sticky bits in
the port are correct interoperability with Andrew Morton's patches to
cleanup and extend the writeback and readahead code, among other things.
This patch reinstates Rik's (active, inactive dirty, inactive clean)
LRU list logic with the rmap information used for proper selection of pages
for eviction and better page aging. It seems to do a pretty good job even
for a first porting attempt. A simple, indicative test suite on a 192 MB
PII machine (loading a large image in GIMP, loading other applications,
heightening memory load to moderate swapout, then going back and
manipulating the original Gimp image to test page aging, then closing all
apps to the starting configuration) shows the following:

2.5.22 vanilla:
Total kernel swapouts during test = 29068 kB
Total kernel swapins during test = 16480 kB
Elapsed time for test: 141 seconds

2.5.23-rmap13b:
Total kernel swapouts during test = 40696 kB
Total kernel swapins during test = 380 kB
Elapsed time for test: 133 seconds

Although rmap's page_launder evicts a ton of pages under load, it seems to
swap the 'right' pages, as it doesn't need to swap them back in again.
This is a good sign. [recent 2.4-aa work pretty nicely too]

Various details for the curious or bored:

        - Tested: UP, 16 MB < mem < 256 MB, x86 arch.
          Untested: SMP, highmem, other archs.

          In particular, I didn't even attempt to port rmap-related
          changes to 2.5's arch/arm/mm/mm-armv.c.

        - page_launder() is coarse and tends to clean/flush too
          many pages at once. This is known behavior, but seems slightly
          worse in 2.5 for some reason.

        - pf_gfp_mask() doesn't exist in 2.5, nor does PF_NOIO. I have
          simply dropped the call in try_to_free_pages() in vmscan.c, but
          there is probably a way to reinstate its logic
          (i.e. avoid memory balancing I/O if the current task
          can't block on I/O). I didn't even attempt it.

        - Writeback: instead of forcing reinstating a page on the
          inactive list when !PageActive, page->mapping, !Pagedirty, and
          !PageWriteback (see mm/page-writeback.c, fs/mpage.c), I just
          let it go without any LRU list changes. If the page is
          inactive and needs attention, it'll end up on the inactive
          dirty list soon anyway, AFAICT. Seems okay so far, but that
          may be flawed/sloppy reasoning... We could always look at the
          page flags and reinstate the page to the appropriate LRU list
          (i.e. inactive clean or dirty) if this turns out to be a
          problem...

        - Make shrink_[i,d,dq]cache_memory return the result of
          kmem_cache_shrink(), not simply 0. Seems pointless to waste
          that information, since we're getting it for free. Rik's patch
          wants that info anyway...

        - Readahead and drop_behind: With the new readahead code, we have
          some choices regarding under what circumstances we choose to
          drop_behind (i.e. only drop_behind if the reads look really
          sequential, etc...). This patch blindly calls drop_behind at
          the conclusion of page_cache_readahead(). Hopefully the
          drop_behind code correctly interprets the new readahead indices.
          It *seems* to behave correctly, but a quick look by another
          pair of eyes would be reassuring.

        - A couple of trivial rmap cleanups for Rik:
                a) Semicolon day! System fails to boot if rmap debugging
                   is enabled in rmap.c. Fix is to remove the extraneous
                   semicolon in page_add_rmap():

                                if (!ptep_to_mm(ptep)); <--

                b) The pte_chain_unlock/lock() pair between the tests for
                   "The page is in active use" and "Anonymous process
                   memory without backing store" in vmscan.c seems
                   unnecessary.

                c) Drop PG_launder page flag, ala current 2.5 tree.

                d) if(page_count(page)) == 0) ---> if(!page_count(page))
                   and things like that...

        - To be consistent with 2.4-rmap, this patch includes a
          minimal BIO-ified port of Andrew Morton's read-latency2 patch
          (i.e. minus the elvtune ioctl stuff) to 2.5, from his patch
          sets. This adds about 7 kB to the patch.

        - The patch also includes compilation fixes:
        (2.5.22)
              drivers/scsi/constants.c (undeclared integer variable)
              drivers/pci/pci-driver.c (unresolved symbol in pcmcia_core)
        (2.5.23)
              include/linux/smp.h (define cpu_online_map for UP)
              kernel/ksyms.c (export default_wake_function for modules)
              arch/i386/i386_syms.c (export ioremap_nocache for modules)

Hope this is of use to someone! It's certainly been a fun and
instructive exercise for me so far. ;)

I'll attempt to keep up with the 2.5 and rmap changes, fix inevitable
bugs in porting, and will upload regular patches to the above URL, at
least until the usual VM suspects start paying more attention to 2.5.
I'll post a quick changelog to the list occasionally if and when any
changes are significant, i.e. other then boring hand patching and
diffing.

Comments, feedback & patches always appreciated!

Craig Kulesa
Steward Observatory, Univ. of Arizona

-
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 : Sun Jun 23 2002 - 22:00:19 EST