Messages in this thread | | | Date | Mon, 12 Apr 2004 08:46:48 -0700 | From | "Martin J. Bligh" <> | Subject | Re: [PATCH] anobjrmap 9 priority mjb tree |
| |
>> Any chance of you doing the same comparison between 2.6.5-aa5 >> 2.6.5-aa5 minus prio-tree? (Well, needn't be -aa5, whatever comes to >> hand. Looks like "patch -p1 -R < prio-tree" mostly works, just some >> rejects in mm/mmap.c itself, let me know if I can help out on that.) >> >> If -aa is okay, I hope so, then it's surely some stupidity from me. > > Good idea. Not sure how easy it'll be to back prio_tree out, but I can > surely do aa5, which would give us a good clue still. Might not be until > this time tommorow though.
Hmm. it's halfway between the two. There does seem to be less sem contention, though the profile ticks in __down isn't really an accurate measure. I'll try to think of some more accurate way to instrument sems (maybe I can do rdtsc before and after taking the sem, and then hack the profiling code to stash delta increment indexed by the callers address). Meanwhile, maybe you can decipher something from the attatched and appended ...
M.
SDET 128 (see disclaimer) Throughput Std. Dev 2.6.5-mjb1 100.0% 1.2% 2.6.5-mjb1-prio 73.6% 0.0% 2.6.5-aa5 86.4% 1.9%
Full profile attatch (to match yesterday's).
diffprofile from mjb1+prio to -aa5 11084 102.0% find_get_page 4976 101.8% clear_page_tables 4482 690.6% schedule 2322 1745.9% pgd_alloc 2036 41.2% atomic_dec_and_lock 1816 0.0% page_add_rmap 1623 16.0% __d_lookup 1428 0.0% __set_page_dirty_buffers 1275 0.0% anon_vma_unlink 1234 25.3% path_lookup 1101 25.0% remove_shared_vm_struct 990 53.7% .text.lock.dec_and_lock 921 0.0% find_get_pages_tag 883 0.0% anon_vma_link 791 52.2% follow_mount 783 250.2% unmap_vmas 764 0.0% strnlen_user 588 32.3% file_move 572 36.8% proc_pid_stat 563 57.9% lookup_mnt 543 55.4% .text.lock.file_table 498 48.0% .text.lock.dcache 488 69.0% flush_signal_handlers 467 215.2% .text.lock.base 460 27.9% kmem_cache_free ... -217 -100.0% __do_softirq -220 -100.0% page_update_anon_rmap -228 -57.6% flush_tlb_page -230 -69.9% sys_close -233 -100.0% direct_strncpy_from_user -233 -7.8% copy_process -235 -21.1% kmap_atomic -254 -37.5% sched_clock -257 -59.6% rb_insert_color -318 -47.7% pte_alloc_map -330 -99.1% find_get_pages -332 -31.5% find_vma -340 -43.8% __vma_prio_tree_remove -356 -29.7% vma_link -369 -100.0% direct_strnlen_user -401 -100.0% page_add_anon_rmap -470 -8.6% do_no_page -610 -8.1% __copy_to_user_ll -619 -100.0% radix_tree_lookup -923 -97.4% set_page_dirty -1133 -12.6% release_pages -1406 -53.7% __block_prepare_write -1553 -14.5% page_remove_rmap -1555 -100.0% page_add_obj_rmap -1753 -38.9% do_page_fault -2269 -68.7% __copy_user_intel -2959 -44.8% do_anonymous_page -3789 -100.0% .text.lock.filemap -7671 -53.6% __wake_up -8214 -36.8% copy_page_range -9948 -62.1% do_wp_page -14120 -100.0% finish_task_switch -27763 -55.8% __down -64017 -12.0% default_idle -107386 -11.9% total [unhandled content-type:application/octet-stream] | |