-
Andrew Morton authored
From: Ravikiran G Thirumalai <kiran@in.ibm.com> The routine disk_round_stats showed up considerably under oprofile for high disk io load (four processes doing dd to the same disk (different partitions) on a 4 way). This is because the counter in_flight which is per-cpu right now gets read every time disk_round_stats gets called. Per cpu counters like disk statistics improve write speed, but reads are slow (since all cpus' local counter values have to be read and summed up). Considering the fact that in_flight counter is modified post disk_round_stats (which reads the in_flight counter) it is better not to per-cpu this counter. Following patch does just that. Below is the profile comparison before and after the change. This was on a 4 way PIII Xeon, 1G ram, 2.6.0-test4-mm2. Before: c010aa60 2910109 92.2249 poll_idle c0275340 23208 0.73549 __copy_to_user_ll c02753b0 11191 0.354657 __copy_from_user_ll c0114aa0 7168 0.227163 mark_offset_tsc c011ad10 6767 0.214455 schedule c011a2b0 6741 0.213631 load_balance c0138890 6710 0.212648 __generic_file_aio_write_nolock c011d302 4683 0.14841 .text.lock.sched c02e4b50 4533 0.143656 ahc_linux_isr c029cec0 3582 0.113518 disk_round_stats c0119b40 3509 0.111205 try_to_wake_up c029d320 3306 0.104771 __make_request c01567d0 3300 0.104581 __block_write_full_page c0156c00 3299 0.104549 __block_prepare_write After: c010aa60 2777940 92.1302 poll_idle c0275340 23479 0.778679 __copy_to_user_ll c02753b0 10943 0.362924 __copy_from_user_ll c0114aa0 7022 0.232884 mark_offset_tsc c0138890 6988 0.231757 __generic_file_aio_write_nolock c011ad10 6607 0.219121 schedule c011d302 5771 0.191395 .text.lock.sched c02e4a60 4458 0.147849 ahc_linux_isr c011a2b0 3921 0.13004 load_balance c01567d0 3569 0.118366 __block_write_full_page c029d2a0 3540 0.117404 __make_request ... c029ceb0 311 0.0103143 disk_round_stats c011d5b0 299 0.00991631 remove_wait_queue
046dbb49