[Bug 229914] New: HDDs working slow
https://bugzilla.novell.com/show_bug.cgi?id=229914 Summary: HDDs working slow Product: openSUSE 10.2 Version: RC 5 Platform: 32bit OS/Version: Other Status: NEW Severity: Normal Priority: P5 - None Component: Basesystem AssignedTo: bnc-team-screening@forge.provo.novell.com ReportedBy: bjoern.wahl@hospital-borken.de QAContact: qa@suse.de Hello ! I just installed OpenSuse 10.2 and found my HDD working really slow. HDDcontoller hp dc5100MT 00:1f.2 IDE interface: Intel Corporation 82801FB/FW (ICH6/ICH6W) SATA Controller (rev 03) (prog-if 8f [Master SecP SecO PriP PriO]) Subsystem: Hewlett-Packard Company Unknown device 300c Flags: bus master, 66MHz, medium devsel, latency 0, IRQ 201 I/O ports at 1818 [size=8] I/O ports at 1830 [size=4] I/O ports at 1820 [size=8] I/O ports at 1834 [size=4] I/O ports at 14f0 [size=16] Capabilities: [70] Power Management version 2 When installing my PC I found the HDD working really slow so I tested it with hdparm -Tt /dev/sda (S-ATA) # hdparm -Tt /dev/sda /dev/sda: Timing cached reads: 1440 MB in 2.00 seconds = 719.95 MB/sec Timing buffered disk reads: 186 MB in 3.03 seconds = 61.47 MB/sec HDD: /dev/sda: ATA SAMSUNG HD080HJ/ ZH10 So I tought it would be the HDD working really slow so I installed a different HDD (ide) # hdparm -Tt /dev/hda /dev/hda: Timing cached reads: 1446 MB in 2.00 seconds = 722.77 MB/sec Timing buffered disk reads: 174 MB in 3.03 seconds = 57.37 MB/sec Das ist mir alles zu langsam. # hdparm /dev/hda /dev/hda: multcount = 16 (on) IO_support = 1 (32-bit) unmaskirq = 0 (off) using_dma = 1 (on) keepsettings = 0 (off) readonly = 0 (off) readahead = 256 (on) geometry = 65535/16/63, sectors = 78165360, start = 0 At my second PC HP DX6100 SUSE 9.3 there is no problem: HDDcontroller hp DX6100: 0000:00:1f.2 IDE interface: Intel Corporation 82801FB/FW (ICH6/ICH6W) SATA Controller (rev 03) (prog-if 8f [Master SecP SecO PriP PriO]) Subsystem: Hewlett-Packard Company: Unknown device 3005 Flags: bus master, 66Mhz, medium devsel, latency 0, IRQ 201 I/O ports at 3818 [size=8] I/O ports at 3830 [size=4] I/O ports at 3820 [size=8] I/O ports at 3834 [size=4] I/O ports at 34f0 [size=16] Capabilities: [70] Power Management version 2 # hdparm /dev/hdb /dev/hdb: multcount = 16 (on) IO_support = 0 (default 16-bit) unmaskirq = 0 (off) using_dma = 1 (on) keepsettings = 0 (off) readonly = 0 (off) readahead = 256 (on) geometry = 19457/255/63, sectors = 160041885696, start = 0 # hdparm -Tt /dev/hdb /dev/hdb: Timing cached reads: 3188 MB in 2.00 seconds = 1591.85 MB/sec Timing buffered disk reads: 48 MB in 3.02 seconds = 15.88 MB/sec Thats really slow (especially at the cached reads) alt = 1591.85 MB/sec neu = 722.77 MB/sec 2neu= 719.95 MB/sec As I could not believe that this is a software bug, I installed OpenSuse 10.1 And it really is faster: # hdparm -Tt /dev/sda /dev/sda: Timing cached reads: 2964 MB in 2.00 seconds = 1483.99 MB/sec Timing buffered disk reads: 180 MB in 3.00 seconds = 59.97 MB/sec # hdparm /dev/sda /dev/sda: IO_support = 0 (default 16-bit) HDIO_GET_UNMASKINTR failed: Inappropriate ioctl for device HDIO_GET_DMA failed: Inappropriate ioctl for device HDIO_GET_KEEPSETTINGS failed: Inappropriate ioctl for device readonly = 0 (off) readahead = 1024 (on) geometry = 9729/255/63, sectors = 156301488, start = 0 So it seems to me that this is a bug, isn`t it ? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 ------- Comment #1 from mhorvath@novell.com 2006-12-20 05:52 MST ------- *** Bug 229922 has been marked as a duplicate of this bug. *** -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 mhorvath@novell.com changed: What |Removed |Added ---------------------------------------------------------------------------- AssignedTo|bnc-team- |kernel-maintainers@forge.provo.novell.com |screening@forge.provo.novell| |.com | -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 gregkh@novell.com changed: What |Removed |Added ---------------------------------------------------------------------------- AssignedTo|kernel- |hare@novell.com |maintainers@forge.provo.nove| |ll.com | -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 hare@novell.com changed: What |Removed |Added ---------------------------------------------------------------------------- AssignedTo|hare@novell.com |teheo@novell.com -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 teheo@novell.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO Info Provider| |bjoern.wahl@hospital-borken.de ------- Comment #2 from teheo@novell.com 2007-01-08 10:45 MST ------- Happy new year. Your bug report is a bit confusing. Can you sum it up for me? And please report /var/log/boot.msg and the results of 'hdparm -I /dev/[hs]dX'. Thanks. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 bjoern.wahl@hospital-borken.de changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |NEW Info Provider|bjoern.wahl@hospital- | |borken.de | ------- Comment #3 from bjoern.wahl@hospital-borken.de 2007-01-12 01:54 MST ------- Created an attachment (id=112607) --> (https://bugzilla.novell.com/attachment.cgi?id=112607&action=view) infos needed 1) Installed OpenSuse 101 on an hp dc5100 everything working fine 2) Installed OpenSuse 102 hdd working really slow hdparm -tT supports that feeling 3) Installed OpenSuse 101 again an hdparm -tT says the add is faster -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 teheo@novell.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO Info Provider| |bjoern.wahl@hospital-borken.de ------- Comment #4 from teheo@novell.com 2007-01-12 17:34 MST ------- Okay, here's the reason why I'm still confused. * In the first report, you're mixing /dev/sdX and /dev/hdX. They are driven by two completely separate driver stacks, so you need to isolate them. * In the info you posted in the previous comment, suse 10.2 shows normal performance of 61.68MB/s buffered disk reads. In the first report, the problematic device is /dev/hdb but it isn't present on boot messages of both 10.1 and 10.2. If you're talking about performance difference in 'Timing cached reads', that's not a driver problem. It's determined by VM. ISTR related discussion on LKML but cannot find it at the moment, but I would be surprised if that has any bearing on real world performance. Keep in mind that hdparm's cached read test is as synthetic as a benchmark can get. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 ------- Comment #5 from bjoern.wahl@hospital-borken.de 2007-01-15 02:57 MST ------- Created an attachment (id=112945) --> (https://bugzilla.novell.com/attachment.cgi?id=112945&action=view) More info Okay, here's the reason why I'm still confused. * In the first report, you're mixing /dev/sdX and /dev/hdX. They are driven by two completely separate driver stacks, so you need to isolate them. Thats right, I just wanted to point out that hdparm shows that the hdd (sata) is really slow.. The bug is only about sata hdds OpenSuse 10.2. * In the info you posted in the previous comment, suse 10.2 shows normal performance of 61.68MB/s buffered disk reads. In the first report, the problematic device is /dev/hdb but it isn't present on boot messages of both 10.1 and 10.2. Well, what was supprising me is that the hdd (sda) seems to be slower when using 10.2 than using 10.1. 10.2: /dev/sda: Timing cached reads: 1440 MB in 2.00 seconds = 719.95 MB/sec Timing buffered disk reads: 186 MB in 3.03 seconds = 61.47 MB/sec 10.1: /dev/sda: Timing cached reads: 2964 MB in 2.00 seconds = 1483.99 MB/sec Timing buffered disk reads: 180 MB in 3.00 seconds = 59.97 MB/sec Look at the cached Reads!!! If you're talking about performance difference in 'Timing cached reads', that's not a driver problem. Correct! It's determined by VM. What das VM mean ? Virual Machine ? ISTR related discussion on LKML What does ISTR mean and LKML ?? but cannot find it at the moment, but I would be surprised if that has any bearing on real world performance. Keep in mind that hdparm's cached read test is as synthetic as a benchmark can get. I also thought about that. But first I had the subjectiv impression that the hdd was much slower and than i started hdparm to check if my impression was right. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 ------- Comment #6 from teheo@novell.com 2007-01-15 03:16 MST ------- VM is virtual memory. ISTR is "I seem to recall". LKML is "Linux kernel mailing list" where the guru penguins hang out. :-) Anyways, that number doesn't mean much by itself and doesn't necessarily implicate real world performance impact. Can you give a shot at kernel.org 2.6.19 kernel and see how it performs in hdparm test? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 ------- Comment #7 from bjoern.wahl@hospital-borken.de 2007-01-15 03:35 MST ------- Created an attachment (id=112951) --> (https://bugzilla.novell.com/attachment.cgi?id=112951&action=view) nothing Ok, if you mean if I could trie installing the newest Vanilla-Kernel and trie hdparm again, I can and will do that..... -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 ------- Comment #8 from teheo@novell.com 2007-01-15 03:43 MST ------- Yeap, that's what I meant. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 ------- Comment #9 from bjoern.wahl@hospital-borken.de 2007-01-16 06:17 MST ------- Created an attachment (id=113114) --> (https://bugzilla.novell.com/attachment.cgi?id=113114&action=view) nothing Ok, here are the results from Germany....;-) # uname -a Linux linux-h4hs 2.6.19.2-default #1 SMP Tue Jan 16 14:11:14 CET 2007 i686 i686 i386 GNU/Linux # hdparm -tT /dev/sda /dev/sda: Timing cached reads: 1512 MB in 2.00 seconds = 755.44 MB/sec Timing buffered disk reads: 184 MB in 3.03 seconds = 60.78 MB/sec still bad results at "cached reads"..... so what ? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 teheo@novell.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |NEW Info Provider|bjoern.wahl@hospital- | |borken.de | ------- Comment #10 from teheo@novell.com 2007-01-16 06:27 MST ------- I was wondering whether the problem disappeared in 2.6.19. This is a VM / (possibly) readahead issue which is still present on the latest stable vanilla kernel. I'll have this bug reassigned to someone more familiar with VM and its history. Thanks for your patience. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 lmb@novell.com changed: What |Removed |Added ---------------------------------------------------------------------------- AssignedTo|kernel- |npiggin@novell.com |maintainers@forge.provo.nove| |ll.com | -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 ------- Comment #12 from bjoern.wahl@hospital-borken.de 2007-01-17 05:54 MST ------- Created an attachment (id=113350) --> (https://bugzilla.novell.com/attachment.cgi?id=113350&action=view) notihing Assuming that hdparm shows the wrong results I tested copying the dvd-image on the harddisk, and there a can not find any difference between 10.2 and 10.1 so maybe it is really only a bug concerning hdparm an not the real world performance.... I keep an eye on that and here are my results... opensuse10-2-kernel2.6.19.2 linux-h4hs:/tmp # date && cp openSUSE-10.2-GM-DVD-i386.iso openSUSE-10.2-GM-DVD-i386.iso2 && date Mi 17. Jan 12:16:47 CET 2007 Mi 17. Jan 12:19:45 CET 2007 linux-h4hs:/tmp # date && cp openSUSE-10.2-GM-DVD-i386.iso openSUSE-10.2-GM-DVD-i386.iso2 && date Mi 17. Jan 12:20:04 CET 2007 Mi 17. Jan 12:23:08 CET 2007 linux-h4hs:/tmp # date && cp openSUSE-10.2-GM-DVD-i386.iso openSUSE-10.2-GM-DVD-i386.iso2 && date Mi 17. Jan 12:23:09 CET 2007 Mi 17. Jan 12:26:09 CET 2007 linux-h4hs:/tmp # uname -a Linux linux-h4hs 2.6.19.2-default #1 SMP Tue Jan 16 14:11:14 CET 2007 i686 i686 i386 GNU/Linux linux-h4hs:/tmp # hdparm -Tt /dev/sda /dev/sda: Timing cached reads: 1448 MB in 2.00 seconds = 724.01 MB/sec Timing buffered disk reads: 184 MB in 3.01 seconds = 61.04 MB/sec opensuse10-2-kernel2.6.18.2 linux-h4hs:/tmp # date && cp openSUSE-10.2-GM-DVD-i386.iso openSUSE-10.2-GM-DVD-i386.iso2 && date Mi 17. Jan 12:29:43 CET 2007 Mi 17. Jan 12:33:00 CET 2007 linux-h4hs:/tmp # date && cp openSUSE-10.2-GM-DVD-i386.iso openSUSE-10.2-GM-DVD-i386.iso2 && date Mi 17. Jan 12:33:30 CET 2007 Mi 17. Jan 12:36:29 CET 2007 linux-h4hs:/tmp # date && cp openSUSE-10.2-GM-DVD-i386.iso openSUSE-10.2-GM-DVD-i386.iso2 && date Mi 17. Jan 12:36:31 CET 2007 Mi 17. Jan 12:39:29 CET 2007 linux-h4hs:/tmp # uname -a Linux linux-h4hs 2.6.18.2-34-default #1 SMP Mon Nov 27 11:46:27 UTC 2006 i686 i686 i386 GNU/Linux linux-h4hs:/tmp # hdparm -Tt /dev/sda /dev/sda: Timing cached reads: 1468 MB in 2.00 seconds = 733.63 MB/sec Timing buffered disk reads: 184 MB in 3.01 seconds = 61.05 MB/sec opensuse10-1-kernel2.6.16.13-4 linux-v77u:/tmp # date && cp openSUSE-10.2-GM-DVD-i386.iso openSUSE-10.2-GM-DVD-i386.iso.2 && date Mi Jan 17 13:27:08 CET 2007 Mi Jan 17 13:30:14 CET 2007 linux-v77u:/tmp # date && cp openSUSE-10.2-GM-DVD-i386.iso openSUSE-10.2-GM-DVD-i386.iso.2 && date Mi Jan 17 13:31:22 CET 2007 Mi Jan 17 13:34:29 CET 2007 linux-v77u:/tmp # date && cp openSUSE-10.2-GM-DVD-i386.iso openSUSE-10.2-GM-DVD-i386.iso.2 && date Mi Jan 17 13:34:38 CET 2007 Mi Jan 17 13:37:42 CET 2007 linux-v77u:/tmp # uname -a Linux linux-v77u 2.6.16.13-4-smp #1 SMP Wed May 3 04:53:23 UTC 2006 i686 i686 i386 GNU/Linux linux-v77u:/tmp # hdparm -tT /dev/sda /dev/sda: Timing cached reads: 2884 MB in 2.00 seconds = 1442.79 MB/sec Timing buffered disk reads: 178 MB in 3.02 seconds = 58.99 MB/sec linux-v77u:/tmp # -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 npiggin@novell.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |ASSIGNED ------- Comment #13 from npiggin@novell.com 2007-01-21 21:54 MST ------- Interesting issue. I'm not sure off the top of my head what could be slowing down the cached reads so much, assuming hdparm is not buggy. I'll try reproducing here, but in the meantime, can you get kernel profiles of a slow run vs a fast run? (boot with profile=1 on the cmdline, run `readprofile -r` before hdparm, and `readprofile | sort -nr > profile.out` afterwards. Thanks -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 ------- Comment #14 from bjoern.wahl@hospital-borken.de 2007-01-22 02:04 MST ------- Created an attachment (id=114100) --> (https://bugzilla.novell.com/attachment.cgi?id=114100&action=view) profiles txt Attached you find the results for a slow pc... -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 ------- Comment #15 from bjoern.wahl@hospital-borken.de 2007-01-22 03:47 MST ------- Created an attachment (id=114115) --> (https://bugzilla.novell.com/attachment.cgi?id=114115&action=view) profile from fast os opensuse10.1 attached you find the results of a fast os 10.1 opensuse -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 ------- Comment #16 from npiggin@novell.com 2007-01-22 18:10 MST ------- Created an attachment (id=114301) --> (https://bugzilla.novell.com/attachment.cgi?id=114301&action=view) pagecache profile test script OK, the profiles from hdparm are unclear as to what the problem is. Half of the issue with hdparm is that it does a variable amount of work, so the profiles aren't measuring equal things. Attached is a script to test pagecache performance. If you could boot with profile=1, and run it as root in an empty directory, that might shed some light on things. Send the resulting profile.out file, and the output of the script, for both the slow and fast kernels. Thanks! -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 ------- Comment #17 from bjoern.wahl@hospital-borken.de 2007-01-25 09:03 MST ------- Created an attachment (id=115144) --> (https://bugzilla.novell.com/attachment.cgi?id=115144&action=view) nothing The slow os (Opensuse10.2) result: # ./test 39.798991617 40.893670745 linux-yytz:/tmp/test # cat profile.out 531 gesamt 0,0003 275 mwait_idle 4,9107 155 __copy_to_user_ll 0,7110 17 sysenter_past_esp 0,1405 13 find_get_page 0,2321 10 do_generic_mapping_read 0,0091 9 vfs_read 0,0258 7 put_page 0,1400 7 dnotify_parent 0,0787 6 _spin_lock 0,4000 5 memcpy 0,1250 4 inotify_inode_queue_event 0,0192 3 sys_read 0,0291 2 sys_write 0,0194 2 rw_verify_area 0,0140 2 kmap_atomic 0,0172 2 __generic_file_aio_read 0,0044 2 do_page_fault 0,0016 1 vfs_write 0,0028 1 touch_atime 0,0070 1 page_cache_readahead 0,0024 1 page_address 0,0078 1 mktime 0,0079 1 generic_file_aio_read 0,0141 1 do_sync_read 0,0040 1 __d_lookup 0,0046 1 __copy_from_user_ll 0,0046 1 _atomic_dec_and_lock 0,0147 0 *unknown* linux-yytz:/tmp/test # -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 ------- Comment #18 from bjoern.wahl@hospital-borken.de 2007-01-26 02:00 MST ------- Created an attachment (id=115317) --> (https://bugzilla.novell.com/attachment.cgi?id=115317&action=view) nothing Result for a fast system (opensuse10.1) # ./test 27.010811000 28.559914000 linux-3pge:/tmp/test # cat profile.out 737 gesamt 0,0004 386 mwait_idle 7,4231 160 __copy_to_user_ll 0,7339 25 audit_syscall_exit 0,0315 21 sysenter_past_esp 0,1736 15 _spin_lock 1,0000 15 find_get_page 0,2679 11 current_kernel_time 0,1897 9 do_generic_mapping_read 0,0081 9 dnotify_parent 0,1011 8 vfs_write 0,0245 8 audit_syscall_entry 0,0280 7 fget_light 0,0693 5 vfs_read 0,0154 5 current_fs_time 0,0581 4 touch_atime 0,0280 4 sys_write 0,0404 4 put_page 0,0800 4 memcpy 0,1000 4 do_page_fault 0,0030 3 rw_verify_area 0,0227 3 __generic_file_aio_read 0,0068 3 do_syscall_trace 0,0083 3 audit_filter_syscall 0,0203 2 vsnprintf 0,0018 2 syscall_call 0,1818 2 file_read_actor 0,0089 1 vm_acct_memory 0,0192 1 vfs_getattr 0,0067 1 unmap_vmas 0,0008 1 timespec_trunc 0,0185 1 sys_read 0,0101 1 syscall_exit_work 0,0312 1 page_address 0,0078 1 number 0,0017 1 __link_path_walk 0,0003 1 kmap_atomic 0,0092 1 inotify_inode_queue_event 0,0032 1 __handle_mm_fault 0,0005 1 generic_file_read 0,0058 1 do_path_lookup 0,0017 1 copy_process 0,0002 0 *unknown* -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 ------- Comment #19 from npiggin@novell.com 2007-01-28 19:01 MST ------- OK, 10.2 is completing in 1.1 seconds, while 10.1 is completing in 1.5, with the profile showing that 10.1 is using more CPU to do the same amount of work. So the problem does not seem to be reading from pagecache performance. Maybe hdparm is just doing something silly? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 ------- Comment #20 from bjoern.wahl@hospital-borken.de 2007-01-29 00:32 MST ------- Created an attachment (id=115813) --> (https://bugzilla.novell.com/attachment.cgi?id=115813&action=view) nothing Well it`s absolutely possible thats it`s only a bug in hdparm. I just also had the feeling that it`s slower.... Maybe you just keep that in mind if something else in this direction arises. Thanks so far! -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 ------- Comment #21 from npiggin@novell.com 2007-01-29 23:57 MST ------- OK, well thanks for reporting -- hdparm does have a significant and unexplained drop in numbers. The only difference that hdparm should be doing is reading through the buffer cache rather than pagecache (although in Linux 2.6 kernels this is basically the same thing, it could be a slight difference). I will attach another patch which you can run with the argument of your block device (ie. /dev/sda) and see if the slowdown shows there. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 ------- Comment #22 from npiggin@novell.com 2007-01-29 23:59 MST ------- Created an attachment (id=116056) --> (https://bugzilla.novell.com/attachment.cgi?id=116056&action=view) buffercache tester -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914 npiggin@novell.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |NEEDINFO Info Provider| |bjoern.wahl@hospital-borken.de -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=229914#c23
Stephan Kulow
participants (1)
-
bugzilla_noreply@novell.com