http://bugzilla.novell.com/show_bug.cgi?id=590242 http://bugzilla.novell.com/show_bug.cgi?id=590242#c1 Stefan Seyfried changed: What |Removed |Added ---------------------------------------------------------------------------- AssignedTo|bnc-team-screening@forge.pr |snwint@novell.com |ovo.novell.com | --- Comment #1 from Stefan Seyfried 2010-03-22 17:28:51 CET --- Did you eventually switch to libx86 or similar CPU emulation? It's especially annoing since hwinfo --framebuffer is called during mkinitrd, apparently for every kernel image, which can lead to quite some waiting ;) -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=590242 http://bugzilla.novell.com/show_bug.cgi?id=590242#c3 --- Comment #3 from Stefan Seyfried 2010-03-23 09:11:14 CET --- Created an attachment (id=349949) --> (http://bugzilla.novell.com/attachment.cgi?id=349949) output of "mdt --modes --show dump" Yes. Each mode seems to take ~500 ms to get. Additionally, both hwinfo and mdt seem to use obscene amounts of memory: strolchi:~ # /usr/bin/time mdt --modes > /dev/null 2.69user 17.20system 0:25.80elapsed 77%CPU (0avgtext+0avgdata 2355200maxresident)k 16inputs+0outputs (1major+3039779minor)pagefaults 0swaps -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=590242 http://bugzilla.novell.com/show_bug.cgi?id=590242#c5 Stefan Seyfried changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |NEW InfoProvider|seife@novell.slipkontur.de | --- Comment #5 from Stefan Seyfried 2010-03-23 12:40:24 CET --- Yes, 32bits, non-PAE, kernel-desktop strolchi:~ # /usr/bin/time mdt --mode 0x0101 0x0101[bb]: 640x480+640, 8 bpp, fb: 0x98000000, a0000.7/a0000.0: 64k 0.35user 0.83system 0:01.86elapsed 63%CPU (0avgtext+0avgdata 2355200maxresident)k 0inputs+0outputs (0major+147363minor)pagefaults 0swaps strolchi:~ # /usr/bin/time hwinfo --framebuffer > /dev/null 3.63user 16.65system 0:32.91elapsed 61%CPU (0avgtext+0avgdata 1585856maxresident)k 7600inputs+0outputs (5major+3039276minor)pagefaults 0swaps So mdt --mode 0x0101 also takes long (almost two seconds) and uses lots of ram. hwinfo --framebuffer in fact does not need *that much* of it (still enough ;) -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=590242 http://bugzilla.novell.com/show_bug.cgi?id=590242#c7 Steffen Winterfeldt changed: What |Removed |Added ---------------------------------------------------------------------------- Priority|P5 - None |P3 - Medium Status|NEW |ASSIGNED --- Comment #7 from Steffen Winterfeldt 2010-03-23 13:57:03 CET --- Ok, I can reproduce it. I'll check. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=590242 http://bugzilla.novell.com/show_bug.cgi?id=590242#c Steffen Winterfeldt changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |snwint@novell.com AssignedTo|snwint@novell.com |kernel-maintainers@forge.pr | |ovo.novell.com -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=590242 http://bugzilla.novell.com/show_bug.cgi?id=590242#c10 Jeff Mahoney changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |jeffm@novell.com --- Comment #10 from Jeff Mahoney 2010-03-23 14:59:07 UTC --- Wait what? -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=590242 http://bugzilla.novell.com/show_bug.cgi?id=590242#c12 --- Comment #12 from Steffen Winterfeldt 2010-03-23 16:48:22 CET --- Yes, but it shouldn't swap; that's the point. It didn't do it before. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=590242 http://bugzilla.novell.com/show_bug.cgi?id=590242#c14 --- Comment #14 from Steffen Winterfeldt 2010-03-24 11:46:29 CET --- You are right; when I replace malloc() with mmap() the problem is gone. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=590242 http://bugzilla.novell.com/show_bug.cgi?id=590242#c16 --- Comment #16 from Steffen Winterfeldt 2010-03-25 17:52:47 CET --- Created an attachment (id=350606) --> (http://bugzilla.novell.com/attachment.cgi?id=350606) test run logs I also can't reproduce this in a factory chroot with 11.2 kernel. That's why I assigned it to kernel in the first place. Fact is, though, that running factory on a factory kernel shows a difference between using malloc() and mmap(). The attached log (taken in a vmware vm) shows the time between mmap() calls increasing because it starts swapping if the memory is used up. If I take the swap away, the process gets oom-killed. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=590242 http://bugzilla.novell.com/show_bug.cgi?id=590242#c18 Steffen Winterfeldt changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |ASSIGNED InfoProvider|snwint@novell.com | --- Comment #18 from Steffen Winterfeldt 2010-03-25 18:03:26 CET --- Created an attachment (id=350615) --> (http://bugzilla.novell.com/attachment.cgi?id=350615) updated test program In case I'm missing something obvious, here is the test program I used. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=590242 http://bugzilla.novell.com/show_bug.cgi?id=590242#c20 --- Comment #20 from Steffen Winterfeldt 2010-03-25 18:13:52 CET ---

Doesn't the small extra mmap() size make a difference?

No. Using the values from the malloc() log for mmap() shows no difference. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=590242 http://bugzilla.novell.com/show_bug.cgi?id=590242#c22 --- Comment #22 from Steffen Winterfeldt 2010-03-26 11:02:13 CET --- I had it running in the rescue system from our install dvd. I don't think there's anything unusual in the environment there. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=590242 http://bugzilla.novell.com/show_bug.cgi?id=590242#c24 --- Comment #24 from Steffen Winterfeldt 2010-03-26 14:04:33 CET --- The difference is MALLOC_PERTURB_ = E vs. 69. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=590242 http://bugzilla.novell.com/show_bug.cgi?id=590242#c26 Petr Baudis changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |NEEDINFO InfoProvider| |snwint@novell.com --- Comment #26 from Petr Baudis 2010-05-10 21:42:21 UTC --- Yes, anything non-numeric is alias for 0 aka "don't perturb" (perturb will fill allocated/freed space with non-zero bytes). What on your system is setting these environment variables? My system has nothing of this. We should get rid of them if this is some OpenSUSE default now. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=590242 http://bugzilla.novell.com/show_bug.cgi?id=590242#c28 Petr Baudis changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |ASSIGNED CC| |pbaudis@novell.com InfoProvider|snwint@novell.com | AssignedTo|pbaudis@novell.com |coolo@novell.com --- Comment #28 from Petr Baudis 2010-05-11 08:44:51 UTC --- Okay, then I'm reassigning this to Coolo to decide whether we want to change or document the defaults in any way. Thank you all. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=590242 http://bugzilla.novell.com/show_bug.cgi?id=590242#c30 --- Comment #30 from Marcus Meissner 2010-05-28 14:44:43 UTC --- it also shows that hwinfo --framebuffer does way too much allocations ;) -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=590242 http://bugzilla.novell.com/show_bug.cgi?id=590242#c31 Marcus Meissner changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED --- Comment #31 from Marcus Meissner 2010-06-23 11:44:45 UTC --- malloc checking is now off in rc1. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.