[Bug 446194] New: mono: powertop flamage ...
https://bugzilla.novell.com/show_bug.cgi?id=446194 Summary: mono: powertop flamage ... Product: openSUSE 11.1 Version: Beta 4 Platform: Other OS/Version: Other Status: NEW Severity: Major Priority: P5 - None Component: Mono AssignedTo: bnc-team-mono@forge.provo.novell.com ReportedBy: mmeeks@novell.com QAContact: qa@suse.de CC: dick@novell.com Found By: --- So - it seems all mono apps like to wake up the CPU remarkably regularly, which is not only bad news for battery life, but a sign of some strange busy-looping (surely) ? f-spot seems to wake up 40x / second, beagled only 20x. powertop has a helpful hint compiled in that says this: "Suggestion: Disable or remove 'beagle' from your system. Beagle is the program that indexes for easy desktop search, however it's not very efficient and costs a significant amount of battery life." Clearly when quiescent (as it is), beagle shouldn't be doing anything - much less using CPU - surely it should just be sitting waiting on the inotify socket ? I understood there was a fix coming in this area some time back - did that not make it for Mono 2.0 ? also - what do we poll here for ? it looks like we call clock_gettime and go back to sleep again - why ? surely we can do a pthread_cond_timedwait with a better time ? Any ideas Dick ? -- 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.
https://bugzilla.novell.com/show_bug.cgi?id=446194
User dick@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=446194#c1
Dick Porter
https://bugzilla.novell.com/show_bug.cgi?id=446194
User miguel@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=446194#c2
Miguel de Icaza
https://bugzilla.novell.com/show_bug.cgi?id=446194
User rkumpera@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=446194#c3
--- Comment #3 from Rodrigo Kumpera
https://bugzilla.novell.com/show_bug.cgi?id=446194
User rkumpera@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=446194#c5
--- Comment #5 from Rodrigo Kumpera
https://bugzilla.novell.com/show_bug.cgi?id=446194
User mmeeks@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=446194#c6
--- Comment #6 from Michael Meeks
2.A)since thread A in blocked waiting on a condvar, we first need to wakeup it.
Clearly :-)
if we deliver a signal and do a pthread_cond_broadcast there is big chance of hitting a non-reentrant code path. Posix clearly states that mutex and condvar related functions are not signal context safe.
Fair enough; I read the glibc/nptl/pthread_cond_broadcast code again - and indeed _broadcast is not signal safe - since it takes a lock before doing the futex syscalls to determine if there are any waiters :-) and I guess the signal could occur in that section, sigh [ of course, since our use-case is not performance critical, and we have no need to take that lock it's a shame ... ] - we could of course, clobber the glibc ABI and do the lll_futex_wake call ourselves inline without that protection - sucky; but perhaps less sucky than re-writing & maintaining a copy of the locking primitives from glibc. -- 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.
https://bugzilla.novell.com/show_bug.cgi?id=446194
User mmeeks@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=446194#c7
--- Comment #7 from Michael Meeks
https://bugzilla.novell.com/show_bug.cgi?id=446194
User rkumpera@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=446194#c8
--- Comment #8 from Rodrigo Kumpera
https://bugzilla.novell.com/show_bug.cgi?id=446194
User rkumpera@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=446194#c10
Rodrigo Kumpera
https://bugzilla.novell.com/show_bug.cgi?id=446194
User lupus@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=446194#c11
--- Comment #11 from Paolo Molaro
https://bugzilla.novell.com/show_bug.cgi?id=446194
User glunardi@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=446194#c12
Guy Lunardi
https://bugzilla.novell.com/show_bug.cgi?id=446194
User miguel@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=446194#c13
--- Comment #13 from Miguel de Icaza
https://bugzilla.novell.com/show_bug.cgi?id=446194
User glunardi@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=446194#c14
--- Comment #14 from Guy Lunardi
https://bugzilla.novell.com/show_bug.cgi?id=446194
User mmeeks@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=446194#c15
--- Comment #15 from Michael Meeks
participants (1)
-
bugzilla_noreply@novell.com