![]() ![]() Here's dummy code that reveals that behavior: _declspec(noinline) int get_alignment(void * ptr) > The (maybe) related bugzilla entry is just odd.I'm writing some low level code and I noticed some unnecessary level of indirection of a function call and I simply don't get what's the reason for that indirection. To: Peter Zijlstra, Linus Torvalds +Cc: Ingo Molnar, Linux Kernel Mailing List 8:28 ` Peter Zijlstra 0 siblings, 1 reply 83+ messages in thread 6:49 ` Peter Zijlstra 8:05 ` Michel Dänzer Model name : AMD A10-7850K Radeon R7, 12 Compute Cores 4C+8Gīut we can start the same debugging session I suppose.Ĭould you run with this patch on provide us with the dmesg after boot? You've not got the ancient dual P4 Xeon Bruno's got. The real fix in that thread I'll also add this to make the scheduler less that's not going to tell us what's wrong with your machine, as Inbox this morning), it was a question of the arch code supplyingĬompletely 'broken' topology information, and the scheduler trusting it So we've had this other thread where the same Michel to that earlier)Īnd that seems to be sorted now (just found positive feedback in my > single word of bitmap, which triggers all the simple code. > And his config file even has NR_CPUS being 32, so it should be a > be properly allocated with the proper addition of cpumask_size(). > That would imply an allocation error, but all the sched groups seem to > where it *just* did a cpumask_clear(), and now the BUG_ON() triggers > + BUG_ON(!cpumask_empty(sched_group_cpus(sg))) > that the BUG_ON() in his added patch triggers: > The (maybe) related bugzilla entry is just odd. > On Tue, at 8:53 PM, Michel Dänzer wrote: On Tue, at 09:21:40PM -0700, Linus Torvalds wrote: To: Linus Torvalds +Cc: Michel Dänzer, Ingo Molnar, Linux Kernel Mailing List 8:05 ` Michel Dänzer 0 siblings, 1 reply 83+ messages in threadįrom: Peter Zijlstra 6:49 UTC ( / raw) 4:21 ` Linus Torvalds 6:49 ` Peter Zijlstra Single word of bitmap, which triggers all the simple code. That would imply an allocation error, but all the sched groups seem toīe properly allocated with the proper addition of cpumask_size().Īnd his config file even has NR_CPUS being 32, so it should be a Where it *just* did a cpumask_clear(), and now the BUG_ON() triggers + BUG_ON(!cpumask_empty(sched_group_cpus(sg))) That the BUG_ON() in his added patch triggers: The (maybe) related bugzilla entry is just odd. > Just happened again with the same change on top of 3.16-rc6. To: Michel Dänzer +Cc: Peter Zijlstra, Ingo Molnar, Linux Kernel Mailing List 6:49 ` Peter Zijlstra 0 siblings, 1 reply 83+ messages in threadįrom: Linus Torvalds 4:21 UTC ( / raw) 3:53 ` Michel Dänzer 4:21 ` Linus Torvalds ![]() I hope this problem is on the radar as a showstopper for 3.16. Just happened again with the same change on top of 3.16-rc6. ![]() > seemed to last longer than before, but maybe that was just luck. > I'm afraid it happened again with 3.16-rc5 plus the above change. To: Peter Zijlstra +Cc: Ingo Molnar, linux-kernel, Linus Torvalds 4:21 ` Linus Torvalds 0 siblings, 1 reply 83+ messages in thread 6:13 ` Michel Dänzer 3:53 ` Michel Dänzer Seemed to last longer than before, but maybe that was just luck. I'm afraid it happened again with 3.16-rc5 plus the above change. 3:53 ` Michel Dänzer 0 siblings, 1 reply 83+ messages in thread 9:29 ` Michel Dänzer 6:13 ` Michel Dänzer Libre software enthusiast | Mesa and X developer Test any better fix you guys come up with. There's a long weekend coming up for me, but after that I'll be happy to To: Peter Zijlstra +Cc: Ingo Molnar, linux-kernel 6:13 ` Michel Dänzer 0 siblings, 1 reply 83+ messages in thread 7:58 ` Random panic in load_balance() with 3.16-rc Peter 9:29 ` Michel Dänzer * Re: Random panic in load_balance() with 3.16-rc config and /proc/cpuinfo attached as well let me know if you > while the box was basically sitting idle. > noticed any pattern as to when it happens at least once it happened > it's legible) randomly while running 3.16-rc4 and -rc5. > I've been running into the panic captured in the attached picture (hope To: Michel Dänzer +Cc: Ingo Molnar, linux-kernel 9:29 ` Michel Dänzer 0 siblings, 1 reply 83+ messages in threadįrom: Peter Zijlstra 7:58 UTC ( / raw) Re: Random panic in load_balance() with 3.16-rc LKML Archive on help / color / mirror / Atom feed * Re: Random panic in load_balance() with 3.16-rc ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |