ipc/sem: play nicer with large nsops allocations
authorDavidlohr Bueso <dave@stgolabs.net>
Fri, 8 Sep 2017 23:17:52 +0000 (16:17 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Sat, 9 Sep 2017 01:26:51 +0000 (18:26 -0700)
commite4243b8062c13b8cb3d91695dc353cb9e6a0da25
treed938e7b91914cef64560cdd0ad28ec510ae97e63
parent8419e64a0b734a1f98a07fc7c489495bebc6e33a
ipc/sem: play nicer with large nsops allocations

Replacing semop()'s kmalloc for kvmalloc was originally proposed by
Manfred on the premise that it can be called for large (than order-1)
sizes.  For example, while Oracle recommends setting SEMOPM to a _minimum_
of 100, some distros[1] encourage the setting to be a factor of the amount
of db tasks (PROCESSES), which can get fishy for large systems (easily
going beyond 1000).

[1] An Example of Semaphore Settings
https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/5/html/Tuning_and_Optimizing_Red_Hat_Enterprise_Linux_for_Oracle_9i_and_10g_Databases/sect-Oracle_9i_and_10g_Tuning_Guide-Setting_Semaphores-An_Example_of_Semaphore_Settings.html

So let's just convert this to kvmalloc, just like the rest of the
allocations we do in ipc.  While the fallback vmalloc obviously involves
more overhead, this by far the uncommon path, and it's better for the user
than just erroring out with kmalloc.

Link: http://lkml.kernel.org/r/20170803184136.13855-2-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
ipc/sem.c