When a thread hints that it is in a spin-wait loop (eg, x86 REP NOP insn),
commit4a06d64ce514d188f01455c1d2d9cbd7996c7c79
authorJulian Seward <jseward@acm.org>
Wed, 24 May 2017 13:42:33 +0000 (24 13:42 +0000)
committerJulian Seward <jseward@acm.org>
Wed, 24 May 2017 13:42:33 +0000 (24 13:42 +0000)
tree6a35ba17092d1b4e05681164a7846dcf0bac568e
parent2cc0e531ecddea7bcface7071e02a01fc10517c0
When a thread hints that it is in a spin-wait loop (eg, x86 REP NOP insn),
deschedule the thread much sooner (300 evchecks) than previously (1000
evchecks).  Once the thread is spinning, there's absolutely no point in
continuing with it, since only a different thread can get it out of the loop,
so we need to drop the current thread ASAP.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@16412
coregrind/m_scheduler/scheduler.c