diff options
| author | Simon Marlow <marlowsd@gmail.com> | 2012-05-16 09:56:12 +0100 |
|---|---|---|
| committer | Simon Marlow <marlowsd@gmail.com> | 2012-05-16 15:53:55 +0100 |
| commit | 66839957a376dbe85608822c1820eb6c99210883 (patch) | |
| tree | 7c984875861ef394c8b5a81b2347be98b3e66b87 /rts | |
| parent | 7fe110aaf5d464b7934e1429b943d862a6da8c06 (diff) | |
| download | haskell-66839957a376dbe85608822c1820eb6c99210883.tar.gz | |
Set the context_switch flag in yield#
yieldThread hasn't been working for a while: unless we set the
context_switch flag to indicate that the current time slice is over,
the RTS scheduler just runs the same thread again. Spotted by Andreas
Voellmy (thanks!).
Diffstat (limited to 'rts')
| -rw-r--r-- | rts/PrimOps.cmm | 5 |
1 files changed, 5 insertions, 0 deletions
diff --git a/rts/PrimOps.cmm b/rts/PrimOps.cmm index 93ef23a71e..2ab26f7ba4 100644 --- a/rts/PrimOps.cmm +++ b/rts/PrimOps.cmm @@ -665,6 +665,11 @@ stg_forkOnzh stg_yieldzh { + // when we yield to the scheduler, we have to tell it to put the + // current thread to the back of the queue by setting the + // context_switch flag. If we don't do this, it will run the same + // thread again. + Capability_context_switch(MyCapability()) = 1 :: CInt; jump stg_yield_noregs; } |