#5282: Bizarre results from -P profiler on OS X
---------------------------------+------------------------------------------
Reporter: bos | Owner:
Type: bug | Status: new
Priority: high | Milestone: 7.2.1
Component: Runtime System | Version: 7.0.3
Keywords: | Testcase:
Blockedby: | Difficulty:
Os: Unknown/Multiple | Blocking:
Architecture: x86_64 (amd64) | Failure: Incorrect result at runtime
---------------------------------+------------------------------------------
Comment(by scpmw):
A few more tidbits of information, from debugging under Mac OS: The signal
seems to be called from strangely predictable places. For single-threaded
RTS, it keeps getting called from `blockUserSignals` and
`unblockUserSignals`. When using `--install-signal-handlers=no`, suddenly
all calls come from the `getrusage` call in `getProcessCPUTime`. I have
verified that using a global status variable: All offending timer calls
happen while in that function.
Another interesting puzzle piece: The way the itimer is initialized seems
to be the reason that it is specific to `-threaded`: If I put "`|| 1`" at
the end of Itimer.c:83, I get correct results for non-threaded builds.
So... Signal masking causing SIGVTALARM to fire? Hm.
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/5282#comment:14>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler
_______________________________________________
Glasgow-haskell-bugs mailing list
Glasgow-haskell-bugs@haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
