Closed
Bug 493316
Opened 15 years ago
Closed 15 years ago
Incorrect execution on bug 460050 testcase with MAX_BRANCHES=4
Categories
(Core :: JavaScript Engine, defect)
Tracking
()
RESOLVED
DUPLICATE
of bug 493657
People
(Reporter: bzbarsky, Assigned: gal)
References
(Blocks 1 open bug)
Details
BUILD: optimized js shell build from t-m rev a75d552b0c64 with the following local changes: a) The v2 patch from bug 490364 applied 2) MAX_BRANCHES changed from 32 to 4 in jstracer.cpp STEPS TO REPRODUCE: Run the full go game js shell testcase from bug 460050. That would be attachment 377491 [details]. Do this with -j and without. EXPECTED RESULTS: Identical output with and without jit enabled. ACTUAL RESULTS: Different results with jit enabled and jit disabled. ADDITIONAL INFORMATION: The problem doesn't seem to get triggered with MAX_BRANCHES=2 or MAX_BRANCHES=32. I can reproduce it with MAX_BRANCHES=8, though. The output in that case is identical to the MAX_BRANCHES=4 output.
Flags: blocking1.9.1?
Reporter | ||
Comment 1•15 years ago
|
||
Oh, and in a debug build the output for MAX_BRANCHES=8 is also incorrect, and different from what it is in an opt build...
Updated•15 years ago
|
Assignee: general → graydon
Updated•15 years ago
|
Flags: blocking1.9.1? → blocking1.9.1+
Assignee | ||
Comment 2•15 years ago
|
||
Graydon is having some sort of canadian holiday. Stealing until then.
Assignee: graydon → gal
Assignee | ||
Comment 3•15 years ago
|
||
I can't reproduce this with a TM tip macosx debug shell. bz, can you check whether this still happens?
Reporter | ||
Comment 4•15 years ago
|
||
The steps from comment 0 still reproduce the bug, yes (just tested with MAX_BRANCHES=4). That's in the optimized shell, also on OSX, and my tree is at rev 1806b6bc084b. Let me try debug shell.
Reporter | ||
Comment 5•15 years ago
|
||
OK, a debug shell from that same rev with MAX_BRANCHES=4 shows incorrect output as well.
Assignee | ||
Comment 6•15 years ago
|
||
We have fixed https://bugzilla.mozilla.org/show_bug.cgi?id=493657, and that bug is a good explanation for why this bug occurred only with MAXBRANCHES=4. 493657 restored a wrong function object when side-exiting from within a trace. With a lower limit on max branches that happened more frequently. I haven't tested the full test case yet, but I am pretty certain this is fixed now.
Assignee | ||
Comment 7•15 years ago
|
||
Ok, I just verified that this is a dup of 493657 now. I need b+ for 493657, then we can dup this.
Assignee | ||
Updated•15 years ago
|
Status: NEW → RESOLVED
Closed: 15 years ago
Resolution: --- → DUPLICATE
You need to log in
before you can comment on or make changes to this bug.
Description
•