1309246 - bug909997.js test failure on sparc due set JS_CODEGEN_NONE

Status: RESOLVED FIXED

(Core :: JavaScript Engine: JIT, defect, P5)

Description

[Petr Sumbera]

User Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:49.0) Gecko/20100101 Firefox/49.0
Build ID: 20160922113459

Steps to reproduce:

This is the only js test failing:

FAIL - ion/bug909997.js
TEST-UNEXPECTED-FAIL | js/src/jit-test/tests/ion/bug909997.js | /builds2/psumbera/userland-ff-45.4esr/components/desktop/firefox/firefox-45.4.0esr/js/src/jit-test/tests/ion/bug909997.js:31:5 Error: Assertion failed: got 0, expected 1 (code 3, args "")

This is becuase sparc code generation is missing (JS_CODEGEN_NONE=1) and JS_GetGlobalJitCompilerOption() will always return 0:

JS_GetGlobalJitCompilerOption(JSRuntime* rt, JSJitCompilerOption opt)
{
printf("JS_GetGlobalJitCompilerOption\n");
#ifndef JS_CODEGEN_NONE
..
#endif
    return 0;
}
  



Expected results:

Since I'm new here I would like to ask for some comments on this.

What are the consequences for FF on sparc than?
Is it expected that bug909997.js test will fail or it's another bug beside missing sparc support for code generation?
Any adivce/pointers if I decide to try to add missing sparc support?

Comment 1

[Tooru Fujisawa [:arai]]

sounds like, we need to make it possible to detect JS_CODEGEN_NONE by testing function,
and skip the test on JS_CODEGEN_NONE.

nbp, how do you think?

Comment 2

[Nicolas B. Pierron [:nbp]]

(In reply to Petr Sumbera from comment #0)
> What are the consequences for FF on sparc than?

There is none.  This function handles reading the flags used by the JIT, but when the JIT is disabled this function does nothing.

> Any adivce/pointers if I decide to try to add missing sparc support?

Either add a new testing function [1], or a new enumerated value which is always supposed to return 1, in the JS_GetGlobalJitCompilerOption function.  Thus, in the test case, add a condition, and call the skip() function if the jit is disabled.


[1] http://searchfox.org/mozilla-central/source/js/src/builtin/TestingFunctions.cpp#3847-3851

Comment 3

[Petr Sumbera]

Attachment: Bug1309246.patch

Comment 4

[Nicolas B. Pierron [:nbp]]

Comment on attachment 8799890 [details] [diff] [review]
Bug1309246.patch

Review of attachment 8799890 [details] [diff] [review]:
-----------------------------------------------------------------

Nice!

Comment 5

[Nicolas B. Pierron [:nbp]]

The current patch just modifies the test case to skip it if Baseline is disabled (or all jit are disabled).
I think we can safely land this without any try-server run.

Comment 6

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/5758ad60c4d8
avoids bug909997.js test failure on systems with JS_CODEGEN_NONE set. r=nbp

Comment 7

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/5758ad60c4d8