1047010 - OdinMonkey: improve asm.js profiling data

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

[Luke Wagner [:luke]]

Attachment: better-ffis

Pretty easy to do and useful for profiling.

Comment 1

[Luke Wagner [:luke]]

Attachment: better-signal-handlers

With this tiny patch, time spent in the out-of-bounds and slow-script handlers will be appropriately represented as such.

Comment 2

[Douglas Crosher [:dougc]]

Comment on attachment 8465711 [details] [diff] [review]
better-ffis

Review of attachment 8465711 [details] [diff] [review]:
-----------------------------------------------------------------

Looks good.

::: js/src/asmjs/AsmJSModule.h
@@ +829,5 @@
>      }
> +    bool addCodeRange(CodeRange::Kind kind, uint32_t begin, uint32_t end) {
> +        return codeRanges_.append(CodeRange(kind, begin, end));
> +    }
> +    bool addCodeRange(CodeRange::Kind kind, uint32_t begin, uint32_t pret, uint32_t end) {

s/pret/pRet/ ?

Comment 3

[Douglas Crosher [:dougc]]

Comment on attachment 8465765 [details] [diff] [review]
better-signal-handlers

Review of attachment 8465765 [details] [diff] [review]:
-----------------------------------------------------------------

Looks good.

Just out of curiosity, does the interrupt frame include the profiler SIGPROF interrupts?

Comment 4

[Luke Wagner [:luke]]

(In reply to Douglas Crosher [:dougc] from comment #3)
> Just out of curiosity, does the interrupt frame include the profiler SIGPROF
> interrupts?

Nope: when the signal or suspend-thread + get-thread-state calls stop the thread, the context retrieved is the point of execution when the asynchronous suspension happened so, if we were executing asm.js code, the pc will be in asm.js code.  The case handled here is executing asm.js code, then you take a fault for out-of-bounds or slow-script-interrupt, then you are in the handler and THEN you get suspended for profiling.  Madness.

Comment 5

[Luke Wagner [:luke]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/9dc9557dca8e
https://hg.mozilla.org/integration/mozilla-inbound/rev/ca9fa41e977c

Comment 6

[Ed Morley [:emorley]]

https://hg.mozilla.org/mozilla-central/rev/ca9fa41e977c
https://hg.mozilla.org/mozilla-central/rev/9dc9557dca8e