1243754 - TEST-UNEXPECTED-FAIL | valgrind-test | 4,096 bytes in 1 blocks are definitely lost at malloc / js_malloc / new_ / getOrCreateChunk

Status: RESOLVED FIXED

(Core :: JavaScript Engine: JIT, defect)

Description

[Hannes Verschore [:h4writer]]

Fallout from landing of shard stubs in bug 1241087

Comment 1

[Hannes Verschore [:h4writer]]

Attachment: Free the stubspace

Small patch. I failed to account for the different way BaselineScript and IonScript are destroyed. As a result the fallbackspace wasn't free'ed. Manually freeing here.

Comment 2

[Jan de Mooij [:jandem]]

Comment on attachment 8713534 [details] [diff] [review]
Free the stubspace

Review of attachment 8713534 [details] [diff] [review]:
-----------------------------------------------------------------

::: js/src/jit/Ion.cpp
@@ +1280,5 @@
>  {
>      script->unlinkFromRuntime(fop);
>      // Frees the potential event we have set.
>      script->traceLoggerScriptEvent_ = TraceLoggerEvent();
> +    script->fallbackStubSpace_.free();

Instead of these last two lines, can we use fop->delete_(script) below, instead of free_?

Comment 3

[Jan de Mooij [:jandem]]

Comment on attachment 8713534 [details] [diff] [review]
Free the stubspace

Clearing per comment 2.

Comment 4

[Hannes Verschore [:h4writer]]

Attachment: Delete the IonScript

Indeed a little bit more elegant.

Comment 5

[Jan de Mooij [:jandem]]

Comment on attachment 8715774 [details] [diff] [review]
Delete the IonScript

Review of attachment 8715774 [details] [diff] [review]:
-----------------------------------------------------------------

Thanks!

Comment 6

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/374422755fcc

Comment 7

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/374422755fcc