779017 - always create ScriptSource

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

[:Benjamin Peterson]

JSScript's scriptSource_ pointer should never be NULL. If the source is not available, scriptSource_ should just be a shell. Adding the source code should be a different method. This will allow other items that are shared per compilation unit like filename and the source map to be moved to the ScriptSource struct.

Comment 1

[:Benjamin Peterson]

Attachment: give every script a ScriptSource

Comment 2

[Jason Orendorff [:jorendorff]]

Comment on attachment 647397 [details] [diff] [review]
give every script a ScriptSource

Review of attachment 647397 [details] [diff] [review]:
-----------------------------------------------------------------

Looks good except for a few nits. r=me with those fixed.

::: js/src/frontend/BytecodeCompiler.cpp
@@ +245,5 @@
>  {
>      if (!CheckLength(cx, length))
>          return false;
> +    ScriptSource *ss = cx->new_<ScriptSource>();
> +    AutoAttachToRuntime attacher(cx->runtime, ss);

CompileScript() checks for null ss. This should too.

::: js/src/jsscript.cpp
@@ +1243,3 @@
>      if (!ownSource) {
>          const size_t nbytes = length * sizeof(jschar);
> +        data.compressed = static_cast<unsigned char *>(cx->runtime->malloc_(nbytes));

Instead use cx->malloc_(nbytes) which reports the error on failure.

@@ +1363,4 @@
>              return false;
>      }
> +    if (!xdr->codeBytes(data.compressed, byteLen)) {
> +        xdr->cx()->free_(data.compressed);

This should only free this memory if we allocated it in the first place (that is, if mode is XDR_DECODE).

Also: if we are decoding, it should probably return length_, compressedLength_, and data.compressed to a consistent state before returning, right? (This ScriptSource is already attached to a script, I think, so it'll be destroyed eventually.) Maybe this can be rewritten to first decode to local variables, then call setSource() only on success.

Comment 3

[:Benjamin Peterson]

https://hg.mozilla.org/integration/mozilla-inbound/rev/e6a683016c73

Comment 4

[Ed Morley [:emorley]]

https://hg.mozilla.org/mozilla-central/rev/e6a683016c73