1144692 - Fix --disable-ion builds

Status: RESOLVED FIXED

(Core :: JavaScript Engine: JIT, defect)

Description

[Benjamin Bouvier [:bbouvier] (inactive)]

Attachment: fix-ion-builds.patch

I think ted asked for this recently on jsapi...
This patch adds a few missing methods and changes the registers in none/ so that register set assertions don't blow up (context: not a single jit-test was passing with the --disable-ion option because of the register set assertions in code() in jit/Registers.h).

Comment 1

[Hannes Verschore [:h4writer]]

Comment on attachment 8579387 [details] [diff] [review]
fix-ion-builds.patch

Review of attachment 8579387 [details] [diff] [review]:
-----------------------------------------------------------------

r+ with requested changes.

::: js/src/jit/none/MacroAssembler-none.h
@@ +61,4 @@
>  
> +static MOZ_CONSTEXPR_VAR Register JSReturnReg_Type = { Registers::r0 };
> +static MOZ_CONSTEXPR_VAR Register JSReturnReg_Data = { Registers::r0 };
> +static MOZ_CONSTEXPR_VAR Register JSReturnReg = { Registers::r0 };

I would prefer to keep this "invalid_reg"!

Instead can you move the definition of "EntryTempMask" inside the only functions that needs it? So inside "generateArgumentsChecks"?

That solves it locally.

Comment 2

[Benjamin Bouvier [:bbouvier] (inactive)]

(In reply to Hannes Verschore [:h4writer] from comment #1)
> Instead can you move the definition of "EntryTempMask" inside the only
> functions that needs it? So inside "generateArgumentsChecks"?
Indeed simpler!

(no try build, this isn't actually tested on tbpl)
https://hg.mozilla.org/integration/mozilla-inbound/rev/ec8f020d47ae

Comment 3

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

https://hg.mozilla.org/mozilla-central/rev/ec8f020d47ae