[webkit-reviews] review granted: [Bug 235388] [JSC] Fix non-JIT Windows LLInt : [Attachment 449532] Patch

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Wed Jan 19 17:29:20 PST 2022


Mark Lam <mark.lam at apple.com> has granted Yusuke Suzuki <ysuzuki at apple.com>'s
request for review:
Bug 235388: [JSC] Fix non-JIT Windows LLInt
https://bugs.webkit.org/show_bug.cgi?id=235388

Attachment 449532: Patch

https://bugs.webkit.org/attachment.cgi?id=449532&action=review




--- Comment #3 from Mark Lam <mark.lam at apple.com> ---
Comment on attachment 449532
  --> https://bugs.webkit.org/attachment.cgi?id=449532
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=449532&action=review

r=me with fixes.

> Source/JavaScriptCore/llint/LowLevelInterpreter32_64.asm:131
> +    elsif C_LOOP or C_LOOP_WIN
> +	   cloopCallSlowPath3 function, a0, a1, a2

This should come first because C_LOOP can have X86 or X86_WIN enabled too (or
any other CPU arch).  I'm surprised that it works for cCall2 (don't know why). 
Maybe should fix that too for consistency.

> Source/JavaScriptCore/llint/LowLevelInterpreter64.asm:165
> +	   addp 64, sp

Maybe this `addp` should come after the register moves below.  r0 is still
pointing to the reserved stack position, right?  They should remain protected
by sp until after we move the results out into registers.


More information about the webkit-reviews mailing list