Bug 150595 – [FTL] Fix the build with LLVM 3.7

• Component: New Bugs
• Status: RESOLVED FIXED

[Csaba Osztrogonác]

build error:

../../Source/JavaScriptCore/llvm/library/LLVMExports.cpp: In function 'JSC::LLVMAPI* initializeAndGetJSCLLVMAPI(void (*)(const char*, ...) volatile, bool*)':
../../Source/JavaScriptCore/llvm/library/LLVMExports.cpp:132:17496: error: invalid conversion from 'LLVMOpaqueValue* (*)(LLVMBuilderRef, LLVMTypeRef, unsigned int, const char*) {aka LLVMOpaqueValue* (*)(LLVMOpaqueBuilder*, LLVMOpaqueType*, unsigned int, const char*)}' to 'LLVMOpaqueValue* (*)(LLVMBuilderRef, LLVMTypeRef, LLVMValueRef, unsigned int, const char*) {aka LLVMOpaqueValue* (*)(LLVMOpaqueBuilder*, LLVMOpaqueType*, LLVMOpaqueValue*, unsigned int, const char*)}' [-fpermissive]

[Csaba Osztrogonác]

Created attachment 264147 [details]
Patch

WIP patch, it fixed the build with LLVM 3.7, but we need a fix works with 3.6 and 3.7 too.

[Csaba Osztrogonác]

Created attachment 264150 [details]
Patch

[Csaba Osztrogonác]

Created attachment 264204 [details]
Patch

EFL EWS was flakey, let's see again.

[Darin Adler]

Comment on attachment 264204 [details]
Patch

I think Filip should review this.

[Csaba Osztrogonác]

(In reply to comment #4)
> Comment on attachment 264204 [details]
> Patch
> 
> I think Filip should review this.

Filip, could you possibly check it?

[Csaba Osztrogonác]

Ping?

[Filip Pizlo]

Comment on attachment 264204 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=264204&action=review

> Source/JavaScriptCore/llvm/LLVMAPI.h:50
> +
> +    // LLVMBuildLandingPad has different prototype with LLVM 3.7 and prior versions.
> +#if LLVM_VERSION_MAJOR >= 4 || (LLVM_VERSION_MAJOR == 3 && LLVM_VERSION_MINOR >= 7)
> +    LLVMValueRef (*BuildLandingPad)(LLVMBuilderRef B, LLVMTypeRef Ty, unsigned NumClauses, const char *Name);
> +#else
> +    LLVMValueRef (*BuildLandingPad)(LLVMBuilderRef B, LLVMTypeRef Ty, LLVMValueRef PersFn, unsigned NumClauses, const char *Name);
> +#endif

Let's just remove this function instead.  We don't call it, and it's a mistake for it to be in the API struct.

[Csaba Osztrogonác]

Created attachment 265037 [details]
Patch

OK, let's remove it. Should we do a general cleanup for other functions too, or just leave them unchaged until we run into any issue?

[Filip Pizlo]

(In reply to comment #8)
> Created attachment 265037 [details]
> Patch
> 
> OK, let's remove it. Should we do a general cleanup for other functions too,
> or just leave them unchaged until we run into any issue?

Yes, I think that would be very useful!

[WebKit Commit Bot]

Comment on attachment 265037 [details]
Patch

Clearing flags on attachment: 265037

Committed r192164: <http://trac.webkit.org/changeset/192164>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.