[Webkit-unassigned] [Bug 154323] Remove LLVM dependencies from WebKit

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Wed Feb 17 09:36:58 PST 2016


https://bugs.webkit.org/show_bug.cgi?id=154323

--- Comment #7 from Filip Pizlo <fpizlo at apple.com> ---
(In reply to comment #6)
> (In reply to comment #4)
> > It would be a lot better to just write our own arm disassembler.
> 
> It can't be serious that writing a brand new disassembler
> is better choice than using a working one. 
>  
> > The issue with the llvm one is that it goes through the same soft linking
> > nonsense that we do for the rest of llvm. 
> 
> It's not the most beautiful think I've ever seen, but it works.
> But I still don't understand why its existence is an issue?
> 
> Do you have a political reason that everything related
> to LLVM should be removed from the trunk immediately?
> I can understand that Apple don't want to use it,
> in this case just remove it from your build system
> and set USE_LLVM_DISASSEMBLER to 0 on your platform.
> 
> But I don't understand why do you want to force Linux 
> users to drop a working debugging possibility?
> 
> > So, I'm not sure how you're proposing to keep it of gtk/efl 
> > removed the dependency from the build system.
> We only removed LLVM based FTL JIT support from the build system.
> 
> LLVM disassembler is optional and still can be enabled explicitly:
> build-jsc --cmakeargs="-DUSE_LLVM_DISASSEMBLER=1"

Wow first of all, politics has nothing to do with this. I'm trying to determine if we will need to keep the llvm soft linking madness or not. That is how we currently use llvm API. You would have to change the llvm disassembler if you wanted it to no longer depend on the soft linking. 

Second, I have no idea how you think the llvm disassembler helps you on arm traditional. If you look in LLVMDisassembler.cpp you will see that it doesn't have support for arm traditional - only arm64 and x86. 

Also, any support for using the llvm disassembler on x86 should be deleted, regardless of the arm traditonal situation. 

Can you provide some technical assistance instead of acting up?  I'm arguing for the deletion of llvm dependencies on technical grounds based on considering the details of how it's used, and you're coming at me with imaginary hypotheticals like arm traditional.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.webkit.org/pipermail/webkit-unassigned/attachments/20160217/c6377fa3/attachment.html>


More information about the webkit-unassigned mailing list