Gabriel Baraldi
Gabriel Baraldi
I guess to be clear, we exploit similar things in julia i.e assuming consts don't change. It's just that our UB surface area is a lot smaller than C's which...
This is mostly for testing on 1.10 but it's likely this need a similar fix on 1.11 and up
I believe it was the force pushes to the base branch that this targets?
Need clang + lld. And make sure your build starts from scratch
Btw, since you are changing this code, is the linked list an efficient enough data structure for this? I haven't measured the overhead of sweeping through this so I'm not...
Could this expose parallelism for sweeping the big objs?
On the libc thing the answer is the very fun (it depends on the malloc implementation). Though we could move it to the background thread regardless (from what i've read...
This fixes the llvmpasses tests ```diff diff --git a/test/llvmpasses/pipeline-prints.ll b/test/llvmpasses/pipeline-prints.ll index babd26c797..ecb7095302 100644 --- a/test/llvmpasses/pipeline-prints.ll +++ b/test/llvmpasses/pipeline-prints.ll @@ -298,12 +298,12 @@ attributes #2 = { inaccessiblemem_or_argmemonly } ; COM: Loop...
The windows build is segfaulting in ``` #0 0x00007ff8e4f70de5 in ntdll!RtlVirtualUnwind () from /c/Windows/SYSTEM32/ntdll.dll #1 0x00007ff8e4f70ccb in ntdll!RtlVirtualUnwind () from /c/Windows/SYSTEM32/ntdll.dll #2 0x00007ff8e3ee3bd8 in RtlVirtualUnwind () from /c/Windows/System32/KERNEL32.DLL #3 0x00007ff8bb29cfed...
This doesn't have the new LLVM build yet. I was fixing the embedding thing mostly