[Libomptarget] Rework image checking further #76120
Conversation
jhuber6
requested review from
jdoerfert,
JonChesterfield,
jplehr,
saiislam and
shiltian
| llvm::getLazyIRModule(std::move(MB), Diagnostic, Context, | ||
| /*ShouldLazyLoadMetadata=*/true); | ||
| auto ModuleOrErr = getLazyBitcodeModule(MemoryBufferRef(Buffer, ""), Context, | ||
| /*ShouldLazyLoadMetadata=*/false); |
There was a problem hiding this comment.
Can the construction of the MemoryBufferRef fail?
I probably don't know enough about the different APIs. Can you tell me why the last parameter here was changed to false?
There was a problem hiding this comment.
No, a memory buffer reference is just a reference to existing memory. The second parameter is the buffer's identifier. E.g. if it was opened from a file it would be the filename.
There was a problem hiding this comment.
Thanks for explaining, and sorry for confusion. I was asking for the parameter about "ShouldLazyLoadMetadata". Why has that changed to false?
There was a problem hiding this comment.
Oh, good catch, that was unintentional. We're only interested in the Triple so we should skip loading the entire table.
Summary: In the future, we may have more checks for different kinds of inputs, e.g. SPIR-V. This patch simply reworks the handling to be more generic and do the magic detection up-front. The checks inside the routines are now asserts so we don't spend time checking this stuff over and over again. This patch also tweaked the bitcode check. I used a different function to get the Lazy-IR module now, as it returns the raw expected value rather than the SM diganostic. No functionality change intended.
Lands and reverts: 64f0681 [Libomptarget] Rework image checking further (llvm#76120) Change-Id: I1420dcaaceb16986d3c2d3f4c0cfc2bc35333fce
Summary:
In the future, we may have more checks for different kinds of inputs,
e.g. SPIR-V. This patch simply reworks the handling to be more generic
and do the magic detection up-front. The checks inside the routines are
now asserts so we don't spend time checking this stuff over and over
again.
This patch also tweaked the bitcode check. I used a different function
to get the Lazy-IR module now, as it returns the raw expected value
rather than the SM diganostic.
No functionality change intended.