Address comments and update testcases.

Thank you for mentioning this issue, the reason why we added MAI->needsFunctionDescriptors() are as follows:

Fix typos

minor change

A bug is caught that IFunc is not supported on AIX, so we shouldn't have cast a GlobalValue to a GlobalObject for all PPC target.

Update comments and fix typo in the testcase;

delete extra ExternalSymSDNode in code;
modify func name;
update the testcase;

provide full context for diff;

Update two testcases;

Because the best way to test which memcpy version we pick when there exists a user-declared one, we need to test the relocation of the call. So this patch would be moved to needs revision section, waiting for relocation patch to land first.

Address the comments;

ping.

Update the pr to address user-defined/user-declared lib function issue;
Add more testcases;

Clean the formatting issue

Refactor the code

Minor changes

Rebase on latest master;
Address 1st round reviews;

I couldn't apply your patch on latest master branch, do you mind updating it?

Since this is a NFC patch, does it make sense to add NFC on patch title?

Just a reminder, since I landed the patch: https://reviews.llvm.org/D70243. You may need to update three testcases there.

I don't think llvm actually try to put int directly into TC entry, it always put a label in there. But please correct me if I'm wrong.

In D70243#1750779, @jasonliu wrote:

Some observation for constant pool lowering:
We put all the constant pool data into rodata section, and create one TC entry per constant pool reference.
It hinders the ability for linker to garbage collect unused data. (Generates them into a separate RO section helps a bit, but won't fully solve this issue)
And we would also create a lot of TC entries for constant pool that might not get cleanup.

For comparison, gcc generates the value of constant pool data directly into the TC entry. It's faster and garbage collectable if not used.

In D70243#1748063, @jasonliu wrote:

When I apply your patch to master, it didn't apply cleanly. So please rebase.
And when I did some work to manually apply it, and run through clang-format, clang-format changed several places.
I suspect it's due to the tabs vs space issue.
I suggest to update this patch with after running the clang-format on it.

Rebase the patch on the latest master;
Fix some formatting issues;

Update testcases;
Address 1st round reviews.

minor changes

Thank you for your fix!

In D68341#1711822, @hubert.reinterpretcast wrote:

LGTM. Thanks.

update the 64 bit testcase

Minor changes & add zeroext to 64bit testcase

Update testcases: split into simpler ones

correct code format & the testcase

Rebase on latest master after the PPCAsmPrinter cleanup NFC patch landed.

Address 2nd round comments

Address commments & add const for lookUpOrCreateTOCEntry & TOC MapVector<const MCSymbol *, MCSymbol *>

Address 2nd round comments;
split out variable name NFC patch;
update testcases;

Move variable 'SRE' into assertion & rename function name to 'getMCSymbolForTOCPseudoMO'

Rebase on latest master to pick up 'isDarwin' related patch

LGTM. Thank you Jinsong for putting up this patch. I found this warning as well, and I was planning to fix it in one of my PR which is also related to Sean's lowering work and will be landed in the future. But it's certainly nice to have this fixed earlier.