Test the symbol resolution related to defined foo@@v1 and foo@v1 in object files/shared objects, undefined foo@v1, and the interaction with --wrap.
Details
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
lld/test/ELF/symver.s | ||
---|---|---|
2 | Perhaps worth a top-level comment saying what this test is intended to cover? | |
16 | ||
23 | ||
28 | (I think this is more traditional punctuation for this style of message) | |
90–91 | Do you need a similar test-case for --wrap=foo@v1 (and/or --wrap=foo@@v1)? There's an interesting interaction between the two things with your follow-up patch, which probably needs exploring in tests. | |
94–97 | Not that it really matters, but in all the other cases, the prefixes are CHECK*. Should this be CHECK6 for consistency? |
As noted elsewhere, my symbol versioning knowledge is lacking. I'm happy with what I've seen, but I could easily have missed things that don't make sense, so best to get someone else to sign off.
lld/test/ELF/symver.s | ||
---|---|---|
96 | Nit: it looks weird that -shared has only one dash, when all other options have two. Probably should standardise on one or the other? Same applies elsewhere. |
lld/test/ELF/symver.s | ||
---|---|---|
96 | The idea is that -shared is the most common form (the compiler driver option is -shared) so I use it. For -pie, I prefer the form as well. For many other long options, I use two dashes. |
Perhaps worth a top-level comment saying what this test is intended to cover?