- Match SPE "DWARF" register numbers to GCC's instead of official DWARF documentation.
- Increase the register count to 148 from 112, even though the upper registers aren't saved (yet).
Details
Details
- Reviewers
- None
- Group Reviewers
Restricted Project Restricted Project
Diff Detail
Diff Detail
- Repository
- rG LLVM Github Monorepo
| Time | Test | |
|---|---|---|
| 210 ms | LLVM.Other::Unknown Unit Message (" ") Script:
--
: 'RUN: at line 19'; /mnt/disks/ssd0/agent/workspace/amd64_debian_testing_clang8/build/bin/opt -O3 -opt-bisect-limit=0 < /mnt/disks/ssd0/agent/workspace/amd64_debian_testing_clang8/llvm/test/Other/opt-bisect-legacy-pass-manager.ll | /mnt/disks/ssd0/agent/workspace/amd64_debian_testing_clang8/build/bin/llc -O3 -opt-bisect-limit=0
|
Event Timeline
Comment Actions
Im surprised that no tests need changing here. Could you please add a clang/llvm test to validate that the generated information matches what libunwind will start looking for?
Comment Actions
I'm looking for any existing tests for PowerPC that might do this for the GPRs and FPRs, but don't see any. Do you know of any tests already? Or tests for other archs I can model one on?
Comment Actions
The tests that I was requesting were for clang/LLVM. Testing libunwind is challenging as it requires runtime tests, and that means that in order to test it effectively, you need each type of machine. I suppose that in theory, you could build a VM based on LLVM's JIT, cross-compile the library, binary translate the code and test it - but that seems incredibly fragile.
Comment Actions
Turns out GCC generates 1200-range register values. This still needs work to avoid the error, but it means giving libunwind special knowledge of the SPE register set.