These tests fail if you build without the x86 llvm backend.
Either because they use an x86 triple or try to backtrace which
requires some x86 knowledge to see all frames.
Details
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
(not adding a reviewer for each file type since most are pretty straightforward I think)
lldb/test/Shell/SymbolFile/Breakpad/unwind-via-raSearch.test | ||
---|---|---|
45 | Same thing happens here if you don't have x86 enabled, this last frame is missing. | |
lldb/test/Shell/SymbolFile/Breakpad/unwind-via-stack-win-no-memory-info.yaml | ||
13 | @labath In this case, without the x86 backend we do get the first 3 frames just not the kernel32.dll frame. (lldb) target symbols add ../llvm-project/lldb/test/Shell/SymbolFile/Breakpad/Inputs/unwind-via-raSearch.syms symbol file '/home/david.spickett/llvm-project/lldb/test/Shell/SymbolFile/Breakpad/Inputs/unwind-via-raSearch.syms' has been added to '/home/david.spickett/build-llvm-aarch64/foo/foo.exe' (lldb) thread backtrace * thread #1 * frame #0: 0x000b1092 foo.exe`many_pointer_args + 2 frame #1: 0x000b1079 foo.exe`call_many + 105 frame #2: 0x000b1085 foo.exe`main + So my guess is that the last frame requires x86 specific knowledge to work out. If you think that's not the case or would prefer I investigate first I can remove it from this patch. |
lldb/test/Shell/SymbolFile/Breakpad/unwind-via-stack-win-no-memory-info.yaml | ||
---|---|---|
13 | Nah, that's fine. I'm pretty sure that is what's happening, and overall, I would actually say lldb should flat out refuse to open binaries/core files/etc. for which it does not have an appropriate back end.... |
Same thing happens here if you don't have x86 enabled, this last frame is missing.