Hi list, For a tool I'm working on I'd like to know where instructions end up in compiled executables by looking them up in the symbol table. Therefore, I've been trying to use the llvm.pcmarker Intrinsic, but I've had no luck. When I pass .bc files with this intrinsic into llc, I'm given this error: LLVM ERROR: Cannot yet select: 0x2026938: ch PCMarker 0x1f0c678, 0x20268b0 I've attempted this with programs targeting both x86 and ppc, with the same results. Is this intrinsic unimplemented at this time? If so, what kind of effort would it take to implement this? -Nathan McCauley -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20100208/3c501b55/attachment.html>
On Mon, Feb 8, 2010 at 1:13 PM, Nathan McCauley <nathan.mccauley at gmail.com> wrote:> Hi list, > > For a tool I'm working on I'd like to know where instructions end up in > compiled executables by looking them up in the symbol table. Therefore, > I've been trying to use the llvm.pcmarker Intrinsic, but I've had no luck. > When I pass .bc files with this intrinsic into llc, I'm given this error: > > LLVM ERROR: Cannot yet select: 0x2026938: ch PCMarker 0x1f0c678, 0x20268b0 > > I've attempted this with programs targeting both x86 and ppc, with the same > results.Neither architecture support the intrinsic.> Is this intrinsic unimplemented at this time? If so, what kind of effort > would it take to implement this?Not much. Add a pseudo instruction to the backend which generates a unique label when codegened, at least for linkers that don't reorder regions delineated by global symbols (linux should be safe). Andrew
Hi, I'm also working on a tool in which I'd like to know the instruction PCs. So for now, the intrinsic is still not supported on x86? -Ang -- View this message in context: http://llvm.1065342.n5.nabble.com/State-of-llvm-pcmarker-Intrinsic-tp31677p46619.html Sent from the LLVM - Dev mailing list archive at Nabble.com.