Hello,
is there any way to debug a program in LLVM-IR line-by-line (i.e. with gdb)?
The problem is, i have a program in human-readable-intermediate representation that segfaults, when executed. I want to know, which line in the IR causes this.
Any help is appreciated.
Best regards,
Sebastian
Are you running this in a JIT? If so this will help: http://llvm.org/docs/DebuggingJITedCode.html.
Garrison
Unfortunately, the current gdb support doesn't include line numbers.
It'll tell you what function it crashed in and the stack trace though.
Line numbers will probably come when the llvm-mc project is merged into the JIT.
I suspected something like that. 
I was hoping i could run the program and step throught IR-Code line by line, maybe even have a look at the variables.
Well, what about compiling it into assembler-code. Is there any way to know which IR-line generated which assembler line(s)? (I do not need any optimizations or whatever, if that helps...)
Sebastian
Sebastian Schlunke <schlunki@cs.tu-berlin.de> writes:
I suspected something like that.
I was hoping i could run the program and step throught IR-Code line by line, maybe even have a look at the variables.
Well, what about compiling it into assembler-code. Is there any way
to know which IR-line generated which assembler line(s)? (I do not
need any optimizations or whatever, if that helps...)
What if you write a program that
1) adds a new GlobalVariable
2) prepends each instruction in the program with a volatile store to
that GlobalVariable using a unique value (line number or instruction
count).
When the program crashes you can use gdb to read the contents of that
GlobalVariable to see the last IR instruction that was executed? 
This is cool. It would seem nice if we had IRBuilder<...> conditionally do this for every instruction.
Is this way out there?
Garrison
Have you tried using 'lli' with --force-interpreter? Then you can step
through the interpreter line by line, which is sort of liking stepping
through the IR line by very-slow-line. 
- Daniel