llvm dev - Feb 2016 - Getting MachineInstr opcode mnemonics as strings

Hello all,

Is there an easy way to get the human-readable opcode mnemonic (e.g., 
"MOV32ri64", "CMP32ri8", "JLE_1") for a
MachineInstr? I am writing a
backend analysis pass for security research, where the idea is for a 
researcher to examine the output of my pass and identify instructions 
from it for use in an attack. Right now I'm representing unique 
instructions with nondescript numeric symbols, like "i28" for the 28th
instruction; but it would be nice to give the reader something more 
helpful like "i28_MOV32ri64".

MachineInstr doesn't have a getName() function like many other LLVM 
IR/MachineIR objects; is there anything roughly equivalent to this? I'm 
already using MachineInstr::print() for debugging output, but it's far 
too verbose for this purpose. I thought about trying to parse the opcode 
mnemonic out of the print() output, until I realized just how nasty that 
would be (not to mention that it prints to a raw_ostream, not an 
in-memory string).


Sincerely,
Ethan Johnson

-- 
Ethan J. Johnson
Computer Science PhD Student, Systems group, University of Rochester
ejohns48 at cs.rochester.edu
ethanjohnson at acm.org
PGP public key available from public directory or on request

On Tue, Feb 16, 2016 at 9:30 PM, Ethan J. Johnson via llvm-dev <
llvm-dev at lists.llvm.org> wrote:
> Hello all,
>
> Is there an easy way to get the human-readable opcode mnemonic (e.g.,
> "MOV32ri64", "CMP32ri8", "JLE_1") for a
MachineInstr? I am writing a
> backend analysis pass for security research, where the idea is for a
> researcher to examine the output of my pass and identify instructions from
> it for use in an attack. Right now I'm representing unique instructions
> with nondescript numeric symbols, like "i28" for the 28th
instruction; but
> it would be nice to give the reader something more helpful like
> "i28_MOV32ri64".
>
> MachineInstr doesn't have a getName() function like many other LLVM
> IR/MachineIR objects; is there anything roughly equivalent to this? I'm
> already using MachineInstr::print() for debugging output, but it's far
too
> verbose for this purpose. I thought about trying to parse the opcode
> mnemonic out of the print() output, until I realized just how nasty that
> would be (not to mention that it prints to a raw_ostream, not an in-memory
> string).
>
The name (as well as some other information) is in MCInstrInfo; if you have
any kind of *InstrInfo (as, e.g., X86InstrInfo isa TargetInstrInfo isa
MCInstrInfo), you should be able to do:

  TII->getName(MI->getOpcode());

-Ahmed

> Sincerely,
> Ethan Johnson
>
> --
> Ethan J. Johnson
> Computer Science PhD Student, Systems group, University of Rochester
> ejohns48 at cs.rochester.edu
> ethanjohnson at acm.org
> PGP public key available from public directory or on request
>
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20160216/02734018/attachment.html>

On 02/17/2016 12:49 AM, Ahmed Bougacha wrote:
>
> On Tue, Feb 16, 2016 at 9:30 PM, Ethan J. Johnson via llvm-dev 
> <llvm-dev at lists.llvm.org <mailto:llvm-dev at
lists.llvm.org>> wrote:
>
>     Hello all,
>
>     Is there an easy way to get the human-readable opcode mnemonic
>     (e.g., "MOV32ri64", "CMP32ri8", "JLE_1")
for a MachineInstr? I am
>     writing a backend analysis pass for security research, where the
>     idea is for a researcher to examine the output of my pass and
>     identify instructions from it for use in an attack. Right now I'm
>     representing unique instructions with nondescript numeric symbols,
>     like "i28" for the 28th instruction; but it would be nice to
give
>     the reader something more helpful like "i28_MOV32ri64".
>
>     MachineInstr doesn't have a getName() function like many other
>     LLVM IR/MachineIR objects; is there anything roughly equivalent to
>     this? I'm already using MachineInstr::print() for debugging
>     output, but it's far too verbose for this purpose. I thought about
>     trying to parse the opcode mnemonic out of the print() output,
>     until I realized just how nasty that would be (not to mention that
>     it prints to a raw_ostream, not an in-memory string).
>
>
> The name (as well as some other information) is in MCInstrInfo; if you 
> have any kind of *InstrInfo (as, e.g., X86InstrInfo isa 
> TargetInstrInfo isa MCInstrInfo), you should be able to do:
>
>   TII->getName(MI->getOpcode());
> -Ahmed
That's exactly what I was looking for - thanks!

Ethan Johnson

-- 
Ethan J. Johnson
Computer Science PhD student, Systems group, University of Rochester
ejohns48 at cs.rochester.edu
ethanjohnson at acm.org
PGP public key available from public directory or on request

-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20160217/8754e827/attachment.html>