Re: [Valgrind-developers] V-Bit Verification

[Julian S. <js...@ac...>]

> > The second thing that is required is a way to write an
> > instruction that passes its operands though an IROp that is
> > specified in a test program, and no others -- so they can be
> > tested individually.  One way we could achieve this i to extend
> > the "Special" instruction mechanism (eg, see guest_amd64_toIR.c
> > and search for "Special") so that it will accept a
> > pseudo-instruction with a specified IR op in the middle of it.
> 
> Interesting idea with the advantage that you don't have to retarget it.

The _toIR.c's for the architectures that we'd want to test on would
require some minor modification.  Also, we would have to specify
these pseudo instructions to take operands from memory and write
results back to memory, since without that we'd have to write
target-specific assembly to get the operands to/from specific
registers.

So, yes, could be retargetable with a little care.

J