>> On this page I see:
>>
>> For ptracex or ptrace64: If the debuggee is a 32-bit process, the
>> options that refer to GPRs or SPRs fail with errno = EIO
>>
>> And I see that in the patch, you call ptrace64 to read registers
>> (PTT_READ_GPRS) for 32 bit processes. How does that work?
>
> I think Simon is correct here. For GPRS access, I understand only
> the following options are supported:
>
> ptrace: 32- or 64-bit GDB, 32-bit inferior
> ptracex: 32-bit GDB, 64-bit inferior
> ptrace64: 64-bit GDB, 64-bit inferior
>
> As I understand, the current set of indirections is intended to
> cover that space - for a 32-bit inferior, you call ptrace, and
> for a 64-bit inferior you call either ptracex or ptrace64,
> depending on whether GDB is built as 64-bit.
>Ideally, we wouldn't just guess. Do you run the testsuite on AIX? If
>so, you should be able to run it in a mode that builds the test programs
>as 32-bit programs. For example:

> $ make check RUNTESTFLAGS='--target_board "unix/-m32"

Yes, I missed this in the documentation. Ulrich and you are right. Appreciate the same for correcting me. I have addressed this and ran the test cases in 32 bit and 64-bit mode on AIX to make sure the changes are functionally correct.

>> Why do you think we can simplify this? I guess we could make
>> possibly make the assumption that GDB itself is always built
>> as 64-bit these days, but then we still have to select between
>> ptrace and ptrace64 at the least …

>I think that requiring GDB to be built as 64-bit program would be
>reasonable (if so, you should add a static check in the source for that).

Yes, from GDB - 13 we are building GDB on AIX as 64-bit application only. We can adjust the code now with this information. I have done some clean up due to this.