This is the mail archive of the gdb-patches@sourceware.org mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [RFA 3/5] New port: CR16: gdb port


On 01/23/2013 02:22 PM, Kaushik Phatak wrote:

>>> +  "r0r1_orig",
>> This too looks like a ptrace detail escaping all the
>> way to the user, similar to the gdbserver issues.
>> Any reason not to split those up?  I think it'd be nicer.	
> This is a kernel scratch register and it part of PT_REGS.
> Would be OK to leave it as is as it may help in debugging?

Debugging gdb, or a user debugging an application?
If users expect to see that as a pair, fine.  If users/scripts
driving gdb would always want to read the values separately,
then I'd think it better to have separate (r0_orig,
r1_orig), or even hide one of them -- does the syscall abi
use both r0/r1 for syscall return, or is one of them in
orig just because that's how ptrace pushes registers
into the buffer?  But it's up to you, really.  The
orig_ registers aren't normally user visible.
I'm just pointing at things that looked fishy on first
sight, and it looked inconsistent to not have a pair
for r0/r1 but have it for the orig_ version.

-- 
Pedro Alves


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]