[patch] gdbarch_syscall_pc_increment

[Yao Qi]

On 12/12/2012 11:42 PM, Aleksandar Ristovski wrote:
>
> This is generic for a given OS that happens to increment instruction
> pointer to allow user code to e.g. set errno.
>
> I provided only arm implementation, but other target cpus would need the
> same if they implement software single stepping.
>
> Increment is cpu specific for a given architecture.
>

'software single step' is implemented differently in the backend of each 
port and your 'syscall_pc_increment' depends on the arch as well,  so a 
gdbarch hook is not needed here.

'gdbarch' stands for a certain general architecture, such as arm, mips, 
and etc.  'gdbarch_tdep' contains the details of the cpus under this 
architecture.

>
>> >
>> >If I understand your problem correctly, you have to define your own
>> >function 'arm_neutrino_syscall_next_pc' in your file
>> >arm-neutrino-tdep.c, and install it on function pointer
>> >'syscall_next_pc' (in 'struct gdbarch_tdep' in arm-tdep.h) in
>> >'arm_neutrino_init_abi'.  Please have a look on how 'syscall_next_pc' is
>> >set in arm-linux-tdep.c.  Then you can compute the pc for your own os in
>> >'arm_neutrino_syscall_next_pc'.  Hope it helps.
>
> No, the destination is not a single address as we do not know the
> outcome of the syscall. It may come back with the instruction pointer of
> the next instruction after 'svc' but also 4 bytes later (4 bytes in our
> case, some other kernel may implement it differently).

You may need a specified field in 'struct gdbarch_tdep', for example,

   /* Do a post fix of a software single step.  */
   void (*software_single_step_fixup) (struct frame_info *frame);

In arm_software_single_step,

@@ -5242,6 +5242,31 @@ arm_software_single_step (struct frame_i
    next_pc = arm_get_next_pc (frame, get_frame_pc (frame));
    arm_insert_single_step_breakpoint (gdbarch, aspace, next_pc);

+  tdep = gdbarch_tdep (gdbarch);
+  if (tdep->software_single_step_fixup)
+    tdep->software_single_step_fixup (frame);
+

and you need to initialize field 'software_single_step_fixup' somewhere, 
and do what you need in it.

b.t.w, I don't see how 'set_gdbarch_syscall_pc_increment' is called in 
your patch.

-- 
Yao (齐尧)