[PATCH 3/8] GDB, gdbserver: aarch64-linux: Initial Guarded Control Stack support

[Thiago Jung Bauermann]

Hello Luis,

Thiago Jung Bauermann <thiago.bauermann@linaro.org> writes:

> Luis Machado <luis.machado@arm.com> writes:
>
>> On 6/8/25 02:03, Thiago Jung Bauermann wrote:
>>> @@ -4590,6 +4643,46 @@ aarch64_gdbarch_init (struct gdbarch_info info, struct
>> gdbarch_list *arches)
>>>      int first_w_regnum = num_pseudo_regs;
>>>      num_pseudo_regs += 31;
>>>  
>>> +  const struct tdesc_feature *feature_gcs
>>> +      = tdesc_find_feature (tdesc, "org.gnu.gdb.aarch64.gcs");
>>> +  int first_gcs_regnum = -1;
>>> +  /* Add the GCS registers.  */
>>> +  if (feature_gcs != nullptr)
>>> +    {
>>> +      first_gcs_regnum = num_regs;
>>> +      /* Validate the descriptor provides the mandatory GCS registers and
>>> +	 allocate their numbers.  */
>>> +      for (i = 0; i < ARRAY_SIZE (aarch64_gcs_register_names); i++)
>>> +	valid_p &= tdesc_numbered_register (feature_gcs, tdesc_data.get (),
>>> +					    first_gcs_regnum + i,
>>> +					    aarch64_gcs_register_names[i]);
>>> +
>>> +      num_regs += i;
>>> +    }
>>> +
>>> +  if (!valid_p)
>>> +    return nullptr;
>>> +
>>> +  const struct tdesc_feature *feature_gcs_linux
>>> +      = tdesc_find_feature (tdesc, "org.gnu.gdb.aarch64.gcs.linux");
>>> +  int first_gcs_linux_regnum = -1;
>>> +  /* Add the GCS Linux registers.  */
>>> +  if (feature_gcs_linux != nullptr)
>>> +    {
>>> +      first_gcs_linux_regnum = num_regs;
>>> +      /* Validate the descriptor provides the mandatory GCS Linux registers
>>> +	 and allocate their numbers.  */
>>> +      for (i = 0; i < ARRAY_SIZE (aarch64_gcs_linux_register_names); i++)
>>> +	valid_p &= tdesc_numbered_register (feature_gcs_linux, tdesc_data.get (),
>>> +					    first_gcs_linux_regnum + i,
>>> +					    aarch64_gcs_linux_register_names[i]);
>>> +
>>> +      /* This feature depends on the GCS feature.  */
>>> +      valid_p &= feature_gcs != nullptr;
>>> +
>>> +      num_regs += i;
>>> +    }
>>> +
>>
>> Given we have a Linux-specific part of the target description XML,
>> shouldn't it be handled in aarch64-linux-tdep.c instead? That's what
>> amd64 seems to do. Though s390 opted to do it in s390-tdep.c as
>> well. No strong preference here, as long as the Linux-specific bit
>> doesn't get in the way of bare metal handling.
>
> Do you mean handling it in aarch64_linux_init_abi? I'll try it out and
> see how it looks. I agree it makes more sense to do it in
> aarch64-linux-tdep.c

I gave up on this because I don't fully understand how
aarch64_gdbarch_init and aarch64_linux_init_abi interact (and also
tdesc_use_registers, which I think is also relevant here). I prefer to
be a bit conservative at this point.

-- 
Thiago