[PATCH 11/12] gdb: make displaced stepping implementation capable of managing multiple buffers

[Simon Marchi]

On 2020-11-24 8:41 p.m., Pedro Alves wrote:
> A few minor comments below.  Otherwise LGTM.

I forgot to reply to the other comments.

> On 11/10/20 9:46 PM, Simon Marchi via Gdb-patches wrote:
>> -      return DISPLACED_STEP_PREPARE_STATUS_ERROR;
>> +  for (displaced_step_buffer &candidate : m_buffers)
>> +    {
>> +      bool bp_in_range = breakpoint_in_range_p (aspace, candidate.addr, len);
>> +      bool is_free = candidate.current_thread == nullptr;
>> +
>> +      if (!bp_in_range)
>> +        {
>> +          if (is_free)
>
> tabs vs spaces.

Done.

>> +	    {
>> +	      buffer = &candidate;
>> +	      break;
>> +	    }
>> +	  else
>> +	    {
>> +	      /* This buffer would be suitable, but it's used right now.  */
>> +	      fail_status = DISPLACED_STEP_PREPARE_STATUS_UNAVAILABLE;
>> +	    }
>> +        }
>
> tabs vs spaces.

Done.

>> +  struct displaced_step_buffer
>> +  {
>> +    displaced_step_buffer (CORE_ADDR addr)
>
> "explicit".

Done.

>
>> +      : addr (addr)
>> +    {}
>> +
>> +    const CORE_ADDR addr;
>> +
>> +    /* The original PC of the instruction currently begin stepped.  */
>
> "begin" -> "being" ?

Indeed.

Thanks,

Simon