Re: [PATCH v2] gdb/corelow: mark bytes unavailable when reading from unavailable mapping

[Simon Marchi <simon.marchi@polymtl.ca>]

On 3/4/26 11:36 AM, Tom Tromey wrote:
>>>>>> "Simon" == simon marchi <simon.marchi@polymtl.ca> writes:
> 
> Simon> The main motivation for this change is to nicely support "lightweight"
> Simon> core files on ROCm (more on this below), but I think that the change
> Simon> also makes sense for regular core files.
> 
> FWIW this all makes sense to me.
> 
> Simon> diff --git a/gdb/corelow.c b/gdb/corelow.c
> Simon> index a28a707c293a..57d05504fa3c 100644
> Simon> --- a/gdb/corelow.c
> Simon> +++ b/gdb/corelow.c
> Simon> @@ -1470,7 +1470,8 @@ core_target::xfer_partial (enum target_object object, const char *annex,
> Simon>  		if (xfer_status == TARGET_XFER_OK)
> Simon>  		  return TARGET_XFER_OK;
>  
> Simon> -		return TARGET_XFER_E_IO;
> Simon> +		*xfered_len = len;
> Simon> +		return TARGET_XFER_UNAVAILABLE;
> Simon>  	      }
> Simon>  	  }
>  
> Just before this is a comment ending with
> 
> 	   If that fails, but the access is within an unavailable region,
> 	   then the access itself should fail.  */
> 
> but perhaps this could be reworded to not say "fail" but something about
> unavailability.
> 
> Though honestly this is kind of a nit since the code speaks for itself.

It's an easy enough fix and it's good to be precise.  Changed it to:

	   If that fails, but the access is within an unavailable region,
	   then report the bytes as unavailable.  */

> Anyway I think this is good.
> Approved-By: Tom Tromey <tom@tromey.com>

Thanks, pushed.

Simon