[PATCH] Add SVR4 psABI specific parser for AUXV entries

[Simon Marchi]

On 2020-04-07 10:14 p.m., Kamil Rytarowski wrote:
> NetBSD and OpenBSD always use an int to store the type as
> defined in the SVR4 psABI specifications rather than long
> as assumed by the default parser.
> 
> Define svr4_auxv_parse() that shares code with default_auxv_parse().
> 
> Remove obsd_auxv_parse() and switch OpenBSD to svr4_auxv_parse().
> Remove not fully accurate comment from obsd-tdep.c.
> 
> Use svr4_auxv_parse() on NetBSD.
> 
> gdb/ChangeLog:
> 
> 	* auxv.h (svr4_auxv_parse): New.
> 	* auxv.c (default_auxv_parse): Split into default_auxv_parse
> 	and generic_auxv_parse.
> 	(svr4_auxv_parse): Add.
> 	* obsd-tdep.c: Include "auxv.h".
> 	(obsd_auxv_parse): Remove.
> 	(obsd_init_abi): Remove comment.
> 	(obsd_init_abi): Change set_gdbarch_auxv_parse passed argument
> 	from `obsd_auxv_parse' to `svr4_auxv_parse'.
> 	* nbsd-tdep.c: Include "auxv.h".
> 	(nbsd_init_abi): Call set_gdbarch_auxv_parse.

Thanks, just a few minor comments.

> diff --git a/gdb/auxv.c b/gdb/auxv.c
> index c13d7a22eb9..240f32236b3 100644
> --- a/gdb/auxv.c
> +++ b/gdb/auxv.c
> @@ -248,34 +248,64 @@ memory_xfer_auxv (struct target_ops *ops,
>    return procfs_xfer_auxv (readbuf, writebuf, offset, len, xfered_len);
>  }
> 
> -/* Read one auxv entry from *READPTR, not reading locations >= ENDPTR.
> -   Return 0 if *READPTR is already at the end of the buffer.
> -   Return -1 if there is insufficient buffer for a whole entry.
> -   Return 1 if an entry was read into *TYPEP and *VALP.  */
> -int
> -default_auxv_parse (struct target_ops *ops, gdb_byte **readptr,
> -		   gdb_byte *endptr, CORE_ADDR *typep, CORE_ADDR *valp)
> +/* Generic auxv_parse body shared with default_auxv_parse() and
> +   svr4_auxv_parse().  */

Instead of saying that it's used by this and that function (which is doomed
to become obsolete very easily anyway), say what is the particularity of this
function, compared to other auxv_parse functions: it takes the size of the
auxv type field as a parameter.

> +
> +static int
> +generic_auxv_parse (struct gdbarch *gdbarch, gdb_byte **readptr,
> +		    gdb_byte *endptr, CORE_ADDR *typep, CORE_ADDR *valp,
> +		    int sizeof_auxv_type)
>  {
> -  const int sizeof_auxv_field = gdbarch_ptr_bit (target_gdbarch ())
> -				/ TARGET_CHAR_BIT;
> -  const enum bfd_endian byte_order = gdbarch_byte_order (target_gdbarch ());
> +  struct type *ptr_type = builtin_type (gdbarch)->builtin_data_ptr;
> +  const int sizeof_auxv_val = TYPE_LENGTH (ptr_type);
> +  enum bfd_endian byte_order = gdbarch_byte_order (gdbarch);
>    gdb_byte *ptr = *readptr;
> 
>    if (endptr == ptr)
>      return 0;
> 
> -  if (endptr - ptr < sizeof_auxv_field * 2)
> +  if (endptr - ptr < 2 * sizeof_auxv_val)
>      return -1;
> 
> -  *typep = extract_unsigned_integer (ptr, sizeof_auxv_field, byte_order);
> -  ptr += sizeof_auxv_field;
> -  *valp = extract_unsigned_integer (ptr, sizeof_auxv_field, byte_order);
> -  ptr += sizeof_auxv_field;
> +  *typep = extract_unsigned_integer (ptr, sizeof_auxv_type, byte_order);
> +  ptr += sizeof_auxv_val;      /* Alignment.  */

Instead of just "Alignment", please add a comment explaining that even if the
auxv type takes less space than an auxv value, there is padding after the type
such that the value is aligned on a multiple of its size (and this is why we
advance by `sizeof_auxv_val` and not `sizeof_auxv_type`.  I think that can
be surprising for somebody who does not know this fact.

> +  *valp = extract_unsigned_integer (ptr, sizeof_auxv_val, byte_order);
> +  ptr += sizeof_auxv_val;
> 
>    *readptr = ptr;
>    return 1;
>  }
> 
> +/* Read one auxv entry from *READPTR, not reading locations >= ENDPTR.
> +   Return 0 if *READPTR is already at the end of the buffer.
> +   Return -1 if there is insufficient buffer for a whole entry.
> +   Return 1 if an entry was read into *TYPEP and *VALP.  */

Since this is the implementation of an exported function, replace the comment
here with:

  /* See auxv.h.  */

and document in the header file.  I think the documentation you have put in the
header file for svr4_auxv_parse is good, it explains what the function does and
how it's different from default_auxv_parse.  I don't think it's particularly
useful to describe all parameters here, since it's a function that implements
an interface.  The parameters should be documented where the interface is
defined.

> +int
> +default_auxv_parse (struct target_ops *ops, gdb_byte **readptr,
> +		   gdb_byte *endptr, CORE_ADDR *typep, CORE_ADDR *valp)

This last line is not properly aligned on the parenthesis.

> +{
> +  struct type *ptr_type = builtin_type (target_gdbarch ())->builtin_data_ptr;
> +  const int sizeof_auxv_type = TYPE_LENGTH (ptr_type);
> +
> +  return generic_auxv_parse (target_gdbarch (), readptr, endptr, typep, valp,
> +			     sizeof_auxv_type);
> +}
> +
> +/* Read one auxv entry from *READPTR, not reading locations >= ENDPTR.
> +   Return 0 if *READPTR is already at the end of the buffer.
> +   Return -1 if there is insufficient buffer for a whole entry.
> +   Return 1 if an entry was read into *TYPEP and *VALP.  */

Same comment about the comment.

> +int
> +svr4_auxv_parse (struct gdbarch *gdbarch, gdb_byte **readptr,
> +		   gdb_byte *endptr, CORE_ADDR *typep, CORE_ADDR *valp)

This last line is not properly aligned on the parenthesis.

Simon