Re: vdso handling

[Pedro Alves <palves@redhat.com>]

On 03/28/2014 11:00 PM, Alan Modra wrote:

> I believe the intent of rounding to a page was to pick up the file
> and program headers at the start of a file and section headers at the
> end, so let's do just that.  On top of my last patch:

Agreed.  This works for me.  Thanks!

> -	if (i_phdrs[i].p_align > 1)
> +	/* Extend the beginning of the first pt_load to cover file
> +	   header and program headers.  */
> +	if (first_phdr == &i_phdrs[i])

Minor nit:  Perhaps the comment could say "first pt_load
if it covers offset 0"?  The computation below confused me a little
until I scrolled up and realized that first_phdr is only set if
the first segment covers offset 0, not whatever the first segment
is.  (I'd even consider renaming it to zero_phdr or
zero_offset_phdr, but with the comment I'd already be
super happy).


On the GDB patch, sorry for not noticing earlier, but:

> +static int
> +find_vdso_size (CORE_ADDR vaddr, unsigned long size,
> +		int read ATTRIBUTE_UNUSED, int write ATTRIBUTE_UNUSED,
> +		int exec ATTRIBUTE_UNUSED, int modified ATTRIBUTE_UNUSED,
> +		void *data)
> +{

Please don't use ATTRIBUTE_UNUSED under gdb/, it'd be flagged by the
ARI as a regression:

gdb/contrib/gdb_ari.sh:

BEGIN { doc["ATTRIBUTE_UNUSED"] = "\
Do not use ATTRIBUTE_UNUSED, do not bother (GDB is compiled with -Werror and, \
consequently, is not able to tolerate false warnings.  Since -Wunused-param \
produces such warnings, neither that warning flag nor ATTRIBUTE_UNUSED \
are used by GDB"
    category["ATTRIBUTE_UNUSED"] = ari_regression
}
/(^|[^_[:alnum:]])ATTRIBUTE_UNUSED([^_[:alnum:]]|$)/ {
    fail("ATTRIBUTE_UNUSED")
}

-- 
Pedro Alves