Re: [PATCH][SPARC] Fix GNU_Sparc_HWCAPS and GNU_Sparc_HWCAPS2 tag values

[Jose E. Marchesi <jose dot marchesi at oracle dot com>]

    On Wed, 2016-06-08 at 16:29 +0200, Jose E. Marchesi wrote:
    > Hi people!  Just a little patch to make readelf -A to print a nice list
    > of comma-separated hwcap names, instead of rubbish :)
    > 
    > 2016-06-08  Jose E. Marchesi  <jose.marchesi@oracle.com>
    > 
    > 	* sparc_attrs.c (sparc_check_object_attribute): Fix the
    > 	calculation of GNU_Sparc_HWCAPS and GNU_Sparc_HWCAPS2 tags values.
    
    A simple testcase for inclusion into tests/run-readelf-A.sh would be
    appreciated.

I will add a test.
    
    > diff --git a/backends/sparc_attrs.c b/backends/sparc_attrs.c
    > index e95b577..505a01d 100644
    > --- a/backends/sparc_attrs.c
    > +++ b/backends/sparc_attrs.c
    > @@ -41,33 +41,60 @@ sparc_check_object_attribute (Ebl *ebl __attribute__ ((unused)),
    >  			      const char *vendor, int tag, uint64_t value,
    >  			      const char **tag_name, const char **value_name)
    >  {
    > +  static const char *hwcaps[30] =
    > +    {
    > +      "mul32", "div32", "fsmuld", "v8plus", "popc", "vis", "vis2",
    > +      "asi_blk_init", "fmaf", NULL, "vis3", "hpc", "random", "trans", "fjfmau",
    > +      "ima", "asi_cache_sparing", "aes", "des", "kasumi", "camellia",
    > +      "md5", "sha1", "sha256", "sha512", "mpmul", "mont", "pause",
    > +      "cbcond", "crc32c"
    > +    };
    > +
    > +  static const char *hwcaps2[11] =
    > +    {
    > +      "fjathplus", "vis3b", "adp", "sparc5", "mwait", "xmpmul",
    > +      "xmont", "nsec", "fjathhpc", "fjdes", "fjaes"
    > +    };
    > +
    > +  /* NAME must be large enough to contain up to N comma-separated list
    > +     of atributes in the arrays defined above.  */
    > +  static char name [30 * 16 + 30];
    > +  name[0] = '\0';
    
    This approach is not thread-safe. But if we keep the current interface
    (which expects it can return a pointer to a non-mutable static string in
    value_name) I don't see a good alternative. Given that it is only used
    in readelf, which isn't multi-threaded, I think it is fine for now.

Ok.  I will drop a note in the code so we don't forget that.
    
    Given that you have made more than 10 commits now we probably should
    give you commit access and appoint you sparc maintainer. Do you have a
    FAS account? https://admin.fedoraproject.org/accounts

Yes, I am jemarch there.