Hi Jose,
On Tue, Aug 09, 2016 at 01:58:30PM +0200, Jose E. Marchesi wrote:
> Ok for master?
OK with the following nitpicks below.
> commit c7e007a9f3bd550a2f00e41f303f23175af0dcca
> Author: Jose E. Marchesi <jose.march...@oracle.com>
> Date: Tue Aug 9 04:30:37 2016 -0700
>
> sparc: fix the printing of hw capabilities object attributes.
>
> The GNU_Sparc_HWCAPS and GNU_Sparc_HWCAPS2 object attributes comprise a
> set of hardware capabilities that may (or not) be present in the target
> machine for which the object was compiled. This patch adds the support
> for printing a nicely formatted comma-separated list with the selected
> hw capabilities.
>
> Signed-off-by: Jose E. Marchesi <jose.march...@oracle.com>
>
> @@ -41,33 +41,62 @@ 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[32] =
> + {
> + "mul32", "div32", "fsmuld", "v8plus", "popc", "vis", "vis2",
> "asi_blk_init", "fmaf",
> + "vis3", "hpc", "random", "trans", "fjfmau", "ima",
> + "asi_cache_sparing", "aes", "des", "kasumi", "camellia", "md5", "sha1",
> + "sha256", "sha512", "mpmul", "mont", "pause", "cbcond", "crc32c",
> "resv30",
> + "resv31"
> + };
> +
> +
> + static const char *hwcaps2[32] =
> + {
> + "fjathplus", "vis3b", "adp", "sparc5", "mwait", "xmpmul", "xmont",
> "nsec",
> + "resv8", "resv9" , "resv10", "resv11", "fjathhpc", "fjdes", "fjaes",
> "resv15",
> + "resv16", "resv17", "resv18", "resv19", "resv20", "resv21", "resv22",
> "resv23",
> + "resv24", "resv25", "resv26", "resv27", "resv28", "resv29", "resv30",
> "resv31",
> + };
If possible try to keep lines < 80 chars please.
> + char *s = name;
> + for (cap = 0; cap < 32; cap++)
> + if (value & (1 << cap))
Use (1U << cap) since signed left shift is undefined if you move 1 into
the signed bit.
> + {
> + if (*s != '\0')
> + s = strcat (s, ",");
> + s = strcat (s, caps[cap] ? caps[cap] : "unknown");
Given the arrays above "unknown" can never happen?
> diff --git a/tests/testfilesparc64attrs.o.bz2
> b/tests/testfilesparc64attrs.o.bz2
> new file mode 100644
> index 0000000..7be7f88
> Binary files /dev/null and b/tests/testfilesparc64attrs.o.bz2 differ
I trust this is as described. But if you could use git format-patch next
time that would be appreciated.
Thanks,
Mark
