On 2019-Feb-15, Kyotaro HORIGUCHI wrote:
> I understand that the behavior of __builtin_c[tl]z(0) is
> undefined from the reason, they convert to bs[rf]. So if we use
> these builtins, additional check is required.
>
> https://gcc.gnu.org/onlinedocs/gcc/Other-Builtins.html
Okay -- the functions check for a 0 argument:
+static inline int
+pg_rightmost_one32(uint32 word)
+{
+ int result = 0;
+
+ Assert(word != 0);
+
+#ifdef HAVE__BUILTIN_CTZ
+ result = __builtin_ctz(word);
+#else
+ while ((word & 255) == 0)
+ {
+ word >>= 8;
+ result += 8;
+ }
+ result += rightmost_one_pos[word & 255];
+#endif /* HAVE__BUILTIN_CTZ */
+
+ return result;
+}
so we're fine.
> And even worse lzcntx is accidentially "fallback"s to bsrx on
> unsupported CPUs, which leads to bogus results.
> __builtin_clzll(8) = 3, which should be 60.
I'm not sure I understand this point. Are you saying that if we use
-mlzcnt to compile, then the compiler builtin is broken in platforms
that don't support the lzcnt instruction? That's horrible. Let's stay
away from -mlzcnt then.
--
Álvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
