On Sat, 7 Nov 2015, Konstantin Belousov wrote:
On Sat, Nov 07, 2015 at 09:25:32PM +1100, Bruce Evans wrote:
...
I put intptr_t in <sys/types.h> long ago, since it was more needed and
less magic than intmax_t. This was obfuscated by moving it to
<sys/_stdint.h> and including that in various places. intmax_t is
still only in <sys/stdint.h> which is much larger. It and uintmax_t
should be together with intptr_t. That is more polluting in theory
but less in practice.
In other words, do you suggest the following change to fix the compilation ?
diff --git a/sys/sys/_stdint.h b/sys/sys/_stdint.h
index d0f9249..a0fe0ad 100644
--- a/sys/sys/_stdint.h
+++ b/sys/sys/_stdint.h
@@ -78,5 +78,13 @@ typedef __intptr_t intptr_t;
typedef __uintptr_t uintptr_t;
#define _UINTPTR_T_DECLARED
#endif
+#ifndef _INTMAX_T_DECLARED
+typedef __intmax_t intmax_t;
+#define _INTMAX_T_DECLARED
+#endif
+#ifndef _UINTMAX_T_DECLARED
+typedef __uintmax_t uintmax_t;
+#define _UINTMAX_T_DECLARED
+#endif
#endif /* !_SYS__STDINT_H_ */
...
Yes, but some source file is apparently not including <sys/systm.h>.
It is probably trying to be too smart and only including <sys/types.h>.
This bug was detected as a side effect of <sys/sysctl.h> growing a
dependency on <sys/systm.h>.
The most exotic type used in <sys/sysctl.h> is counter_u64_t. The magic
for getting this defined without much namespace pollution seems to be
that it is only used in a macro that is only used by source files that
include <sys/counter.h>. Macros often work like that. Here intmax_t
is always needed since it is in an unconditional declaration.
Bruce
_______________________________________________
svn-src-all@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"
