1024097 - jemalloc assumes that having inttypes implies something about the existence of _Bool

Status: RESOLVED FIXED

(Core :: Memory Allocator, defect)

Description

[Jeff Muizelaar [:jrmuizel]]

clang-cl will build without having inttypes but does have _Bool

Comment 1

[Mike Hommey [:glandium]]

I guess it has _Bool but doesn't define true, false and __bool_true_false_are_defined?

Comment 2

[(no longer active)]

The issue is that it supports _Bool as a built-in type, so the following won't compile:

typedef BOOL _Bool;

Comment 3

[Mike Hommey [:glandium]]

Note that was not what i was asking.

Comment 4

[Mike Hommey [:glandium]]

Let's put it another way: does the following work?

diff --git a/memory/jemalloc/src/include/msvc_compat/C99/stdbool.h b/memory/jemalloc/src/include/msvc_compat/C99/stdbool.h
--- a/memory/jemalloc/src/include/msvc_compat/C99/stdbool.h
+++ b/memory/jemalloc/src/include/msvc_compat/C99/stdbool.h
@@ -1,16 +1,18 @@
 #ifndef stdbool_h
 #define stdbool_h
 
 #include <wtypes.h>
 
 /* MSVC doesn't define _Bool or bool in C, but does have BOOL */
 /* Note this doesn't pass autoconf's test because (bool) 0.5 != true */
+#ifndef __clang__
 typedef BOOL _Bool;
+#endif
 
 #define bool _Bool
 #define true 1
 #define false 0
 
 #define __bool_true_false_are_defined 1
 
 #endif /* stdbool_h */

Comment 5

[Jeff Muizelaar [:jrmuizel]]

(In reply to Mike Hommey [:glandium] from comment #4)

It does.

Comment 6

[Mike Hommey [:glandium]]

Attachment: Allow jemalloc to build with clang-cl

Comment 7

[Mike Hommey [:glandium]]

https://github.com/jemalloc/jemalloc/commit/c521df5dcf7410898cabdcb556f919535cf16d19

Comment 8

[Mike Hommey [:glandium]]

Note, this issue makes me think there might be other problems lingering with clang-cl because of the discrepancy between the MSVC version it advertizes, and the MSVC headers version being different, in more than Mozilla code.

Comment 9

[Mike Hommey [:glandium]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/2cdbb55e9ad9

Comment 10

[(no longer active)]

(In reply to comment #8)
> Note, this issue makes me think there might be other problems lingering with
> clang-cl because of the discrepancy between the MSVC version it advertizes, and
> the MSVC headers version being different, in more than Mozilla code.

Yeah, agreed.

Comment 11

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/2cdbb55e9ad9