1441558 - MinGW debug build does not run with the Poison IO Interposer

Status: RESOLVED FIXED

(Core :: XPCOM, enhancement, P1)

Description

[Tom Ritter [:tjr] (OOTO until 4/30 at least)]

The debug build crashes on run with this; whatever we're doing here I don' think MinGW can handle.

Comment 1

[Nathan Froyd [:froydnj]]

I have zero problems with disabling the IO poisoning stuff on mingw.

Comment 2

[Tom Ritter [:tjr] (OOTO until 4/30 at least)]

Looks good: https://treeherder.mozilla.org/#/jobs?repo=try&revision=51f1075496b60505d7a7b90ed43a2c760d185294

Comment 3

[Tom Ritter [:tjr] (OOTO until 4/30 at least)]

Attachment: Bug 1441558 Do not compile the Poison IO Interposer on MinGW

Review commit: https://reviewboard.mozilla.org/r/223510/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/223510/

Comment 4

[Nathan Froyd [:froydnj]]

Comment on attachment 8954419 [details]
Bug 1441558 Do not compile the Poison IO Interposer on MinGW

https://reviewboard.mozilla.org/r/223510/#review229600

::: xpcom/build/moz.build:32
(Diff revision 1)
>  
>  if CONFIG['OS_ARCH'] == 'WINNT':
>      EXPORTS += ['nsWindowsDllInterceptor.h']
>      EXPORTS.mozilla += ['perfprobe.h']
> +    SOURCES += ['perfprobe.cpp']
> +    if CONFIG['CC_TYPE'] != 'gcc':

I was going to suggest `CONFIG['CC_TYPE'] == 'msvc'`, but I guess we want the interposer for clang-cl builds, since those are more msvc-ish.  OK, the check as written makes sense.

Comment 5

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/8c6ddf756d2a
Do not compile the Poison IO Interposer on MinGW r=froydnj

Comment 6

[Noemi Erli[:noemi_erli]]

https://hg.mozilla.org/mozilla-central/rev/8c6ddf756d2a