[Bug 46694] New: widl crashes in dlls/atl during build on ppc64el

https://bugs.winehq.org/show_bug.cgi?id=46694

Bug ID: 46694
           Summary: widl crashes in dlls/atl during build on ppc64el
           Product: Wine
           Version: 4.2
          Hardware: Other
                OS: Linux
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: -unknown
          Assignee: [e-mail]
          Reporter: [e-mail]
      Distribution: ---

Created attachment 63657
  --> https://bugs.winehq.org/attachment.cgi?id=63657
Backtrace

Building on ppc64el fails with a segmentation fault in widl while attempting to
build dlls/atl.  Backtrace attached.

[Bug 46694] widl crashes in dlls/atl during build on ppc64el

https://bugs.winehq.org/show_bug.cgi?id=46694

Timothy Pearson  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Blocks|                            |46330

[Bug 46694] widl crashes in dlls/atl during build on ppc64el

https://bugs.winehq.org/show_bug.cgi?id=46694

--- Comment #1 from Timothy Pearson  ---
There is a questionable statement in tools/widl/write_msft.c, line 2162:

if (-1 < tunion->typelib_idx)

When widl crashes, typelib_idx is == -1, which as I understand it indicates the
index is invalid?  However, the above statement does not trap the invalid
index.

Changing the statement to read as follows resolves the crash:

if (-1 <= tunion->typelib_idx)

Is this change correct in the context of widl design?

[Bug 46694] widl crashes in dlls/atl during build on ppc64el

https://bugs.winehq.org/show_bug.cgi?id=46694

Zebediah Figura  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]

--- Comment #2 from Zebediah Figura  ---
(In reply to Timothy Pearson from comment #1)
> Changing the statement to read as follows resolves the crash:
>
> if (-1 <= tunion->typelib_idx)
>
> Is this change correct in the context of widl design?

Sorry for the late reply, but no. -1 is meant as a sentinel to say that the
index hasn't been assigned yet.
> From the backtrace, I guess the union's name is probably NULL. But I don't know

why this would happen on ppc64el specifically. If this bug is still present in
current Wine (in particular I wonder if 732f67ed3 made a difference), can you
please attach a log of the output generated when running the offending command
with the additional option "-d 0x3f"?

[Bug 46694] widl crashes in dlls/atl during build on ppc64el

[WineHQ Bugzilla]

https://bugs.winehq.org/show_bug.cgi?id=46694

Fabian Maurer  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]

--- Comment #3 from Fabian Maurer  ---
Since power pc support got removed, can this be closed?

[Bug 46694] widl crashes in dlls/atl during build on ppc64el

[WineHQ Bugzilla]

https://bugs.winehq.org/show_bug.cgi?id=46694

--- Comment #4 from Timothy Pearson  ---
Why was ppc64le support removed?

[Bug 46694] widl crashes in dlls/atl during build on ppc64el

[WineHQ Bugzilla]

https://bugs.winehq.org/show_bug.cgi?id=46694

--- Comment #5 from Fabian Maurer  ---
Don't ask me, I only saw 61e210f1967a4237d07f5c791433b99f49e78113. There#s 3
commits called "Get rid of PowerPC support"