[Bug 793103] v4l2: compilation failure with mmap/mmap64 signature miss-match

[GStreamer (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=793103

--- Comment #25 from Michael Tretter <m.tretter at pengutronix.de> ---
(In reply to Nicolas Dufresne (stormer) from comment #23)
> (In reply to Michael Tretter from comment #21)
> > The patch fixes the problem, but there are further issues:
> > 
> > My glibc actually has mmap64 with off_t as 8 bytes, but was not used,
> > because configure is missing a check check for largefile support. I attach a
> > patch that adds the check. The size of off_t is now reported as 8 bytes.
> > 
> > I still don't think that it is a good idea to copy from libv4l2 and use the
> > GLib types for the prototypes. The function pointer prototypes should
> > exactly match the prototypes of the function that is actually called.
> 
> Have you read my patch ? I've change the offset to off_t. The other types
> does not matter, glib or not, they will be the same.
> 
> > Therefore, ALL the pointers in V4l2Object that might point to syscalls
> > should use the prototypes of the syscall and the libv4l2 calls should be
> > wrapped accordingly.
> 
> Again, read my last patch properly.

I read your patch and I absolutely agree that it is the right solution. I also
understand that the GLib types are the same and do not matter. No objection to
that. However, I think it would be cleaner (only cleaner, not more correct or
whatever), to use exactly the same types as the signature of the syscall, if it
is a function pointer to a syscall.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.