Let me be the first to say "Bad Chase..." about the html... This is a clean
version of the previous message.
I believe this is the correct fix. I am guessing that the arch options I
had selected when test compiling this patch did not actually compile this
section of the code. If that fixes all the compilation errors you have, it
seems valid to me.
Thanks
--
Chase Douglas
IBM Linux Technology Center
[EMAIL PROTECTED]
(614) 456-0455 T/L 268-7011
[EMAIL PROTECTED] wrote: -----
>Hi Guys,
>
>With fresh download from RC2 I've hit a compile error:
>
> CC libc/stdlib/malloc/malloc.o
>libc/stdlib/malloc/malloc.c:47: error: conflicting types for
'__malloc_mmb_heap'
>
>libc/stdlib/malloc/malloc.h:80: error: previous declaration of
'__malloc_mmb_heap' was here
>
>so to fix the conflict, I've changed malloc.h to match malloc.c :
>
>
>diff --git a/libc/stdlib/malloc/malloc.h b/libc/stdlib/malloc/malloc.h
>
>index f49ed34..657a67c 100644
>--- a/libc/stdlib/malloc/malloc.h
>+++ b/libc/stdlib/malloc/malloc.h
>@@ -77,7 +77,7 @@ extern struct malloc_mmb *__malloc_mmapped_blocks;
> /* A heap used for allocating malloc_mmb structures. We could allocate
>
> them from the main heap, but that tends to cause heap fragmentation in
> annoying ways. */
>-extern struct heap __malloc_mmb_heap;
>+extern struct heap_free_area *__malloc_mmb_heap;
>
> /* Define MALLOC_MMB_DEBUGGING to cause malloc to emit debugging info
about
>
> about mmap block allocation/freeing by the `uclinux broken munmap'
code
>
>But this then causes:
> CC libc/stdlib/malloc/free.o
>libc/stdlib/malloc/free.c: In function 'free_to_heap':
>libc/stdlib/malloc/free.c:178: warning: passing argument 2 of
'free_to_heap' from incompatible pointer type
>
> CC libc/stdlib/malloc/realloc.o
>
>which I can fix with :
>
>diff --git a/libc/stdlib/malloc/free.c b/libc/stdlib/malloc/free.c
>index fd29690..44b2f89 100644
>--- a/libc/stdlib/malloc/free.c
>+++ b/libc/stdlib/malloc/free.c
>
>@@ -175,7 +175,7 @@ free_to_heap (void *mem, struct heap_free_area *heap,
malloc_mutex_t *heap_lock)
> __pthread_mutex_unlock (heap_lock);
>
> /* Release the descriptor block we used. */
>
>- free_to_heap (mmb, &__malloc_mmb_heap,
&__malloc_mmb_heap_lock);
>+ free_to_heap (mmb, __malloc_mmb_heap,
&__malloc_mmb_heap_lock);
>
> /* Do the actual munmap. */
>
> munmap ((void *)mmb_start, mmb_end - mmb_start);
>
>
>But I haven't looked into why this has changed type or anything like that
- Is this the right thing to do ? and if so - well there it is :)
>
>--
>Cheers
>Kieran Bingham
_______________________________________________
uClibc mailing list
[email protected]
http://busybox.net/cgi-bin/mailman/listinfo/uclibc
