Reworded the comment about __wrap_free detection of the allocator used to
allocate the pointer (it can free a pointer created by either the host
malloc(), kmalloc() or vmalloc()).
Signed-off-by: Paolo 'Blaisorblade' Giarrusso <[email protected]>
---
linux-2.6.10-rc-paolo/arch/um/kernel/main.c | 9 ++++++---
1 files changed, 6 insertions(+), 3 deletions(-)
diff -puN arch/um/kernel/main.c~uml-wrap_free arch/um/kernel/main.c
--- linux-2.6.10-rc/arch/um/kernel/main.c~uml-wrap_free 2004-11-16 02:19:08.289619800 +0100
+++ linux-2.6.10-rc-paolo/arch/um/kernel/main.c 2004-11-16 02:19:08.292619344 +0100
@@ -222,13 +222,16 @@ void __wrap_free(void *ptr)
* physical memory - kmalloc/kfree
* kernel virtual memory - vmalloc/vfree
* anywhere else - malloc/free
- * If kmalloc is not yet possible, then the kernel memory regions
- * may not be set up yet, and the variables not initialized. So,
- * free is called.
+ * If kmalloc is not yet possible, then either high_physmem and/or
+ * end_vm are still 0 (as at startup), in which case we call free, or
+ * we have set them, but anyway addr has not been allocated from those
+ * areas. So, in both cases __real_free is called.
*
* CAN_KMALLOC is checked because it would be bad to free a buffer
* with kmalloc/vmalloc after they have been turned off during
* shutdown.
+ * XXX: However, we sometimes shutdown CAN_KMALLOC temporarily, so
+ * there is a possibility for memory leaks.
*/
if((addr >= uml_physmem) && (addr < high_physmem)){
_