Re: [Qemu-devel] [PATCH] use qemu_malloc and friends consistently

[Jamie Lokier <jamie@shareable.org>]

jcd@tribudubois.net wrote:
> 
> ----- "Jamie Lokier" <jamie@shareable.org> a écrit :
> 
> > jcd@tribudubois.net wrote:
> > > Hi Kevin,
> > > 
> > > Thanks for pointing this. I guess it just sounds strange to me that
> > > somebody would want to alloc 0 bytes. But why not ...
> > 
> > Something that nobody pointed out is that sometimes you can have
> > zero-length structures these days.  E.g. if you have a struct
> > containing a few spinlocks in the Linux kernel, it will be zero
> > length
> > when built on UP targets.  I doubt if QEMU has any such structures,
> > but it's the sort of thing that shouldn't break
> > qemu_malloc(sizeof(some_type)).
> 
> This is true but the kernel API is also making a clear distinction between the NULL returned value for allocation error and the specific ZERO_SIZE_PTR value returned for kmalloc(0, XXX). Things don't get mixed ...

Ooh, smart.

That crossed my mind for qemu_malloc() too: return (void*)16, will
SEGFAULT just like NULL but is non-NULL.

-- Jamie