Re: [Qemu-devel] [Qemu-Devel] [PATCH] Changed malloc to g_malloc, free to g_free in linux-user/qemu.h

[Stefan Hajnoczi <stefanha@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 1779 bytes --]

On Wed, Mar 30, 2016 at 09:58:41PM +0530, haris iqbal wrote:
> On Wed, Mar 30, 2016 at 7:39 PM, Stefan Hajnoczi <stefanha@gmail.com> wrote:
> > On Thu, Mar 24, 2016 at 12:02:03AM +0530, Md Haris Iqbal wrote:
> >> Signed-off-by: Md Haris Iqbal <haris.phnx@gmail.com>
> >> ---
> >>  linux-user/qemu.h | 4 ++--
> >>  1 file changed, 2 insertions(+), 2 deletions(-)
> >>
> >> diff --git a/linux-user/qemu.h b/linux-user/qemu.h
> >> index 26b0ba2..3c3fd15 100644
> >> --- a/linux-user/qemu.h
> >> +++ b/linux-user/qemu.h
> >> @@ -381,7 +381,7 @@ static inline void *lock_user(int type, abi_ulong guest_addr, long len, int copy
> >>  #ifdef DEBUG_REMAP
> >>      {
> >>          void *addr;
> >> -        addr = malloc(len);
> >> +        addr = g_malloc(len);
> >>          if (copy)
> >>              memcpy(addr, g2h(guest_addr), len);
> >>          else
> >> @@ -407,7 +407,7 @@ static inline void unlock_user(void *host_ptr, abi_ulong guest_addr,
> >>          return;
> >>      if (len > 0)
> >>          memcpy(g2h(guest_addr), host_ptr, len);
> >> -    free(host_ptr);
> >> +    g_free(host_ptr);
> >>  #endif
> >>  }
> >
> > If I understand correctly either lock_user() or lock_user_string() may
> > be followed by unlock_user().  If you change unlock_user() to g_free()
> > then you also need to change lock_user_string() to g_malloc() to avoid a
> > malloc()/g_free() mismatch.
> 
> lock_user_string() does not use malloc itself, but calls lock_user()
> from itself.

You are right.  The reason I got confused is because there are 3
versions of lock_user(): linux-user, bsd-user, and
include/exec/softmmu-semi.h.

Please send equivalent patches for the other 2 versions as well.  This
way they stay consistent.

Thanks,
Stefan

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 473 bytes --]