hi mike,

El Thu, Jan 21, 2010 at 06:01:47PM -0500 Mike Frysinger ha dit:

On Thursday 21 January 2010 16:29:24 Matthias Kaehlcke wrote:

...

smc911x_initialize(): remove unecessary call to free() and return 0 in case of failure instead of -1

Signed-off-by: Matthias Kaehlcke matthias@kaehlcke.net

drivers/net/smc911x.c | 3 +-- 1 files changed, 1 insertions(+), 2 deletions(-)

diff --git a/drivers/net/smc911x.c b/drivers/net/smc911x.c index 5d51406..f2b5895 100644 --- a/drivers/net/smc911x.c +++ b/drivers/net/smc911x.c @@ -242,8 +242,7 @@ int smc911x_initialize(u8 dev_num, int base_addr)

dev = malloc(sizeof(*dev)); if (!dev) {

• free(dev);
  

OK

...

• return -1;
  

• return 0;
  

this is an error path, so i think -1 is correct. if you're out of memory, increase your malloc region.

that's what I thought in the first place, but Ben Warren told me that in the initialize function the return value indicates the number of devices that were initialized (see http://lists.denx.de/pipermail/u-boot/2010-January/066859.html)

best regards

Hi Ben,

El Thu, Jan 21, 2010 at 11:36:41PM -0800 Ben Warren ha dit:

On Thu, Jan 21, 2010 at 10:18 PM, Matthias Kaehlcke matthias@kaehlcke.net wrote:

 El Thu, Jan 21, 2010 at 06:01:47PM -0500 Mike Frysinger ha dit:
 > On Thursday 21 January 2010 16:29:24 Matthias Kaehlcke wrote:
 > > smc911x_initialize(): remove unecessary call to free() and
 > > return 0 in case of failure instead of -1
 > >
 > > Signed-off-by: Matthias Kaehlcke <matthias@kaehlcke.net>
 > > ---
 > >  drivers/net/smc911x.c |    3 +--
 > >  1 files changed, 1 insertions(+), 2 deletions(-)
 > >
 > > diff --git a/drivers/net/smc911x.c b/drivers/net/smc911x.c
 > > index 5d51406..f2b5895 100644
 > > --- a/drivers/net/smc911x.c
 > > +++ b/drivers/net/smc911x.c
 > > @@ -242,8 +242,7 @@ int smc911x_initialize(u8 dev_num, int
 base_addr)
 > >
 > >     dev = malloc(sizeof(*dev));
 > >     if (!dev) {
 > > -           free(dev);
 >
 > OK
 >
 > > -           return -1;
 > > +           return 0;
 >
 > this is an error path, so i think -1 is correct.  if you're out of
 memory,
 > increase your malloc region.

 that's what I thought in the first place, but Ben Warren told me that
 in the initialize function the return value indicates the number of
 devices that were initialized (see
 http://lists.denx.de/pipermail/u-boot/2010-January/066859.html)

I agree that this is confusing. If the following xxx_eth_initialize() return codes seem reasonable, I'll send a patch tomorrow that applies it to the logic in net/eth.c: -2: reserved (magic value, used to return from __def_eth_init()) -1: error 0: no devices added, no error 1+: number of devices added

to me it seems reasonable and less confusing than the current logic.

i can collaborate in the task of applying it to the driver code