Hi Richard,

On Thu, 25 Oct 2012 21:09:52 +0800 Richard Yang <weiyang@linux.vnet.ibm.com> wrote:
>
> On Thu, Oct 25, 2012 at 02:28:54PM +1100, Stephen Rothwell wrote:
> >
> >After merging the akpm tree, today's linux-next build (powerpc
> >ppc64_defconfig) produced these warnings:
> >
> >drivers/infiniband/hw/cxgb3/cxio_resource.c: In function '__cxio_init_resource_fifo':
> >drivers/infiniband/hw/cxgb3/cxio_resource.c:62:3: warning: comparison of distinct pointer types lacks a cast [enabled by default]
> >drivers/infiniband/hw/cxgb3/cxio_resource.c:74:4: warning: comparison of distinct pointer types lacks a cast [enabled by default]
> >drivers/infiniband/hw/cxgb3/cxio_resource.c:81:4: warning: comparison of distinct pointer types lacks a cast [enabled by default]
> >drivers/infiniband/hw/cxgb3/cxio_resource.c:86:4: warning: comparison of distinct pointer types lacks a cast [enabled by default]
> >drivers/infiniband/hw/cxgb3/cxio_resource.c:89:7: warning: comparison of distinct pointer types lacks a cast [enabled by default]
> >drivers/infiniband/hw/cxgb3/cxio_resource.c: In function 'cxio_init_qpid_fifo':
> >drivers/infiniband/hw/cxgb3/cxio_resource.c:123:4: warning: comparison of distinct pointer types lacks a cast [enabled by default]
> >drivers/infiniband/hw/cxgb3/cxio_resource.c: In function 'cxio_hal_get_resource':
> >drivers/infiniband/hw/cxgb3/cxio_resource.c:184:6: warning: comparison of distinct pointer types lacks a cast [enabled by default]
> >drivers/infiniband/hw/cxgb3/cxio_resource.c: In function 'cxio_hal_put_resource':
> >drivers/infiniband/hw/cxgb3/cxio_resource.c:193:2: warning: comparison of distinct pointer types lacks a cast [enabled by default]
> >drivers/infiniband/hw/cxgb3/cxio_resource.c:193:2: warning: comparison of distinct pointer types lacks a cast [enabled by default]
> >drivers/infiniband/hw/cxgb3/cxio_resource.c:193:2: warning: comparison of distinct pointer types lacks a cast [enabled by default]
> >
> >Probably introduced by commit "include/linux/kfifo.h: replace open-coded
> >type check code with typecheck()".
> 
> You test this on ppc64 platform?

I use a ppc64 cross compiler.

> As I reading the code, the second parameter of kfifo_in() should be type
> "const void *". 

That is probably true.

> Hmm. I am not sure this error is there before the patch?

There was no warning before the patch.   There are also lots of similar
new warnings in a x86_64 allmodconfig build.

Given that your patch just exposes already existing problems, we need
patches that fix those problems.
-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au