linux-next: build warnings after merge of the rdma tree

linux-next: build warnings after merge of the rdma tree

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 2555 bytes --]

Hi all,

After merging the rdma tree, today's linux-next build (x86_64
allmodconfig) produced these warnings:

In file included from include/linux/printk.h:336:0,
                 from include/linux/kernel.h:14,
                 from arch/x86/include/asm/percpu.h:45,
                 from arch/x86/include/asm/current.h:6,
                 from include/linux/mutex.h:14,
                 from drivers/infiniband/hw/cxgb4/iw_cxgb4.h:34,
                 from drivers/infiniband/hw/cxgb4/cq.c:33:
drivers/infiniband/hw/cxgb4/cq.c: In function '__c4iw_poll_cq_one':
include/linux/dynamic_debug.h:127:3: warning: 'cqe.u.gen.wrid_hi' may be used uninitialized in this function [-Wmaybe-uninitialized]
   __dynamic_pr_debug(&descriptor, pr_fmt(fmt), \
   ^~~~~~~~~~~~~~~~~~
drivers/infiniband/hw/cxgb4/cq.c:674:16: note: 'cqe.u.gen.wrid_hi' was declared here
  struct t4_cqe cqe;
                ^~~
In file included from include/linux/printk.h:336:0,
                 from include/linux/kernel.h:14,
                 from arch/x86/include/asm/percpu.h:45,
                 from arch/x86/include/asm/current.h:6,
                 from include/linux/mutex.h:14,
                 from drivers/infiniband/hw/cxgb4/iw_cxgb4.h:34,
                 from drivers/infiniband/hw/cxgb4/cq.c:33:
include/linux/dynamic_debug.h:127:3: warning: 'cqe.u.gen.wrid_low' may be used uninitialized in this function [-Wmaybe-uninitialized]
   __dynamic_pr_debug(&descriptor, pr_fmt(fmt), \
   ^~~~~~~~~~~~~~~~~~
drivers/infiniband/hw/cxgb4/cq.c:674:16: note: 'cqe.u.gen.wrid_low' was declared here
  struct t4_cqe cqe;
                ^~~
In file included from include/linux/printk.h:336:0,
                 from include/linux/kernel.h:14,
                 from arch/x86/include/asm/percpu.h:45,
                 from arch/x86/include/asm/current.h:6,
                 from include/linux/mutex.h:14,
                 from drivers/infiniband/hw/cxgb4/iw_cxgb4.h:34,
                 from drivers/infiniband/hw/cxgb4/cq.c:33:
include/linux/dynamic_debug.h:127:3: warning: 'cqe.len' may be used uninitialized in this function [-Wmaybe-uninitialized]
   __dynamic_pr_debug(&descriptor, pr_fmt(fmt), \
   ^~~~~~~~~~~~~~~~~~
drivers/infiniband/hw/cxgb4/cq.c:674:16: note: 'cqe.len' was declared here
  struct t4_cqe cqe;
                ^~~

Introduced by commit

  4ab39e2f98f2 ("RDMA/cxgb4: Make c4iw_poll_cq_one() easier to analyze")

Again, I can't easily tell if these are false positives or not.

-- 
Cheers,
Stephen Rothwell

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: linux-next: build warnings after merge of the rdma tree

[Jason Gunthorpe]

On Tue, Jul 10, 2018 at 11:33:42AM +1000, Stephen Rothwell wrote:
> Hi all,
> 
> After merging the rdma tree, today's linux-next build (x86_64
> allmodconfig) produced these warnings:
> 
> In file included from include/linux/printk.h:336:0,
>                  from include/linux/kernel.h:14,
>                  from arch/x86/include/asm/percpu.h:45,
>                  from arch/x86/include/asm/current.h:6,
>                  from include/linux/mutex.h:14,
>                  from drivers/infiniband/hw/cxgb4/iw_cxgb4.h:34,
>                  from drivers/infiniband/hw/cxgb4/cq.c:33:
> drivers/infiniband/hw/cxgb4/cq.c: In function '__c4iw_poll_cq_one':
> include/linux/dynamic_debug.h:127:3: warning: 'cqe.u.gen.wrid_hi' may be used uninitialized in this function [-Wmaybe-uninitialized]
>    __dynamic_pr_debug(&descriptor, pr_fmt(fmt), \
>    ^~~~~~~~~~~~~~~~~~
> drivers/infiniband/hw/cxgb4/cq.c:674:16: note: 'cqe.u.gen.wrid_hi' was declared here
>   struct t4_cqe cqe;
>                 ^~~
> In file included from include/linux/printk.h:336:0,
>                  from include/linux/kernel.h:14,
>                  from arch/x86/include/asm/percpu.h:45,
>                  from arch/x86/include/asm/current.h:6,
>                  from include/linux/mutex.h:14,
>                  from drivers/infiniband/hw/cxgb4/iw_cxgb4.h:34,
>                  from drivers/infiniband/hw/cxgb4/cq.c:33:
> include/linux/dynamic_debug.h:127:3: warning: 'cqe.u.gen.wrid_low' may be used uninitialized in this function [-Wmaybe-uninitialized]
>    __dynamic_pr_debug(&descriptor, pr_fmt(fmt), \
>    ^~~~~~~~~~~~~~~~~~
> drivers/infiniband/hw/cxgb4/cq.c:674:16: note: 'cqe.u.gen.wrid_low' was declared here
>   struct t4_cqe cqe;
>                 ^~~
> In file included from include/linux/printk.h:336:0,
>                  from include/linux/kernel.h:14,
>                  from arch/x86/include/asm/percpu.h:45,
>                  from arch/x86/include/asm/current.h:6,
>                  from include/linux/mutex.h:14,
>                  from drivers/infiniband/hw/cxgb4/iw_cxgb4.h:34,
>                  from drivers/infiniband/hw/cxgb4/cq.c:33:
> include/linux/dynamic_debug.h:127:3: warning: 'cqe.len' may be used uninitialized in this function [-Wmaybe-uninitialized]
>    __dynamic_pr_debug(&descriptor, pr_fmt(fmt), \
>    ^~~~~~~~~~~~~~~~~~
> drivers/infiniband/hw/cxgb4/cq.c:674:16: note: 'cqe.len' was declared here
>   struct t4_cqe cqe;
>                 ^~~
> 
> Introduced by commit
> 
>   4ab39e2f98f2 ("RDMA/cxgb4: Make c4iw_poll_cq_one() easier to analyze")
> 
> Again, I can't easily tell if these are false positives or not.

What compiler is producing these? I got nothing from 0-day build
service or my local gcc-7..

They are false positives and I guess we need to put the
uninitialized_var back that was hiding them.

Also curious that the powerpc compiler gets a different set..

Thanks,
Jason

Re: linux-next: build warnings after merge of the rdma tree

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 595 bytes --]

Hi Jason,

On Mon, 9 Jul 2018 21:41:57 -0600 Jason Gunthorpe <jgg@mellanox.com> wrote:
> 
> What compiler is producing these? I got nothing from 0-day build
> service or my local gcc-7..

The x86 compiler is a v7.3.1 cross compiler hosted on PowerPC LE and
built from sources.

> They are false positives and I guess we need to put the
> uninitialized_var back that was hiding them.

Rats.  :-(

> Also curious that the powerpc compiler gets a different set..

The powerpc builds are done with:

$ gcc --version
gcc (Debian 7.3.0-21) 7.3.0

-- 
Cheers,
Stephen Rothwell

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: linux-next: build warnings after merge of the rdma tree

[Jason Gunthorpe]

On Tue, Jul 10, 2018 at 02:05:12PM +1000, Stephen Rothwell wrote:
> Hi Jason,
> 
> On Mon, 9 Jul 2018 21:41:57 -0600 Jason Gunthorpe <jgg@mellanox.com> wrote:
> > 
> > What compiler is producing these? I got nothing from 0-day build
> > service or my local gcc-7..
> 
> The x86 compiler is a v7.3.1 cross compiler hosted on PowerPC LE and
> built from sources.

Curiouser and curiouser.. Mine is

gcc version 7.3.0 (Ubuntu 7.3.0-21ubuntu1~16.04) 

> > They are false positives and I guess we need to put the
> > uninitialized_var back that was hiding them.
> 
> Rats.  :-(

Indeed.. I dislike that macro.

Jason