On Wed, Aug 25, 2021 at 10:11:45AM -0700, Nick Desaulniers wrote:

On Wed, Aug 25, 2021 at 6:44 AM Pavel Skripkin [off-list ref] wrote:

quoted

On 8/25/21 4:34 PM, Dan Carpenter wrote:

quoted

On Wed, Aug 25, 2021 at 04:02:26PM +0300, Pavel Skripkin wrote:

quoted

quoted

This is not related to your patch.  Ignore it.

quoted

quoted

vim +2022 drivers/staging/r8188eu/hal/usb_halinit.c

quoted

   2020                case HW_VAR_BCN_VALID:
   2021                        /* BCN_VALID, BIT(16) of REG_TDECTRL = BIT(0) of REG_TDECTRL+2, write 1 to clear, Clear by sw */

quoted

2022                         u8 tmp;

Hm, I don't know anything about ARM compilers, so should I wrap this code
block with {}?

Yep.

quoted

My local gcc 11.1.1 (x86_64) does not produce any warnings/errors

You should figure out whats up with that because it shouldn't compile
with the gcc options that the kernel uses.

AFAIK, at least 2 guys except me in this CC list compiled my series without
errors/warnings. Maybe, staging tree is missing some Makefile updates?


I'll resend series this evening anyway, but this is strange....

Hm...  In my version of GCC the error is:

drivers/staging/r8188eu/hal/usb_halinit.c:1870:3: error: a label can only be part of a statement and a declaration is not a statement

That's a different error from what I was expecting.  It's caused by
having a declaration directly after a case statement.  The warning that
I was expecting was from -Wdeclaration-after-statement and it looks
like this:

warning: ISO C90 forbids mixed declarations and code [-Wdeclaration-after-statement]

You really should try investigate why this compiles for you because
something is going wrong.  It should not build without a warning.

Looks like it's bug in gcc 11.1.1. I've rebuilt this module with gcc 10
(gcc-10 (SUSE Linux) 10.3.1 20210707 [revision
048117e16c77f82598fca9af585500572d46ad73]) and build fails with error
described above


My default gcc is

gcc (SUSE Linux) 11.1.1 20210721 [revision
076930b9690ac3564638636f6b13bbb6bc608aea]


Any idea? :)

The original report said the build was with clang-14, which is near
top of tree and unreleased. It's possible that that build had a bug
that hopefully was reverted.

No, Clang is right to refuse to build the code.  It's GCC 11.1.1 which
is wrong because it allows it.  I'm not C language expert but we're
using the -Wdeclaration-after-statement because we don't want this.

regards,
dan carpenter