[PATCH] checkpatch: fix false positive on casting to double pointer

[PATCH] checkpatch: fix false positive on casting to double pointer

[Scott J. Goldman]

Signed-off-by: Scott J. Goldman <scottjg@vmware.com>
---
 scripts/checkpatch.pl         |    4 +++-
 tests/t/t9190-double-pointers |    6 ++++++
 2 files changed, 9 insertions(+), 1 deletions(-)
 create mode 100644 tests/t/t9190-double-pointers

diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl
index 87bbb8b..687bd6f 100755
--- a/scripts/checkpatch.pl
+++ b/scripts/checkpatch.pl
@@ -1982,7 +1982,9 @@ sub process {
 					 $op eq '*' or $op eq '/' or
 					 $op eq '%')
 				{
-					if ($ctx =~ /Wx[^WCE]|[^WCE]xW/) {
+					if ($op eq '*' && $cc =~ /^\*/) {
+						# double pointer is ok
+					} elsif ($ctx =~ /Wx[^WCE]|[^WCE]xW/) {
 						ERROR("need consistent spacing around '$op' $at\n" .
 							$hereptr);
 					}
diff --git a/tests/t/t9190-double-pointers b/tests/t/t9190-double-pointers
new file mode 100644
index 0000000..3f3f7f9
--- /dev/null
+++ b/tests/t/t9190-double-pointers
@@ -0,0 +1,6 @@
+pass "foo **x;"
+pass "x = **y;"
+pass "x = (foo **)y;"
+
+err='"(foo ** )" should be "(foo **)"'
+fail	"x = (foo ** )y;"	"$err"
-- 
1.6.0.3

Re: [PATCH] checkpatch: fix false positive on casting to double pointer

[Andy Whitcroft]

On Thu, Jun 17, 2010 at 9:26 PM, Scott J. Goldman <scottjg@vmware.com> wrote:
>
> Signed-off-by: Scott J. Goldman <scottjg@vmware.com>
> ---
>  scripts/checkpatch.pl         |    4 +++-
>  tests/t/t9190-double-pointers |    6 ++++++
>  2 files changed, 9 insertions(+), 1 deletions(-)
>  create mode 100644 tests/t/t9190-double-pointers
>
> diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl
> index 87bbb8b..687bd6f 100755
> --- a/scripts/checkpatch.pl
> +++ b/scripts/checkpatch.pl
> @@ -1982,7 +1982,9 @@ sub process {
>                                         $op eq '*' or $op eq '/' or
>                                         $op eq '%')
>                                {
> -                                       if ($ctx =~ /Wx[^WCE]|[^WCE]xW/) {
> +                                       if ($op eq '*' && $cc =~ /^\*/) {
> +                                               # double pointer is ok
> +                                       } elsif ($ctx =~ /Wx[^WCE]|[^WCE]xW/) {
>                                                ERROR("need consistent spacing around '$op' $at\n" .
>                                                        $hereptr);

Hrm, I would expect any number of levels of pointers to be detected
already correctly.

/me goes check.

-apw

Re: [PATCH] checkpatch: fix false positive on casting to double pointer

[Andy Whitcroft]

> Hrm, I would expect any number of levels of pointers to be detected
> already correctly.
>
> /me goes check.

We already have tests for both types which are passing generally, in
t730-* and t7020-* so I think this is more likely a failure to detect
the type as a type.  If you could send on the sample that triggers
this I will have a look.

-apw

RE: [PATCH] checkpatch: fix false positive on casting to double pointer

[Scott Goldman]

> > Hrm, I would expect any number of levels of pointers to be detected
> > already correctly.
> >
> > /me goes check.
>
> We already have tests for both types which are passing generally, in
> t730-* and t7020-* so I think this is more likely a failure to detect
> the type as a type.  If you could send on the sample that triggers
> this I will have a look.
> 
> -apw

Yikes! you're right, lousy tests on my part. The part of my patch that triggered it was:
+  return snmp_fold_field((void __percpu **)net->mib.udp_statistics,
+           UDP_MIB_RCVBUFERRORS) +
+     snmp_fold_field((void __percpu **)net->mib.udplite_statistics,
+           UDP_MIB_RCVBUFERRORS);


ERROR: need consistent spacing around '*' (ctx:WxO)
#115: FILE: drivers/net/vmxnet3/vmxnet3_drv.c:1784:
+       return snmp_fold_field((void __percpu **)net->mib.udp_statistics,
                                              ^
Maybe caused by the __percpu attr?

[PATCH] checkpatch: fix false positives involving __percpu attribute

[Scott J. Goldman]

Signed-off-by: Scott J. Goldman <scottjg@vmware.com>
---
 scripts/checkpatch.pl      |    3 ++-
 tests/t/t7005-attr-matcher |    2 ++
 2 files changed, 4 insertions(+), 1 deletions(-)

diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl
index 87bbb8b..d701956 100755
--- a/scripts/checkpatch.pl
+++ b/scripts/checkpatch.pl
@@ -150,7 +150,8 @@ our $Attribute	= qr{
 			____cacheline_aligned|
 			____cacheline_aligned_in_smp|
 			____cacheline_internodealigned_in_smp|
-			__weak
+			__weak|
+			__percpu
 		  }x;
 our $Modifier;
 our $Inline	= qr{inline|__always_inline|noinline};
diff --git a/tests/t/t7005-attr-matcher b/tests/t/t7005-attr-matcher
index a50405f..738da4c 100755
--- a/tests/t/t7005-attr-matcher
+++ b/tests/t/t7005-attr-matcher
@@ -31,6 +31,8 @@ attr	"____cacheline_internodealigned_in_smp"
 
 attr	"__weak"
 
+attr	"__percpu"
+
 attr	"__user"
 attr	"__kernel"
 attr	"__force"
-- 
1.6.0.3