Re: false negative in checpatch for asm/ vs. linux/

Previous thread: [PATCH] macintosh: qindfarm_smu_sat: semaphore to mutex by Daniel Walker on Tuesday, April 29, 2008 - 12:43 pm. (1 message)

Next thread: [git pull] PCI updates for 2.6.26 by Jesse Barnes on Tuesday, April 29, 2008 - 1:12 pm. (6 messages)
To: <apw@...>
Cc: LKML <linux-kernel@...>
Date: Tuesday, April 29, 2008 - 12:49 pm

Hi Andy,

I think the following:

#warn if <asm/foo.h> is #included and <linux/foo.h> is available (uses RAW line)
if ($tree && $rawline =~ m{^.\#\s*include\s*\<asm\/(.*)\.h\>}) {
my $checkfile = "$root/include/linux/$1.h";
if (-f $checkfile && $1 ne 'irq') {
WARN("Use #include <linux/$1.h> instead of <asm/$1.h>\n" .
$herecurr);
}
}

in checkpatch.pl can lead to false positives for asm/ includes from within
linux/ files? e.g., I have a patch which moves the asm/hugetlb.h #include up to
the top of include/linux/hugetlb.h, but I get:

[09:46:27]nacc@arkanoid:~/linux/views/linux-2.6-work$ git show HEAD | ./scripts/checkpatch.pl -
WARNING: Use #include <linux/hugetlb.h> instead of <asm/hugetlb.h>
#17: FILE: include/linux/hugetlb.h:8:
+#include <asm/hugetlb.h>

Thanks,
Nish

--
Nishanth Aravamudan <nacc@us.ibm.com>
IBM Linux Technology Center
--

To: Nishanth Aravamudan <nacc@...>
Cc: LKML <linux-kernel@...>
Date: Thursday, May 1, 2008 - 6:24 am

Yes, clearly stupid and trivial to detect. Will be sorted in the 0.19.

Thanks for the report.

-apw
--

Previous thread: [PATCH] macintosh: qindfarm_smu_sat: semaphore to mutex by Daniel Walker on Tuesday, April 29, 2008 - 12:43 pm. (1 message)

Next thread: [git pull] PCI updates for 2.6.26 by Jesse Barnes on Tuesday, April 29, 2008 - 1:12 pm. (6 messages)