Re: [PATCH 0/2] check_headers.sh with hunk exceptions (lib/list_sort.c tools/ copy)

[Namhyung Kim]

Hi Arnaldo,

On Wed, Oct 02, 2024 at 02:29:52PM -0300, Arnaldo Carvalho de Melo wrote:
> On Thu, Oct 03, 2024 at 01:21:02AM +0800, Kuan-Wei Chiu wrote:
> > On Mon, Sep 30, 2024 at 05:21:34PM -0300, Arnaldo Carvalho de Melo wrote:
> > > From: Arnaldo Carvalho de Melo <acme@xxxxxxxxxx>
> > > 	Please take a look, as per tools/include/uapi/README we carry
> > > copies of kernel files for various reasons and check when it drifts, in
> > > this case we need another way to accept diffs while checking, its all
> > > spelled out in the individual patches, please ack.
> 
> > > Arnaldo Carvalho de Melo (2):
> > >   tools check_headers.sh: Add check variant that excludes some hunks
> > >   perf tools: Cope with differences for lib/list_sort.c copy from the kernel
> 
> > LGTM. For the series:
>  
> > Acked-by: Kuan-Wei Chiu <visitorckw@xxxxxxxxx>
> 
> Adding it to that cset before pushing to the written in stone
> perf-tools/perf-tools branch.

Oh, I thought you wanted to have it in the perf-tools-next branch.
But it's ok to go through perf-tools.  I'll drop this patchset from
tmp.perf-tools-next.

Thanks,
Namhyung

>  
> > While reviewing the patches, I noticed that there was already a
> > difference between lib/list_sort.c and tools/lib/list_sort.c regarding
> > an additional #include <linux/bug.h>. This prompted me to investigate
> > the reason for this discrepancy. From what I can see, both files only
> > seem to require only three headers:
> 
> > #include <linux/compiler.h> /* for likely() macro */
> > #include <linux/export.h> /* for EXPORT_SYMBOL() macro */
> > #include <linux/list_sort.h> /* for list_sort() and linux/types.h */
>  
> > I'll check the git history and run build tests to confirm. If only
> > these headers are needed, I'll submit a cleanup patch.
> 
> tools/ is a sidecar or sorts for the kernel, that tries to add value to
> kernel developers while not getting in their way.
> 
> Sometimes things we discover while using more widely things that were
> designed for use in the kernel source may be of help to kernel
> developers, if this is one such case, great!
> 
> But IIRC that linux/bug.h discrepancy, without further checking, was
> something already somehow accepted via:
> 
> +check lib/list_sort.c                '-I "^#include <linux/bug.h>"'
> 
> in the cset that introduced the copy:
> 
>   92ec3cc94c2cb60d ("tools lib: Adopt list_sort() from the kernel sources")
> 
> - Arnaldo