Re: [PATCH v4] kbuild: rewrite check-local-export in sh/awk

[Masahiro Yamada <masahiroy@kernel.org>]

On Wed, Sep 28, 2022 at 2:38 PM Masahiro Yamada <masahiroy@kernel.org> wrote:
>
> On Wed, Sep 28, 2022 at 7:26 AM Randy Dunlap <rdunlap@infradead.org> wrote:
> >
> > Hi,
> >
> > On 9/6/22 02:28, Masahiro Yamada wrote:
> > > On Sun, Sep 4, 2022 at 1:01 AM Masahiro Yamada <masahiroy@kernel.org> wrote:
> > >>
> > >> On Thu, Sep 1, 2022 at 6:03 AM Owen Rafferty <owen@owenrafferty.com> wrote:
> > >>>
> > >>> Signed-off-by: Owen Rafferty <owen@owenrafferty.com>
> > >>> ---
> > >>
> > >>
> > >> Please input something in the commit log.
> > >>
> > >> I think the benchmark in v2 is worth mentioning
> > >> because "awk is faster than bash" is one benefit
> > >> of applying this patch.
> > >>
> > >>
> > >
> > >
> > >
> > > Applied to linux-kbuild. Thanks.
> > >
> > >
> > > (V5 was not delivered to ML somehow,
> > > but I found it in my mailbox.)
> >
> > Yeah, I haven't seen that one either.
> >
> > For whatever is in linux-next-20220927, I am seeing something
> > unpleasant. I'm not positive that it's due to this patch, so I'm
> > still checking/testing (but I'm about to leave home for awhile so
> > I wanted to go ahead and let people know about this).
> >
> > I do N number of randconfig builds in a script (say 10).
> > What I am seeing is that when an 'nm' error happens, the
> > script is Terminated and not continued. E.g., if the error
> > is on randconfig build #4, builds 5-10 are never started.
> > The controlling script dies.
> >
>
>
> Hmm, I have not yet observed such an error.
>
> If it happens depending on a particular configuration,
> please share the .config file.
>
>


Owen provided the following fix-up.
I see his mail in my mailbox, but not in ML.
(Does the ML reject mails from non-subscribers?)




diff --git a/scripts/check-local-export b/scripts/check-local-export
index 0c049ff44aca..f90b5a9c67b3 100755
--- a/scripts/check-local-export
+++ b/scripts/check-local-export
@@ -8,6 +8,7 @@
 # EXPORT_SYMBOL should be used for global symbols.

 set -e
+pid=$$

 # If there is no symbol in the object, ${NM} (both GNU nm and llvm-nm) shows
 # 'no symbols' diagnostic (but exits with 0). It is harmless and hidden by
@@ -20,7 +21,7 @@ set -e
 # Then, the following line will be simpler:
 #   { ${NM} --quiet ${1} || kill 0; } |

-{ ${NM} ${1} 2>/dev/null || { echo "${0}: ${NM} failed" >&2; kill 0; } } |
+{ ${NM} ${1} 2>/dev/null || { echo "${0}: ${NM} failed" >&2; kill $pid; } } |
 ${AWK} -v "file=${1}" '
 BEGIN {




I squashed the diff to the original patch.
Thanks.




-- 
Best Regards
Masahiro Yamada