[OE-core] [PATCH] mdadm: fix gcc8 maybe-uninitialized/format-overflow warning

Adrian Bunk bunk at stusta.de
Wed Mar 20 05:21:15 UTC 2019


On Tue, Mar 19, 2019 at 11:51:47PM -0400, Khem Raj wrote:
> On Fri, Mar 15, 2019 at 6:45 AM Adrian Bunk <bunk at stusta.de> wrote:
> >
> > On Tue, Mar 12, 2019 at 05:41:58PM +0800, changqing.li at windriver.com wrote:
> > > From: Changqing Li <changqing.li at windriver.com>
> > >
> > > while compiled with -Werror=maybe-uninitialized/-Werror=format-overflow=,
> > > it failed
> > >
> > > [snip]
> > > | Incremental.c: In function 'Incremental_container':
> > > | Incremental.c:1593:3: error: 'mdfd' may be used uninitialized in this function [-Werror=maybe-uninitialized]
> > > | close(mdfd);
> > > | ^~~~~~~~~~~
> > >
> > > [snip]
> > > super-intel.c: In function 'apply_takeover_update':
> > > | super-intel.c:9615:15: error: '%d' directive writing between 1 and 11 bytes into a region of size 7 [-Werror=format-overflow=]
> > > | " MISSING_%d", du->index);
> > > | ^~
> > >...
> >
> > I am seeing these warnings only with -Og, are you also seeing them with
> > -Og (DEBUG_OPTIMIZATION) only?
> >
> > If this is true, I would consider
> > https://sources.debian.org/src/mdadm/4.1-2/debian/patches/debian-no-Werror.diff/
> > a better workaround.
> >
> This seems a broader brush, I really dont like to relegate Werror if
> we dont have to, because it will force us
> to fix the code.

How are we getting such fixes properly reviewed?

What actually happens is that the easiest change that silences a warning 
gets applied.

And for such bogus -Og only warnings there was no problem in the code,
meaning regression risks by OE/Yocto-only patches without fixing anything.

> However, I am seeing it fail with clang now
> 
> https://errors.yoctoproject.org/Errors/Details/233618/

This is caused by the "fix" for the -Og gcc warning.

Which looks bogus, and might be harmful.

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed



More information about the Openembedded-core mailing list