[dpdk-dev] [PATCH 02/20] ethdev: allow drivers to return error on close

[Stephen Hemminger]

On Thu, 24 Sep 2020 10:12:46 +0100
Bruce Richardson <bruce.richardson at intel.com> wrote:

> On Thu, Sep 24, 2020 at 12:35:58AM +0200, Thomas Monjalon wrote:
> > 24/09/2020 00:02, Stephen Hemminger:  
> > > On Wed, 23 Sep 2020 23:52:07 +0200
> > > Thomas Monjalon <thomas at monjalon.net> wrote:
> > >   
> > > > > > > > Yes it is a step in the right direction.
> > > > > > > > Changing the API means adding some error checks in every apps.
> > > > > > > > The reason for not doing is that I did not commit for this task,
> > > > > > > > and I am not the author of the deprecation notice.
> > > > > > > > We can discuss how to follow up once this series is merged.      
> > > > > > > 
> > > > > > > Not sure what application can really do if dev_close fails other than call rte_exit()?      
> > > > > > 
> > > > > > At least, it can show a message to the user.    
> > > > > 
> > > > > Agree, but no code checks return from close() system call now.
> > > > > At least testpmd should be updated; the examples are fine.    
> > > > 
> > > > How do you explicitly silence the static code analyzers?
> > > > Do you recommend adding (void) in front of the call to close()?  
> > > 
> > > Coverity usually gives something like, "return value of foo() is not checked
> > > but checked 10 other places"  
> > 
> > Yes so it should be silenced with (void) or ret =
> >   
> Does coverity still give the warning if the majority of cases don't
> actually check the return value?

It looks like coverity only looks at some functions and if majority is checking.
For example, it never nags about just calling close().