Re: [PATCH for-4.15 5/5] tools/xenstored: Silence coverity when using xs_state_* structures

[Jürgen Groß <jgross@xxxxxxxx>]

On 26.02.21 09:57, Julien Grall wrote:
> Hi Juergen,
>
> On 26/02/2021 07:10, Jürgen Groß wrote:
> > On 25.02.21 18:41, Julien Grall wrote:
> > > From: Julien Grall <jgrall@xxxxxxxxxx>
> > >
> > > Coverity will report unitialized values for every use of xs_state_*
> > > structures in the save part. This can be prevented by using the [0]
> > > rather than [] to define variable length array.
> > >
> > > Coverity-ID: 1472398
> > > Coverity-ID: 1472397
> > > Coverity-ID: 1472396
> > > Coverity-ID: 1472395
> > > Signed-off-by: Julien Grall <jgrall@xxxxxxxxxx>
> > Sorry, but Coverity is clearly wrong here.
> I saw what Andrew wrote but neither of you really provided enough 
> information to infer the same. Care to provide more details?
> > Should we really modify our code to work around bugs in external
> > static code analyzers?
> I don't think it is OK to have 866 issues (and counting) and keep 
> ignoring them because Coverity may be wrong. We should fix them one way 
> or another. If this means telling Coverity they are reporting false 
> positive, then fine.
> But for that, I first needs a bit more details why they are clearly wrong.
Lets put it this way: why is a[0] not critical, but a[] is?

Semantically there is no difference, so Coverity MUST be wrong in
some way (either a[] is really not critical, or a[0] should be
critical).

Juergen

Attachment: OpenPGP_0xB0DE9DD628BF132F.asc
Description: application/pgp-keys

Attachment: OpenPGP_signature
Description: OpenPGP digital signature