Re: [Qemu-devel] [PATCH] checkpatch: Disallow glib asserts in main code

[Daniel P. Berrange]

On Fri, Apr 28, 2017 at 02:45:32PM +0100, Dr. David Alan Gilbert wrote:
> * Daniel P. Berrange (address@hidden) wrote:
> > On Thu, Apr 27, 2017 at 05:55:26PM +0100, Dr. David Alan Gilbert (git) 
> > wrote:
> > > From: "Dr. David Alan Gilbert" <address@hidden>
> > > 
> > > Glib commit a6a875068779 (from 2013) made many of the glib assert
> > > macros non-fatal if a flag is set.
> > > This causes two problems:
> > >   a) Compilers moan that your code is unsafe even though you've
> > >      put an assert in before the point of use.
> > >   b) Someone evil could, in a library, call
> > >      g_test_set_nonfatal_assertions() and cause our assertions in
> > >      important places not to fail and potentially allow memory overruns.
> > > 
> > > Ban most of the glib assertion functions (basically everything except
> > > g_assert and g_assert_not_reached) except in tests/
> > > 
> > > This makes checkpatch gives an error such as:
> > > 
> > >   ERROR: Use g_assert or g_assert_not_reached
> > >   #77: FILE: vl.c:4725:
> > >   +    g_assert_cmpstr("Chocolate", >, "Cheese");
> > 
> > Or could we perhaps instead undo the damage via a hack like
> > 
> >  #define g_assert_cmpint g_assert_cmpint_orig
> >  #define g_assert_cmpint(x, y, z) \
> >      g_assert_cmpint_orig(x, y,x); \
> >      abort()
> 
> I'd be kind of OK adding a q_assert_cmpint if you wanted,
> but I think we shouldn't change the semantics of a public
> name.

Personally I think it would be worth having them - the whole point of
these more specific g_assert_* macros is that they provide clearer
error messages when they're triggered, so I prefer their use generally


Regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|