[PATCH] Fix crash when assiging to a property with an invalid type

[Olivier Goffart]

On Wednesday 30 July 2014 14:02:51 jahanian wrote:
> On Jul 30, 2014, at 1:43 PM, jahanian <fjahanian at apple.com> wrote:
> > On Jul 30, 2014, at 1:21 PM, Olivier Goffart <ogoffart at kde.org> wrote:
> >> It does crash for me (linux, r214316).
> >> In debug mode, there is an assert. In release mode it segfaults.
> > 
> > Strange, I tested it against a debuggable compiler (with assert on).
> > Let me investigate a bit more why it doesn’t crash for me.
> 
> OK. I missed that it was an ObjectiveC++ test. Both patches look good to me.


Thanks.
I committed them to trunk as r214734 and r214735.
I believe they should be merged in the 3.5 branch since they fix regression 
against 3.4, what do you think?



> > 
> >>> I don’t see a crash with TOT clang:
> >>> 
> >>> t.m:3:3: error: unknown type name 'A'
> >>> A* response; // expected-error {{unknown type name 'A'}}
> >>> ^
> >>> t.m:8:11: error: unknown type name 'A'
> >>> @property A* response;  // expected-error {{unknown type name 'A'}}
> >>> 
> >>>         ^
> >>> 
> >>> t.m:13:16: error: expected a type
> >>> - (void) foo :(A*) a   // expected-error {{expected a type}}
> >>> 
> >>>              ^
> >>> 
> >>> t.m:15:17: error: assignment to readonly property
> >>> self.response = a;
> >>> ~~~~~~~~~~~~~ ^ ~
> >>> 4 errors generated.
> >>> 
> >>> - Fariborz
> >>> 
> >>> On Jul 30, 2014, at 12:42 AM, Olivier Goffart <ogoffart at kde.org> wrote:
> >>>> Hi,
> >>>> 
> >>>> Attached you will find a fix for a crash I have when parsing objective
> >>>> C
> >>>> with missing include.
> >>>> This is a regression against clang 3.4 which did not crash.
> >>>> 
> >>>> My approach here is naive and just fixes the symptoms (the crash).
> >>>> 
> >>>> Regards
> > 
> > _______________________________________________
> > cfe-commits mailing list
> > cfe-commits at cs.uiuc.edu
> > http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits