Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 11 Feb 2014 23:39:33 +0100
From:      Tijl Coosemans <tijl@FreeBSD.org>
To:        Matthias Andree <mandree@FreeBSD.org>
Cc:        freebsd-ports@freebsd.org
Subject:   Re: PATCH: Re: graphics/rawtherapee: r342622 crashes on HEAD
Message-ID:  <20140211233933.483089df@kalimero.tijl.coosemans.org>
In-Reply-To: <52FAA0AD.1070703@FreeBSD.org>
References:  <20140209223228.GR80056@ithaqua.etoilebsd.net> <bfdcc4891e05f898e12591cb17be5527.squirrel@mx.waitman.net> <20140209223830.GS80056@ithaqua.etoilebsd.net> <53abb5a25f86f9c10fabcabb83e4157d.squirrel@mx.waitman.net> <20140210104037.7dcaf6b0@X220.alogt.com> <52F895F3.6070606@FreeBSD.org> <20140211010834.384f8a34@X220.alogt.com> <52F95186.9010503@FreeBSD.org> <20140211192707.GB1943@elch.exwg.net> <52FA8A6B.7020504@gwdg.de> <20140211205507.GC1943@elch.exwg.net> <52FA91B3.2080201@gwdg.de> <52FAA0AD.1070703@FreeBSD.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, 11 Feb 2014 23:14:05 +0100 Matthias Andree wrote:
> Am 11.02.2014 22:10, schrieb Rainer Hurling:
>> Am 11.02.2014 21:55, schrieb Christoph Moench-Tegeder:
>>> My base is 10.0-RELEASE, and the port is "totally recent"
>>> (rawtherapee-4.0.12_1, Makefile r343191). converters/libiconv
>>> hadn't been installed until "right now".
>>> OTOH I'm very careful with using umlauts in file names etc. (been bitten
>>> once too often), even as my environment is UTF-8, so I might
>>> have missed that issue when testing.
>> 
>> Thanks for this info. So perhaps your situation is like the one,
>> mandree@ observed on his 10.0 box: no crashes without the patch?
> 
> I am also using 10.0-RELEASE.
> 
> So the question is if someone has (had) problems with rawtherapee on
> 10-*STABLE* (OSVERSION >= 1000500) while glib20 is still at
> glib-2.36.3_1?  Please speak up if you are one of them!
> 
> Because if so that might suggest we are facing regressions
> from 10-RELEASE to 10-STABLE and
> from 10-RELEASE to 11-STABLE.
> 
> That might warrant investigation.

10-RELEASE shipped with a faulty implementation of WCHAR_T conversion,
so it doesn't return an error but it doesn't do the right thing either.
This implementation has been backed out again after the release, so
iconv_open returns -1 again when you request WCHAR_T conversion.
The crash happens when an application doesn't test for errors and
assumes the return value is a valid pointer.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20140211233933.483089df>