Since the current lensfun in Ubuntu is only half-fixed, I'll upload a
version which adds sentinels to all enums (that way lensfun should again
behave as if it were compiled with -4.3).
For reference, see the attached patch.
Cheers,
Stefan.
** Attachment added: 20-enum-sentinel.dpatch
This bug was fixed in the package lensfun - 0.2.3-0ubuntu6
---
lensfun (0.2.3-0ubuntu6) karmic; urgency=low
* Add sentinels for other enums to 20-enum-sentinel, LP: #433368
(second try).
-- Stefan Potyra sistp...@ubuntu.com Wed, 23 Sep 2009 20:59:58
+0200
** Changed in:
hm... a closer look at ufraw reveals that the same out of bounds accessing is
used for different enumeration values.
I must admit, that -- while that's legal from a C perspective -- I think it's
not a too good idea to use out of bounds values for the enums in ufraw.
I guess it would be a better
The patch fixes UFRaw and I'll commit it to cvs (I'm one of the UFRaw
developers).
I still think the problem should be kept open until it is fixed in
lensfun. A library should never be allowed to break on valid C code -
even if it is obfuscated.
--
[regression] lensfun 0.2.3-0ubuntu4 freezes
I still think the problem should be kept open until it is fixed in lensfun. A
library should never be allowed to break on valid
C code - even if it is obfuscated.
I already mailed Andrew Zabolotny (the lensfun author) the link to this
bug report. But commits to lensfun cvs seem very
I'll try to import new CVS code into my PPA tonight...
--
[regression] lensfun 0.2.3-0ubuntu4 freezes UFRaw.
https://bugs.launchpad.net/bugs/433368
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
--
ubuntu-bugs mailing list
Pascal, thanks a lot, that's what I wanted to do next ;).
--
[regression] lensfun 0.2.3-0ubuntu4 freezes UFRaw.
https://bugs.launchpad.net/bugs/433368
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
--
ubuntu-bugs mailing list
I just updated my PPA, with code directly from CVS without any further
patches, and everything seems dandy...
--
[regression] lensfun 0.2.3-0ubuntu4 freezes UFRaw.
https://bugs.launchpad.net/bugs/433368
You received this bug notification because you are a member of Ubuntu
Bugs, which is
This bug was fixed in the package lensfun - 0.2.3-0ubuntu5
---
lensfun (0.2.3-0ubuntu5) karmic; urgency=low
* Add 20-enum-sentinel patch to have g++-4.4 not remove the otherwise
unreachable code handling out-of-bounds enum values via the
c-wrapper functions. LP: #433368.
The bug is still present in lensfun 0.2.3 0ubuntu5.
** Changed in: lensfun (Ubuntu)
Status: Fix Released = Confirmed
--
[regression] lensfun 0.2.3-0ubuntu4 freezes UFRaw.
https://bugs.launchpad.net/bugs/433368
You received this bug notification because you are a member of Ubuntu
Bugs,
is this http://gcc.gnu.org/PR36917 ?
--
[regression] lensfun 0.2.3-0ubuntu4 freezes UFRaw.
https://bugs.launchpad.net/bugs/433368
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
--
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
Yes, it is. So gcc is behaving correctly there. Moving back to lensfun.
** Package changed: gcc-4.4 (Ubuntu) = lensfun (Ubuntu)
** Changed in: lensfun (Ubuntu)
Assignee: (unassigned) = StefanPotyra (sistpoty)
--
[regression] lensfun 0.2.3-0ubuntu4 freezes UFRaw.
So here for some language legalese:
ufraw calls the c method, where it's (to my understanding) legal to treat an
enum as an int. It's not nice to exceed the bounds, but it's legal.
In lensfun, the C function is a wrapper around the c++ method. There the
compiler may reduce the size of the enum
** Attachment added: ufraw-lensfun-ltrace
http://launchpadlibrarian.net/32110912/ufraw-lensfun-ltrace
--
[regression] lensfun 0.2.3-0ubuntu4 freezes UFRaw.
https://bugs.launchpad.net/bugs/433368
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed
** Attachment added: ufraw-lensfun-strace
http://launchpadlibrarian.net/32110918/ufraw-lensfun-strace
--
[regression] lensfun 0.2.3-0ubuntu4 freezes UFRaw.
https://bugs.launchpad.net/bugs/433368
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed
I can verify this problem. UFRaw with LensFun stopped working for me as
well, after updating to 0ubuntu4. After reverting to 0ubuntu3 everything
worked again.
This problem can be checked without manually building UFRaw... I maintain a
PPA, with UFRaw CVS checkouts with LensFun support:
** Changed in: lensfun (Ubuntu)
Status: New = Confirmed
--
[regression] lensfun 0.2.3-0ubuntu4 freezes UFRaw.
https://bugs.launchpad.net/bugs/433368
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
--
ubuntu-bugs mailing list
Hi,
hm... interesting, thanks for the bug report.
I very much doubt that it's related to the source changes I did (see
http://launchpadlibrarian.net/32071063/lensfun_0.2.3-0ubuntu3_0.2.3-0ubuntu4.diff.gz)
to be honest.
It might however be related that -0ubuntu3 binaries were back then built
Wow, usually my wild guesses are complete and utter nonsense. This time,
it looks like I was right though.
Attaching disassembly of lfLens::GetTCAModelDesc(lfTCAModel, char
const**, lfParameter const***), both from 0.2.3-0ubuntu4 (compiled with
gcc-4.4) and from unstable's 0.2.3-2. The source of
** Attachment added: lf_gcc-4.4.S
http://launchpadlibrarian.net/32120375/lf_gcc-4.4.S
--
[regression] lensfun 0.2.3-0ubuntu4 freezes UFRaw.
https://bugs.launchpad.net/bugs/433368
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
--
** Attachment added: lf_gcc-4.3.S
http://launchpadlibrarian.net/32120392/lf_gcc-4.3.S
--
[regression] lensfun 0.2.3-0ubuntu4 freezes UFRaw.
https://bugs.launchpad.net/bugs/433368
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
--
Though I believe that this is a bug in the gcc-4.4 (I faintly recall
that POSIX says that enum's are treated as ints, at least in C... but my
memory may be wrong in this regard), I guess that the loop in ufraw
should really be fixed to pass only values as defined by the enum to
lensfun.
**
(hmpf, I'd like to make this bug affect both gcc-4.4 and lensfun, but I
fail... can anyone who knows where to click do this please? Thanks!)
--
[regression] lensfun 0.2.3-0ubuntu4 freezes UFRaw.
https://bugs.launchpad.net/bugs/433368
You received this bug notification because you are a member of
** Description changed:
After updating lensfun from 0.2.3-0ubuntu3 to 0.2.3-0ubuntu4 I have been
hit by a severe regression on Karmic (amd64).
When rebuilding the UFRaw cvs code base configured by ./configure
--with-lensfun the resulting ufraw binary freezes with no error
messages
24 matches
Mail list logo
