Liferea dies with sqlite 3.5 - Debian

This is a discussion on Liferea dies with sqlite 3.5 - Debian ; tag 446050 help thanks The current version of liferea in Debian unstable (1.4.3-1), as well as the latest upstream version (1.4.5) fail if used together with libsqlite3-0 from experimental (3.5.1-1). The program works correctly with the versions in testing and ...

+ Reply to Thread
Results 1 to 5 of 5

Thread: Liferea dies with sqlite 3.5

  1. Liferea dies with sqlite 3.5

    tag 446050 help
    thanks

    The current version of liferea in Debian unstable (1.4.3-1), as well
    as the latest upstream version (1.4.5) fail if used together with
    libsqlite3-0 from experimental (3.5.1-1). The program works correctly
    with the versions in testing and unstable (3.4.2-1, 3.4.2-2).

    The following error mesage is printed to the console:

    *** glibc detected *** /usr/bin/liferea-bin: free(): invalid pointer: 0x08514040 ***

    This error apparently kills the db thread in liferea, so that it never
    gets to the point of showing the UI.

    If I set MALLOC_CHECK_=2 then glibc calls abort on detecting this
    error. Using gdb to set a breakpoint on abort I get a most unhelpful
    backtrace:

    Breakpoint 2, 0xb7322ff6 in abort () from /lib/libc.so.6
    (gdb) bt
    #0 0xb7322ff6 in abort () from /lib/libc.so.6
    #1 0xb7363c15 in ?? () from /lib/libc.so.6
    #2 0x00000001 in ?? ()
    #3 0xb7e86bb4 in ?? () from /usr/lib/libsqlite3.so.0
    #4 0xbfc50588 in ?? ()
    #5 0xb7e86bb4 in ?? () from /usr/lib/libsqlite3.so.0
    #6 0x00000000 in ?? ()

    The addresses inside libsqlite3.so.0 are in the data section of the
    library (they're consistently offset 0xbb4 into it, actually) from
    which I deduce the stack is getting corrupted.

    Any help on debugging this or pointers on how to track it would be appreciated.

    --
    Rodrigo Gallardo
    GPG-Fingerprint: 7C81 E60C 442E 8FBC D975 2F49 0199 8318 ADC9 BC28

    -----BEGIN PGP SIGNATURE-----
    Version: GnuPG v1.4.6 (GNU/Linux)

    iD8DBQFHDZmyAZmDGK3JvCgRAtrXAJ0e95fhVc5Y/wa6/vyiCQChp/ls6QCdFvrN
    ctCiqfa50ZNPjF1CE0+caxg=
    =H3MC
    -----END PGP SIGNATURE-----


  2. Re: Liferea dies with sqlite 3.5

    Luis Rodrigo Gallardo Cruz writes:

    > The current version of liferea in Debian unstable (1.4.3-1), as well as
    > the latest upstream version (1.4.5) fail if used together with
    > libsqlite3-0 from experimental (3.5.1-1). The program works correctly
    > with the versions in testing and unstable (3.4.2-1, 3.4.2-2).


    > The following error mesage is printed to the console:


    > *** glibc detected *** /usr/bin/liferea-bin: free(): invalid pointer: 0x08514040 ***


    > This error apparently kills the db thread in liferea, so that it never
    > gets to the point of showing the UI.


    > If I set MALLOC_CHECK_=2 then glibc calls abort on detecting this
    > error. Using gdb to set a breakpoint on abort I get a most unhelpful
    > backtrace:


    Have you tried running it under valgrind? It's usually way better at
    figuring out things like this.

    --
    Russ Allbery (rra@debian.org)


    --
    To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
    with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

  3. Re: [Liferea-devel] Liferea dies with sqlite 3.5

    Luis Rodrigo Gallardo Cruz wrote:
    > tag 446050 help
    > thanks
    >
    > The current version of liferea in Debian unstable (1.4.3-1), as well
    > as the latest upstream version (1.4.5) fail if used together with
    > libsqlite3-0 from experimental (3.5.1-1). The program works correctly
    > with the versions in testing and unstable (3.4.2-1, 3.4.2-2).


    This has been reported upstream:
    http://sourceforge.net/tracker/index...05&atid=581684

    I guess you may want to follow-up there too.

    Best,
    Emilio

    >
    > The following error mesage is printed to the console:
    >
    > *** glibc detected *** /usr/bin/liferea-bin: free(): invalid pointer: 0x08514040 ***
    >
    > This error apparently kills the db thread in liferea, so that it never
    > gets to the point of showing the UI.
    >
    > If I set MALLOC_CHECK_=2 then glibc calls abort on detecting this
    > error. Using gdb to set a breakpoint on abort I get a most unhelpful
    > backtrace:
    >
    > Breakpoint 2, 0xb7322ff6 in abort () from /lib/libc.so.6
    > (gdb) bt
    > #0 0xb7322ff6 in abort () from /lib/libc.so.6
    > #1 0xb7363c15 in ?? () from /lib/libc.so.6
    > #2 0x00000001 in ?? ()
    > #3 0xb7e86bb4 in ?? () from /usr/lib/libsqlite3.so.0
    > #4 0xbfc50588 in ?? ()
    > #5 0xb7e86bb4 in ?? () from /usr/lib/libsqlite3.so.0
    > #6 0x00000000 in ?? ()
    >
    > The addresses inside libsqlite3.so.0 are in the data section of the
    > library (they're consistently offset 0xbb4 into it, actually) from
    > which I deduce the stack is getting corrupted.
    >
    > Any help on debugging this or pointers on how to track it would be appreciated.



    -----BEGIN PGP SIGNATURE-----
    Version: GnuPG v1.4.6 (GNU/Linux)

    iD8DBQFHDe8vhTV17EoIsv4RAuuvAJ90BpG/NY7fO/dMnUTLg8qm0CWY/wCfZayG
    lPlcVl81G3qsnf1BqCKmWq4=
    =gXX9
    -----END PGP SIGNATURE-----


  4. Re: Liferea dies with sqlite 3.5

    Hi,
    * Luis Rodrigo Gallardo Cruz [2007-10-11 11:54]:
    > tag 446050 help
    > thanks
    >
    > The current version of liferea in Debian unstable (1.4.3-1), as well
    > as the latest upstream version (1.4.5) fail if used together with
    > libsqlite3-0 from experimental (3.5.1-1). The program works correctly
    > with the versions in testing and unstable (3.4.2-1, 3.4.2-2).
    >
    > The following error mesage is printed to the console:
    >
    > *** glibc detected *** /usr/bin/liferea-bin: free(): invalid pointer: 0x08514040 ***

    [...]
    The reason might be the change in sqlite3_free. In prior
    versions it did the same like free but now it does some
    additional mutex stuff for example. Maybe some free needs to
    be changed to sqlite3_free (you'll see in a backtrace).
    Kind regards
    Nico
    --
    Nico Golde - http://ngolde.de - nion@jabber.ccc.de - GPG: 0x73647CFF
    For security reasons, all text in this mail is double-rot13 encrypted.

    -----BEGIN PGP SIGNATURE-----
    Version: GnuPG v1.4.6 (GNU/Linux)

    iD8DBQFHDfo1HYflSXNkfP8RAp88AJ9dvZAKB+hsmpRdGPyq1x 0vHG9JTgCgnZ8T
    b9xoFCPlLhlSQW2/mY5OGGI=
    =2Lwx
    -----END PGP SIGNATURE-----


  5. Bug#446050: Liferea dies with sqlite 3.5

    On 10/11/07, Nico Golde wrote:
    > Hi,
    > * Luis Rodrigo Gallardo Cruz [2007-10-11 11:54]:
    > > tag 446050 help
    > > thanks
    > >
    > > The current version of liferea in Debian unstable (1.4.3-1), as well
    > > as the latest upstream version (1.4.5) fail if used together with
    > > libsqlite3-0 from experimental (3.5.1-1). The program works correctly
    > > with the versions in testing and unstable (3.4.2-1, 3.4.2-2).
    > >
    > > The following error mesage is printed to the console:
    > >
    > > *** glibc detected *** /usr/bin/liferea-bin: free(): invalid pointer: 0x08514040 ***

    > [...]
    > The reason might be the change in sqlite3_free. In prior
    > versions it did the same like free but now it does some
    > additional mutex stuff for example. Maybe some free needs to
    > be changed to sqlite3_free (you'll see in a backtrace).


    This was the critical hint. Thanks again!

    Fixed upstream with Liferea release 1.4.5b.
    Patch against 1.4.5 is attached.

    Best Regards,
    Lars


+ Reply to Thread