could not load plugin 'crystalspace.window.x.extf86vm'

[kickvb]

Hi all.

I try to run CS on my new Debian testing system (the 'etch' distribution).
When I launch walktest or any other test program (except g2dtest which works perfectly), the system generates the following message then raise a segmentation fault:

WARNING: could not load plugin 'crystalspace.window.x.extf86vm'

The error is most presumably linked to the fact that the X11 client distributed with Debian 'etch' is no more XFree86 and has been replaced by X.org.
This is due to the fact that the licence of XFree86 has been changed recently, and do no more fit with the views of the open-source community. In response to that, most of the GNU/Linux distributions such as Debian, Fedora, Slackware, FreeBSD,... have migrated to the X.org X11 client. You can find some info on that via http://www.debian-administration.org/articles/185

Can someone tell what has to be done in order to run CS on X.org?

Thanks a lot

[sunshine]

A lot of people are using CS on X.org without problem, so that should not specifically be a problem. I don't presently have the code in front of me, but I believe that this building of the 'extf86vm' plugin is controlled by a configure check. Can you look at the output of your configure run to see what it reported for the various X-related checks. If this one fails, look in config.log to see why it failed. At any rate, the lack of this plugin should not cause the program to crash. It would be useful if you could post a backtrace from the debugger showing the actual crash.

Also report what compiler version you are using. There is a known problem, for instance, where gcc 4.0.1 generates buggy code which crashes CS applications.

[kickvb]

Thanks a lot Eric for your comments.

The lack of the xextf86vm plugin is no more a problem, I simply missed to install the libxxf86vm-dev package, and I'm really sorry about that.

I was also effectively using gcc 4.0.1, and I'm still very sorry about that.
So I tried with gcc-3.3 but the problem persists and all applications except g2dtest appears to raise a segfault.

To change the compiler, I installed the version 3.3.6 by using apt-get, then used "export CC=gcc-3.3" to change the compiler to be used, then "configure" and "jam" on a new cvs checkout. The config.log file confirms that gcc-3.3 is used:

configure:1620: checking for gcc
configure:1646: result: gcc-3.3
configure:1890: checking for C compiler version
configure:1893: gcc-3.3 --version </dev/null >&5
gcc-3.3 (GCC) 3.3.6 (Debian 1:3.3.6-7)

I tried to see the backtrace on various applications, all of them were generating a segfault in the csEngine::ShineLights method, while walktest was crashing during csLoader::ParseSector.
Here are some backtraces:

walktest
--------
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread -1210898208 (LWP 8506)]
0xb6d7d3ef in ~csRef (this=0xb6faf91d) at ref.h:184
184         CSREF_TRACK_DECREF (obj, this);
(gdb) bt full
#0  0xb6d7d3ef in ~csRef (this=0xb6faf91d) at ref.h:184
No locals.
#1  0xb6d740dd in csLoader::ParseSector (this=0xd8000000,
    ldr_context=0x2b6faf9, node=0xc0000000, ssource=0x2b6faf9)
    at plugins/csparser/csloader.cpp:5027
        meshname = 0x4008177a <Address 0x4008177a out of bounds>
        mesh = {obj = 0xe8000000}
        child = {obj = 0x2000000}
        value = 0x2b6faf9 <Address 0x2b6faf9 out of bounds>
        id = 3892314132
        do_culler = 120
        culplugname = 0x2b6faf9 <Address 0x2b6faf9 out of bounds>
        sector = (class iSector *) 0x60000000
        secname = 0x2000000 <Address 0x2000000 out of bounds>
        culler_params = {obj = 0x2b6faf9}
        it = {obj = 0x40000000}
Previous frame inner to this frame (corrupt stack?)


pathtut
-------
0xb6fa0c68 in csEngine::ShineLights (this=0x8148668, region=0x0, meter=0x0)
    at plugins/engine/3d/engine.cpp:1465
1465        if (s->GetMovable ()->GetSectors ()->GetCount () <= 0 &&
(gdb) bt full
#0  0xb6fa0c68 in csEngine::ShineLights (this=0x8148668, region=0x0, meter=0x0)
    at plugins/engine/3d/engine.cpp:1465
        s = (class iMeshWrapper *) 0x85312f8
        current = {lm_version = 3, normal_light_level = 128, ambient_red = 20,
  ambient_green = 20, ambient_blue = 20, cosinus_factor = 0, lightmap_size = 0}
        reason = 0xb709d574 "no 'lm_precalc_info' found in cache"
        num_meshes = 1
        failed_meshes = {count = 0, capacity = 0, threshold = 16, root = 0x0}
        light_count = 3
        failed = 0
        max_failed_meshes = 4
        start = 3070368477
        cm = (class iCacheManager *) 0x85307e4
        data = {obj = 0x0}
        lit = {obj = 0x8530a6c}
        stop = 3220901160
        do_relight = true
        l = (class iLight *) 0x85313cc
        sn = 0
#1  0xb6f9983f in csEngine::Prepare (this=0x8148668, meter=0x0)
    at plugins/engine/3d/engine.cpp:1177
No locals.
#2  0x0804bf41 in PathTut::Initialize (this=0x8133df0, argc=1, argv=0xbffb1044)
---Type <return> to continue, or q <return> to quit---
    at apps/tutorial/pathtut/pathtut.cpp:236
        walls = {obj = 0x85312f8}
        walls_state = {obj = 0x8530b30}
        ll = (class iLightList *) 0x852f9a8
        txtmgr = (class iTextureManager *) 0x0
        m = {m11 = 0, m12 = -2.70834316e-05, m13 = -2.99482199e-05,
  m21 = 9.80908925e-45, m22 = -2.7361617e-05, m23 = 1.40129846e-45,
  m31 = -2.72453763e-05, m32 = 1.40129846e-45, m33 = -2.70834316e-05}
        spstate = {obj = 0x80f2810}
        txt = (class iTextureWrapper *) 0x813707c
        tm = (class iMaterialWrapper *) 0x852f770
        ws = {obj = 0x8530ef8}
        light = {obj = 0x85321f0}
        g2d = (class iGraphics2D *) 0x30
        imeshfact = {obj = 0xbffb0ed0}
#3  0x0804c595 in main (argc=1, argv=0xbffb1044)
    at apps/tutorial/pathtut/pathtut.cpp:385
        object_reg = (class iObjectRegistry *) 0x813707c


simple1
-------
0xb6f97c68 in csEngine::ShineLights (this=0x8146620, region=0x0, meter=0x0)
    at plugins/engine/3d/engine.cpp:1465
1465        if (s->GetMovable ()->GetSectors ()->GetCount () <= 0 &&
(gdb) bt full
#0  0xb6f97c68 in csEngine::ShineLights (this=0x8146620, region=0x0, meter=0x0)
    at plugins/engine/3d/engine.cpp:1465
        s = (class iMeshWrapper *) 0x852f490
        current = {lm_version = 3, normal_light_level = 128, ambient_red = 20,
  ambient_green = 20, ambient_blue = 20, cosinus_factor = 0, lightmap_size = 0}
        reason = 0xb7094574 "no 'lm_precalc_info' found in cache"
        num_meshes = 1
        failed_meshes = {count = 0, capacity = 0, threshold = 16, root = 0x0}
        light_count = 3
        failed = 0
        max_failed_meshes = 4
        start = 3070331613
        cm = (class iCacheManager *) 0x852e97c
        data = {obj = 0x0}
        lit = {obj = 0x852ec04}
        stop = 3220855944
        do_relight = true
        l = (class iLight *) 0x852f564
        sn = 0
#1  0xb6f9083f in csEngine::Prepare (this=0x8146620, meter=0x0)
    at plugins/engine/3d/engine.cpp:1177
No locals.
#2  0x0804bb05 in Simple::Application (this=0xbffa5d98)
    at apps/tutorial/simple1/simple1.cpp:199
        g2d = (class iGraphics2D *) 0x810cca4
#3  0x0805bc1c in csApplicationFramework::Start ()
    at libs/cstool/csapplicationframework.cpp:71
No locals.
#4  0x0805bdd3 in csApplicationFramework::Main (this=0xbffa5d98, argc=1,
    argv=0xbffa5e94) at libs/cstool/csapplicationframework.cpp:124
        iReturn = 0
#5  0x0804da3f in csApplicationRunner<Simple>::Run (argc=1, argv=0xbffa5e94)
    at csapplicationframework.h:433
        app = {<csApplicationFramework> = {<csInitializer> = {<No data fields>}, _vptr.csApplicationFramework = 0x80e070c, static m_Ptr = 0xbffa5d98,
    static m_FoundationStringName = 0x80e1adc "crystalspace.libcsappframe",
    static m_ApplicationStringName = 0x80e0c18 "CrystalSpace.Simple1",
    restartFlag = false,
    static object_reg = 0x813507c}, <csBaseEventHandler> = {<scfImplementation1<csBaseEventHandler,iEventHandler>> = {<scfImplementation<csBaseEventHandler>> = {<iBase> = {_vptr.iBase = 0x80e0750}, scfObject = 0xbffa5da0, scfRefCount = 2,
        scfParent = 0x0,
        scfWeakRefOwners = 0x0}, <iEventHandler> = {<No data fields>}, <No data fields>}, queue = {obj = 0x812d064}, static pmfnTriggers = {{
        __pfn = 0x51 <scfImplementation<csBaseEventHandler>::RemoveRefOwner(iBase**)+65>, __delta = 0}, {
        __pfn = 0x39 <scfImplementation<csBaseEventHandler>::RemoveRefOwner(iBase**)+41>, __delta = 0}, {
        __pfn = 0x3d <scfImplementation<csBaseEventHandler>::RemoveRefOwner(iBas---Type <return> to continue, or q <return> to quit---
e**)+45>, __delta = 0}, {
        __pfn = 0x41 <scfImplementation<csBaseEventHandler>::RemoveRefOwner(iBase**)+49>, __delta = 0}, {
        __pfn = 0x45 <scfImplementation<csBaseEventHandler>::RemoveRefOwner(iBase**)+53>, __delta = 0}, {
        __pfn = 0x49 <scfImplementation<csBaseEventHandler>::RemoveRefOwner(iBase**)+57>, __delta = 0}, {
        __pfn = 0x4d <scfImplementation<csBaseEventHandler>::RemoveRefOwner(iBase**)+61>, __delta = 0}, {
        __pfn = 0x2d <scfImplementation<csBaseEventHandler>::RemoveRefOwner(iBase**)+29>, __delta = 0}, {
        __pfn = 0x31 <scfImplementation<csBaseEventHandler>::RemoveRefOwner(iBase**)+33>, __delta = 0}, {
        __pfn = 0x35 <scfImplementation<csBaseEventHandler>::RemoveRefOwner(iBase**)+37>, __delta = 0}, {
        __pfn = 0x29 <scfImplementation<csBaseEventHandler>::RemoveRefOwner(iBase**)+25>, __delta = 0}, {
        __pfn = 0x25 <scfImplementation<csBaseEventHandler>::RemoveRefOwner(iBase**)+21>, __delta = 0}, {__pfn = 0, __delta = 0}, {__pfn = 0, __delta = 0}, {
        __pfn = 0, __delta = 0}, {__pfn = 0, __delta = 0}, {__pfn = 0,
        __delta = 0}, {__pfn = 0, __delta = 0}}}, engine = {obj = 0x8146658},
  loader = {obj = 0x8148768}, g3d = {obj = 0x810c21c}, kbd = {
    obj = 0x8107a24}, vc = {obj = 0x812cc4c}, view = {obj = 0x852bcf4},
  room = 0x852da68, rotX = -1.95599079, rotY = 4.27349594e-34}
        result = 0
        again = 8
#6  0x0804b666 in main (argc=1, argv=0xbffa5e94)
    at apps/tutorial/simple1/simple1.cpp:263
No locals.


Any ideas?

[sunshine]

Hmm, as I recall, ShineLights() is the function in which the gcc 4.0.1 botch causes a crash, so this looks like the same problem. Some features from the gcc 4.x line are backported to gcc 3.x, including bugs. This may be the case in this instance. It is reported that gcc 4.0.2 and 4.1.x fix the problem. Is it possible for you to try one of those versions?

[kickvb]

Alright, perfect!

I re-compiled everything with gcc 4.02 and everything works great now.
For the history, I also tried with gcc 2.95 but it leads to the same segfaults.

Thanks a lot Eric!