murphyd segfaults

[pohly]

murphyd segfaults here while using the D-Bus API. "Here" is some kind of SUSE flavor (not sure exactly, ask if it matters; it has g++ (SUSE Linux) 4.7.2 20130108 [gcc-4_7-branch revision 195012]), murphyd compiled from source (latest as of today, fb72a9f1) with "-g -O2".

Looks like a jump to invalid memory in gdb in valgrind:

Program received signal SIGSEGV, Segmentation fault. 0x000000000066c450 in ?? () (gdb) where

0 0x000000000066c450 in ?? ()

1 0x0000000000000000 in ?? ()

==25826== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 2 from 2) vex amd64->IR: unhandled instruction bytes: 0x6D 0x65 0x6D 0x6F 0x74 0x6F 0x6F 0x0 vex amd64->IR: REX=0 REX.W=0 REX.R=0 REX.X=0 REX.B=0 vex amd64->IR: VEX=0 VEX.L=0 VEX.nVVVV=0x0 ESC=NONE vex amd64->IR: PFX.66=0 PFX.F2=0 PFX.F3=0 ==25833== valgrind: Unrecognised instruction at address 0x793dbc0. ==25833== at 0x793DBC0: ??? ==25833== Your program just tried to execute an instruction that Valgrind ==25833== did not recognise. There are two possible reasons for this. ==25833== 1. Your program has a bug and erroneously jumped to a non-code ==25833== location. If you are running Memcheck and you just saw a ==25833== warning about a bad jump, it's probably your program's fault. ==25833== 2. The instruction is legitimate but Valgrind doesn't handle it, ==25833== i.e. it's Valgrind's fault. If you think this is the case or ==25833== you are not sure, please let us know and we'll try to fix it.

It works when I compile with "-g", i.e. without optimization.

To reproduce: $ murphyd -f -c murphy-lua.conf & $ kill-murphy.py

Setup:

$ cat testing/murphy-lua.conf
#set resolver-ruleset 'src/resolver/test-input'
load-plugin lua config="/home/nightly/testing/murphy-ref.lua"

nightly@syncev:~$ cat testing/murphy-ref.lua
m = murphy.get()

-- try loading console plugin
m:try_load_plugin('console')

-- m:try_load_plugin('console', 'webconsole', {
--                               address = 'wsck:127.0.0.1:3000/murphy',
--                               httpdir = 'src/plugins/console',
-- --                              sslcert = 'src/plugins/console/console.crt',
-- --                              sslpkey = 'src/plugins/console/console.key'
--          })

-- m:try_load_plugin('systemd')

-- load a test plugin
if m:plugin_exists('test.disabled') then
    m:load_plugin('test', {
                       string2  = 'this is now string2',
                       boolean2 = true,
                       int32 = -981,
                       double = 2.73,
                       object = {
                           foo = 1,
                           bar = 'bar',
                           foobar = 3.141,
                           barfoo = 'bar foo',
                           array = { 'one', 'two', 'three',
                                     { 1, 'two', 3, 'four' } },
                           yees = true,
                           noou = false
                       }
                 })
--    m:load_plugin('test', 'test2')
--    m:info("Successfully loaded two instances of test...")
end

-- load the dbus plugin if it exists
-- if m:plugin_exists('dbus') then
--     m:load_plugin('dbus')
-- end

-- load glib plugin, ignoring any errors
-- m:try_load_plugin('glib')

-- load the native resource plugin
if m:plugin_exists('resource-native') then
    m:load_plugin('resource-native')
    m:info("native resource plugin loaded")
else
    m:info("No native resource plugin found...")
end

-- load the dbus resource plugin
if m:plugin_exists('resource-dbus') then
    m:try_load_plugin('resource-dbus', {
        dbus_bus = "session",
        dbus_service = "org.Murphy",
        dbus_track = true,
        default_zone = "driver",
        default_class = "implicit"
      })
    m:info("dbus resource plugin loaded")
else
    m:info("No dbus resource plugin found...")
end

-- load the WRT resource plugin
if m:plugin_exists('resource-wrt') then
    m:try_load_plugin('resource-wrt', {
                          address = "wsck:127.0.0.1:4000/murphy",
                          httpdir = "src/plugins/resource-wrt",
--                          sslcert = 'src/plugins/resource-wrt/resource.crt',
--                          sslpkey = 'src/plugins/resource-wrt/resource.key'
                      })
else
    m:info("No WRT resource plugin found...")
end

-- load the domain control plugin if it exists
if m:plugin_exists('domain-control') then
    m:load_plugin('domain-control')
else
    m:info("No domain-control plugin found...")
end

-- -- load the domain control plugin if it exists
-- if m:plugin_exists('domain-control') then
--     m:try_load_plugin('domain-control', 'wrt-export', {
--         external_address = '',
--         internal_address = '',
--         wrt_address = "wsck:127.0.0.1:5000/murphy",
--         httpdir     = "src/plugins/domain-control"
--     })
-- else
--     m:info("No domain-control plugin found...")
-- end

-- define application classes
application_class { name="implicit" , priority=0 , modal=false, share=true , order="lifo" }

-- define zone attributes
zone.attributes {
}

-- define zones
zone {
     name = "driver"
}

-- define resource classes
resource.class {
     name = "audio_playback",
     shareable = true,
     attributes = {
         role = { mdb.string, "music", "rw" },
         pid = { mdb.string, "<unknown>", "rw" },
         policy = { mdb.string, "relaxed", "rw" }
     }
}

-- SyncEvolution resources: one per runtest.py
-- Some tests can run in parallel. Those resources are shareable.
for i,v in pairs {
    -- compiling the source on one platform
    "compile",

    -- checking out source
    "libsynthesis",
    "syncevolution",
    "activesyncd",

    -- local tests
    "evolution",
    "dbus",
    "pim",
    } do
    resource.class {
        name = v,
        shareable = true
    }
end

-- TODO (in runtests.py): some of these resources overlap
for i,v in pairs {
    -- tests involving unique peers
    "googlecalendar",
    "googlecontacts",
    "owndrive",
    "yahoo",
    "oracle",
    "davical",
    "apple",
    "googleeas",
    "exchange",
    "edsfile",
    "edseds",
    "edsxfile",
    "davfile",
    "edsdav",
    "mobical",
    "memotoo",
    } do
    resource.class {
        name = v,
        shareable = false
    }
end

-- test for creating selections: don't remove, murphyd won't start without it
-- (E: Failed to enable resolver autoupdate.)
mdb.select {
           name = "audio_owner",
           table = "audio_playback_owner",
           columns = {"application_class"},
           condition = "zone_name = 'driver'",
}

$ cat testing/kill-murphy.py
#! /usr/bin/python -u

gobject = None
try:
    import gobject
except ImportError:
    try:
         from gi.repository import GObject as gobject
    except ImportError:
         pass
import dbus
from dbus.mainloop.glib import DBusGMainLoop
DBusGMainLoop(set_as_default=True)
bus = dbus.SessionBus()
loop = gobject.MainLoop()
murphy = dbus.Interface(bus.get_object('org.Murphy', '/org/murphy/resource'), 'org.murphy.manager')
path = murphy.createResourceSet()
resourceset = dbus.Interface(bus.get_object('org.Murphy', path), 'org.murphy.resourceset')
resourceset.addResource('libsynthesis') # This resource must exist, otherwise this script hangs!
resourceset.request()
while resourceset.getProperties()['status'] == 'pending':
    loop.get_context().iteration(True)
resourceset.release()
resourceset.delete()

[pohly]

While I am at it: murphyd refuses to start when I remove the (useless) mdb.select { "audio_owner" } from the murphy-ref.lua.

It says: E: Failed to enable resolver autoupdate. then quits.

[pohly]

And one more question related to the Python test script: if the resource name is something that hasn't been defined, the script just hangs waiting for a status change forever. Is that intentional? Wouldn't it be better to report that as an error, ideally by throwing an error in request()?

[ipuustin]

Thanks for the bug report. I'll take a look at the D-Bus segfault. I think the autoupdate enabling failure would warrant a completely new issue!

[klihub]

Thanks, I'll take a look at that. My initial guess is that it bails out if the generated resolver ruleset is empty, but I'll take a closer look.

Cheers, kli

On 10 Dec 2013, at 22:02, Patrick Ohly notifications@github.com wrote:

While I am at it: murphyd refuses to start when I remove the (useless) mdb.select { "audio_owner" } from the murphy-ref.lua.

It says: E: Failed to enable resolver autoupdate. then quits.

— Reply to this email directly or view it on GitHub.

[ipuustin]

Alright, I think I found the bug. Could you try the latest master to see if the problem persists? Thanks.

[pohly]

Yes, master seems to work now even when compiled with -O2.

[ipuustin]

Ok, I'm closing this issue.