Re: Bug#712848: FTBFS on kfreebsd-*: test-suite hangs and is killed after a timeout
The test-suite for glib2.0 fails to complete on kfreebsd-* as can be seen at [1]. On both kfreebsd-amd64 and kfreebsd-i386 the test-suite is killed after 150 min of inactivity. We would appreciate any help and insight from the kfreebsd to fix those failures on kfreebsd-*. Some observations (kfreebsd-amd64): * it hangs in /gdbus/proxy/async when "commented out", the builds continues up to /gdbus/codegen-peer-to-peer: g_dbus_connection_real_closed: Remote peer vanished with error: Underlying GIOStream returned 0 bytes on an async read (g-io-error-quark, 0). Exiting. FAIL GTester: last random seed: R02Se8737fe1b38820e4006b87deae6daa93 * some failures might be due to new kqueue module Overview of changes from GLib 2.35.1 to 2.35.2 GIO now has kqueue support for GFileMonitor (BSDs, Mac OS) Petr --- gio/tests/gdbus-proxy.c +++ gio/tests/gdbus-proxy.c @@ -930,7 +930,9 @@ g_test_add_func ("/gdbus/proxy", test_proxy); g_test_add_func ("/gdbus/proxy/no-properties", test_no_properties); g_test_add_func ("/gdbus/proxy/wellknown-noauto", test_wellknown_noauto); +#ifndef __FreeBSD_kernel__ g_test_add_func ("/gdbus/proxy/async", test_async); +#endif ret = g_test_run(); -- To UNSUBSCRIBE, email to debian-bsd-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/alpine.lnx.2.00.1306201655190.19...@contest.felk.cvut.cz
Re: Bug#712848: FTBFS on kfreebsd-*: test-suite hangs and is killed after a timeout
Hi, On 20/06/13 18:41, Petr Salinger wrote: >> The test-suite for glib2.0 fails to complete on kfreebsd-* as can be >> seen at [1]. On both kfreebsd-amd64 and kfreebsd-i386 the test-suite is >> killed after 150 min of inactivity. > >> We would appreciate any help and insight from the kfreebsd to fix those >> failures on kfreebsd-*. > > Some observations (kfreebsd-amd64): > > * it hangs in /gdbus/proxy/async > > when "commented out", the builds continues up to > > /gdbus/codegen-peer-to-peer: g_dbus_connection_real_closed: Remote peer > vanished with error: Underlying GIOStream returned 0 bytes on an async read > (g-io-error-quark, 0). Exiting. > FAIL > GTester: last random seed: R02Se8737fe1b38820e4006b87deae6daa93 > > * some failures might be due to new kqueue module >Overview of changes from GLib 2.35.1 to 2.35.2 >GIO now has kqueue support for GFileMonitor (BSDs, Mac OS) > This attempt didn't hang: https://buildd.debian.org/status/fetch.php?pkg=glib2.0&arch=kfreebsd-amd64&ver=2.36.3-2&stamp=1372553037 One test failed though: TEST: file... (pid=67485) /file/basic: OK /file/parent:OK /file/child: OK /file/type: OK /file/replace-load: OK /file/replace-cancel:OK /file/async-delete: OK /file/copy-preserve-mode:OK ** GLib-GIO:ERROR:/«PKGBUILDDIR»/./gio/tests/file.c:452:test_create_delete: assertion failed (data->monitor_created == 1): (0 == 1) /file/async-create-delete/0: FAIL May well be due to the kqueue support for file monitor. Help fixing it on kfreebsd is very welcome. Thanks, Emilio -- To UNSUBSCRIBE, email to debian-bsd-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/51d29975.2060...@debian.org
Re: Bug#712848: FTBFS on kfreebsd-*: test-suite hangs and is killed after a timeout
On 02/07/13 11:12, Emilio Pozuelo Monfort wrote: > Hi, > > On 20/06/13 18:41, Petr Salinger wrote: >>> The test-suite for glib2.0 fails to complete on kfreebsd-* as can be >>> seen at [1]. On both kfreebsd-amd64 and kfreebsd-i386 the test-suite is >>> killed after 150 min of inactivity. >> >>> We would appreciate any help and insight from the kfreebsd to fix those >>> failures on kfreebsd-*. >> >> Some observations (kfreebsd-amd64): >> >> * it hangs in /gdbus/proxy/async >> >> when "commented out", the builds continues up to >> >> /gdbus/codegen-peer-to-peer: g_dbus_connection_real_closed: Remote peer >> vanished with error: Underlying GIOStream returned 0 bytes on an async read >> (g-io-error-quark, 0). Exiting. >> FAIL >> GTester: last random seed: R02Se8737fe1b38820e4006b87deae6daa93 >> >> * some failures might be due to new kqueue module >>Overview of changes from GLib 2.35.1 to 2.35.2 >>GIO now has kqueue support for GFileMonitor (BSDs, Mac OS) >> > > This attempt didn't hang: > > https://buildd.debian.org/status/fetch.php?pkg=glib2.0&arch=kfreebsd-amd64&ver=2.36.3-2&stamp=1372553037 > > One test failed though: > > TEST: file... (pid=67485) > /file/basic: OK > /file/parent:OK > /file/child: OK > /file/type: OK > /file/replace-load: OK > /file/replace-cancel:OK > /file/async-delete: OK > /file/copy-preserve-mode:OK > ** > GLib-GIO:ERROR:/«PKGBUILDDIR»/./gio/tests/file.c:452:test_create_delete: > assertion failed (data->monitor_created == 1): (0 == 1) > /file/async-create-delete/0: FAIL > > May well be due to the kqueue support for file monitor. Help fixing it on > kfreebsd is very welcome. The sparc build has been fixed now (upload pending) so this is going to block migration of glib to testing, which blocks gtk+ 3.8 and many stuff that needs gtk+. I don't have much time these days so help looking at this is very appreciated. I wonder if some bsd port has a local patch to fix this. Thanks, Emilio -- To UNSUBSCRIBE, email to debian-bsd-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/51d5ce34.1030...@debian.org
Re: Bug#712848: FTBFS on kfreebsd-*: test-suite hangs and is killed after a timeout
GLib-GIO:ERROR:/?PKGBUILDDIR?/./gio/tests/file.c:452:test_create_delete: assertion failed (data->monitor_created == 1): (0 == 1) /file/async-create-delete/0: FAIL May well be due to the kqueue support for file monitor. Help fixing it on kfreebsd is very welcome. This test is guarded by: /* This test doesn't work with GPollFileMonitor, because it assumes * that the monitor will notice a create immediately followed by a * delete, rather than coalescing them into nothing. */ if (!strcmp (G_OBJECT_TYPE_NAME (data->monitor), "GPollFileMonitor")) { g_print ("skipping test for this GFileMonitor implementation"); goto skip; } The kqueue support might have the same limit. I do not know whether is better to use kqueue via gamin or kqueue directly in glib2.0. Petr -- To UNSUBSCRIBE, email to debian-bsd-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/alpine.lnx.2.00.1307090844540.12...@contest.felk.cvut.cz
Re: Bug#712848: FTBFS on kfreebsd-*: test-suite hangs and is killed after a timeout
I'm seeing some new errors in the testsuite now, such as: > (/home/steven/glib2.0-2.40.0/debian/build/deb/gio/tests/.libs/lt-mimeapps:67182): > GVFS-RemoteVolumeMonitor-WARNING **: cannot open directory > /usr/share/gvfs/remote-volume-monitors: Error opening directory > '/usr/share/gvfs/remote-volume-monitors': No such file or directory > # Start of appinfo tests > Trace/breakpoint trap > # GVFS-RemoteVolumeMonitor-FATAL-WARNING: cannot open directory > /usr/share/gvfs/remote-volume-monitors: Error opening directory > '/usr/share/gvfs/remote-volume-monitors': No such file or directory > ERROR: mimeapps - missing test plan > ERROR: mimeapps - exited with status 133 (terminated by signal 5?) happening for a few tests, all with the same FATAL-WARNING. Some open questions (I'll try to figure out myself if I get time) : * what's a FATAL-WARNING? * what's supposed to create /usr/share/gvfs/remote-volume-monitors? Regards, -- Steven Chamberlain ste...@pyro.eu.org -- To UNSUBSCRIBE, email to debian-bsd-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/53cd0a5f.4080...@pyro.eu.org
Re: Bug#712848: FTBFS on kfreebsd-*: test-suite hangs and is killed after a timeout
Control: tags -1 + patch Hi, There was no progress on this bug upstream yet: https://bugzilla.gnome.org/753378 But now we have too many packages in sid waiting on glib2.0 >= 2.44, or indirectly. Please could we use the attached patch as a workaround (as was done in FreeBSD), disabling kqueue, until the bug is fixed. Thanks, Regards, -- Steven Chamberlain ste...@pyro.eu.org --- a/debian/rules +++ b/debian/rules @@ -92,6 +92,12 @@ --disable-gtk-doc endif +ifeq ($(DEB_HOST_ARCH_OS), kfreebsd) + # Disable kqueue on kfreebsd, until https://bugzilla.gnome.org/753378 + # can be fixed. + DEB_CONFIGURE_EXTRA_FLAGS += "ac_cv_header_sys_event_h=no" +endif + DEB_CONFIGURE_FLAGS_deb := \ --enable-gtk-doc \ --enable-static \ signature.asc Description: Digital signature
Re: Bug#712848: FTBFS on kfreebsd-*: test-suite hangs and is killed after a timeout
Hi, I propose the attached patch for this. Upstream hasn't replied on this issue in their tracker in months, but for kfreebsd we need to do something, as a large backlog of packages are BD-Uninstallable in sid due to this. I think it is a recursion from the GUnixMountMonitor constructor, to a GLocalFileMonitor on /etc/fstab, and into GUnixMountMonitor again, now with a mutex already held, so it deadlocks. https://bugzilla.gnome.org/page.cgi?id=traceparser/trace.html&trace_id=235354 That mutex in glocalfile.c:g_local_file_find_enclosing_mount() doesn't seem necessary any more IMHO. Inside it, only 'mount' is modified, but that's just a stack variable local to this function. When klass->get_mount_for_mount_path is called, it's given one const parameter and the other is unused, so they're unchanged. 'klass' doesn't seem it could be modified either inside that function. It doesn't recurse infinitely, but seems to work correctly and pass the testsuite after this change. The FreeBSD project already applied my patch in their ports tree, and their users seem happy with it. If you don't want to apply the patch, because you have concerns it could create a regression on Linux, the mutex lock/unlock could be wrapped in #ifndef __FreeBSD_kernel__ instead. But I'm surprised the deadlock issue hasn't been seen on Linux, as the same generic "unix" implementation is used there as for kfreebsd. Thanks. Regards, -- Steven Chamberlain ste...@pyro.eu.org --- a/gio/gunionvolumemonitor.c +++ b/gio/gunionvolumemonitor.c @@ -591,11 +591,7 @@ mount = NULL; if (klass->get_mount_for_mount_path) -{ - g_rec_mutex_lock (&the_volume_monitor_mutex); - mount = klass->get_mount_for_mount_path (mount_path, cancellable); - g_rec_mutex_unlock (&the_volume_monitor_mutex); -} +mount = klass->get_mount_for_mount_path (mount_path, cancellable); /* TODO: How do we know this succeeded? Keep in mind that the native * volume monitor may fail (e.g. not being able to connect to signature.asc Description: Digital signature
Re: Bug#712848: FTBFS on kfreebsd-*: test-suite hangs and is killed after a timeout
Hi, Thanks for applying that fix to glib2.0; some ~200 packages have built now on kfreebsd-* in sid due to this! Regards, -- Steven Chamberlain ste...@pyro.eu.org signature.asc Description: Digital signature