Hello community,

here is the log from the commit of package udev for openSUSE:12.1 checked in at 
2011-11-08 20:12:34
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Comparing /work/SRC/openSUSE:12.1/udev (Old)
 and      /work/SRC/openSUSE:12.1/.udev.new (New)
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

Package is "udev", Maintainer is "[email protected]"

Changes:
--------
--- /work/SRC/openSUSE:12.1/udev/udev.changes   2011-10-24 13:30:32.000000000 
+0200
+++ /work/SRC/openSUSE:12.1/.udev.new/udev.changes      2011-11-08 
20:13:22.000000000 +0100
@@ -1,0 +2,5 @@
+Tue Nov  8 15:27:02 UTC 2011 - [email protected]
+
+- Added upstream patch to avoid deadlocks, bnc#722858
+
+-------------------------------------------------------------------

New:
----
  udev.git-7944a13a2b8516d5fa354703789f5988cb4299b6.patch

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

Other differences:
------------------
++++++ udev.spec ++++++
--- /var/tmp/diff_new_pack.3kOznm/_old  2011-11-08 20:13:22.000000000 +0100
+++ /var/tmp/diff_new_pack.3kOznm/_new  2011-11-08 20:13:22.000000000 +0100
@@ -43,6 +43,8 @@
 Source60:       boot.udev
 #Patch from upstream to make bluetooth input devices working again.
 Patch1:         udev.git-617746e09795575c6258dd075ee7f0a44ce61e1e.patch
+#PATCH-FIX-UPSTREAM bnc#722858
+Patch2:         udev.git-7944a13a2b8516d5fa354703789f5988cb4299b6.patch
 
 # Upstream First - Policy:
 # Never add any patches to this package without the upstream commit id
@@ -101,6 +103,7 @@
 %prep
 %setup -q
 %patch1 -p1
+%patch2 -p1
 
 %build
 # prevent man pages from re-building (xmlto)

++++++ udev.git-7944a13a2b8516d5fa354703789f5988cb4299b6.patch ++++++
commit 7944a13a2b8516d5fa354703789f5988cb4299b6
Author: Steve Langasek <[email protected]>
Date:   Sat Oct 8 02:14:09 2011 -0700

    udevd: exit - process events before signals in worker
    
    When a worker receives both a signal and a udev event in the same epoll_wait
    run, the event must be processed first because the udev parent considers the
    event already dispatched.  If we process the signal first and exit, udevd
    times out after 60 seconds waiting for a response from an already-dead
    worker.
    
    Ref: https://bugs.launchpad.net/bugs/818177
    Signed-off-by: Steve Langasek <[email protected]>

diff --git a/udev/udevd.c b/udev/udevd.c
index 77aec9d..b65b53f 100644
--- a/udev/udevd.c
+++ b/udev/udevd.c
@@ -347,6 +347,7 @@ static void worker_new(struct event *event)
                                for (i = 0; i < fdcount; i++) {
                                        if (ev[i].data.fd == fd_monitor && 
ev[i].events & EPOLLIN) {
                                                dev = 
udev_monitor_receive_device(worker_monitor);
+                                               break;
                                        } else if (ev[i].data.fd == fd_signal 
&& ev[i].events & EPOLLIN) {
                                                struct signalfd_siginfo fdsi;
                                                ssize_t size;
-- 
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to