|
Matej Habrnal |
ba94a9 |
From 10a3ff45b150fe426624080abf903242af3ac201 Mon Sep 17 00:00:00 2001
|
|
Matej Habrnal |
ba94a9 |
From: Jakub Filak <jfilak@redhat.com>
|
|
Matej Habrnal |
ba94a9 |
Date: Tue, 23 Feb 2016 15:40:17 +0100
|
|
Matej Habrnal |
ba94a9 |
Subject: [PATCH] CCpp: turn off compat cores
|
|
Matej Habrnal |
ba94a9 |
|
|
Matej Habrnal |
ba94a9 |
Disable compat cores because with systemd>=229, RLIMIT_CORE is
|
|
Matej Habrnal |
ba94a9 |
'unlimited' by default and RLIMIT_CORE!=0 was the signal to create a
|
|
Matej Habrnal |
ba94a9 |
core file in the old way. So, abrt-ccpp-hook creates the old way cores
|
|
Matej Habrnal |
ba94a9 |
everywhere on file systems by default and users don't know about it
|
|
Matej Habrnal |
ba94a9 |
because they did not enable it manually via 'ulimit -c unlimited'.
|
|
Matej Habrnal |
ba94a9 |
|
|
Matej Habrnal |
ba94a9 |
We could restore the old RLIMIT_CORE value by creating a systemd
|
|
Matej Habrnal |
ba94a9 |
configuration file with 'DefaultLimitCORE=0' but that would not be a
|
|
Matej Habrnal |
ba94a9 |
good solution for our users:
|
|
Matej Habrnal |
ba94a9 |
* you have to reboot, otherwise the configuration has no effect
|
|
Matej Habrnal |
ba94a9 |
* and therefore it is impossible to set the new value when
|
|
Matej Habrnal |
ba94a9 |
abrt-ccpp.service is stopped and the users who expect
|
|
Matej Habrnal |
ba94a9 |
RTLIMIT_CORE='unlimited' would be confused
|
|
Matej Habrnal |
ba94a9 |
|
|
Matej Habrnal |
ba94a9 |
We should probably adopt the behaviour of systemd-coredump. However,
|
|
Matej Habrnal |
ba94a9 |
that means that abrtd must not drop any problem it receives and should
|
|
Matej Habrnal |
ba94a9 |
mark those problems that would be dropped as NOT-REPORTABLE to prevent
|
|
Matej Habrnal |
ba94a9 |
users from filing bug reports for unfixable problems.
|
|
Matej Habrnal |
ba94a9 |
|
|
Matej Habrnal |
ba94a9 |
Find more details at:
|
|
Matej Habrnal |
ba94a9 |
https://github.com/systemd/systemd/issues/2643
|
|
Matej Habrnal |
ba94a9 |
|
|
Matej Habrnal |
ba94a9 |
Signed-off-by: Jakub Filak <jfilak@redhat.com>
|
|
Matej Habrnal |
ba94a9 |
---
|
|
Matej Habrnal |
ba94a9 |
src/hooks/CCpp.conf | 5 ++++-
|
|
Matej Habrnal |
ba94a9 |
1 file changed, 4 insertions(+), 1 deletion(-)
|
|
Matej Habrnal |
ba94a9 |
|
|
Matej Habrnal |
ba94a9 |
diff --git a/src/hooks/CCpp.conf b/src/hooks/CCpp.conf
|
|
Matej Habrnal |
ba94a9 |
index 025189a..edf66b7 100644
|
|
Matej Habrnal |
ba94a9 |
--- a/src/hooks/CCpp.conf
|
|
Matej Habrnal |
ba94a9 |
+++ b/src/hooks/CCpp.conf
|
|
Matej Habrnal |
ba94a9 |
@@ -7,7 +7,10 @@
|
|
Matej Habrnal |
ba94a9 |
# If the original template string starts with "|", the string "core" is used
|
|
Matej Habrnal |
ba94a9 |
# instead of the template.
|
|
Matej Habrnal |
ba94a9 |
# For more information about naming core dump files see 'man 5 core'.
|
|
Matej Habrnal |
ba94a9 |
-MakeCompatCore = yes
|
|
Matej Habrnal |
ba94a9 |
+#
|
|
Matej Habrnal |
ba94a9 |
+# 'no' because of RLIMIT_CORE='unlimited' since systemd-229:
|
|
Matej Habrnal |
ba94a9 |
+# https://github.com/systemd/systemd/issues/2643
|
|
Matej Habrnal |
ba94a9 |
+MakeCompatCore = no
|
|
Matej Habrnal |
ba94a9 |
|
|
Matej Habrnal |
ba94a9 |
# Do you want a copy of crashed binary be saved?
|
|
Matej Habrnal |
ba94a9 |
# (useful, for example, when _deleted binary_ segfaults)
|
|
Matej Habrnal |
ba94a9 |
--
|
|
Matej Habrnal |
ba94a9 |
2.5.0
|
|
Matej Habrnal |
ba94a9 |
|