Project

General

Profile

Bug #1231

ipsec stop produces errors on systems that don't support subsecond sleep

Added by Saso Slavicic almost 6 years ago. Updated almost 6 years ago.

Status:
Closed
Priority:
Low
Assignee:
Tobias Brunner
Category:
starter
Target version:
5.4.0
Start date:
08.12.2015
Due date:
Estimated time:
Affected version:
5.3.5
Resolution:
Fixed

Description

Subsecond sleep is not supported in busybox.
On OpenWRT, stopping ipsec produces errors:

  1. /etc/init.d/ipsec stop
    Stopping strongSwan IPsec...
    sleep: invalid number '0.1'
    sleep: invalid number '0.1'
    sleep: invalid number '0.1'
    sleep: invalid number '0.1'
    sleep: invalid number '0.1'

Associated revisions

Revision 7f527156 (diff)
Added by Tobias Brunner almost 6 years ago

ipsec: Fix stop command on systems where sleep(1) only supports integers

Fixes #1231.

History

#1 Updated by Tobias Brunner almost 6 years ago

  • Status changed from New to Feedback

Does sleep 0.1 2>/dev/null suppress the message?

#2 Updated by Saso Slavicic almost 6 years ago

Yes.

#3 Updated by Tobias Brunner almost 6 years ago

So a simple fix would be to just suppress the error. Or we could do something like this in case detecting and reporting the error takes less than 100ms:

diff --git a/src/ipsec/_ipsec.in b/src/ipsec/_ipsec.in
index 89c7ef753e46..86f7c640ef02 100644
--- a/src/ipsec/_ipsec.in
+++ b/src/ipsec/_ipsec.in
@@ -259,7 +259,7 @@ stop)
             loop=110
             while [ $loop -gt 0 ] ; do
                 kill -0 $spid 2>/dev/null || break
-                sleep 0.1
+                sleep 0.1 2>/dev/null || sleep 1
                 loop=$(($loop - 1))
             done
             if [ $loop -eq 0 ]

#4 Updated by Saso Slavicic almost 6 years ago

Sleeping for 1s should also decrease loop by additional 9 iterations...just in case so that it doesn't hang for 110s before issuing kill.

#5 Updated by Tobias Brunner almost 6 years ago

So maybe something like this:

diff --git a/src/ipsec/_ipsec.in b/src/ipsec/_ipsec.in
index 89c7ef753e46..7da0fd36d6d1 100644
--- a/src/ipsec/_ipsec.in
+++ b/src/ipsec/_ipsec.in
@@ -259,7 +259,12 @@ stop)
             loop=110
             while [ $loop -gt 0 ] ; do
                 kill -0 $spid 2>/dev/null || break
-                sleep 0.1
+                sleep 0.1 2>/dev/null
+                if [ $? -ne 0 ]
+                then
+                    sleep 1
+                    loop=$(($loop - 9))
+                fi
                 loop=$(($loop - 1))
             done
             if [ $loop -eq 0 ]

#6 Updated by Saso Slavicic almost 6 years ago

There shouldn't be any possibility for the loop to be overshot, that last line probably needs to be changed to '-le'.
But I'm probably nitpicking already... :-)

#7 Updated by Tobias Brunner almost 6 years ago

There shouldn't be any possibility for the loop to be overshot, that last line probably needs to be changed to '-le'.

You mean in case sleep 0.1 only fails occasionally?

#8 Updated by Saso Slavicic almost 6 years ago

Yep...I have no idea if that can happen, but rather be safe than sorry as it doesn't really cost anything.

#9 Updated by Tobias Brunner almost 6 years ago

  • Tracker changed from Issue to Bug
  • Subject changed from subsecond sleep not supported to ipsec stop produces errors on systems that don't support subsecond sleep
  • Category set to starter
  • Status changed from Feedback to Closed
  • Assignee set to Tobias Brunner
  • Target version set to 5.4.0
  • Resolution set to Fixed

Yep...I have no idea if that can happen, but rather be safe than sorry as it doesn't really cost anything.

I guess it's not very likely, but I agree. I pushed the fix to master.

Also available in: Atom PDF