FS#20191 - [openssh] upstream config files modified on install
Attached to Project:
Arch Linux
Opened by Pete (tam1138) - Thursday, 15 July 2010, 21:14 GMT
Last edited by Guillaume ALAUX (galaux) - Monday, 17 January 2011, 20:43 GMT
Opened by Pete (tam1138) - Thursday, 15 July 2010, 21:14 GMT
Last edited by Guillaume ALAUX (galaux) - Monday, 17 January 2011, 20:43 GMT
|
Details
As long as svn can remember, the openssh PKGBUILD contains
within the following code that alters the default upstream
config files:
#adjust our config files sed -i \ -e 's|^#ListenAddress 0.0.0.0|ListenAddress 0.0.0.0|g' \ -e 's|^#UsePAM no|UsePAM yes|g' \ -e 's|^#ChallengeResponseAuthentication yes|ChallengeResponseAuthentication no|g' \ $startdir/pkg/etc/ssh/sshd_config sed -i -e 's|^# Host \*|Host *|g' $startdir/pkg/etc/ssh/ssh_config echo "HashKnownHosts yes" >> $startdir/pkg/etc/ssh/ssh_config echo "StrictHostKeyChecking ask" >> $startdir/pkg/etc/ssh/ssh_config #ssh_config sed -i \ -e 's|^# Host \*|Host *|g' \ ${pkgdir}/etc/ssh/ssh_config Of these, two are no-ops (ListenAddress and StrictHostKeyChecking), one is a duplicated no-op (Host), and one could argue---which I would---that the others should be left at their defaults and the onus put on the user to configure them correctly. Though I could see that UsePAM and ChallengeResponseAuthentication could be considered sufficiently widely-used to modify them as above, but HashKnownHosts, at least, is not something that is closing a huge security hole. Plus, it screws up my bash-completion. I have attached a dreadfully complicated patch for those who wish to avoid typing 15dd. (You're on your own if you're using that... other editor.) |
This task depends upon
Closed by Guillaume ALAUX (galaux)
Monday, 17 January 2011, 20:43 GMT
Reason for closing: Fixed
Additional comments about closing: Fixed in 5.6p1-2
Monday, 17 January 2011, 20:43 GMT
Reason for closing: Fixed
Additional comments about closing: Fixed in 5.6p1-2
openssh-PKGBUILD.patch
The only "bug" is the duplicate "Host *" entry in the PKGBUILD.
This has been fixed, according to my ABS tree.