Holger Hans Peter Freyther <holger <at> freyther.de> writes:
>
> The following two patches are compile tested only. The first one is
> to avoid work when systemd is compiled without audit support. The
> second is starting to remove dynamic allocations from the
> dispatch_message_real method.
I just went ahead and added #if 0/#endif around the variables I do not
need for my usecase. The CPU usage is noticable down, still more than
the busybox syslog implementation.
> _COMM
> _EXE
> _CMDLINE
> _SYSTEMD_CGROUP
server_dispatch_message():
1. shortened_cgroup_path(ucred->pid);
Open /proc/PID/cgroup and then the cgroup..
Open /proc/1/cgroup and then the cgroup
2. Eventually call server_driver_message which will
1. call dispatch_message_real().. which will..
1. call shorted_cgroup_path(ucred->pid);
Open /proc/pid...
Open /proc/1
3. call dispatch_message_real()
1. shorted_cgroup_path(ucred->pid);
Open /proc/PID/cgroup and then the cgroup..
Open /proc/1/cgroup ...
Just adding the path to the dispatch_message_real signature is a bit
tricky as the server_driver_message is creating a ucred with the pid,
gid,uid of the journald process...
Is it really needed to determine the cgroup of /proc/1 on every message?
Couldn't journald keep the variables for its identity around?
_______________________________________________
systemd-devel mailing list
systemd-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/systemd-devel
