The other day I was running the stock fedora kernel on my ip
forwarding setup, to see what the performance was, and the performance
wasn't very good.
system is S5520HC dual socket 2.93GHz Xeon 5570 (Nehalem) with 3 quad
port 82580 adapters (12 ports). Traffic is bidirectional 64 byte
packets being forwarded and received on each port, basically port to
port routing. I am only using 12 flows currently.
The driver is igb, and I am using an affinity script that lines up
each pair of ports that are forwarding traffic into optimal
configurations for cache locality. I am also disabling
remote_node_defrag_ratio to stop cross node traffic.
With the fedora default kernel from F14 it appears that
CONFIG_NETFILTER=y means that I cannot unload all of netfilter even if
I stop iptables service.
perf showed netfilter being prominent, and removing it gives me much
higher throughput. Is there a reason CONFIG_NETFILTER=y ? Isn't it a
good thing to be able to disable netfilter if you want to?
ptrace_report_clone() uses child->pid, this is obviously wrong unless
the tracer is from the global namespace. Change it to use task_pid_vnr().
This is still not right, we should use the tracer's namespace, not
parent's. But this matches upstream, and at least this works if they
are from the same namespace.
Reported-by: Robin Green <greenrd(a)greenrd.org>
Signed-off-by: Oleg Nesterov <oleg(a)redhat.com>
@@ -403,7 +403,7 @@ static u32 ptrace_report_clone(u32 actio
- ctx->eventmsg = child->pid;
+ ctx->eventmsg = task_pid_vnr(child);
* We shouldn't stop now, inside the do_fork() path.
* We will stop later, before return to user-mode.
We've had a number of panics reported by people using cifs mounts on
F15. See the following reported bugs:
I'd like to see both of the following patches added to 188.8.131.52
and above until they make it into the stable series. The second
may also be appropriate for F16, but it may be simpler to just
wait for the stable series to catch up there.
The patches are pretty straightforward fixes, let me know if
anything with them isn't clear.
Pavel Shilovsky (1):
CIFS: Fix ERR_PTR dereference in cifs_get_root
Steve French (1):
[CIFS] possible memory corruption on mount
fs/cifs/cifsfs.c | 10 ++++++----
fs/cifs/connect.c | 3 ++-
2 files changed, 8 insertions(+), 5 deletions(-)
Thanks for working on this 3.0.x for f15.
Just to let you know I've been testing 184.108.40.206-0.fc15.x86_64 (for a
few hours) and so far its working well (lenovo W520 I7 sandy bridge
laptop). Will keep you posted of any issues should they arise.
Moving forward with our plan to release Fedora 16 with the 3.1 kernel, I
pushed 3.1-rc2 into the F16 branch this morning. Dave and I have both
been running it in rawhide for a bit now, and it's been fairly stable.
There will, of course, be bugs but getting some testing on the rc's will
hopefully allow us to get them resolved before 3.1 final is out. I will
submit a build for f16 updates-testing a bit later this week.
NOTE: This will not impact the Alpha release. That will go out with
Another attempt. This version tries to decouple utrace and ptrace.
This way it is much simpler to follow the upstream changes, afaics.
- The single-stepping updates in ptrace_resume() can race
This was fixed by 20/20, but I noticed that this patch
is buggy right before sending.
- Perhaps PTRACE_SYSEMU/TIF_SYSCALL_EMU logic was broken,
I need to recheck.