Hello,
after reading emails from Philip Guenther and Mark Kettenis, doing some RTFM on
locking in OpenBSD kernel I have a new patch. I call it as a smp-step-0.
Patch introduces a KERNEL_LOCK() to PF. Many dances with KERNEL_LOCK() happens
in pf_test(). From future work point of view there are
/large snip
Assuming the locking in MULTIPROCESSOR goes like:
interrupt grabs splsoftnet - ip_input - PF grabs KERNEL_LOCK()
We need to take care of ioctl() call path and purge thread. Those need to
get synchronize with packets using KERNEL_LOCK(). They should not to mess
with
On 27/08/15(Thu) 14:19, Alexandr Nedvedicky wrote:
/large snip
Assuming the locking in MULTIPROCESSOR goes like:
interrupt grabs splsoftnet - ip_input - PF grabs KERNEL_LOCK()
We need to take care of ioctl() call path and purge thread. Those need to
get synchronize with packets
On 27/08/15(Thu) 11:10, Alexandr Nedvedicky wrote:
On Wed, Aug 26, 2015 at 06:12:10PM +0200, Mark Kettenis wrote:
Date: Wed, 26 Aug 2015 17:30:14 +0200
From: Alexandr Nedvedicky alexandr.nedvedi...@oracle.com
Hello,
I'm not sure I got everything right in Calgary. So this patch
On Wed, Aug 26, 2015 at 06:12:10PM +0200, Mark Kettenis wrote:
Date: Wed, 26 Aug 2015 17:30:14 +0200
From: Alexandr Nedvedicky alexandr.nedvedi...@oracle.com
Hello,
I'm not sure I got everything right in Calgary. So this patch should
roughly illustrates how I think we should start
Hello,
I'm not sure I got everything right in Calgary. So this patch should roughly
illustrates how I think we should start moving forward to make PF MULTIPROCESSOR
friendly. It's quite likely my proposal/way is completely off, I'll be happy if
you put me back to ground.
The brief summary of