Hi,
I am working on a diff to run UDP input in parallel. Btrace kstack
analysis shows that SIP hash for PCB lookup is quite expensive.
When running in parallel we get lock contention on the PCB table
mutex.
So it results in better performance to calculate the hash value
before taking the mutex.
On Tue, 07 Feb 2023 20:35:10 -0700, Todd C. Miller wrote:
> On Tue, 07 Feb 2023 17:17:02 -0700, Todd C. Miller wrote:
>
> > Yes, the bug is that the number is not displayed. The following
> > diff fixes that but there is still a bug because the resulting line
> > also lacks a line number. In
On Thu, Jun 22, 2023 at 02:25:08PM +0200, Claudio Jeker wrote:
> Here is a diff to refactor and update dvrmpd to use the new ibuf API.
> Instead of doing the checksum calculations in various places do it in
> send_packet() as a central place.
Reads ok, and again:
> Could people using dvrmpd test
Here is a diff to refactor and update dvrmpd to use the new ibuf API.
Instead of doing the checksum calculations in various places do it in
send_packet() as a central place.
Could people using dvrmpd test this diff, thanks
--
:wq Claudio
Index: ask_nbrs2.c
On Thu, Jun 22, 2023 at 11:15:31AM +0200, Theo Buehler wrote:
> On Thu, Jun 22, 2023 at 10:26:29AM +0200, Claudio Jeker wrote:
> > Another ibuf_seek() -> ibuf_set_n16() conversions.
> > While there also switch to ibuf_data() and ibuf_size().
>
> Reads fine. ok. a test would indeed be nice.
I did
On Thu, Jun 22, 2023 at 10:26:29AM +0200, Claudio Jeker wrote:
> Another ibuf_seek() -> ibuf_set_n16() conversions.
> While there also switch to ibuf_data() and ibuf_size().
Reads fine. ok. a test would indeed be nice.
Another ibuf_seek() -> ibuf_set_n16() conversions.
While there also switch to ibuf_data() and ibuf_size().
Anyone still uses ripd that can test this?
--
:wq Claudio
Index: auth.c
===
RCS file: /cvs/src/usr.sbin/ripd/auth.c,v