locks and kernel randomness...

[Warner Losh]

> On Feb 25, 2015, at 1:20 PM, John Baldwin <jhb at FreeBSD.org> wrote:
> 
> On Wednesday, February 25, 2015 10:56:59 AM Konstantin Belousov wrote:
>> The cost of the proposed patch, of course, is not the several
>> thousands of instructions in the rebalance. The problem with it is the
>> introduction of the new spinlock, which will be used in many places
>> after the introduction. The cost of the new and often used spinlock is
>> the increase of both interrupt latency and interrupt handler jitter and
>> cpu switch jitter.
>> 
>> So neither buildworld timing, nor network throughput are adequate
>> to estimate the change.  It is system unresponsivness and loss of
>> the realtime behaviour up to some degree.
>> 
>> I thought that it was obvious, at least after spinlocks were mentioned,
>> but apparently it is not, since proposals to measure the patch effect
>> by benchmarking buildworld or passing the traffic are made.
> 
> +1
> 
> The only thing I will add is that in general this makes the system more
> fragile and complex as well.  Please just stay with a regular mutex and change
> the scheduler to not use random() (which seems to be in progress?).  I'm not
> sure why we needed the extra 40 messages in this thread after that was
> effectively said the first time.

If you’d like to be helpful, please review https://reviews.freebsd.org/D1968
Harrison Grundy wrote it, and I put it in phabricator for him while he sorts
out his own account.

In a fit of passive aggressiveness, I’ve added everybody who commented on
this thread as reviewers, except Bruce and Harrison who don’t have accounts
yet.

Warner

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 842 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <http://lists.freebsd.org/pipermail/freebsd-arch/attachments/20150225/bf662a79/attachment.sig>