7061, 7062 Time Wait Improvements

7061, 7062 Time Wait Improvements

[Robert Mustacchi]

https://us-east.manta.joyent.com/rmustacc/public/webrevs/7062/index.html

Work by Patrick Mooney. See the older thread "Feedback wanted - Overhaul
TIME_WAIT expiration"​ for more background.

Re: [developer] 7061, 7062 Time Wait Improvements

[Garrett D'Amore]

[-- Attachment #1: Type: text/plain, Size: 896 bytes --]

I think I looked at this before,  In any event I’ve gone back over it
again.  I’m not super thrilled with use of ddi_get_lbolt64(), mostly
because I think we’ve tried pretty hard to avoid referencing lbolt in other
projects (tickless work did a lot of cleanups there), but I guess its ok.
 (A better alternative, IMO, would be gethrtime().  Once upon a time
ddi_get_lbolt() was faster, but with modern illumos / Solaris that’s no
longer true, since there is no longer a global lbolt variable.)

Modulo that relatively minor complaint, it looks good, and is well
documented to boot.  Thanks.

On Mon, Jun 6, 2016 at 10:40 AM, Robert Mustacchi <rm@joyent.com> wrote:

> https://us-east.manta.joyent.com/rmustacc/public/webrevs/7062/index.html
> 
> Work by Patrick Mooney. See the older thread "Feedback wanted - Overhaul
> TIME_WAIT expiration"​ for more background.
> 

[-- Attachment #2: Type: text/html, Size: 1508 bytes --]

Re: [developer] 7061, 7062 Time Wait Improvements

[Patrick Mooney]

[-- Attachment #1: Type: text/plain, Size: 1543 bytes --]

I used lbolt to be consistent with the surrounding code.  Switching it to
use hrtime wouldn't be difficult, but I would propose that be part of a
larger (separate) effort to clean up the inet stack where possible.

Thanks for taking a look.

On 6 June 2016 at 14:33, Garrett D'Amore <garrett@damore.org> wrote:

> I think I looked at this before,  In any event I’ve gone back over it
> again.  I’m not super thrilled with use of ddi_get_lbolt64(), mostly
> because I think we’ve tried pretty hard to avoid referencing lbolt in other
> projects (tickless work did a lot of cleanups there), but I guess its ok.
>  (A better alternative, IMO, would be gethrtime().  Once upon a time
> ddi_get_lbolt() was faster, but with modern illumos / Solaris that’s no
> longer true, since there is no longer a global lbolt variable.)
>
> Modulo that relatively minor complaint, it looks good, and is well
> documented to boot.  Thanks.
>
> On Mon, Jun 6, 2016 at 10:40 AM, Robert Mustacchi <rm@joyent.com> wrote:
>
>> https://us-east.manta.joyent.com/rmustacc/public/webrevs/7062/index.html
>>
>> Work by Patrick Mooney. See the older thread "Feedback wanted - Overhaul
>> TIME_WAIT expiration"​ for more background.
>>
>>
>> http://www.listbox.com
>>
>
> *illumos-developer* | Archives
> <https://www.listbox.com/member/archive/182179/=now>
> <https://www.listbox.com/member/archive/rss/182179/26936138-1705e021> |
> Modify
> <https://www.listbox.com/member/?&>
> Your Subscription <http://www.listbox.com>
>

[-- Attachment #2: Type: text/html, Size: 2201 bytes --]

Re: [developer] 7061, 7062 Time Wait Improvements

[Josef 'Jeff' Sipek]

FWIW, it looks like this got 7061 bundled in by accident.

Jeff.

On Mon, Jun 06, 2016 at 10:40:07 -0700, Robert Mustacchi wrote:
> https://us-east.manta.joyent.com/rmustacc/public/webrevs/7062/index.html
> 
> Work by Patrick Mooney. See the older thread "Feedback wanted - Overhaul
> TIME_WAIT expiration"​ for more background.
> 

-- 
All parts should go together without forcing.  You must remember that the
parts you are reassembling were disassembled by you.  Therefore, if you
can’t get them together again, there must be a reason.  By all means, do not
use a hammer.
                — IBM Manual, 1925

Re: [developer] 7061, 7062 Time Wait Improvements

[Patrick Mooney]

It's intentional.  Being fairly intertwined, the two changes were put
back together.

On 6 June 2016 at 15:06, Josef 'Jeff' Sipek <jeffpc@josefsipek.net> wrote:
> FWIW, it looks like this got 7061 bundled in by accident.
>
> Jeff.
>
> On Mon, Jun 06, 2016 at 10:40:07 -0700, Robert Mustacchi wrote:
>> https://us-east.manta.joyent.com/rmustacc/public/webrevs/7062/index.html
>>
>> Work by Patrick Mooney. See the older thread "Feedback wanted - Overhaul
>> TIME_WAIT expiration" for more background.
>>
> 
> --
> All parts should go together without forcing.  You must remember that the
> parts you are reassembling were disassembled by you.  Therefore, if you
> can’t get them together again, there must be a reason.  By all means, do not
> use a hammer.
>                 — IBM Manual, 1925
> 

Re: [developer] 7061, 7062 Time Wait Improvements

[Josef 'Jeff' Sipek]

On Mon, Jun 06, 2016 at 15:07:50 -0500, Patrick Mooney wrote:
> It's intentional.  Being fairly intertwined, the two changes were put
> back together.

Sigh... I should stop reviewing for the day.  I confused the 7061 change
with the keepalive tcp_tunables.c change (7064) in the other review.

Jeff.

> On 6 June 2016 at 15:06, Josef 'Jeff' Sipek <jeffpc@josefsipek.net> wrote:
> > FWIW, it looks like this got 7061 bundled in by accident.
> >
> > Jeff.
> >
> > On Mon, Jun 06, 2016 at 10:40:07 -0700, Robert Mustacchi wrote:
> >> https://us-east.manta.joyent.com/rmustacc/public/webrevs/7062/index.html
> >>
> >> Work by Patrick Mooney. See the older thread "Feedback wanted - Overhaul
> >> TIME_WAIT expiration" for more background.
> >>
> > 
> > --
> > All parts should go together without forcing.  You must remember that the
> > parts you are reassembling were disassembled by you.  Therefore, if you
> > can’t get them together again, there must be a reason.  By all means, do not
> > use a hammer.
> >                 — IBM Manual, 1925
> > 

-- 
Failure is not an option,
It comes bundled with your Microsoft product.

Re: [developer] 7061, 7062 Time Wait Improvements

[Garrett D'Amore]

[-- Attachment #1: Type: text/plain, Size: 1730 bytes --]

That sounds quite reasonable to me.  Thanks. ;-)

On Mon, Jun 6, 2016 at 12:56 PM, Patrick Mooney <patrick.mooney@joyent.com>
wrote:

> I used lbolt to be consistent with the surrounding code.  Switching it to
> use hrtime wouldn't be difficult, but I would propose that be part of a
> larger (separate) effort to clean up the inet stack where possible.
>
> Thanks for taking a look.
>
> On 6 June 2016 at 14:33, Garrett D'Amore <garrett@damore.org> wrote:
>
>> I think I looked at this before,  In any event I’ve gone back over it
>> again.  I’m not super thrilled with use of ddi_get_lbolt64(), mostly
>> because I think we’ve tried pretty hard to avoid referencing lbolt in other
>> projects (tickless work did a lot of cleanups there), but I guess its ok.
>>  (A better alternative, IMO, would be gethrtime().  Once upon a time
>> ddi_get_lbolt() was faster, but with modern illumos / Solaris that’s no
>> longer true, since there is no longer a global lbolt variable.)
>>
>> Modulo that relatively minor complaint, it looks good, and is well
>> documented to boot.  Thanks.
>>
>> On Mon, Jun 6, 2016 at 10:40 AM, Robert Mustacchi <rm@joyent.com> wrote:
>>
>>> https://us-east.manta.joyent.com/rmustacc/public/webrevs/7062/index.html
>>>
>>> Work by Patrick Mooney. See the older thread "Feedback wanted - Overhaul
>>> TIME_WAIT expiration"​ for more background.
>>>
>>>
>>> http://www.listbox.com
>>>
>>
>> *illumos-developer* | Archives
>> <https://www.listbox.com/member/archive/182179/=now>
>> <https://www.listbox.com/member/archive/rss/182179/26936138-1705e021> |
>> Modify
>> <https://www.listbox.com/member/?&>
>> Your Subscription <http://www.listbox.com>
>>
>
>

[-- Attachment #2: Type: text/html, Size: 2714 bytes --]