Re: FCP-0101: Deprecating most 10/100 Ethernet drivers

classic Classic list List threaded Threaded
101 messages Options
123456
Reply | Threaded
Open this post in threaded view
|

Re: FCP-0101: Deprecating most 10/100 Ethernet drivers

Brooks Davis-2
On Thu, Oct 04, 2018 at 05:07:20PM +0200, Joel Dahl wrote:

> On Wed, Oct 03, 2018 at 09:05:16PM +0000, Brooks Davis wrote:
> > The criteria for exception are:
> >  - Popular in applications where it is likely to be deployed beyond the
> >    support lifetime of FreeBSD 12 (late 2023).
> >    - 5 reports of uses in the wild on machines running FreeBSD 12 will be
> >      deemed satisfy the "popular"
> >      requirement.
>
> Why doesn't reports of uses on machines running FreeBSD 10/11 count? I don't
> get it. 12.0 isn't even out yet, and most of our users are probably not
> running CURRENT. As I wrote in an earlier email, I have lots of these cards
> running in production - and most of them are on FreeBSD 11. They'll
> likely be upgraded to 12.1 in the future (but probably not 12.0 - I usually
> skip .0 releases). But doing the jump to CURRENT/12 now is just out of the
> question - these are production systems after all.
For the current poll, good faith intent to upgrade is fine.

-- Brooks

signature.asc (465 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: FCP-0101: Deprecating most 10/100 Ethernet drivers

Rodney W. Grimes-4
> On Thu, Oct 04, 2018 at 05:07:20PM +0200, Joel Dahl wrote:
> > On Wed, Oct 03, 2018 at 09:05:16PM +0000, Brooks Davis wrote:
> > > The criteria for exception are:
> > >  - Popular in applications where it is likely to be deployed beyond the
> > >    support lifetime of FreeBSD 12 (late 2023).
> > >    - 5 reports of uses in the wild on machines running FreeBSD 12 will be
> > >      deemed satisfy the "popular"
> > >      requirement.
> >
> > Why doesn't reports of uses on machines running FreeBSD 10/11 count? I don't
> > get it. 12.0 isn't even out yet, and most of our users are probably not
> > running CURRENT. As I wrote in an earlier email, I have lots of these cards
> > running in production - and most of them are on FreeBSD 11. They'll
> > likely be upgraded to 12.1 in the future (but probably not 12.0 - I usually
> > skip .0 releases). But doing the jump to CURRENT/12 now is just out of the
> > question - these are production systems after all.
>
> For the current poll, good faith intent to upgrade is fine.

What I am finding very bothersome at this point is that a great
miss understanding has been conveyed onto the users by the
statement that "core has discussed this and we plan to proceed
as proposed"

From a posting by Warner that statement is incorrect, this WHOLE
fcp-101 is up for discussion and shaping.  Right here above is an example
of one thing that needs to be corrected in the FSP, the criteria
is incorrectly stated if infact as "good faith intenet to upgrade
is fine."

I also saw another person state that the "5" user number appears
to be very arbitrary.  I agree.

We should NOT be taking the pole until the FCP itself is approved...
as altering the FCP could greatly effect the outcome of that pole.

--
Rod Grimes                                                 [hidden email]
_______________________________________________
[hidden email] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-arch
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: FCP-0101: Deprecating most 10/100 Ethernet drivers

Brooks Davis-2
In reply to this post by Warner Losh
On Wed, Oct 03, 2018 at 08:18:15PM -0600, Warner Losh wrote:

> On Wed, Oct 3, 2018 at 7:58 PM Rodney W. Grimes <
> [hidden email]> wrote:
>
> > > >>> Please direct replies to freebsd-arch <<<
> > >
> > > FCP-01010 (https://github.com/freebsd/fcp/blob/master/fcp-0101.md)
> > > outlines a plan to deprecate most 10/100 Ethernet drivers in FreeBSD 12
> > > and remove them in FreeBSD 13 to reduce the burden of maintaining and
> > > improving the network stack.  We have discussed this within the
> > > core team and intend to move forward as proposed.  We are solictiting
> >
> > Since when did a FCP become a core only process????  Why was this
> > dicussed within core and not within the larger community?
>
> Core hasn't approved this FCP, we're in the community discussion phase now.
> If you have substantive comments, please comment. It's a proposal, and it's
> being discussed now.
As Warner says, this FCP isn't approved.  Core's discussion centers on
two points.  First, we'd like to encourage the use the FCP process to
solicit feedback and document decisions.  That means using it so we can
grind the sharp edges off and understand what works and doesn't work.
Second, core believes that we support too much stuff and do a bad job of
removing old things.  Threads like this are a way to get more data prior
to adding deprecation annotations (another way to collect data).

Thus far we've found some that clearly should be on the exception list
which is exactly as expected.

-- Brooks

signature.asc (465 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

RE: FCP-0101: Deprecating most 10/100 Ethernet drivers

Cy Schubert-4
In reply to this post by Robert Clausecker
I'm willing to help out with rl(4) as I have one here. Others, not scheduled for removal, that I can help one way or another are are NICs, including wireless, currently installed here.

---
Sent using a tiny phone keyboard.
Apologies for any typos and autocorrect.
Also, this old phone only supports top post. Apologies.

Cy Schubert
<[hidden email]> or <[hidden email]>
The need of the many outweighs the greed of the few.
---

-----Original Message-----
From: Poul-Henning Kamp
Sent: 04/10/2018 07:03
To: Warner Losh
Cc: Alexey Dokuchaev; Brooks Davis; FreeBSD-STABLE Mailing List; FreeBSD Net; [hidden email]; [hidden email]
Subject: Re: FCP-0101: Deprecating most 10/100 Ethernet drivers

--------
In message <[hidden email]>
, Warner Losh writes:

>Most of these drivers have had dozens or hundreds of commits each over the
>years to keep up with the API changes. This acts as a tax on innovation
>because it's such a pain in the back side to change all the drivers in the
>tree.

As one who has been there, a couple of times: SECONDED!

It is particular unpleasant when you have no way to test the changes.

--
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
[hidden email]         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.
_______________________________________________
[hidden email] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-arch
To unsubscribe, send any mail to "[hidden email]"

_______________________________________________
[hidden email] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-arch
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: FCP-0101: Deprecating most 10/100 Ethernet drivers

Warner Losh
In reply to this post by Rodney W. Grimes-4
On Thu, Oct 4, 2018 at 9:35 AM Rodney W. Grimes <
[hidden email]> wrote:

> > On Thu, Oct 04, 2018 at 05:07:20PM +0200, Joel Dahl wrote:
> > > On Wed, Oct 03, 2018 at 09:05:16PM +0000, Brooks Davis wrote:
> > > > The criteria for exception are:
> > > >  - Popular in applications where it is likely to be deployed beyond
> the
> > > >    support lifetime of FreeBSD 12 (late 2023).
> > > >    - 5 reports of uses in the wild on machines running FreeBSD 12
> will be
> > > >      deemed satisfy the "popular"
> > > >      requirement.
> > >
> > > Why doesn't reports of uses on machines running FreeBSD 10/11 count? I
> don't
> > > get it. 12.0 isn't even out yet, and most of our users are probably not
> > > running CURRENT. As I wrote in an earlier email, I have lots of these
> cards
> > > running in production - and most of them are on FreeBSD 11. They'll
> > > likely be upgraded to 12.1 in the future (but probably not 12.0 - I
> usually
> > > skip .0 releases). But doing the jump to CURRENT/12 now is just out of
> the
> > > question - these are production systems after all.
> >
> > For the current poll, good faith intent to upgrade is fine.
>
> What I am finding very bothersome at this point is that a great
> miss understanding has been conveyed onto the users by the
> statement that "core has discussed this and we plan to proceed
> as proposed"
>
> From a posting by Warner that statement is incorrect, this WHOLE
> fcp-101 is up for discussion and shaping.


For the record, I never said anything to the contrary. Stop putting words
in my mouth. It's not helpful. I said it was in the community feedback
phase. That's part of the process: changing things as the community gives
feedback.


> Right here above is an example
> of one thing that needs to be corrected in the FSP, the criteria
> is incorrectly stated if infact as "good faith intenet to upgrade
> is fine."
>

That's part of the community feedback process. We add things, we adjust
things. I never once said anything to the contrary in this thread.


> I also saw another person state that the "5" user number appears
> to be very arbitrary.  I agree.
>

It's totally arbitrary. What's your point? We have to start somewhere, and
so far the data is splitting nicely between 0 or 1 users and > 5 if my
counts are correct. It appears, so far, to be a useful first order sorting
function.


> We should NOT be taking the pole until the FCP itself is approved...
> as altering the FCP could greatly effect the outcome of that pole.
>

I disagree. We can run the two in parallel unless we hit something major.
So far, I've seen nothing that suggests the polling done so far is invalid.

Warner
_______________________________________________
[hidden email] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-arch
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

RE: FCP-0101: Deprecating most 10/100 Ethernet drivers

Cy Schubert-4
In reply to this post by Robert Clausecker
Not that I'm arguing to keep ed(4), I have three ne2000 PCI cards in my desk here. Sure one could insert one in a current machine, but why?

---
Sent using a tiny phone keyboard.
Apologies for any typos and autocorrect.
Also, this old phone only supports top post. Apologies.

Cy Schubert
<[hidden email]> or <[hidden email]>
The need of the many outweighs the greed of the few.
---

-----Original Message-----
From: Warner Losh
Sent: 04/10/2018 07:45
To: [hidden email]
Cc: [hidden email]
Subject: Re: FCP-0101: Deprecating most 10/100 Ethernet drivers

Well, I'd wager its 100 cards that nobody is currently using (ed) has no
value. Even 1000 different cards or 10,000. The ed(4) driver likely
supported in excess of 1000 different cards because so many people made
ne-2000 compatible cards... in the early 1990s. However, since nobody has
ISA or PC Card (not CardBus) interfaces anymore (those machines topped out
around 32-64MB, which FreeBSD no longer works well on), the benefit to the
project is quite low. Even the 'newer' PC Card versions that were 10/100
couldn't get more than about 10-12Mbps due to ISA/PC Card bus speed
limitations. The PCI versions were never popular (I had to hunt a bunch for
them 10 years ago when I was finishing up my activities on the driver for
my vast PC Card collection to find an example to test), and even it had
trouble beyond 20Mbps because it wasn't DMA'd. The ED driver was a solid
driver last time I tried it, but when I can plug in dozens of 100Mbps or
1Gbps cards into the same CardBus slot and those cost < $10 now, there's
very little return on programmer time to keeping this one going.

However, having said all that, if we can document 5 real users of this card
on machines running FreeBSD 12, it will meet the criteria for remaining,
just like any other driver.... So far we've found 0, while we have found
many other users of other drivers.

Warner

On Thu, Oct 4, 2018 at 8:27 AM Robert Clausecker <[hidden email]> wrote:

> I have a machine with FreeBSD 2.2.8 running with such an interface, but
> none with FreeBSD 12, so you do have a point here.  However, I am not
> sure if it's a good idea to kill this driver; it's good for over 100
> different cards according to the man page, so surely there are some
> users left.
>
> Yours,
> Robert Clausecker
>
> On Wed, Oct 03, 2018 at 05:45:18PM -0600, Warner Losh wrote:
> > On Wed, Oct 3, 2018 at 3:54 PM Robert Clausecker <[hidden email]> wrote:
> >
> > > I request that ed(4)
> >
> >
> > How many FreeBSD 12.0 machines do you have running with this interface?
> >
> > QEMU does support this interface, but also supports the Intel E1000
> series
> > (em/igb), so it's not necessarily needed for QEMU.
> >
> > Warner
>
> --
> ()  ascii ribbon campaign - for an 8-bit clean world
> /\  - against html email  - against proprietary attachments
> _______________________________________________
> [hidden email] mailing list
> https://lists.freebsd.org/mailman/listinfo/freebsd-arch
> To unsubscribe, send any mail to "[hidden email]"
>
_______________________________________________
[hidden email] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-arch
To unsubscribe, send any mail to "[hidden email]"

_______________________________________________
[hidden email] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-arch
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

RE: FCP-0101: Deprecating most 10/100 Ethernet drivers

Cy Schubert-4
In reply to this post by Robert Clausecker
People need to submit patches then. OTOH, they can all be moved to ports. IMO, when pkgbase becomes a reality, much of this will become moot. People will be able to mix and match base and ports packages.

---
Sent using a tiny phone keyboard.
Apologies for any typos and autocorrect.
Also, this old phone only supports top post. Apologies.

Cy Schubert
<[hidden email]> or <[hidden email]>
The need of the many outweighs the greed of the few.
---

-----Original Message-----
From: Alexey Dokuchaev
Sent: 04/10/2018 08:11
To: Warner Losh
Cc: Mark Linimon; [hidden email]; [hidden email]; FreeBSD-STABLE Mailing List; FreeBSD Net
Subject: Re: FCP-0101: Deprecating most 10/100 Ethernet drivers

On Thu, Oct 04, 2018 at 08:43:33AM -0600, Warner Losh wrote:
> As far as I know, none of the drivers listed could do 1Gbps.

Right.  My point was that original proposal put 10/100 drivers into one
basket, which is IMHO not fair: 10Mbps cards are rarely seen and used,
100mbps are not, just like 1000bps ones.

That said, I'm okay with deorbiting NICs that cannot do more than 10mbps.
Cards that can do at least 100mbps should stay.  Following up on Ricks'
question, seeing a good example of modernization a certain driver would
help interested people/hw owners to keep drivers for their cards viable.

./danfe
_______________________________________________
[hidden email] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-arch
To unsubscribe, send any mail to "[hidden email]"

_______________________________________________
[hidden email] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-arch
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: FCP-0101: Deprecating most 10/100 Ethernet drivers

Warner Losh
In reply to this post by Robert Clausecker
On Thu, Oct 4, 2018 at 9:59 AM Rodney W. Grimes <
[hidden email]> wrote:

> > On Thu, Oct 4, 2018 at 9:46 AM Cy Schubert <[hidden email]>
> > wrote:
> >
> > > I'm willing to help out with rl(4) as I have one here. Others, not
> > > scheduled for removal, that I can help one way or another are are NICs,
> > > including wireless, currently installed here.
> > >
> >
> > There's an iflib man page that's a decent place to start. The API has
> > evolved over time, so corrections to the man page would be welcome (and
> > committed as quickly as the freeze allows). I'm reading through the
> current
> > iflib drivers to see which one would be best to recommend.
>
> Nothing in the current state of the "freeze" would block a
> man page correction.
>

All commits, no matter how trivial, require re@ approval. That necessarily
slows things down, hence my phrase "as quickly as the freeze allows."

Warner
_______________________________________________
[hidden email] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-arch
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: FCP-0101: Deprecating most 10/100 Ethernet drivers

Brooks Davis-2
In reply to this post by Robert Clausecker
>>> Please direct replies to freebsd-arch <<<

A few points of clarification:

Rod correctly points out that this message makes it look like the FCP is
a done deal as written.  This is not the case and we welcome feedback
on the entire proposal.  IMO, soliciting input on the list of drivers
along with the proposed process is a way to keep discussion concrete so
we will proceed with both.

It was asked: when does iflib conversion need to occur to save a driver?
My proposed plan it to proceed with deprecation notices of otherwise
unpopular drivers, but conversion can come in and remove those notices
at and upto (or even after) removal from the tree.

In an effort to save some email, we will be moving rl(4) to the list of
drivers to STAY as it has proved itself to be popular.  A few others
appear to be well on their way so keep the reports coming.

Thanks,
Brooks

P.S. As a person who has edited every driver in the tree multiple times
in the last year (mostly in an external tree), I will consider this
process successful even if we keep the majority of listed drivers in the
tree.

On Wed, Oct 03, 2018 at 09:05:16PM +0000, Brooks Davis wrote:

> >>> Please direct replies to freebsd-arch <<<
>
> FCP-01010 (https://github.com/freebsd/fcp/blob/master/fcp-0101.md)
> outlines a plan to deprecate most 10/100 Ethernet drivers in FreeBSD 12
> and remove them in FreeBSD 13 to reduce the burden of maintaining and
> improving the network stack.  We have discussed this within the
> core team and intend to move forward as proposed.  We are solictiting
> feedback on the list of drivers to be excepted from removal.
>
> The current list of drivers slated for REMOVAL is:
>
> ae, bfe, bm, cs, dme, ed, ep, ex, fe, pcn, rl, sf, smc, sn,
> ste, tl, tx, txp, vx, wb, xe
>
> The current list of drivers that will STAY in the tree is:
>
> dc, ffec, fxpl, hme, le, sis, vr, xl
>
> The criteria for exception are:
>  - Popular in applications where it is likely to be deployed beyond the
>    support lifetime of FreeBSD 12 (late 2023).
>    - 5 reports of uses in the wild on machines running FreeBSD 12 will be
>      deemed satisfy the "popular"
>      requirement.
>  - Required to make a well supported embedded or emulation platform usable.
>  - Ported to use iflib (reducing future maintenance cost.)
>
> Please reply to this message with nominations to the exception list.
>
> The full FCP-0101 is included below.
>
> -- Brooks
>
> ---
> authors: Brooks Davis <[hidden email]>
> state: feedback
> ---
>
> # FCP 101: Deprecation and removal of 10/100 Ethernet drivers
>
> Deprecate most 10 and 10/100Mbps Ethernet drivers and remove them before
> FreeBSD 13.
>
> ## Problem Statement
>
> Each network driver creates drag for the project as we attempt to
> improve the network stack or provide new features such as expanded
> 32-bit compatibility.  For example, the author has edited every single
> NIC driver more than once in the past year to update management (`ioctl`)
> interfaces.  We could improve this situation by converting drivers to
> iflib, but each additional driver takes work.
>
> 10 and 100 megabit Ethernet drivers are largely irrelevant today
> and we have a significant number of them in the tree.  The ones that
> are no longer used and/or are not known to be working need to be
> removed due to the significant ongoing 'tax' on new development.
>
> For at least a decade, most systems (including small embedded
> systems) have shipped with gigabit Ethernet devices and virtual
> machines commonly emulate popular gigabit devices.  We wish to
> retain support for popular physical and virtual devices while
> removing support for uncommon ones.  With a few exceptions these
> drivers are unlikely to be used by our user base by the time FreeBSD
> 12 is obsolete (approximately 2024).
>
> ## Proposed Solution
>
> We propose to deprecate devices which are not sufficiently popular.  This
> will entail:
>  - (October 2018) Send this list to freebsd-net and freebsd-stable.
>  - (Before FreeBSD 12.0-RELEASE - October 2018) Update the manpages and
>    attach routines for each device to be removed and merge those changes
>    to FreeBSD 12.
>  - (One month after FreeBSD 12.0-RELEASE - January 2018) Remind
>    freebsd-net and freebsd-stable users of pending deletion.
>  - (Two months after FreeBSD 12.0-RELEASE - February 2019) Delete deprecated
>    devices.
>
> Through out this process, solicit feedback on additions to the exception
> list and update this document as required.  For a device to be placed on
> the exception list the device must meet one of the following criteria:
>  - Popular in applications where it is likely to be deployed beyond the
>    support lifetime of FreeBSD 12 (late 2023).
>    - 5 reports of uses in the wild on machines running FreeBSD 12 will be
>      deemed satisfy the "popular"
>      requirement.
>  - Required to make a well supported embedded or emulation platform usable.
>  - Ported to use iflib (reducing future maintenance cost.)
>
> ### Exceptions to removal
>
> Device | Reason
> -------|-------------------------------------------------
> ffec   | Onboard Ethernet for Vybrid arm7 boards
> fxp    | Popular device long recommended by the project.
> dc     | Popular device for CardBus card.
> hme    | Built in interface on many supported sparc64 platforms.
> le     | Emulated by QEMU, alternatives don't yet work for mips64.
> sis    | Soekris Engineering net45xx, net48xx, lan1621, and lan1641.
> vr     | Soekris Engineering net5501, some Asus motherboards.
> xl     | Popular device for CardBus card.
>
> Note: USB devices have been excluded from consideration in this round.
>
> ### Device to be removed
>
> ae, bfe, bm, cs, dme, ed, ep, ex, fe, pcn, rl, sf, smc, sn,
> ste, tl, tx, txp, vx, wb, xe
>
> ## Final Disposition
>
> TBD


signature.asc (465 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: FCP-0101: Deprecating most 10/100 Ethernet drivers

Brooks Davis-2
In reply to this post by Warner Losh
On Thu, Oct 04, 2018 at 11:21:03AM -0500, Kevin Day wrote:

>
>
> > On Oct 4, 2018, at 7:58 AM, Warner Losh <[hidden email]> wrote:
> >
> > On Thu, Oct 4, 2018 at 6:53 AM Luciano Mannucci <[hidden email]>
> > wrote:
> >
> >>> While I have no objection for general direction, I have doubts about
> >> removal
> >>> of ste(4) and especially rl(4). These are cheap 100Mbit VERY popular NICs
> >>> sold in enourmous values in certain markets by vendors like D-Link and
> >>> TP-Link using various trade names.
> >>
> >
> > Enough people are using rl that it's off the list by my count.
>
> If you need any extra data, rl is the only one that I'd be sad to lose as well. We have a lot of embedded devices that are still shipping today with rl chips. I could chip in a little to sponsor someone to modernize the driver if needed.
It's definitly on the STAY list at this point.  If you could help get it
update that would be great, especially if it's still shipping (a very
useful datapoint).

-- Brooks

signature.asc (465 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: FCP-0101: Deprecating most 10/100 Ethernet drivers

Bakul Shah
In reply to this post by Robert Clausecker
On Wed, 03 Oct 2018 21:05:16 -0000 Brooks Davis <[hidden email]> wrote:
>
> The current list of drivers slated for REMOVAL is:
>
> ae, bfe, bm, cs, dme, ed, ep, ex, fe, pcn, rl, sf, smc, sn,
> ste, tl, tx, txp, vx, wb, xe
>
> The current list of drivers that will STAY in the tree is:
>
> dc, ffec, fxpl, hme, le, sis, vr, xl

What is the disposition of drivers not on either list?

> 10 and 100 megabit Ethernet drivers are largely irrelevant today
> and we have a significant number of them in the tree.  The ones that
> are no longer used and/or are not known to be working need to be
> removed due to the significant ongoing 'tax' on new development.

I don't understand why there is a "significant ongoing 'tax'
on new development" for old NICs.  Can the internal MI<->MD
interface be evolved in the direction where the MD drivers for
old h/w "just work"? Or is it a hopeless task?
_______________________________________________
[hidden email] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-arch
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: FCP-0101: Deprecating most 10/100 Ethernet drivers

Ian Lepore-3
In reply to this post by tech-lists
On Thu, 2018-10-04 at 10:21 -0600, Warner Losh wrote:

> On Thu, Oct 4, 2018 at 10:15 AM Michelle Sullivan <[hidden email]>
> wrote:
>
> >
> > tech-lists wrote:
> > >
> > >
> > > I'm astonished you're considering removing rl given how common it is.
> > >
> > I'll second that comment - though no disrespect to Brooks.  Brooks as
> > far as I can see is just the messenger.
> >
> Absent good data, one has to make one's best guesses. I guessed wrong here
> in my comments to Brooks about which ones were must keeps. I knew it was
> popular back in the day (~2000), but had thought it's popularity had waned
> much more than it apparently has. I last deployed systems with rl in them
> around 2007, and at the time it was trailing edge gear (the SBCs we used at
> Timing Solutions tended to use popular, but ~5-year-old technology because
> that market segment wanted longevity of spare availability...).
>
> Warner

11 years later, we (Timing Solutions, now a division of Microchip) are
still using SBCs with rl(4) hardware and still shipping software
updates with that driver built into the kernel. We build systems with a
lifespan in the field of 20 years or more, and the stability and
compatibility across OS upgrades over that kind of span is a BIG reason
to use freebsd rather than linux for such things.

-- Ian

_______________________________________________
[hidden email] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-arch
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: FCP-0101: Deprecating most 10/100 Ethernet drivers

Warner Losh
In reply to this post by Bakul Shah
On Thu, Oct 4, 2018 at 11:25 AM Bakul Shah <[hidden email]> wrote:

> On Wed, 03 Oct 2018 21:05:16 -0000 Brooks Davis <[hidden email]>
> wrote:
> >
> > The current list of drivers slated for REMOVAL is:
> >
> > ae, bfe, bm, cs, dme, ed, ep, ex, fe, pcn, rl, sf, smc, sn,
> > ste, tl, tx, txp, vx, wb, xe
> >
> > The current list of drivers that will STAY in the tree is:
> >
> > dc, ffec, fxpl, hme, le, sis, vr, xl
>
> What is the disposition of drivers not on either list?
>

Apart from de, what are they?


> > 10 and 100 megabit Ethernet drivers are largely irrelevant today
> > and we have a significant number of them in the tree.  The ones that
> > are no longer used and/or are not known to be working need to be
> > removed due to the significant ongoing 'tax' on new development.
>
> I don't understand why there is a "significant ongoing 'tax'
> on new development" for old NICs.  Can the internal MI<->MD
> interface be evolved in the direction where the MD drivers for
> old h/w "just work"? Or is it a hopeless task?
>

There's two problems. One is that the current APIs are very much setup for
cut and paste driver construction. This leads to many drivers needing to be
changed more often than necessary as the APIs are evolved. The second is
the nature of the hardware has changed. We've gone from devices that can
handle at most a single packet at the same time to drives that can handle
thousands with some of the TCP stack offloaded into the card. This wide
range of hardware is difficult to program for with the current stack. iflib
is supposed to help (which is the MI/MD thing you're talking about), but in
the end it can likely help only so much before support for old cards holds
back adaptation of new features for new cards. Taken together, the old NICs
in the tree represent a real burden to people trying to innovate (or even
just bug fix) in this area. Add to that the inability to actually test the
hardware in any meaningful way, and you have a situation that needs to
change.

Warner
_______________________________________________
[hidden email] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-arch
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: FCP-0101: Deprecating most 10/100 Ethernet drivers

Brooks Davis-2
In reply to this post by Bakul Shah
On Thu, Oct 04, 2018 at 10:24:13AM -0700, Bakul Shah wrote:

> On Wed, 03 Oct 2018 21:05:16 -0000 Brooks Davis <[hidden email]> wrote:
> >
> > The current list of drivers slated for REMOVAL is:
> >
> > ae, bfe, bm, cs, dme, ed, ep, ex, fe, pcn, rl, sf, smc, sn,
> > ste, tl, tx, txp, vx, wb, xe
> >
> > The current list of drivers that will STAY in the tree is:
> >
> > dc, ffec, fxpl, hme, le, sis, vr, xl
>
> What is the disposition of drivers not on either list?
They weren't considered and nothing changes unless someone points them
and proposes some action.  The document points out that USB devices
were skipped.  Not mentioned were NICs tied to specific architectures.

> > 10 and 100 megabit Ethernet drivers are largely irrelevant today
> > and we have a significant number of them in the tree.  The ones that
> > are no longer used and/or are not known to be working need to be
> > removed due to the significant ongoing 'tax' on new development.
>
> I don't understand why there is a "significant ongoing 'tax'
> on new development" for old NICs.  Can the internal MI<->MD
> interface be evolved in the direction where the MD drivers for
> old h/w "just work"? Or is it a hopeless task?

I've touched every single Ethernet driver by hand multiple times in
that past year in our research tree.  We'll never know how much
modernization isn't being done because it's a pain.  iflib does reduce
this cost, but conversion isn't trivial.  We should work to migrate
drivers that are used and stop wasting time on ones that aren't.

-- Brooks

signature.asc (465 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: FCP-0101: Deprecating most 10/100 Ethernet drivers

Ian Lepore-3
In reply to this post by Ian Lepore-3
On Thu, 2018-10-04 at 11:38 -0600, Warner Losh wrote:

> On Thu, Oct 4, 2018 at 11:26 AM Ian Lepore <[hidden email]> wrote:
>
> >
> > On Thu, 2018-10-04 at 10:21 -0600, Warner Losh wrote:
> > >
> > > On Thu, Oct 4, 2018 at 10:15 AM Michelle Sullivan
> > > wrote:
> > >
> > > >
> > > >
> > > > tech-lists wrote:
> > > > >
> > > > >
> > > > >
> > > > > I'm astonished you're considering removing rl given how common it is.
> > > > >
> > > > I'll second that comment - though no disrespect to Brooks.  Brooks as
> > > > far as I can see is just the messenger.
> > > >
> > > Absent good data, one has to make one's best guesses. I guessed wrong
> > here
> > >
> > > in my comments to Brooks about which ones were must keeps. I knew it was
> > > popular back in the day (~2000), but had thought it's popularity had
> > waned
> > >
> > > much more than it apparently has. I last deployed systems with rl in them
> > > around 2007, and at the time it was trailing edge gear (the SBCs we used
> > at
> > >
> > > Timing Solutions tended to use popular, but ~5-year-old technology
> > because
> > >
> > > that market segment wanted longevity of spare availability...).
> > >
> > > Warner
> > 11 years later, we (Timing Solutions, now a division of Microchip) are
> > still using SBCs with rl(4) hardware and still shipping software
> > updates with that driver built into the kernel. We build systems with a
> > lifespan in the field of 20 years or more, and the stability and
> > compatibility across OS upgrades over that kind of span is a BIG reason
> > to use freebsd rather than linux for such things.
> >
> OK. I'd have thought those SBCs would have gone out of production years
> ago.... It's a good datapoint to know that there's multiple users of
> FreeBSD using these parts in products that are still shipping. That's a
> clear and compelling benefit to the project that offsets the efforts that
> it's taken them to keep things current with rl.
>
> In this case, though, rl is off the list, so that hardware should still be
> good. The only other SBC I was aware of at Timing Solutions was one that
> had an 'ed' chip on it (an ISA realtek part IIRC) that was used in around
> 2001, but in a 'one off' custom setup that I don't think will ever be
> upgraded.... But I have to ask since I know how things worked during my
> time there and systems that 'would never be upgraded' often times were
> later...
>
> I'd also suggest that rl stands in stark contrast to the cs, wb, sn, smc,
> sf, tl, tx and vr drivers, which nobody has mentioned in this thread, and
> which I doubt are in use in any FreeBSD system of any age today.
>
> Warner

I checked all our various kernel configs, and the only one on the list
we still use appears to be rl.

One driver I was surprised to see was not on the list was vte. So I'll
just preemptively mention that we do use that one too.

-- Ian
_______________________________________________
[hidden email] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-arch
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: FCP-0101: Deprecating most 10/100 Ethernet drivers

Bakul Shah
In reply to this post by Warner Losh
On Thu, 04 Oct 2018 11:30:35 -0600 Warner Losh <[hidden email]> wrote:

>
> On Thu, Oct 4, 2018 at 11:25 AM Bakul Shah <[hidden email]> wrote:
>
> > On Wed, 03 Oct 2018 21:05:16 -0000 Brooks Davis <[hidden email]>
> > wrote:
> > >
> > > The current list of drivers slated for REMOVAL is:
> > >
> > > ae, bfe, bm, cs, dme, ed, ep, ex, fe, pcn, rl, sf, smc, sn,
> > > ste, tl, tx, txp, vx, wb, xe
> > >
> > > The current list of drivers that will STAY in the tree is:
> > >
> > > dc, ffec, fxpl, hme, le, sis, vr, xl
> >
> > What is the disposition of drivers not on either list?
> >
>
> Apart from de, what are they?

I have re on on the motherboard on one machine.

Granted this is a very dumb test but...

cd /usr/src/dev
for a in *; do if [ -e $a/if_$a.c ] ; then echo $a; fi; done | wc -l
      86

> > > 10 and 100 megabit Ethernet drivers are largely irrelevant today
> > > and we have a significant number of them in the tree.  The ones that
> > > are no longer used and/or are not known to be working need to be
> > > removed due to the significant ongoing 'tax' on new development.
> >
> > I don't understand why there is a "significant ongoing 'tax'
> > on new development" for old NICs.  Can the internal MI<->MD
> > interface be evolved in the direction where the MD drivers for
> > old h/w "just work"? Or is it a hopeless task?
> >
>
> There's two problems. One is that the current APIs are very much setup for
> cut and paste driver construction. This leads to many drivers needing to be
> changed more often than necessary as the APIs are evolved. The second is
> the nature of the hardware has changed. We've gone from devices that can
> handle at most a single packet at the same time to drives that can handle
> thousands with some of the TCP stack offloaded into the card. This wide
> range of hardware is difficult to program for with the current stack. iflib
> is supposed to help (which is the MI/MD thing you're talking about), but in
> the end it can likely help only so much before support for old cards holds
> back adaptation of new features for new cards. Taken together, the old NICs
> in the tree represent a real burden to people trying to innovate (or even
> just bug fix) in this area. Add to that the inability to actually test the
> hardware in any meaningful way, and you have a situation that needs to
> change.

Thanks for the explanation.
_______________________________________________
[hidden email] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-arch
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: FCP-0101: Deprecating most 10/100 Ethernet drivers

Ian Lepore-3
In reply to this post by Ian Lepore-3
On Thu, 2018-10-04 at 11:58 -0600, Warner Losh wrote:

> On Thu, Oct 4, 2018 at 11:47 AM Ian Lepore <[hidden email]> wrote:
>
> >
> > On Thu, 2018-10-04 at 11:38 -0600, Warner Losh wrote:
> > >
> > > On Thu, Oct 4, 2018 at 11:26 AM Ian Lepore <[hidden email]>
> > > wrote:
> > >
> > > >
> > > >
> > > > On Thu, 2018-10-04 at 10:21 -0600, Warner Losh wrote:
> > > > >
> > > > >
> > > > > On Thu, Oct 4, 2018 at 10:15 AM Michelle Sullivan
> > > > > wrote:
> > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > > tech-lists wrote:
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > > I'm astonished you're considering removing rl given how
> > > > > > > common
> > it is.
> > >
> > > >
> > > > >
> > > > > >
> > > > > > >
> > > > > > >
> > > > > > I'll second that comment - though no disrespect to
> > > > > > Brooks.  Brooks
> > as
> > >
> > > >
> > > > >
> > > > > >
> > > > > > far as I can see is just the messenger.
> > > > > >
> > > > > Absent good data, one has to make one's best guesses. I
> > > > > guessed wrong
> > > > here
> > > > >
> > > > >
> > > > > in my comments to Brooks about which ones were must keeps. I
> > > > > knew it
> > was
> > >
> > > >
> > > > >
> > > > > popular back in the day (~2000), but had thought it's
> > > > > popularity had
> > > > waned
> > > > >
> > > > >
> > > > > much more than it apparently has. I last deployed systems
> > > > > with rl in
> > them
> > >
> > > >
> > > > >
> > > > > around 2007, and at the time it was trailing edge gear (the
> > > > > SBCs we
> > used
> > >
> > > >
> > > > at
> > > > >
> > > > >
> > > > > Timing Solutions tended to use popular, but ~5-year-old
> > > > > technology
> > > > because
> > > > >
> > > > >
> > > > > that market segment wanted longevity of spare
> > > > > availability...).
> > > > >
> > > > > Warner
> > > > 11 years later, we (Timing Solutions, now a division of
> > > > Microchip) are
> > > > still using SBCs with rl(4) hardware and still shipping
> > > > software
> > > > updates with that driver built into the kernel. We build
> > > > systems with a
> > > > lifespan in the field of 20 years or more, and the stability
> > > > and
> > > > compatibility across OS upgrades over that kind of span is a
> > > > BIG reason
> > > > to use freebsd rather than linux for such things.
> > > >
> > > OK. I'd have thought those SBCs would have gone out of production
> > > years
> > > ago.... It's a good datapoint to know that there's multiple users
> > > of
> > > FreeBSD using these parts in products that are still shipping.
> > > That's a
> > > clear and compelling benefit to the project that offsets the
> > > efforts that
> > > it's taken them to keep things current with rl.
> > >
> > > In this case, though, rl is off the list, so that hardware should
> > > still
> > be
> > >
> > > good. The only other SBC I was aware of at Timing Solutions was
> > > one that
> > > had an 'ed' chip on it (an ISA realtek part IIRC) that was used
> > > in around
> > > 2001, but in a 'one off' custom setup that I don't think will
> > > ever be
> > > upgraded.... But I have to ask since I know how things worked
> > > during my
> > > time there and systems that 'would never be upgraded' often times
> > > were
> > > later...
> > >
> > > I'd also suggest that rl stands in stark contrast to the cs, wb,
> > > sn, smc,
> > > sf, tl, tx and vr drivers, which nobody has mentioned in this
> > > thread, and
> > > which I doubt are in use in any FreeBSD system of any age today.
> > >
> > > Warner
> > I checked all our various kernel configs, and the only one on the
> > list
> > we still use appears to be rl.
> >
> > One driver I was surprised to see was not on the list was vte. So
> > I'll
> > just preemptively mention that we do use that one too.
> >
> I'll assume that you've deployed more than 5 of these systems and
> that you
> may someday upgrade them as well?  Which of the Vortex86 processors
> are you
> using, if you can answer that...
>
> Warner

It's a DM&P Vortex86DX on a PCA-6743 board, which you can still buy.

32-bit only, BTW, which is why I hate hearing recent mumblings about
discarding 32-bit x86 support in freebsd.

-- Ian
_______________________________________________
[hidden email] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-arch
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: FCP-0101: Deprecating most 10/100 Ethernet drivers

Eugene Grosbein-10
In reply to this post by Ian Lepore-3
05.10.2018 0:38, Warner Losh wrote:

> I'd also suggest that rl stands in stark contrast to the cs, wb, sn, smc,
> sf, tl, tx and vr drivers, which nobody has mentioned in this thread, and
> which I doubt are in use in any FreeBSD system of any age today.

vr(4) mentioned in the STAY list or else I would be first yelling as
it is still very common in embedded solutions (including integrated ports in my home router).


_______________________________________________
[hidden email] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-arch
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: FCP-0101: Deprecating most 10/100 Ethernet drivers

Eugene Grosbein-10
In reply to this post by Brooks Davis-2
04.10.2018 23:22, Brooks Davis wrote:

> In an effort to save some email, we will be moving rl(4) to the list of
> drivers to STAY as it has proved itself to be popular.  A few others
> appear to be well on their way so keep the reports coming.

And ste(4) please, as these are hardly replaceable two- and four-ports cards.
In many cases it is impossible to replace them without replacement of whole boxes
that have no extra PCI slots.


_______________________________________________
[hidden email] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-arch
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: FCP-0101: Deprecating most 10/100 Ethernet drivers

Poul-Henning Kamp
In reply to this post by Robert Clausecker
--------

>FCP-01010 (https://github.com/freebsd/fcp/blob/master/fcp-0101.md)

Can I open a FCP to rename FCP to FBS (for FreeBSD BikeShed) ?

Guys... most if not all of these emails could have been sent to
directly Brooks without Cc'ing four mailing lists.

Then Brooks could revise his tallies and scores to match informed
reality and _then_ we could discuss if the criteria were sound
on the list(s).

Poul-Henning (singing an almost 20 year old refrain again)

--
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
[hidden email]         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.
_______________________________________________
[hidden email] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-arch
To unsubscribe, send any mail to "[hidden email]"
123456