Date: Fri, 29 Mar 2013 10:42:48 -0600 From: Scott Long <scott4long@yahoo.com> To: Barney Cordoba <barney_cordoba@yahoo.com> Cc: Nick Rogers <ncrogers@gmail.com>, Adrian Chadd <adrian@freebsd.org>, Jeffrey EPieper <jeffrey.e.pieper@intel.com>, "freebsd-net@freebsd.org" <freebsd-net@freebsd.org>, "Clement Hermann \(nodens\)" <nodens2099@gmail.com>, Jack Vogel <jfvogel@gmail.com> Subject: Re: igb and ALTQ in 9.1-rc3 Message-ID: <F024F568-0749-4132-9AB1-5010ED531B04@yahoo.com> In-Reply-To: <1364575092.74048.YahooMailClassic@web121604.mail.ne1.yahoo.com> References: <1364575092.74048.YahooMailClassic@web121604.mail.ne1.yahoo.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Comedy gold. It's been a while since I've seen this much idiocy from = you, Barney. Hopefully the rest of the mailing list will blackhole you, = as I'm about to, and we can all get back to real work. Scott On Mar 29, 2013, at 10:38 AM, Barney Cordoba <barney_cordoba@yahoo.com> = wrote: > it needs a lot more than a patch. It needs to be completely re-thunk >=20 > --- On Fri, 3/29/13, Adrian Chadd <adrian@freebsd.org> wrote: >=20 > From: Adrian Chadd <adrian@freebsd.org> > Subject: Re: igb and ALTQ in 9.1-rc3 > To: "Barney Cordoba" <barney_cordoba@yahoo.com> > Cc: "Jack Vogel" <jfvogel@gmail.com>, "Nick Rogers" = <ncrogers@gmail.com>, "Jeffrey EPieper" <jeffrey.e.pieper@intel.com>, = "freebsd-net@freebsd.org" <freebsd-net@freebsd.org>, "Clement Hermann = (nodens)" <nodens2099@gmail.com> > Date: Friday, March 29, 2013, 12:07 PM >=20 > Barney, > Patches gratefully accepted. >=20 >=20 >=20 > Adrian >=20 >=20 >=20 > On 29 March 2013 08:54, Barney Cordoba <barney_cordoba@yahoo.com> = wrote: >=20 >=20 >=20 >=20 >=20 > --- On Fri, 3/29/13, Pieper, Jeffrey E <jeffrey.e.pieper@intel.com> = wrote: >=20 >=20 >=20 >> From: Pieper, Jeffrey E <jeffrey.e.pieper@intel.com> >=20 >> Subject: RE: igb and ALTQ in 9.1-rc3 >=20 >> To: "Barney Cordoba" <barney_cordoba@yahoo.com>, "Jack Vogel" = <jfvogel@gmail.com>, "Nick Rogers" <ncrogers@gmail.com> >=20 >=20 >> Cc: "freebsd-net@freebsd.org" <freebsd-net@freebsd.org>, "Clement = Hermann (nodens)" <nodens2099@gmail.com> >=20 >=20 >> Date: Friday, March 29, 2013, 11:45 AM >=20 >>=20 >=20 >>=20 >=20 >> -----Original Message----- >=20 >> From: owner-freebsd-net@freebsd.org >=20 >> [mailto:owner-freebsd-net@freebsd.org] >=20 >> On Behalf Of Barney Cordoba >=20 >> Sent: Friday, March 29, 2013 5:51 AM >=20 >> To: Jack Vogel; Nick Rogers >=20 >> Cc: freebsd-net@freebsd.org; >=20 >> Clement Hermann (nodens) >=20 >> Subject: Re: igb and ALTQ in 9.1-rc3 >=20 >>=20 >=20 >>=20 >=20 >>=20 >=20 >> --- On Thu, 3/28/13, Nick Rogers <ncrogers@gmail.com> >=20 >> wrote: >=20 >>=20 >=20 >>> From: Nick Rogers <ncrogers@gmail.com> >=20 >>> Subject: Re: igb and ALTQ in 9.1-rc3 >=20 >>> To: "Jack Vogel" <jfvogel@gmail.com> >=20 >>> Cc: "Barney Cordoba" <barney_cordoba@yahoo.com>, >=20 >> "Clement Hermann (nodens)" <nodens2099@gmail.com>, >=20 >> "freebsd-net@freebsd.org" >=20 >> <freebsd-net@freebsd.org> >=20 >>> Date: Thursday, March 28, 2013, 9:29 PM >=20 >>> On Thu, Mar 28, 2013 at 4:16 PM, Jack >=20 >>> Vogel <jfvogel@gmail.com> >=20 >>> wrote: >=20 >>>> Have been kept fairly busy with other matters, >=20 >> one >=20 >>> thing I could do short >=20 >>>> term is >=20 >>>> change the defines in igb the way I did in the em >=20 >>> driver so you could still >=20 >>>> define >=20 >>>> the older if_start entry. Right now those are >=20 >> based on >=20 >>> OS version and so you >=20 >>>> will >=20 >>>> automatically get if_transmit, but I could change >=20 >> it to >=20 >>> be IGB_LEGACY_TX or >=20 >>>> so, >=20 >>>> and that could be defined in the Makefile. >=20 >>>>=20 >=20 >>>> Would this help? >=20 >>>=20 >=20 >>> I'm currently using ALTQ successfully with the em >=20 >> driver, so >=20 >>> if igb >=20 >>> behaved the same with respect to using if_start instead >=20 >> of >=20 >>> if_transmit >=20 >>> when ALTQ is in play, that would be great. I do not >=20 >>> completely >=20 >>> understand the change you propose as I am not very >=20 >> familiar >=20 >>> with the >=20 >>> driver internals. Any kind of patch or extra >=20 >>> Makefile/make.conf >=20 >>> definition that would allow me to build a 9-STABLE >=20 >> kernel >=20 >>> with an igb >=20 >>> driver that works again with ALTQ, ASAP, would be much >=20 >>> appreciated. >=20 >>>=20 >=20 >>>>=20 >=20 >>>> Jack >=20 >>>>=20 >=20 >>>>=20 >=20 >>>>=20 >=20 >>>> On Thu, Mar 28, 2013 at 2:31 PM, Nick Rogers >=20 >> <ncrogers@gmail.com> >=20 >>> wrote: >=20 >>>>>=20 >=20 >>>>> On Tue, Dec 11, 2012 at 1:09 AM, Jack Vogel >=20 >> <jfvogel@gmail.com> >=20 >>> wrote: >=20 >>>>>> On Mon, Dec 10, 2012 at 11:58 PM, Gleb >=20 >>> Smirnoff <glebius@freebsd.org> >=20 >>>>>> wrote: >=20 >>>>>>=20 >=20 >>>>>>> On Mon, Dec 10, 2012 at 03:31:19PM >=20 >> -0800, >=20 >>> Jack Vogel wrote: >=20 >>>>>>> J> UH, maybe asking the owner of >=20 >> the >=20 >>> driver would help :) >=20 >>>>>>> J> >=20 >>>>>>> J> ... and no, I've never been >=20 >> aware of >=20 >>> doing anything to stop >=20 >>>>>>> supporting >=20 >>>>>>> altq >=20 >>>>>>> J> so you wouldn't see any >=20 >> commits. If >=20 >>> there's something in the altq >=20 >>>>>>> code >=20 >>>>>>> or >=20 >>>>>>> J> support (which I have nothing >=20 >> to do >=20 >>> with) that caused this no-one >=20 >>>>>>> informed >=20 >>>>>>> J> me. >=20 >>>>>>>=20 >=20 >>>>>>> Switching from if_start to >=20 >> if_transmit >=20 >>> effectively disables ALTQ >=20 >>>>>>> support. >=20 >>>>>>>=20 >=20 >>>>>>> AFAIR, there is some magic >=20 >> implemented in >=20 >>> other drivers that makes them >=20 >>>>>>> modern (that means using >=20 >> if_transmit), but >=20 >>> still capable to switch to >=20 >>>>>>> queueing >=20 >>>>>>> mode if SIOCADDALTQ was casted upon >=20 >> them. >=20 >>>>>>>=20 >=20 >>>>>>>=20 >=20 >>>>>> Oh, hmmm, I'll look into the matter after >=20 >> my >=20 >>> vacation. >=20 >>>>>>=20 >=20 >>>>>> Jack >=20 >>>>>=20 >=20 >>>>> Has there been any progress on resolving this >=20 >>> issue? I recently ran >=20 >>>>> into this problem upgrading my servers from >=20 >> 8.3 to >=20 >>> 9.1-RELEASE and am >=20 >>>>> wondering what the latest recommendation is. >=20 >> I've >=20 >>> used ALTQ and igb >=20 >>>>> successfully for years and it is unfortunate >=20 >> it no >=20 >>> longer works. >=20 >>>>> Appreciate any advice. >=20 >>>>>=20 >=20 >>>=20 >=20 >>> Do yourself a favor and either get a cheap dual port >=20 >> 82571 card or >=20 >>> 2 cards and disable the IGB ports. The igb driver is >=20 >> defective, and until >=20 >>> they back out the new, untested multi-queue stuff you're >=20 >> just neutering >=20 >>> your system trying to use it. >=20 >>>=20 >=20 >>> Frankly this project made a huge mistake by moving >=20 >> forward with multi >=20 >>> queue just for the sake of saying that you support it; >=20 >> without having >=20 >>> any credible plan for implementing it. That nonsense >=20 >> that Bill Macy did >=20 >>> should have been tarballed up and deposited in the trash >=20 >> folder. The >=20 >>> biggest mess in programming history. >=20 >>>=20 >=20 >>> That being said, the solution is not to hack the igb >=20 >> driver; its to make >=20 >>> ALTQ if_transmit compatible, which shouldn't be all that >=20 >> difficult. >=20 >>>=20 >=20 >>> BC >=20 >>=20 >=20 >> I may be misunderstanding what you are saying, but if the >=20 >> solution is, as you say "not to hack the igb driver", then >=20 >> how is it defective in this case? Or are you just directing >=20 >> vitriol toward Intel? Multi-queue is working fine in igb. >=20 >>=20 >=20 >> Jeff >=20 >=20 >=20 > It's defective because it's been poorly implemented and has more bugs >=20 > than a Manhattan hotel bed. Adding queues without a proper plan just = add >=20 > more lock contention. It's not a production-ready driver. >=20 >=20 >=20 > As Jack once said, Intel doesn't care about performance, they're just >=20 > example drivers. igb is an example of how not to do things. >=20 >=20 >=20 > BC >=20 > _______________________________________________ >=20 > freebsd-net@freebsd.org mailing list >=20 > http://lists.freebsd.org/mailman/listinfo/freebsd-net >=20 > To unsubscribe, send any mail to "freebsd-net-unsubscribe@freebsd.org" >=20 >=20 >=20 > _______________________________________________ > freebsd-net@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-net > To unsubscribe, send any mail to "freebsd-net-unsubscribe@freebsd.org"
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?F024F568-0749-4132-9AB1-5010ED531B04>