Created attachment 174968 [details] adds net/intel-em-kmod.diff Hi all, FreeBSD since 10.3 has two issues with its em(4) driver at the moment, which likely also apply to FreeBSD 11, so following the steps of the intel-ixl-kmod package we propose this new port that seems not to show these regressions after thorough testing. One of the issues is tracked here: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=211219 The other issue will be documented this week, affecting at least the 82574L chipset being unstable in netmap mode. A partial fix went into HEAD a month back, but only fixes the netmap mode partially: https://lists.freebsd.org/pipermail/svn-src-head/2016-August/089933.html This can also help users to get newer network drivers faster in the future. :) Cheers, Franco
Created attachment 174969 [details] poudriere testport build log
It's affecting more than you Franco. :) Thanks for submitting this, can't wait to see it included in the tree.
Created attachment 175253 [details] revised patch o changed license to 3-clause as per source code indication (came from the intel-ixl-kmod port) o adjusted package description, does not include igb (came from the distfile README)
Created attachment 175290 [details] slave port example Hi all, I'm the author of net/intel-ixl-kmod and I work for Intel. The ixl port was made to be a master one, that means to serve as a base for other driver ports' (slaves). Here's the example of how this port would look like being a slave to net/intel-ixl-kmod. The patch includes a small change to the ixl port as well to be able to override the MAINTAINER. This approach has some advantages: - You don't have to repeat a lot of code - If anything big changes to all the drivers - it will be changed in the master port and automatically propagated to others. If you would consider this - just substitute your patch with mine. Regards, Sergey
Comment on attachment 175253 [details] revised patch Hi Sergey, Thanks for this! Looks better, works fine. Obsoleting the other patch in favour of yours. :) Cheers, Franco
A commit references this bug: Author: sbruno Date: Thu Oct 6 20:18:31 UTC 2016 New revision: 423442 URL: https://svnweb.freebsd.org/changeset/ports/423442 Log: Change MAINTAINER variable to support using this as a master port for other slave ports. PR: 212828 Submitted by: Sergey Kozlov <kozlov.sergey.404@gmail.com> Reviewed by: Franco Fichtner <franco@opnsense.org> Changes: head/net/intel-ixl-kmod/Makefile
A commit references this bug: Author: sbruno Date: Thu Oct 6 20:20:54 UTC 2016 New revision: 423444 URL: https://svnweb.freebsd.org/changeset/ports/423444 Log: Add kmod port for the exact tar archive from Intel's download site for those folks who don't want to use the FreeBSD supported version. PR: 212828 Submitted by: Sergey Kozlov <kozlov.sergey.404@gmail.com> Reviewed by: Franco Fichtner <franco@opnsense.org> Changes: head/net/Makefile head/net/intel-em-kmod/ head/net/intel-em-kmod/Makefile head/net/intel-em-kmod/distinfo head/net/intel-em-kmod/pkg-descr
Theoretically, this is done now.
Thanks Sean! Sergey, there is a small issue that prevents building the driver under i386 which works fine as far as I can see. It needs to avoid pulling in the restrictions of the ixl port. I used if ! defined(${MASTERDIR}) to work around this, can propose a patch if you want? Cheers, Franco
I've added a patch for review here: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=213283 Cheers, Franco