I had a chance to test it out and the results are weird.
The system is a SuperMicro server 50161-TF with a dual Intel 82574L Gigabit
Ethernet controller on the motherboard. It has a network share using Samba
4.4.2. It is not connected to the internet, only on a small local network.
The Red Hat kernel default driver e1000e will either list millions of errors or
millions of RX drops when the command /sbin/ifconfig is run when trying to copy
GB size files onto a windows desktop. It then will kill the Samba connection.
This has never worked.
The Intel e1000e driver 3.0.4 and then 3.1.0.2 worked with no problems after
compiling on Red Hat 6.1 to 6.7 server edition.
No Intel drivers currently compile on Red Hat 6.8 server.
Changing all instances of vlan_tx_* to skb_vlan_* allows 3.3.3 to compile, but
when tested this driver now displays the same problems that the Red Hat default
driver does as showing millions of errors and then dropping the connection.
Making the same change to the source code to 3.1.0.2, it compiles just fine and
works without problems.
So something changed in 3.3.3 that now breaks the driver. As far as I can tell,
the new Intel 3.3.3 driver has less references to vlan_tx_* functions and does
not work so the issue must be in the differences in how this is handled from
3.1.0.2 to 3.3.3.
I submitted a bug report: Bug 1339056.
Thank you for your help,Spencer
On Monday, May 23, 2016 10:22 AM, Jarod Wilson <ja...@redhat.com> wrote:
I would definitely suggest filing a bug in Red Hat's bugzilla. If the
hardware works with the out of box driver and completely falls down with
the in box driver, that's definitely something that should be fixed. I
was thinking there was maybe an optional feature you wanted, but that's
a clear defect that Red Hat needs to be made aware of and get fixed.
--
Jarod Wilson
Spencer Hall wrote:
> The base e1000e driver that comes in the Red Hat Kernel does not work
> with my motherboard's intel Ethernet ports. It appears to function until
> I try to move any large amount of data over the network and then it
> experiences massive multiple RX drops until the Ethernet drops connection.
>
> This problem disappears once I started building the e1000e module for
> the kernel from sourceforge.
>
> I can get more information on the motherboard and specific Ethernet
> model number on Monday.
>
> Spencer
>
>
> On Saturday, May 21, 2016 11:32 AM, Jarod Wilson <ja...@redhat.com> wrote:
>
>
> The Red Hat backport isn't broken, it just "is". The change simply
> caused a build failure, because some function names changed. An updated
> out of tree driver would need to special case RHEL-6.8 to use the
> updated function names, which match upstream. Out of curiosity, what
> functionality is missing from the in box driver that is required?
>
> --
> Jarod Wilson
>
>
> > On May 21, 2016, at 1:02 PM, Fujinaka, Todd <todd.fujin...@intel.com
> <mailto:todd.fujin...@intel.com>> wrote:
> >
> > I'm a little unclear on your complaint as there are no version
> numbers. Are you telling us the Red Hat backport is broken or are you
> telling us that the standalone driver is broken?
> >
> > There is a new e1000e standalone version coming very soon if that's
> what the issue is.
> >
> > Todd Fujinaka
> > Software Application Engineer
> > Networking Division (ND)
> > Intel Corporation
> > todd.fujin...@intel.com <mailto:todd.fujin...@intel.com>
> > (503) 712-4565
> >
> >
> > -----Original Message-----
> > From: Jarod Wilson [mailto:ja...@redhat.com <mailto:ja...@redhat.com>]
> > Sent: Friday, May 20, 2016 10:01 AM
> > To: Spencer Hall <dragon_bra...@yahoo.com
> <mailto:dragon_bra...@yahoo.com>>
> > Cc: e1000-devel@lists.sourceforge.net
> <mailto:e1000-devel@lists.sourceforge.net>
> > Subject: Re: [E1000-devel] e1000e Compile error on Red Hat 6.8
> >
> >> On Fri, May 20, 2016 at 05:34:22AM +0000, Spencer Hall wrote:
> >> I have been compiling the e1000e module for all the Red Hat 6
> kernels. With the latest release of Red Hat 6.8 with the new kernel
> 2.6.32-642, when trying to build the e1000e module it fails with the
> following error:
> >> implicit declaration of function 'vlan_tx_tag_present'
> >> A similar problem showed up on virtualbox:
> >> See:virtualbox.org/ticket/15411virtualbox.org/changeset/60995/vbox
> >> It seems like a definitions problem, but this is beyond my ability.
> >
> > RHEL-6.8 includes a backport of:
> >
> > commit df8a39defad46b83694ea6dd868d332976d62cc0
> > Author: Jiri Pirko <j...@resnulli.us <mailto:j...@resnulli.us>>
> > Date: Tue Jan 13 17:13:44 2015 +0100
> >
> > net: rename vlan_tx_* helpers since "tx" is misleading there
> >
> > s/vlan_tx_tag_present/skb_vlan_tag_present/
> >
> > --
> > Jarod Wilson
> > ja...@redhat.com <mailto:ja...@redhat.com>
------------------------------------------------------------------------------
Mobile security can be enabling, not merely restricting. Employees who
bring their own devices (BYOD) to work are irked by the imposition of MDM
restrictions. Mobile Device Manager Plus allows you to control only the
apps on BYO-devices by containerizing them, leaving personal data untouched!
https://ad.doubleclick.net/ddm/clk/304595813;131938128;j
_______________________________________________
E1000-devel mailing list
E1000-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/e1000-devel
To learn more about Intel® Ethernet, visit
http://communities.intel.com/community/wired
