Fog 1.1.0 - Intel driver issues - Dell OptiPlex 755
-
No worries, I am happy to test any of the kernels as this is the one show stopper I have found for fog 1.x.x for us.
Just testing a bunch of our desktops -
Dell OptiPlex 745 - Intel 965 Express - Broadcom NetExtreme 57xx NIC- Drivers load fine and IP Address gets assigned quickly
Dell OptiPlex 755 - Q35 Express Chip - 82566DM-2 NIC - Intel Drivers fail to load correctly
Dell OptiPlex 760 - Q45/Q43 Express Chip - 82567-LM3 - Intel Drivers fail to load correctly
Dell OptiPlex 780 - Q45 Express chip - 82567-LM3 - Intel Drivers fail to load correctlyWikipedia has a good page for the Dell OptiPlex that matches up to mine for Chipsets - [url]https://en.wikipedia.org/wiki/Dell_OptiPlex[/url]
Also just installed the new BootMenu.class.php all seems to be working for the iPXE/Fog menu.
I’ve got a few options set in the advanced menu as well and that looks good as well. -
Okay,
So 1806 is released and has the BootMenu.class.php modifications as well as a few others. This should help you with your weirdness on the net0/net1 issues.
When you say fail to load correctly, you mean the DCHP/BOOTP reply not for us xid[blah] message?
So that means the pcie_aspm=off isn’t working?
-
Sorry,
Yes it repeats the message DHCP/BOOTP reply not for us etc. over and over and then fails to load an IP Address via DHCP when loading into fog client system information (compatibility) menu option.
I’m using the Kernel from unofficial Tom Elliott version 3.15.0I updated the BootMenu.class.php before I tested it all again so I’m guessing that pcie_aspm might not be the issue?
-
I’m going to try building two kernels.
First, with only the e1000e drivers loaded, and second with only the e1000 drivers loaded, will you test them?
-
Yep no worries
-
Still building as I decided to clean them to ensure all references to e1000e and e1000 were separate from each other. It’ll likely be about 10 - 20 minutes more, sorry about the delay!
-
No worries,
Thanks for the help, especially since I suspect our timezones are significantly different.Cheers,
-
Based on timing, I’d say you’re correct. Where you located, Australia, Europe? Not that you have to answer or anything.
-
Okay, kernels are ready for download/test if you’re up to it.
[url]https://mastacontrola.com/bzImageE1000[/url] <- intel E1000 not intel E1000E (my bet is on this one)
[url]https://mastacontrola.com/bzImageE1000E[/url] <- intel E1000E not intel E1000
-
ok, I’ll download that now to bzImage
I’m suspecting both versions are the 64bit one? -
Correct
-
Just tested both on the Dell OptiPlex 755
The bzImageE1000 didn’t load any network at all and didn’t report any errors and shows the network ip/mac as blank.
The bzImageE1000E reported the usual Reply not for us, op[2] etc. and then showed the network/mac as blank again -
Dang!
-
I can test them both on the other OptiPlex 760 and OptiPlex 780, but I am suspecting it is the same issue with the E1000E driver.
I can’t really find much on Google about this issue either. -
I’m doing a test based on [url]http://comments.gmane.org/gmane.linux.drivers.e1000.devel/13193[/url]
Hopefully this may work. Maybe not setting the values at all (commenting them out) will work?
[url]https://mastacontrola.com/bzImageE1000E[/url] (ready to try out if you are)
-
testing now.
-
If that one doesn’t work, try just redownload and test please, I’ve made another mod based on the patch and hopefully will get somewhere.
-
Yep same issue, redownloading now.
-
Same issue again.
-
As a temp work around I tried to recompile the old fog kitchensink as 64bit and 32bit to see if that worked.
The kernel compiles okay but it kernel panics on init.xz.
I’m guessing it has to do with the new compression of the init being init.xz and it can’t decompress it when it tries to load.EDIT: Got the old 2.6.39 one working with kitchen sink, not everything works. the SATA drivers don’t seem to work so I can’t see the HDD, DMI doesn’t show any fields either as well as errors about files not being found but the Network worked, and it mentioned when it was loading that it had to fall back to MSI legacy interrupts.
I will test the 3.15.0 kernel but change some driver options to the e1000e driver to see if any of that helps.