These are last couple of comment from the RH rep re the bug ie wont fix, move on
Corinna Vinschen 2017-02-03 04:54:03 EST
(In reply to turandot from comment #9)
> Hi Corinna,
>
> is there a r8168 available within RHEL 7.x? I don't think so...
No, there isn't, of course, since the r8168 driver is covered by
patents incompatible with the Linux source tree.
I refered to it as an alternative since you were already using it
per
https://bugzilla.redhat.com/show_bug.cgi?id=1410994> Therefore I installed the r8168 from elrepo.org and did the following:
> [...]
> Then I once again did the WoL test, now with r6168 loaded:
>
> - init 0
> - Test WoL: FAIL
> - switch off main line
> - wait 10 seconds
> - switch on main line (system does not start automatically)
> - Test WoL: OK
>
> So this driver does not support WoL as well. (The elrepo r8168 for RHEL 6
> does support WoL.)
This isn't quite correct. Both drivers, r8168 from Realtek and r8169
from the Linux source tree support WoL. The problem are some of the NICs
which just don't always work as desired, sometimes even dependent on the
machine they are plugged into. WoL is a delicate issue, susceptible to
all kinds of hardware differences.
Please note that the 8168c model is quite old and EOLed since 2009 as far
as I know. It has been long since replaced by more modern chpsets like
8168evl or 8168f used on many third-party PCIe NICs, the latest of which
is the 8168h.
As I outlined in my replies, I have no trouble with WoL at all with the
8168f NIC I have available in my test setup, neither with RHEL 6.8, nor with
RHEL 7.3. It may be a feasible option to upgrade the NIC to a later version,
especially given the low cost of the device.
> Then I searched the dmesg output for the requested line:
>
> "Realtek PCIe GBE Family Controller mcfg = 1234"
>
> Unfortunately I did not find any output like this. I just see the following:
My bad. I realised belatedly that the above line is only printed if
the driver is running in debug mode. If you want to see the version of
your 8168c model, you have to download the r8168 driver from Realtek,
apply the newly attached patch (8168.diff) and build and install it.
But...
> Did I miss something? Should I compile the kernel with the provided
> patch as per comment 8?
...yes, this would be the better option.
Regards,
Corinna
Corinna Vinschen 2017-02-14 10:16:29 EST
Hi turandot,
We got hold of a 8168c (21) model just as yours. However, even after a
couple of days testing we can't get it to honor WoL requests at all,
independently of using the Linux in-tree 8169 driver, or the
Realtek-provided 8168 driver. We tested this in different hardware
setups as well as in different network setups, known to WoL well using
a Realtek 8168f.
To top it off, the 8168c we got for testing can't WoL after a power
cycle, because it switches itself to 'WoL off' on a cold start.
Other than that, the NIC works as desired. Our conclusion is that the
devices differ in some cases even when using seemingly identical
hardware, and the devices are apparently prone to BIOS or hardware
differences.
Given that we're unable to reproduce this problem with reasonable
effort, I'll close this issue (as well as BZ 1410994) as CANTFIX.
Corinna
Status: ASSIGNED → CLOSED
Resolution: --- → CANTFIX
Last Closed: 2017-02-14 10:16:29