[Intel-wired-lan] Regression: e1000e link not coming up

[Intel-wired-lan] Regression: e1000e link not coming up

[Jan Janssen]

Hi,

since 5.3 my I219-V network interface sometimes doesn't come up when booting
or resuming from suspend. This either forces me to use "ip link down/up" or
reloading the module to get working internet.

I reported this issue previously to netdev at vger.kernel.org where I initially
bisected to a wrong commit. After finding the right one nobody reacted to it,
so now I am trying this again here.

There is also a bugzilla entry for this, but it seems to be ignored by
maintainers too: https://bugzilla.kernel.org/show_bug.cgi?id=205067
For me, this happens with systemd-networkd, but the people in the bugzilla
report are observing this with NetworkManager.

The commit introducing the bug is 59653e6497d16f7ac1d9db088f3959f57ee8c3db
(e1000e: Make watchdog use delayed work) and reverting it did make this bug go
away. Though, it cannot be reverted automatically on the latest git tree from
Torvalds anymore.

I've also tried the master branch on jkirsher/net-queue.git  and
jkirsher/next-queue.git, but the bug is still present there too.

This bug is really annoying and inconvenient, so I would really appreciate if
somebody could fix this bug or at least have the offending commit reverted.

Regards,
Jan

[Intel-wired-lan] Regression: e1000e link not coming up

[Jeff Kirsher]

On Sat, 2019-11-16 at 12:50 +0100, Jan Janssen wrote:
> Hi,
> 
> since 5.3 my I219-V network interface sometimes doesn't come up when
> booting
> or resuming from suspend. This either forces me to use "ip link
> down/up" or
> reloading the module to get working internet.
> 
> I reported this issue previously to netdev at vger.kernel.org where I
> initially
> bisected to a wrong commit. After finding the right one nobody
> reacted to it,
> so now I am trying this again here.

I apologize, this is the correct mailing list for issues with the Intel
wired LAN.


> There is also a bugzilla entry for this, but it seems to be ignored
> by
> maintainers too: https://bugzilla.kernel.org/show_bug.cgi?id=205067
> For me, this happens with systemd-networkd, but the people in the
> bugzilla
> report are observing this with NetworkManager.
> 
> The commit introducing the bug is
> 59653e6497d16f7ac1d9db088f3959f57ee8c3db
> (e1000e: Make watchdog use delayed work) and reverting it did make
> this bug go
> away. Though, it cannot be reverted automatically on the latest git
> tree from
> Torvalds anymore.

I will work with the e1000e driver lead (Sasha Neftin) to work up a
patch that will apply to David Miller's net tree.  I will make sure you
are CC'd on the patch, so you will know when it makes it into the
upstream kernel.

> I've also tried the master branch on jkirsher/net-queue.git  and
> jkirsher/next-queue.git, but the bug is still present there too.
> 
> This bug is really annoying and inconvenient, so I would really
> appreciate if
> somebody could fix this bug or at least have the offending commit
> reverted.
> 
> Regards,
> Jan
> 
> 

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: This is a digitally signed message part
URL: <http://lists.osuosl.org/pipermail/intel-wired-lan/attachments/20191118/2c55ed22/attachment.asc>

[Intel-wired-lan] Regression: e1000e link not coming up

[Neftin, Sasha]

On 11/18/2019 20:29, Jeff Kirsher wrote:
> On Sat, 2019-11-16 at 12:50 +0100, Jan Janssen wrote:
>> Hi,
>>
>> since 5.3 my I219-V network interface sometimes doesn't come up when
>> booting
>> or resuming from suspend. This either forces me to use "ip link
>> down/up" or
>> reloading the module to get working internet.
>>
>> I reported this issue previously to netdev at vger.kernel.org where I
>> initially
>> bisected to a wrong commit. After finding the right one nobody
>> reacted to it,
>> so now I am trying this again here.
> 
> I apologize, this is the correct mailing list for issues with the Intel
> wired LAN.
> 
> 
>> There is also a bugzilla entry for this, but it seems to be ignored
>> by
>> maintainers too: https://bugzilla.kernel.org/show_bug.cgi?id=205067
>> For me, this happens with systemd-networkd, but the people in the
>> bugzilla
>> report are observing this with NetworkManager.
>>
>> The commit introducing the bug is
>> 59653e6497d16f7ac1d9db088f3959f57ee8c3db
>> (e1000e: Make watchdog use delayed work) and reverting it did make
>> this bug go
>> away. Though, it cannot be reverted automatically on the latest git
>> tree from
>> Torvalds anymore.
> 
> I will work with the e1000e driver lead (Sasha Neftin) to work up a
> patch that will apply to David Miller's net tree.  I will make sure you
> are CC'd on the patch, so you will know when it makes it into the
> upstream kernel.
> 
I heard a lot of peoples complain about a problem with using 
delayed_work struct in e1000e. I would suggest reverting this patch.
>> I've also tried the master branch on jkirsher/net-queue.git  and
>> jkirsher/next-queue.git, but the bug is still present there too.
>>
>> This bug is really annoying and inconvenient, so I would really
>> appreciate if
>> somebody could fix this bug or at least have the offending commit
>> reverted.
>>
>> Regards,
>> Jan
>>
>>
> 

[Intel-wired-lan] Regression: e1000e link not coming up

[Neftin, Sasha]

On 11/19/2019 08:12, Neftin, Sasha wrote:
> On 11/18/2019 20:29, Jeff Kirsher wrote:
>> On Sat, 2019-11-16 at 12:50 +0100, Jan Janssen wrote:
>>> Hi,
>>>
>>> since 5.3 my I219-V network interface sometimes doesn't come up when
>>> booting
>>> or resuming from suspend. This either forces me to use "ip link
>>> down/up" or
>>> reloading the module to get working internet.
>>>
>>> I reported this issue previously to netdev at vger.kernel.org where I
>>> initially
>>> bisected to a wrong commit. After finding the right one nobody
>>> reacted to it,
>>> so now I am trying this again here.
>>
>> I apologize, this is the correct mailing list for issues with the Intel
>> wired LAN.
>>
>>
>>> There is also a bugzilla entry for this, but it seems to be ignored
>>> by
>>> maintainers too: https://bugzilla.kernel.org/show_bug.cgi?id=205067
>>> For me, this happens with systemd-networkd, but the people in the
>>> bugzilla
>>> report are observing this with NetworkManager.
>>>
>>> The commit introducing the bug is
>>> 59653e6497d16f7ac1d9db088f3959f57ee8c3db
>>> (e1000e: Make watchdog use delayed work) and reverting it did make
>>> this bug go
>>> away. Though, it cannot be reverted automatically on the latest git
>>> tree from
>>> Torvalds anymore.
>>
>> I will work with the e1000e driver lead (Sasha Neftin) to work up a
>> patch that will apply to David Miller's net tree.? I will make sure you
>> are CC'd on the patch, so you will know when it makes it into the
>> upstream kernel.
>>
> I heard a lot of peoples complain about a problem with using 
> delayed_work struct in e1000e. I would suggest reverting this patch.
Looks reverting of this patch solved also the 
https://bugzilla.kernel.org/show_bug.cgi?id=205047

>>> I've also tried the master branch on jkirsher/net-queue.git? and
>>> jkirsher/next-queue.git, but the bug is still present there too.
>>>
>>> This bug is really annoying and inconvenient, so I would really
>>> appreciate if
>>> somebody could fix this bug or at least have the offending commit
>>> reverted.
>>>
>>> Regards,
>>> Jan
>>>
>>>
>>
> 
> _______________________________________________
> Intel-wired-lan mailing list
> Intel-wired-lan at osuosl.org
> https://lists.osuosl.org/mailman/listinfo/intel-wired-lan

[Intel-wired-lan] Regression: e1000e link not coming up

[Jan Janssen]

On Tuesday, 19 November 2019 07:12:18 CET Neftin, Sasha wrote:
> On 11/18/2019 20:29, Jeff Kirsher wrote:
> > On Sat, 2019-11-16 at 12:50 +0100, Jan Janssen wrote:
> >> Hi,
> >>
> >> since 5.3 my I219-V network interface sometimes doesn't come up when
> >> booting
> >> or resuming from suspend. This either forces me to use "ip link
> >> down/up" or
> >> reloading the module to get working internet.
> >>
> >> I reported this issue previously to netdev at vger.kernel.org where I
> >> initially
> >> bisected to a wrong commit. After finding the right one nobody
> >> reacted to it,
> >> so now I am trying this again here.
> >
> > I apologize, this is the correct mailing list for issues with the Intel
> > wired LAN.
> >
> >> There is also a bugzilla entry for this, but it seems to be ignored
> >> by
> >> maintainers too: https://bugzilla.kernel.org/show_bug.cgi?id=205067
> >> For me, this happens with systemd-networkd, but the people in the
> >> bugzilla
> >> report are observing this with NetworkManager.
> >>
> >> The commit introducing the bug is
> >> 59653e6497d16f7ac1d9db088f3959f57ee8c3db
> >> (e1000e: Make watchdog use delayed work) and reverting it did make
> >> this bug go
> >> away. Though, it cannot be reverted automatically on the latest git
> >> tree from
> >> Torvalds anymore.
> >
> > I will work with the e1000e driver lead (Sasha Neftin) to work up a
> > patch that will apply to David Miller's net tree.  I will make sure you
> > are CC'd on the patch, so you will know when it makes it into the
> > upstream kernel.
>
> I heard a lot of peoples complain about a problem with using
> delayed_work struct in e1000e. I would suggest reverting this patch.
>
> >> I've also tried the master branch on jkirsher/net-queue.git  and
> >> jkirsher/next-queue.git, but the bug is still present there too.
> >>
> >> This bug is really annoying and inconvenient, so I would really
> >> appreciate if
> >> somebody could fix this bug or at least have the offending commit
> >> reverted.
> >>
> >> Regards,
> >> Jan

So... when can we expect a patch revert?