Re: 3.3.x: e1000 and ixgbe hang

Re: 3.3.x: e1000 and ixgbe hang

[Wolfgang Walter]

Am Dienstag, 8. Mai 2012 schrieben Sie:
> On Tue, May 08, 2012 at 11:30:12PM +0200, Wolfgang Walter wrote:
> > I would be very glad if these two patches will make it into the next
> > stable 3.3:
> >
> > http://article.gmane.org/gmane.linux.drivers.e1000.devel/9948
> >
> > Dave Miller said that he will take care of it:
> >
> > http://article.gmane.org/gmane.linux.drivers.e1000.devel/9993
>
> If he said he would take care of it, why are you posting it here again?

a) Because it didn't make it into 3.3.5. The fix (for 3.3.x) is known since 
2012-04-20 and I find it appropriate to remind of it after more than two 
weeks.

b) To document the bug (which makes 3.3. i.e. unusable for us) and that there 
are fixes in the stable mailinglist and not only in e1000-devel (especially 
as it isn't a e1000 specific problem).

c) I met 2 colleague yesterday which had the same problems but didn't know 
there was a fix.

c) Dave Miller is a busy man.

It's unclear how one should use new kernels if it is difficult to find known 
fixes.

Regards,
-- 
Wolfgang Walter
Studentenwerk München
Anstalt des öffentlichen Rechts

Re: 3.3.x: e1000 and ixgbe hang

[David Miller]

From: Wolfgang Walter <wolfgang.walter@stwm.de>
Date: Wed, 9 May 2012 11:07:11 +0200

> c) Dave Miller is a busy man.

Therefore pinging me on crap like this makes me a more busy man, so
don't do it, unless you want it to take even longer for me to get
to it.

Re: 3.3.x: e1000 and ixgbe hang

[Wolfgang Walter]

Am Mittwoch, 9. Mai 2012 schrieben Sie:
> From: Wolfgang Walter <wolfgang.walter@stwm.de>
> Date: Wed, 9 May 2012 11:07:11 +0200
>
> > c) Dave Miller is a busy man.
>
> Therefore pinging me on crap like this makes me a more busy man, so
> don't do it, unless you want it to take even longer for me to get
> to it.

I don't understand why you are so unfriendly.

I didn't sent you any mail nor did I complain to you. If you read my first 
mail on this issue then you will see that.

But when we are at it: I think that when a bug in a stable kernel is found and 
the final fix is known it is bad practice to hord that patch till submission 
without letting stable@ know about it. I don't mean that the patch should go 
into the stable tree immediately. Of course it makes sense to wait some time 
to see if it works in Linus' tree. But why should other people waste their 
time hunting a bug already solved? And why deny people the fix?

Regards,
-- 
Wolfgang Walter
Studentenwerk München
Anstalt des öffentlichen Rechts

Re: 3.3.x: e1000 and ixgbe hang

[Greg KH]

On Wed, May 09, 2012 at 06:57:13PM +0200, Wolfgang Walter wrote:
> Am Mittwoch, 9. Mai 2012 schrieben Sie:
> > From: Wolfgang Walter <wolfgang.walter@stwm.de>
> > Date: Wed, 9 May 2012 11:07:11 +0200
> >
> > > c) Dave Miller is a busy man.
> >
> > Therefore pinging me on crap like this makes me a more busy man, so
> > don't do it, unless you want it to take even longer for me to get
> > to it.
> 
> I don't understand why you are so unfriendly.
> 
> I didn't sent you any mail nor did I complain to you. If you read my first 
> mail on this issue then you will see that.

You were complaining to me, about how David isn't working fast enough to
solve a problem that is affecting you, when he said he would take care
of it when he got the chance to.

You didn't say it in so many words, but that was the general gist of it.
And I don't blame David for being brisk, I think it was really crappy of
you to do that.

> But when we are at it: I think that when a bug in a stable kernel is found and 
> the final fix is known it is bad practice to hord that patch till submission 
> without letting stable@ know about it. I don't mean that the patch should go 
> into the stable tree immediately. Of course it makes sense to wait some time 
> to see if it works in Linus' tree. But why should other people waste their 
> time hunting a bug already solved? And why deny people the fix?

No one is "denying" anyone anything, it's all open, you are free to take
the patch and do with it what you want, if you need to.  Some of us have
other responsibilities and tasks and will get to things as soon as we
possibly can.

In other words, trust us, if you don't, that's fine, then do it
yourself, but to try to route around the documented and well-working
procedures that we have in place, is flat out rude.

greg k-h

Re: 3.3.x: e1000 and ixgbe hang

[David Miller]

From: Wolfgang Walter <wolfgang.walter@stwm.de>
Date: Wed, 9 May 2012 18:57:13 +0200

> But when we are at it: I think that when a bug in a stable kernel is
> found and the final fix is known it is bad practice to hord that
> patch till submission without letting stable@ know about it.

I think you have no idea what goes into vetting patches for -stable.

It can take me days to put together a series, and also I time my
-stable submissions with when Linus pulls my 'net' bug fixes into his
tree since a part of the -stable requirements is presence in Linus's
tree.

I therefore batch, because bleeding out individual fixes one by one to
-stable rarely, if ever, makes sense.

And another part of what goes into vetting a patch is time.  The
longer a patch sits in a non-stable tree getting tested, the more
likely any unwanted bugs and side effects will be caught before the
patch goes into -stable.

And finally I am under no obligation whatsoever to post some kind of
status report to -stable every few days saying when I'll do this or
that.

Re: 3.3.x: e1000 and ixgbe hang

[Wolfgang Walter]

On Wednesday 09 May 2012, David Miller wrote:
> From: Wolfgang Walter <wolfgang.walter@stwm.de>
> Date: Wed, 9 May 2012 18:57:13 +0200
> 
> > But when we are at it: I think that when a bug in a stable kernel is
> > found and the final fix is known it is bad practice to hord that
> > patch till submission without letting stable@ know about it.
> 
> I think you have no idea what goes into vetting patches for -stable.

I can imagine that very well. I wrote and manage a bunch of larger patches 
against wine vanilla for us and to port them regularly to newer versions.

> 
> It can take me days to put together a series, and also I time my
> -stable submissions with when Linus pulls my 'net' bug fixes into his
> tree since a part of the -stable requirements is presence in Linus's
> tree.

I know.

> 
> I therefore batch, because bleeding out individual fixes one by one to
> -stable rarely, if ever, makes sense.

I understand that and as I sad it was not my intention to question that.

> 
> And another part of what goes into vetting a patch is time.  The
> longer a patch sits in a non-stable tree getting tested, the more
> likely any unwanted bugs and side effects will be caught before the
> patch goes into -stable.
> 
> And finally I am under no obligation whatsoever to post some kind of
> status report to -stable every few days saying when I'll do this or
> that.
> 

I didn't say that you should do that.

I just don't see the point why I should not send a mail to stable@. It 
documented that there is a known problem in 3.3.5, the fix for it and that you 
already aware of it and will take care to get it into stable. Nobody had 
reported that on stable@.

I mentioned your mail not to blame you but so nobody thinks I want du submit 
it directly. Everybody reading your mail I linked to could see that you wrote 
it May 1st and therefor could not be in 3.3.5.

But I think it would have been good if someone (not you) had sent a mail as 
early as 2012-04-20 to stable@ so that others could find the solution more 
easily.


Regards,
-- 
Wolfgang Walter
Studentenwerk München
Anstalt des öffentlichen Rechts