2.6.24 regression: Wake On Lan in sky2 broken on Mac mini

2.6.24 regression: Wake On Lan in sky2 broken on Mac mini

[Tino Keitel]

Hi folks,

with 2.6.24-rc8, Wake On LAN doesn't work anymore as it used to with
2.6.23 on my Mac mini Core Duo. I saw that this was reported in
http://bugzilla.kernel.org/show_bug.cgi?id=9721 and on netdev a patch
for the sky2 driver was sent by Stephen Hemminger. This patch fixed WOL
for me after applying it to 2.6.24-rc8.

However, it seems as the patch never made it into the kernel. Instead,
the commit that was suspected to break WOL
(84cd2dfb04d23a961c5f537baa243fa54d0987ac) was reverted
(be63a21c9573fbf88106ff0f030da5974551257b).

Now I tried the 2.6.24 release and noticed that WOL is still broken.
I'll be happy to test any patches that can make it into 2.6.24.1.

Regards,
Tino

Re: 2.6.24 regression: Wake On Lan in sky2 broken on Mac mini

[Mikael Pettersson]

Tino Keitel writes:
 > Hi folks,
 > 
 > with 2.6.24-rc8, Wake On LAN doesn't work anymore as it used to with
 > 2.6.23 on my Mac mini Core Duo. I saw that this was reported in
 > http://bugzilla.kernel.org/show_bug.cgi?id=9721 and on netdev a patch
 > for the sky2 driver was sent by Stephen Hemminger. This patch fixed WOL
 > for me after applying it to 2.6.24-rc8.
 > 
 > However, it seems as the patch never made it into the kernel. Instead,
 > the commit that was suspected to break WOL
 > (84cd2dfb04d23a961c5f537baa243fa54d0987ac) was reverted
 > (be63a21c9573fbf88106ff0f030da5974551257b).
 > 
 > Now I tried the 2.6.24 release and noticed that WOL is still broken.
 > I'll be happy to test any patches that can make it into 2.6.24.1.

1. Wrong mailing list; use netdev (@vger) instead.
2. The reverted commit had much much more serious consequences than
   "wol doesn't work", it actually caused BIOS hangs and failed reboots.

Re: 2.6.24 regression: Wake On Lan in sky2 broken on Mac mini

[Tino Keitel]

On Mon, Jan 28, 2008 at 09:21:30 +0100, Mikael Pettersson wrote:
> Tino Keitel writes:
>  > Hi folks,
>  > 
>  > with 2.6.24-rc8, Wake On LAN doesn't work anymore as it used to with
>  > 2.6.23 on my Mac mini Core Duo. I saw that this was reported in
>  > http://bugzilla.kernel.org/show_bug.cgi?id=9721 and on netdev a patch
>  > for the sky2 driver was sent by Stephen Hemminger. This patch fixed WOL
>  > for me after applying it to 2.6.24-rc8.
>  > 
>  > However, it seems as the patch never made it into the kernel. Instead,
>  > the commit that was suspected to break WOL
>  > (84cd2dfb04d23a961c5f537baa243fa54d0987ac) was reverted
>  > (be63a21c9573fbf88106ff0f030da5974551257b).
>  > 
>  > Now I tried the 2.6.24 release and noticed that WOL is still broken.
>  > I'll be happy to test any patches that can make it into 2.6.24.1.
> 
> 1. Wrong mailing list; use netdev (@vger) instead.

Done.

> 2. The reverted commit had much much more serious consequences than
>    "wol doesn't work", it actually caused BIOS hangs and failed reboots.

Yes, but even with the reverted commit, WOL still doesn't work. I just
tried the patch from the netdev mailing list with the 2.6.24 release
version and now WOL works for me.

Regards,
Tino

Re: 2.6.24 regression: Wake On Lan in sky2 broken on Mac mini

[Stephen Hemminger]

On Mon, 28 Jan 2008 09:48:55 +0100
Tino Keitel <tino.keitel@gmx.de> wrote:

> On Mon, Jan 28, 2008 at 09:21:30 +0100, Mikael Pettersson wrote:
> > Tino Keitel writes:
> >  > Hi folks,
> >  > 
> >  > with 2.6.24-rc8, Wake On LAN doesn't work anymore as it used to with
> >  > 2.6.23 on my Mac mini Core Duo. I saw that this was reported in
> >  > http://bugzilla.kernel.org/show_bug.cgi?id=9721 and on netdev a patch
> >  > for the sky2 driver was sent by Stephen Hemminger. This patch fixed WOL
> >  > for me after applying it to 2.6.24-rc8.
> >  > 
> >  > However, it seems as the patch never made it into the kernel. Instead,
> >  > the commit that was suspected to break WOL
> >  > (84cd2dfb04d23a961c5f537baa243fa54d0987ac) was reverted
> >  > (be63a21c9573fbf88106ff0f030da5974551257b).
> >  > 
> >  > Now I tried the 2.6.24 release and noticed that WOL is still broken.
> >  > I'll be happy to test any patches that can make it into 2.6.24.1.
> > 
> > 1. Wrong mailing list; use netdev (@vger) instead.
> 
> Done.
> 
> > 2. The reverted commit had much much more serious consequences than
> >    "wol doesn't work", it actually caused BIOS hangs and failed reboots.
> 
> Yes, but even with the reverted commit, WOL still doesn't work. I just
> tried the patch from the netdev mailing list with the 2.6.24 release
> version and now WOL works for me.

Patch went to Jeff, but never made it into 2.6.24.  The timing was just wrong.

-- 
Stephen Hemminger <stephen.hemminger@vyatta.com>

Re: 2.6.24 regression: Wake On Lan in sky2 broken on Mac mini

[Ingo Molnar]

* Mikael Pettersson <mikpe@it.uu.se> wrote:

>  > Now I tried the 2.6.24 release and noticed that WOL is still 
>  > broken. I'll be happy to test any patches that can make it into 
>  > 2.6.24.1.
> 
> 1. Wrong mailing list; use netdev (@vger) instead.

lkml is the right mailing list for reporting Linux bugs.

this is an extermely harmful trend i've seen lately: some kernel hackers 
going out on a limb directing the flow of bugreports _away_ from lkml, 
by suggesting to testers that lkml is somehow inappropriate for 
reporting Linux kernel bugs.

It's not even the standard "I Cc:-ed netdev, maybe they are interested 
in this" message but the above, plain incorrect: "this is the wrong 
mailing list" message.

Mikael, what you do is as harmful to Linux as if you were intentionally 
putting bugs into the kernel source. In fact it's more harmful because 
it is irreversible: bugs you put into Linux i can fix and i can review 
all past patches you did to undo the damage - tester attention and 
feedback you redirect we cannot direct back. Stop it!

	Ingo

Re: 2.6.24 regression: Wake On Lan in sky2 broken on Mac mini

[Mikael Pettersson]

Ingo Molnar writes:
 > 
 > * Mikael Pettersson <mikpe@it.uu.se> wrote:
 > 
 > >  > Now I tried the 2.6.24 release and noticed that WOL is still 
 > >  > broken. I'll be happy to test any patches that can make it into 
 > >  > 2.6.24.1.
 > > 
 > > 1. Wrong mailing list; use netdev (@vger) instead.
 > 
 > lkml is the right mailing list for reporting Linux bugs.
 > 
 > this is an extermely harmful trend i've seen lately: some kernel hackers 
 > going out on a limb directing the flow of bugreports _away_ from lkml, 
 > by suggesting to testers that lkml is somehow inappropriate for 
 > reporting Linux kernel bugs.
 > 
 > It's not even the standard "I Cc:-ed netdev, maybe they are interested 
 > in this" message but the above, plain incorrect: "this is the wrong 
 > mailing list" message.
 > 
 > Mikael, what you do is as harmful to Linux as if you were intentionally 
 > putting bugs into the kernel source. In fact it's more harmful because 
 > it is irreversible: bugs you put into Linux i can fix and i can review 
 > all past patches you did to undo the damage - tester attention and 
 > feedback you redirect we cannot direct back.

Ok, I can see how my overly terse statement could be interpreted
in this way, and I apologize for that.

However, it _is_ a fact that there is a proliferation of specialized
mailing lists, and it is also a fact that many developers _only_ read
those lists. I'm in no way defending this behaviour, on the contrary
I probably dislike it as much as you do. But we can't ignore it.

I should of course have written something like "please cc: <whateverlist>"
instead of the stupid "wrong mailing" list comment.

 > Stop it!

Gladly.

Re: 2.6.24 regression: Wake On Lan in sky2 broken on Mac mini

[Tino Keitel]

On Mon, Jan 28, 2008 at 14:37:38 +0100, Mikael Pettersson wrote:

[...]

> However, it _is_ a fact that there is a proliferation of specialized
> mailing lists, and it is also a fact that many developers _only_ read
> those lists. I'm in no way defending this behaviour, on the contrary
> I probably dislike it as much as you do. But we can't ignore it.

I planned to post this to netdev first, but then I thought that it's
more a task for the stable team and/or those people who track
regressions, and I don't know if those people track all the subsystem
mailing lists. Especially as there is a patch that fixes the issue for
me, and the corresponding bugzilla entry is alrealy marked as resolved
and "patch available", so it seems to me as it is just a matter of
incorporating an existing patch into further revisions of 2.6.24.

Regards,
Tino

Re: 2.6.24 regression: Wake On Lan in sky2 broken on Mac mini

[Adrian Bunk]

On Mon, Jan 28, 2008 at 02:43:05PM +0100, Tino Keitel wrote:
> On Mon, Jan 28, 2008 at 14:37:38 +0100, Mikael Pettersson wrote:
> 
> [...]
> 
> > However, it _is_ a fact that there is a proliferation of specialized
> > mailing lists, and it is also a fact that many developers _only_ read
> > those lists. I'm in no way defending this behaviour, on the contrary
> > I probably dislike it as much as you do. But we can't ignore it.
> 
> I planned to post this to netdev first, but then I thought that it's
> more a task for the stable team and/or those people who track
> regressions, and I don't know if those people track all the subsystem
> mailing lists. Especially as there is a patch that fixes the issue for
> me, and the corresponding bugzilla entry is alrealy marked as resolved
> and "patch available", so it seems to me as it is just a matter of
> incorporating an existing patch into further revisions of 2.6.24.

If in doubt, send it to both lists.

> Regards,
> Tino

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed

Re: 2.6.24 regression: Wake On Lan in sky2 broken on Mac mini

[Rafael J. Wysocki]

On Monday, 28 of January 2008, Adrian Bunk wrote:
> On Mon, Jan 28, 2008 at 02:43:05PM +0100, Tino Keitel wrote:
> > On Mon, Jan 28, 2008 at 14:37:38 +0100, Mikael Pettersson wrote:
> > 
> > [...]
> > 
> > > However, it _is_ a fact that there is a proliferation of specialized
> > > mailing lists, and it is also a fact that many developers _only_ read
> > > those lists. I'm in no way defending this behaviour, on the contrary
> > > I probably dislike it as much as you do. But we can't ignore it.
> > 
> > I planned to post this to netdev first, but then I thought that it's
> > more a task for the stable team and/or those people who track
> > regressions, and I don't know if those people track all the subsystem
> > mailing lists. Especially as there is a patch that fixes the issue for
> > me, and the corresponding bugzilla entry is alrealy marked as resolved
> > and "patch available", so it seems to me as it is just a matter of
> > incorporating an existing patch into further revisions of 2.6.24.
> 
> If in doubt, send it to both lists.

And CC to the maintainer of the subsystem in question as well as to his
upstream etc. (if you know who they are).

Greetings,
Rafael

Re: 2.6.24 regression: Wake On Lan in sky2 broken on Mac mini

[Ingo Molnar]

* Mikael Pettersson <mikpe@it.uu.se> wrote:

> Ok, I can see how my overly terse statement could be interpreted in 
> this way, and I apologize for that.
> 
> However, it _is_ a fact that there is a proliferation of specialized 
> mailing lists, and it is also a fact that many developers _only_ read 
> those lists. I'm in no way defending this behaviour, on the contrary I 
> probably dislike it as much as you do. But we can't ignore it.

i'm not worried about that aspect: those developers either write perfect 
code that never needs any tester or user assistance (in which case our 
discussion is moot), or if it's buggy then these developers will 
eventually be replaced with more capable people who are helping their 
testers and users more actively by reading lkml and responding there.

What we must not do is to give in to the splintering and laziness and 
actively _chase people away_ from lkml.

> I should of course have written something like "please cc: 
> <whateverlist>" instead of the stupid "wrong mailing" list comment.

yeah, thanks, that will do fine.

_Helping_ subsystem maintainers become aware of problems and 
cross-Cc:-ing to whatever preferred email alias they use (be that their 
own mailing list address or netdev@) is a necessary and positive feature 
of lkml: almost by definition the user has little idea about what is 
wrong with their kernel, and if the mail is unspecific enough or labeled 
incorrectly then it's easy for a maintainer to miss it. (It's already a 
very good first step when user know that the problem area is their 
kernel to begin with.)

Putting the "weight of initial discovery" on to the user on the other 
hand is actively harmful. It will only result in users picking the wrong 
list and being ignored there, upstream maintainers not getting a full 
picture about what kind of per subsystem bugs people are experiencing, 
etc., etc. The only thing that works in the long run is to make the 
private development email aliases _opt-in_ feature, and to always have a 
main body of information: lkml.

	Ingo