Re: Kernel 2.6.28 regression: Hang after hibernate (patch included)

[Francois Romieu <romieu@fr.zoreil.com>]

Rafael J. Wysocki <rjw@sisk.pl> :
[...]
> Please don't top post, it makes following the discussion really difficult.
> 
> On Friday 02 January 2009, Frank Groeneveld wrote:
> > A kernel without SMP doesn't solve the problem. It gives the same
> > output as before, except without the CPU notices. I also found out
> > that a "vanilla" 2.6.27 does have the same problem. A patched 2.6.27
> > kernel (by Gentoo Linux) doesn't have this problem. I attached that
> > patch, so you might see what the problem is now.
> 
> Not really, the patch (reproduced below for reference) is quite big and does
> a couple of different things.  Also, it should be shown to the maintainer of
> the code in question.

Thanks for the Cc Rafael.

The aforementioned patch matches three commits - mostly bugfixes - which
went in between 2.6.27 and 2.6.28:

commit ccdffb9a88b2907b159538d7bfd6256621db4f84

    r8169: get ethtool settings through the generic mii helper

commit 2857ffb7b8913ef713533ac5783abd70a20529e4

    r8169: additional 8101 and 8102 support

commit 523a609496dbc3897e530db2a2f27650d125ea00

    r8169: fix RxMissed register access

Frank, can you:
1. try the r8169.c file from the 2.6.28 tree with Ubuntu's kernel
2. grep for the XID line printed by the r8169 driver in the lernel's dmesg

If 1. does not work , you can peel the onion with the patchkit available
at http://userweb.kernel.org/~romieu/r8169/2.6.27/ : apply them in
increasing order to your working Ubuntu driver until is stops waking
up correctly (skip patches #0001, #0006 and #0007 as they match the
commits above) or try a binary search.

As a side note for those who remember the early 2.6.27-rc "null mac
address with the r8169 driver", the patchkit does not include the
patches which were included between 2.6.27 and 2.6.28 then later
reverted.

-- 
Ueimor