[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

RE: Applying e1000e patch to the Debian 2.6.26 kernel



Dear Thomas,

   Good afternoon.  I deeply appreciate for this, and I will always try my
very best working with you or any customers.

   Our company web site is www.kingstarusa.com, and yes, my e-mail address
is john@kingstarusa.com.  I sincerely look forward to hearing from you for
any questions or quote requests.  Have a nice weekend.

Best Regards!

John Wu
King Star Computer
1259 Reamwood Avenue
Sunnyvale, CA 94089
TEL: 408-736-8590  ex. 107
FAX: 408-736-4151
John@kingstarusa.com

- Supermicro Preferred System Provider.
- Asus & Tyan Approved System Integrator.
- Intel Premier Channel Partner 2009.
- AMD Solution Provider.
- nVidia Authorized System Integrator, Tesla C1060, S1070, etc.
- Lenovo Approved Solution Partner & Reseller.
- InforTrend FC, SAS, SATA, SCSI, iSCSI Storage Provider.


-----Original Message-----
From: Thomas Goirand [mailto:thomas@goirand.fr] 
Sent: Friday, June 19, 2009 1:52 PM
To: Ben Hutchings
Cc: dann frazier; debian-kernel@lists.debian.org; info@gplhost.com; John Wu
Subject: Re: Applying e1000e patch to the Debian 2.6.26 kernel

Thomas Goirand wrote:
> Ben Hutchings wrote:
>> On Tue, Jun 16, 2009 at 12:50:34AM +0800, Thomas Goirand wrote:
>> [...]
>>> So I'm wondering, did somebody applied some patches for e1000e 
>>> between the release of Lenny 5.01? Or maybe, the e1000e on the 
>>> X8STi-3F is different from the one of my Thinkpad t500? Is there 
>>> even different versions of the board at Intel?
>> [...]
>>  
>> There are many slightly different chips handled by the e1000e driver.  
>> If it was broken for all of them we would have fixed it before 
>> release!  You have to test on the specific model which the kernel 
>> from "lenny" doesn't work on.
>>
>> Ben.
> 
> That is quite what I think as well. Here's what I am up to.
> 
> 1/ The Debian kernel, which USED to have no support for my Thinkpad 
> e1000e works fully now. Someone MUST have done something to the 
> kernel, because I'm 100% sure that at the time of Lenny, it was not
working.
> 2/ On our last test on the Supermicro board, it did FAIL last week.
> 
> Now, I have asked my hardware supplier to put one of these X8STi-F 
> online in order to test with it, he has been very nice, and he said he 
> will. I'll be able to do a full test and report here.
> 
> Thanks to you all for helping me to find out what's going on!
> 
> Thomas

Hi again,

Finally, my hardware supplier in California has been VERY nice, and has made
it possible for me to have access to a server equipped with the Supermicro
X8STi-F motherboard that we want to use later on in production. As the guy
is really awesome, I want to name him publicly.
If you need any kind of hardware for servers, ask to talk to John Wu at
KingstarUSA in California. He will reply to you within minutes, and make
reasonable quotes. He will also be able to support you with any issue (like
now). Drop him a mail if you need, his is as Cc: here. Give him some
business, he deserves it!

Anyway, back to the patch. I have tried to apply all the patches that Ben
sent to me, and this is NOT ENOUGH to have the e1000e working. In fact, we
have to realize that, if it's still the same Ethernet controller name (I
mean e1000e), it's quite a different controller. The one on that motherboard
is a 82574L, which seems to be a more recent model. The driver source in the
kernel has many specificities for it. So if you guys, at the kernel team,
decide to add and maintain a patch, this will NOT be a tiny one just adding
new PCI IDs...

Please read (quickly) the diff file that I have attached, it shows better
than anything else what I'm talking about. It was done the "wrong" way:
taking the sources from 2.6.29 (from backports.org), making a diff -r -u of
the e1000e folder, and apply corrections until it compiles, including some
kernel API changes that I had to backport. It's a bit ugly, but IT DOES work
very well, and I'm quite happy that I have at least the solution to apply
this patch to the Debian kernel sources.

Now, if you still think that there is a possibility to have a patch included
(which would be, to my eyes, a very wise decision), I guess that I'd need to
apply patches from Git again in order to make it cleaner and smaller. I have
searched, and couldn't find the gitweb for the e1000e project.

Ben Hutchings, would you be able to do what you did before: give me a patch
collection that I can apply one by one, until I can see the driver working?
This time I need MORE patches, at least until kernel 2.6.28 (as the driver
appeared to work with the backported Debian CD including
2.6.28 kernel). Then I'll do my best to make a much smaller patch than the
one that I have attached, and hopefully, have it accepted by you guys.

Thanks for your time reading this, and the support of my work here,

Thomas


Reply to: