Supporting thread for the Intel Wired Gigabit Ethernet Driver e1000 port bounty (http://www.os2world.com/content/view/19642/71/).
Already sort of done, available on Intel's site, was also an IBM SWC driver but yes, it wouldnt be a complete uptodate as of 2010 driver.
I have an 2 different Intel Gigabit card's, one requires the Intel driver, the other only works with IBM's Gigabit card driver.
Intel driver e1000_3.61.zip is here :-
http://www.os2site.com/sw/drivers/network/intel/index.html
The one in my \ibmcom\macs directory is dated 2006, so I've obviously got a more up todate one somewhere. Might have come with the card.
The IBM Gigabit driver (uses the Intel Giga chipset) should be available via eComStation's website, there are 6 different intel cards that I know of that work fine. Will have to make a list.
Cheers
These old drivers no matter from where ONLY supports PCI based adapter. This bounty is for newer PCI-E based cards and chipsets. I even have an official answer from intel that there are only old drivers that wont work with newer cards ;)
Intel stated in 2005 when the Pro Set/10 was released, in its changelog, that this would be their last OS/2 driver release, unfortunately.
Ya, I figured all of that out since my post. I'm also looking for the linux source so when I
get back to a few other things I need to update, I'll see if its viable, which having looked
at a few other drivers probably means its not viable for me to do (I have time constraints
and more often than not drivers require extensive work and missing dependancy packages).
but you never know until you look, and see what old OS/2 code has been left behind, maybe.
Add to that I have no idea about PCI-E.....
Cheers
Best starting point would be if we had the last Intel driver source code. Did someone ask Intel about that?
Hi,
I am a bit puzzled what everyone here is talking about. My understanding was, that the bounty stated it clearly:
Quote
Description: Port the linux intel wired Ethernet Driver to eComStation and OS/2 to support the following adapters.
[...]
The requirements are:
* Port them to a Native eCS-OS2 driver
* This driver needs to work with high irq and acpi in general (oemhlp and 'apm')
* Optional: Enable multi port feature
[...]
Source Code Available: Intel Site
...and the "Intel site" is a link to the .tar file.
Did I miss on something? If so, we need to rephrase the bounty...
Cheers,
Thomas
the bounty is quite clear... this link is for the developer as its the source code mentioned in the bounty.
To have old driver source from intel might be a nice addition yes, but like this we need to deal with the legal department. In case someone has a source he can do as normal tech support wont answer any of those questions. Without it its still possible to have a nice port of the driver.
I have a desktop machine with Intel 82578DC Gigabit network adapter. I offer my self as beta tester for any developer interested on this bounty.
I think the bounty system is a good idea, but fear that it will take very long time to collect the amount
necessary for getting a developer interested. Is there a developer out there that could make an offer?
If so, we could perhaps get the money the way it was done for Qt4?
This friday i can not reveal more then that I'm talking to a developer to port the 1000e and e1000e module. (I could be mixing up the names). There are 2 modules in the Linux kernel that support just about all the Intel Gigabit ethernet drivers. A port of the first module should be ready for public beta within 2 1/2 months.
This depends however on the talks I'm having. So I guess this bounty is not realy needed anymore. The source code will be put in public repository at Netlabs.
I'm busy next week working out the final details Mensys should be able to sign the contract hopefully next week.
Roderick Klein
Mensys
Hey Roderick,
...that sounds like good news!
Thanks for the updates!
Cheers,
Thomas
Interesting news about this bounty. First the driver modules are separated, but if we would have a port of both of them nice :)
How came mensys to this idea to support the development of this specific driver? Would be nice when we would have some updates on the progress of the contract and planing. In case this all would take place then we will move to a new bounty and no money is lost.
We had the idea because it makes business sense. End users and companies using eCS mostly have Intel based NIC's. As I typed I would give a reply if there is news. I just mentioned the NIC driver last week. We are busy working out the contract details.
From my point of view the bounty system is a wishing well. I talked to Nick also, who did the Nvidia driver.
The company we are looking todo it with _estimates_ its about 2 month of work the first of 2 Linux modules.
That is full time work. Nick also made a rough estimate and its not a small project!
Hence the thing with these projects you need much more money to then lets say 10 years back. The bounty system does not seem to achieve critical mass to get that money. Also the price tag of Russian and Ukraine developers have gone up. And they have of course every right since they are extremely skilled.
Roderick Klein
Mensys
Hi Roderick,
Quote from: rwklein on 2010.03.24, 15:35:24
From my point of view the bounty system is a wishing well.
You are correct. But it was created more or less to be exactly like that.
It was never intended to be used for "professional" sponsoring with companies and contract things.
Quote
That is full time work. Nick also made a rough estimate and its not a small project!
Then it is perhaps not suited for the bounty system!
Quote
Hence the thing with these projects you need much more money to then lets say 10 years back. The bounty system does not seem to achieve critical mass to get that money. Also the price tag of Russian and Ukraine developers have gone up. And they have of course every right since they are extremely skilled.
Well, heck. The bounty system cannot create that critical mass, because for such a mass of money it would need features to control the delivery which are not available with the bounty system. What that means is: People will not invest LARGE amounts of money because the bounty system doesn't provide the TRUST that "you get what you pay for". There is no guarantee. There is no "what if" rules covered.
In all, the bounty system lacks things that people need for investing into professional project funding.
But: It lacks that because it wasn't built for that purpose.
What we can think about is:
- Set up a project managing like with Qt4
- We will encourage people to send the money to that project (like Qt4)
- We will ask the sponsor(s) of the current bounty here if os2world is allowed to shift the money to that new project
Then, no more argueing about the bounty system. "Obtaining the critical mass" will then be on your side of the fence. ;)
About russian developers:
Yeah, yeah, bla bla... they are extremely skilled. Sure. But if they are too hard to handle and can't live to their promise (regardless how skilled they are) it's worth nothing. I'm sure we can find equally skilled developers in say Romania, Bulgaria, heck, the Faroer Islands if you want. You don't have to stick with Russia if they are not worth their price tag in the end.
I agree that "you get what you pay for": Pay peanuts and you'll get monkeys. But what is the benefit of paying MORE to get BETTER people but in the end, you never know IF or WHEN you get your code??
From my offshore / outsourcing experience I can tell you there are two ways that work:
1) Get your "cheap" workers colocated HERE to guide them (heck, yes: sit next to them and CONTROL them)
2) Leave them over in their country but tell them about FINES that apply if they don't live up to the agreed scope of delivery and timelines
(And get a lawyer that kicks their ass if they don't seem to take work orders and contracts for serious... )
Everything else - sadly enough - is romantic idiotism.
[ This was a free advice from someone who works in offshore software projects since a couple of years. ]
Cheers,
Thomas
I see no problem to have a bounty for this driver port, we had already success with another one but if mensys feels free to take over and make a project for it then all is fine. So how is the progress on it? Contract is on the way and we can see some work soon?
Mike,
Look at the announcement in the OS/2 & eCS Software News.
Updated status on: Gigabit Ethernet and HDA Drivers
Heya,
a bit of update on this:
During todays edition of my frequent lunch chats with Rodrick Klein, I was told that Mensys has successfully finished the development phase of the new Intel (wired) NIC driver (the one in question here). Mensys aims to enter a public beta testing phase for the new driver starting within this week by releasing the driver "Intel e1000e" (e1000e.os2) to the eCS Beta Zone.
An official announcement with an according link and information to test volunteers will follow.
Cheers,
Thomas
Quote from: warpcafe on 2010.06.07, 23:07:35
Mensys aims to enter a public beta testing phase for the new driver starting within this week by releasing the driver "Intel e1000e" (e1000e.os2) to the eCS Beta Zone.
An official announcement with an according link and information to test volunteers will follow.
Thomas, do you know if the testing phase started?
Hi Martin
People on the ecSTestTeam received a post about this recently. Just to update everbody on the "state of play" here is the email:-
Hello,
As promised at Warpstock 2010 in Trier. A new Intel NIC driver would be
coming out. This is based on the source code of e1000e.
It supports the following chipsets:
http://www.os2world.com/content/view/19642/71/
Attached is e1000e.ziq rename it to e1000e.zip.
Let me know if the driver works. You don't have to open tickets.
One outstanding is that the NIC driver in some cases does not get a
DHCP lease this has been reported.
We have 2 systems on which it works so far here.
If it does not work, it could hang on netbind.exe or you have an error
in lantran.log.
If it does not work send the file \ecs\install\rsp\rawdata.000 to me
and lantran.log.
Then I will pass it along to the company developing this NIC driver.
Please don't distribute this driver beyond the this group as first some
more testing is needed.
Roderick
So, looks like it is not quite ready for a public test yet.
Regards
Pete
Mensys has announced their interest on claiming this bounty.
The binaries are available for download at: http://svn.ecomstation.nl/multimac (http://svn.ecomstation.nl/multimac)
The source code is also posted on that site.
Mensys had requested that the founds claimed will be donated back to the "Eclipse.org Standard Windowing Toolkit (SWT) OS/2 Port" bounty.
Any feedback from the community is welcome in this thread. I will be closing the bounty and moving the funds in two weeks.
Thanks
Martin Iturbide
OS2World.com Bounty Admin.
Quote from: miturbide on 2011.04.06, 00:55:37
Mensys has announced their interest on claiming this bounty.
The binaries are available for download at: http://svn.ecomstation.nl/multimac (http://svn.ecomstation.nl/multimac)
The source code is also posted on that site.
Mensys had requested that the founds claimed will be donated back to the "Eclipse.org Standard Windowing Toolkit (SWT) OS/2 Port" bounty.
Any feedback from the community is welcome in this thread. I will be closing the bounty and moving the funds in two weeks.
I believe they are doing a pretty good job on updating NIC drivers.
BTW... any chance of Intel 82579 being supported by e1000 on eCS? It seems it'll be present on my new motherboard and Linux e1000 seems to support it already, but it is not listed on eCS's e1000 wiki. :/
Hi Daniel,
Quote from: djcaetano on 2011.04.06, 16:23:39
Quote from: miturbide on 2011.04.06, 00:55:37
Mensys has announced their interest on claiming this bounty.
The binaries are available for download at: http://svn.ecomstation.nl/multimac (http://svn.ecomstation.nl/multimac)
The source code is also posted on that site.
Mensys had requested that the founds claimed will be donated back to the "Eclipse.org Standard Windowing Toolkit (SWT) OS/2 Port" bounty.
Any feedback from the community is welcome in this thread. I will be closing the bounty and moving the funds in two weeks.
I believe they are doing a pretty good job on updating NIC drivers.
BTW... any chance of Intel 82579 being supported by e1000 on eCS? It seems it'll be present on my new motherboard and Linux e1000 seems to support it already, but it is not listed on eCS's e1000 wiki. :/
The e1000 driver for eCS is pretty easy to build - I'd imagine it would be fairly easy to merge in support for newer chipsets from the Linux module. When you get your board, if it doesn't work, I'll try take a look for you if nobody else steps up.