OS/2, Realtek and NIC drivers

[Ben]

I recently upgraded some of my hardware and my new motherboard has a built-in Realtek PCIe Gigabit NIC.

The RTGNDA13, (version 2.01),  drivers work fine for TCP/IP, however MPTN will not work with it.

I searched online, but could not find a thing.

I heard about some new NIC drivers being developed by Mensys, but I could not find a list.

Has anyone here had any success getting this NIC to work with eCS?


Update:

I could not find a better name for the NIC than that, neither in the manual, nor listed in Windows 7 Device Manager, but the board shows a chip labeled "RTL 8111E".

[ivan]

Which Realtek chip does it use?

[Ben]

Like I said in the above post RTL 8111E. I guess you posted your reply before I updated the post. No matter.

I have since found the driver I heard rumours of, it is the Mensys MultiMac and it lists my NIC.

And strangely enough, I found it on the eComstation BetaZone! a site I had long given up for dead.

Anyway, it works even worse than the Realtek 8169 driver which at least gave me TCP/IP.

Oh, well, unless someone is aware of another driver then I'm dead in the water with respect to that chipset. 8(

[Doug Bissett]

You need to go to the eComStation Software Subscription site. If you do not have that, you do not have access to anything that has been released after your subscription ran out. It must be renewed after one year (the site becomes static, as you describe, if you don't).

The ONLY version of the Multimac Realtek driver that will work with the chipset that you have, is the one from that source. It was, unfortunately released as r8169-0.3.1.wpi, which is the same name, and version, used for an earlier, defective, version of the driver. The one that works, for me, is dated 5/17/2013 (m/d/y). Bldlevel reports:

Code: [Select]

Signature:       @#D Azarewicz:0.3.1#@##1## 17 May 2013 08:46:29     DAZAR1
:::1::SVN253@@Realtek 8169/1Gb PCIe MAC Driver
Vendor:          D Azarewicz
Revision:        0.03.1
Date/Time:       17 May 2013 08:46:29
Build Machine:   DAZAR1
FixPak Version:  SVN253
File Version:    0.3.1
Description:     Realtek 8169/1Gb PCIe MAC Driver
on the r8169.os2 file. If that is what you have, please report that it doesn't work, with as much information as you can supply, at the Multimac bug tracker.

I have three Realtek Gigabit devices. PCI.EXE describes one, on board device, as:
Device 8168h RTL8111/8168B PCI Express Gigabit Ethernet controller
it is, in fact a:
10EC:8168 version 21 (RTL8168c/8111c)
(from RtlChipId.exe, that is available in the Software Subscription betazone).
A second, on board device, is described as:
Device 8168h RTL8111/8168B PCI Express Gigabit Ethernet controller
It is, in fact:
10EC:8168 version 33 (RTL8168evl/8111evl)
which is probably the same as yours.
I don't have the other one installed, but I think it is
10EC:8169 version ?? (RTL8169a/8111a)
(or something similar). It is on a PCI plug in card.

As you can see, Ivan's question is valid, you need to find out which Realtek chip it is, and the only tool, that I know about, that actually tells you that, is RtlChipId.exe, which is found in the betazone under Software Subscription (which is NOT the same as the betazone in the eCS web site).

The third one always worked with the various versions of the Multimac Realtek driver.

The first one started working, before the second one did.

The second one is working properly, with the above mentioned driver, but did some nasty things before the latest driver was available. One of those nasty things is that it disabled the ethernet interface, and the ONLY way to get it turned back on, is to turn off all power to the machine (pull the plug, - and remove the battery if it is a laptop - for a minute, or so). Even windows (and the working Multimac driver) wouldn't re-enable it, before the power was removed. The only other driver, that came close to working, was the old RTGNDA driver.

[Alex Taylor]

There are basically 3 RealTek GbE drivers available (well, 3 1/2).

• The RTGNDA driver (which you have)
• GenMAC with a RealTek driver package
• The MultiMac driver. There are actually two different drivers: the older (and unreliable) R8110 driver, and the newer R8169 driver which was written to replace it.

The RTGNDA driver works reasonably well in some circumstances, but not in others, and IIRC has some critical flaw like not being SMP-safe (or something like that).

The GenMAC driver supposedly works OK if it supports your chipset, but may require a manually-specified MAC address (see below).

The R8110 driver was apparently so badly written that the MultiMac developer decided to simply abandon it rather than try and fix it, and wrote the newer R8169 driver from scratch to replace it. However, if using R8169 you are very strongly recommended to have at least version 0.31, as earlier versions suffered from various other problems.

One other point. In all of the above cases (with maybe the exception of recent versions of R8169), on some chipsets you might not get a usable MAC address after booting. If this is the case (check your LANTRAN.LOG and/or the output of netstat -n, 'physical address' value - if it's all 0 then you have this problem), you need to manually specify the MAC address in your MPTS config. This parameter is driver specific but is usually called 'NETADDRESS'.  Note that not all versions of R8110.OS2 support this option, but you shouldn't be using R8110.OS2 anyway as it's buggy and obsolete.  I don't remember offhand if GenMAC supports it or not.

[Ben]

OK.

Here is what I got that I can add:

According to PCI.exe I have the following;

Vendor 10ECh Realtek Semiconductor Co., Ltd.
Device 8168h RTL8111/8168/8411 PCI Express Gigabit Ethernet Controller
Command 0007h (I/O Access, Memory Access, BusMaster)
Status 0010h (Has Capabilities List, Fast Timing)
Revision 06h, Header Type 00h, Bus Latency Timer 00h
Self test 00h (Self test not supported)
Cache line size 32 Bytes (8 DWords)
PCI Class Network, type Ethernet
Subsystem ID 84321043h P8P67 and other motherboards
Subsystem Vendor 1043h ASUSTeK Computer Inc.
Address 0 is an I/O Port : D800h
Address 1 is not present or not configured!
Address 2 is a Memory Address (64-bit, Prefetchable) : F8FFF000h
Address 4 is a Memory Address (64-bit, Prefetchable) : F8FF8000h
System IRQ 5, INT# A

The driver I downloaded from the BetaZone is this;

r8169-0.3.1.wpi

Which doesn't work.

According to the change log in the included readme it is;

v.0.3.1 16-May-2013 - David Azarewicz

Which would indicate that it is the same as your version, Doug, less one day...

Alex, I do believe that I have a copy of the GenMAC driver here running on one of the machines, with an RTL 8169 chipset, (I have had great success with that chip), working without problem. I haven't tried it on this new one, (RTL8111E), so I'll give that a go now, and see how that works.

Doug, I do not have access to Subscription Services though I have in the past; I do not consider it to be a valid expenditure... though I am sure that there are others who will disagree. Also, I do not believe that I should pay to beta test, anything(rather, I should be paid), though I do understand the reasons why Mensys handles it that way.

With regard to reporting software bugs, if I get the time I will do it, but I have found that in the past it produced results about 1 time in 5. Ergo: time wasted. It could have been inadequate reporting, but I have not once been asked for anything additional so...

At any rate, I'll see how that GenMAC driver does now...

[Doug Bissett]

Device 8168h RTL8111/8168/8411 PCI Express Gigabit Ethernet Controller

Unfortunately, that does NOT tell you what the chip is. Realtek seems to have about 40 different chip versions that PCI will identify as the same thing. They are not the same. You need the RtlChipId.exe program, which is found along side of the "working" driver, in the betazone. Without that information, everybody is guessing about what chip you have. There is a clue, that you have something different, because of the "/8411" entry.

less one day...

Okay, I just checked, and what is in the non SS betazone is what I have. RtlChipId.exe is also there. That means that you should collect the relevant information, and report the problem.

Doug, I do not have access to Subscription Services though I have in the past; I do not consider it to be a valid expenditure...

As for Software Subscription being a "valid expenditure", that depends on your point of view. If you want (some of) the latest stuff, before it becomes available in the "next" release, you pay for Software Subscription. However, you must also look at it as a way to support moving eCS (OS/2) forward. IMO, I can't afford NOT to have SS, it is a way to support the work that is being done, and, you get the "next" release, when it becomes available. Not a bad deal, although it would be better if "next" releases were better managed, and more frequent.

To correct what I wrote earlier (About the "third device"):

Code: [Select]

10EC:8169 version ?? (RTL8169a/8111a)
(or something similar). It is on a PCI plug in card.
should read:

Code: [Select]

10EC:8169 version 3 (RTL8169sb/8111sb)
It is on a PCI plug in card.


At any rate, I'll see how that GenMAC driver does now...

I couldn't get that one to work with the 10EC:8168 version 33 (RTL8168evl/8111evl) device. It did work with the 10EC:8169 version 3 (RTL8169sb/8111sb) device, but only at 100 Mbs. When I tried it at 1000 Mbs, it would trap in GENMAC. You could try updating to the windows driver that comes with it, but that may not work at all. GENMAC is definitely limited, and most newer windows drivers won't work.

BTW, which RTGNDA driver do you have? The one that I was using is v1.13. BLDLEVEL shows:

Code: [Select]

Signature:       @#chagrin:1.13#@Realtek RTL8169 OS/2 NDIS 2.0.1 MAC Driver
Vendor:          chagrin
Revision:        1.13
File Version:    1.13
Description:     Realtek RTL8169 OS/2 NDIS 2.0.1 MAC Driver
on the rtgnda.os2 file. I also had to use some parameters to make it use IRQs above 15, and to add the MAC address.

[walking_x]

on the rtgnda.os2 file. I also had to use some parameters to make it use IRQs above 15, and to add the MAC address.

MAC address parameter must be uncommented in NIF and, in most cases, set up manually. This driver set zero MAC too often
And it works, except using it as netbeui server. Netbeui client and all kinds of netbios over ip work fine.

[Andi B.]

And it works, except ....

Have to repeat myself whenever I read rtgnda as I really had a hard time to find that out -

Do a search on the internet before trying it. Years ago I installed my system with it and got sporadic traps. Took me months till I found out it was this driver. After that others told me this driver is known to sporadically trap systems. Maybe it was Eugene so probably there are some traces about ecomstation.ru...

[Doug Bissett]

Years ago I installed my system with it and got sporadic traps

When I was using RTGNDA, nothing else would work, at all, so I figured that potential traps were better than no networking. FWIW, I used it for a couple of months, and never had a problem with it, other than I couldn't use it with VBox (which didn't matter anyway). I haven't used NETBIOS (not over TCP/IP) for many years. I have an Asus WL-330g that will not work with it. If anybody is interested, rtgnda13.zip comes with source code.

I did take the steps necessary to report the Realtek problem, and work with those who know about such things, to get it fixed. Sitting back, waiting for somebody else to do it, is not an option, although it took a second report to get enough interest to get it done. Apparently, the Linux crowd has been having similar problems with Realtek devices, but there are a lot more of them working on the problem.

[Ben]

You need the RtlChipId.exe program, which is found along side of the "working" driver, in the betazone. Without that information, everybody is guessing about what chip you have. There is a clue, that you have something different, because of the "/8411" entry.

OK.

Done.

Here it is:

"Found Realtek NIC at bus 3:1:0: 10EC:8169 version 3 (RTL8169sb/8110sb)"

At any rate, I'll see how that GenMAC driver does now...

I couldn't get that one to work with the 10EC:8168 version 33 (RTL8168evl/8111evl) device. It did work with the 10EC:8169 version 3 (RTL8169sb/8111sb) device, but only at 100 Mbs. When I tried it at 1000 Mbs, it would trap in GENMAC. You could try updating to the windows driver that comes with it, but that may not work at all. GENMAC is definitely limited, and most newer windows drivers won't work.

I have not had the time yet to test my version on the new machine. All that I can say it that it worked so well these long years that it has been in that I forget it was in there; totally transparent.

Samba won't install under Ubuntu and needs to be beat over the head a few times. I will post the results of the GenMac RTL driver when I have them.

[Doug Bissett]

"Found Realtek NIC at bus 3:1:0: 10EC:8169 version 3 (RTL8169sb/8110sb)"

Interesting... That is the same one that always worked with Multimac, for me, but mine is on a PCI plug in card. It also worked with GENMAC, but only at 100 Mbs.

Samba won't install under Ubuntu and needs to be beat over the head a few times.

Odd. SAMBA comes with Linux Mint (which is Ubuntu under the covers). All it needed was to be configured.

[Andi B.]

At that time I looked into the rtgnda sources but honestly I'm not so masochist these days anymore to try fixing a NIC driver written in assembler. Multimac is the way to go in this case.

The os2 kernel sources would be the only reason for me if available to dig into assembler again.

[Steven Levine]

"Found Realtek NIC at bus 3:1:0: 10EC:8169 version 3 (RTL8169sb/8110sb)"

This chipset is documented as supported by the multimac driver.  I guess it's possible that the PCI-e bus is an issue, but the only way to get this resolved is to submit a ticket  at svn.ecomstation.nl/multimac

[Doug Bissett]

r8169-0.3.1.wpi

Which doesn't work.

I guess it is time to ask the tough question. What, exactly, do you mean by "Which doesn't work"

If you report that "it doesn't work", nobody is going to bother to ask you "What doesn't work"? They also won't bother trying to figure it out. They need to know, exactly, what it is that they are trying to fix, and how to reproduce the problem.

It is also possible, that you have something else that is causing the problem. With more information, we might be able to spot the problem, without bothering the developer, who is overloaded with work already.

NETBIOS *might* be the missing link. As I said, I haven't used that, for years. It was causing problems when I quit using it, and I am sure that it hasn't changed. The last time I tried NETBIOS over TCPIP, it was causing more problems than SAMBA (which is certainly not perfect), but it worked better than NETBIOS, for me, when I was using it.