s for the USB drivers. The beta was not released of eCS 2.2 until all issue's where traced and fixed on our equipment. We had 1 system that would hang on the USB drivers after the last device driver was loaded, this delayed the beta release with 3 days. Just like ACPI the USB drivers have been tested on very broad range of systems. Also the USB drivers shipped in eCS 2.2 beta are not 100% the same compared to Lars Erdmann his drivers. The drivers had been slightly modified by David and tested on his 10 systems at home. Also all systems in the Mensys office where used to test the USB drivers and to test booting from USB CD ROM (another 30 systems or so). And all these systems are of different brands and chip sets BIOS makes etc.
I hate to have to point this out, but Lars knew about the problem with USBEHCD.SYS, days before the release of eCS 2.2 beta. Obviously, somebody decided to ignore the situation. Whether the beta should have been delayed, yet again, or released with the far more stable 186 version of the USB drivers, is a decision that needs to be made. It is true, that a few people had problems with version 186, but they were far fewer than those who had trouble with the 189 version. Then, if you made other changes, that were basically untested by those who did have known problems, I don't feel sorry for you at all. You deserve all of the flack that you are going to get. I do feel sorry that eCS users don't have a good product to test.
Sadly the Realtek dossier is far from simple. And eCS 2.1 also had the same issue!
That the old driver from 2.1 works for you has serious issue's and would cause kernel traps on other systems.
Fair enough, but that is no excuse for releasing a driver that was known, days before release, to have a very serious problem on some systems. I have two systems with different Realtek NICs, and both of them experienced the problem where the interface was disabled, until I did a full power off to get them reset. FWIW, the only driver that works with the Realtek 8168 in my Lenovo ThinkPad L530, is the old RTGNDA driver, which is not included with eCS 2.2. It is not perfect, but working most of the time, is a lot better than not working at all, and causing serious problems that the average user would have no idea what to do about.
In fact, I haven't got around to checking the on board Realtek 8168, that died the first time I tried the old r8110 driver (it was new at the time). The symptoms were exactly the same. It had been working with the GENMAC driver before that (with a few problems). It may actually work if I enable it, and load a good driver in it.
Realtek makes a complete and utter total mess of its cipset and PCI id's it uses.
I realized that a couple of years ago, yet you recently recommended Realtek NICs. When I got the L530, I found no place where they mentioned the wired NIC chipset. I specifically got an Intel WiFi, looking forward to support from Multimac. If I had realized it had a Realtek wired NIC, I probably would not have got it. My older T510 has an Intel wired NIC, and a Realtek WiFi, which is also waiting for Multimac. I guess I will get the T510 repaired (something died in the power control circuits), and switch the Intel WiFi with the Realtek one. The L530 has other problems, so it might as well be totally useless, and live in the windows world.
FWIW, The WiFi in the L530 is an Intel 8086:0891. I can get GENMAC to load three windows drivers, and they all seem to work, but I can find no way to get the radio turned on (there is no physical switch). It acts exactly like a wired NIC with the cable disconnected. I can load the driver, assign an IP address with IFCONFIG, and I can PING it, but that is the end of the road. I think there are a number of other machines, with Intel WiFi NICs, that have the same problem.
Today, I also observed a very interesting phenomenon. I was doing some big file copies, using SAMBA client on the machine with a Realtek 8169 chipset, and the r8169 Multimac driver. It was doing a sustained data transfer at about 9,000 KB/s. At the other end was a SAMBA server running with an NVIDIA chipset, and the Multimac driver. I then started up an RSync data transfer, using the L530, with a Realtek 8168 chipset, and the RTGNDA driver. It was running happily at about 32,000 KB/s, to the same NVIDIA chipset, which was happily handling about 41,000 KB/s data rate. This is, of course on a 1 GB network. Looking on the internet, it is suggested that a 1 GB network should be able to run near 60,000 KB/s. In fact, my antique IBM ThinkPad A22e, running an Intel 100 Mbs NIC, runs at sustained speed of about 8,000 KB/s. There is a serious problem with the Realtek Multimac driver, which needs to be fixed, since it seems to be a popular brand with many manufacturers using it.