On several computers in which I have NetGear FA311/FA312 NICs, I found that I was periodically experiencing either crashes or network stalls depending on the flavor of Windows in use. On machines running Windows 2000 which are susceptible to BSOD's (Blue-Screens-Of-Death), I was getting reports that the problem was due to a DRIVER_IRQL_NOT_LESS_OR_EQUAL bug check occurring in NDIS.SYS, which is part of Windows' network infrastructure. Upon further examination, I discovered these problems were due to a flaw in the drivers for the NetGear FA311 / FA312 NICs, and in the reference drivers provided by National Semiconductor for the DP83815 MacPhyter chipset used in these NetGear NICs.
Based on my research, I was able to write a test program that reliably causes the same BSOD in Windows 2000 on any machine with one of these NICs. When the same test program is ran on a machine running Windows 9x/ME, the computer continues to function, but the network becomes non-operational. I have been able to reproduce the problem with v1.30 and v1.80 of the NetGear drivers, and v5.0.119.0 and v5.0.124.0 of the National Semiconductor drivers. I have also confirmed this problem with numerous other people world wide.
For anyone interestd in the root cause of the problem, it is due to a limitation of the National Semiconductor DP83815 / MacPhyter reference driver. Basically, if a packet is split into too many fragments as it is being passed down the protocol stack, the packet gets thrown away by the driver after the driver has already committed to Windows that it will notify Windows once the packet has been sent. As a result, Windows gets into an infinite loop asking the driver to send the packet.
I have contacted both NetGear and National Semiconductor regarding this issue. After much dialog with NetGear, NetGear refuses to acknowledge that this is a problem with their product. On the other hand, National Semiconductor has acknowledged that they are aware of the problem, and that they expected to release new drivers "in a couple weeks". Unfortunately, that was a few months ago. If you are buying a new network card, I would avoid the cards I have mentioned until the drivers have been fixed. If you have one of these cards and are experiencing such problems, maybe this will shed some light on what is happening. Lastly, if you are concerned about this problem, I would encourage you to contact NetGear or National
Semiconductor.
If you'd like to experiment with my test program, drop me a note at ccooper21@hotmail.com and I'd be glad to send you the program and source code for it.
RWilco
Based on my research, I was able to write a test program that reliably causes the same BSOD in Windows 2000 on any machine with one of these NICs. When the same test program is ran on a machine running Windows 9x/ME, the computer continues to function, but the network becomes non-operational. I have been able to reproduce the problem with v1.30 and v1.80 of the NetGear drivers, and v5.0.119.0 and v5.0.124.0 of the National Semiconductor drivers. I have also confirmed this problem with numerous other people world wide.
For anyone interestd in the root cause of the problem, it is due to a limitation of the National Semiconductor DP83815 / MacPhyter reference driver. Basically, if a packet is split into too many fragments as it is being passed down the protocol stack, the packet gets thrown away by the driver after the driver has already committed to Windows that it will notify Windows once the packet has been sent. As a result, Windows gets into an infinite loop asking the driver to send the packet.
I have contacted both NetGear and National Semiconductor regarding this issue. After much dialog with NetGear, NetGear refuses to acknowledge that this is a problem with their product. On the other hand, National Semiconductor has acknowledged that they are aware of the problem, and that they expected to release new drivers "in a couple weeks". Unfortunately, that was a few months ago. If you are buying a new network card, I would avoid the cards I have mentioned until the drivers have been fixed. If you have one of these cards and are experiencing such problems, maybe this will shed some light on what is happening. Lastly, if you are concerned about this problem, I would encourage you to contact NetGear or National
Semiconductor.
If you'd like to experiment with my test program, drop me a note at ccooper21@hotmail.com and I'd be glad to send you the program and source code for it.
RWilco
Facebook
Twitter