Hi. ----- Original Message ----- From: "Szymon Madejczyk" <[EMAIL PROTECTED]> To: <[EMAIL PROTECTED]> Sent: Friday, May 31, 2002 4:04 PM Subject: [WinPcap-users] About Q-15 in WinPcap FAQ
> Hi, > > Q-15 in WinPcap says : > > >Q-15: Why doesn't WinPcap work on my multiprocessor (SMP) machine? > > > >WinPcap does not support on SMP machines, because we don't have > multiprocessor computers on which to >develop/test it. For this reason, > recent versions have an initialization check that prevents them to load on > SMP >machines. > >Older versions don't have this check, so they are loaded by the operating > system (the driver seems to work) >but the results are generally wrong. > Remember, it is not safe to run these versions under smp machines, since > >hangups of the machine, data corruptions, blue screens can happ > > I'v commented initialization check for processor count in npf.sys sources > and recompiled it. > I'm using it on my SMP machine. A few programs that use winpcap are running > and everythings looks ok. > I've compared results with other programs that don't ude wpcap ant results > are the same. > Do someone know what "the results are generally wrong" exactly mean ? It means that we do not have a SMP to make tests, some users reported strange behaviours (especially wrong timestamps), and in theory the driver can have problems since the callback that receives packets (the so called TAP) can run on both the CPUs, even concurrently, and the shared resources are not locked at all. However, I must repeat, we have not done any test in our labs, since we do NOT have a SMP machine. Stating that on your SMP system the driver seems to work is not sufficient to state that WinPcap is SMP-safe(for example, have you tested winpcap under heavy network load? With many capture instances open at then same time?) GV > > Regards, > Szymek > >
