thank you, can i just confirm something? On Mon, Jun 8, 2009 at 2:13 PM, Peter Memishian<peter.memish...@sun.com> wrote: > > > from http://www.opensolaris.org/jive/thread.jspa?messageID=364896񙅠 > > > > i can guess from stack trace that tcp_send_data() is in the execution path. > > > > but searching through source code there are so many tcp_send_data: > > > > > http://src.opensolaris.org/source/xref/kernel-sockets/ksocket/usr/src/uts/common/inet/tcp/tcp.c#18843 > >
kernel-socket this is the socket implementation that attempt to get rid of STREAM, correct? > > > http://src.opensolaris.org/source/xref/nwam/nwam1/usr/src/uts/common/inet/tcp/tcp.c#18342 > > > > > http://src.opensolaris.org/source/xref/onnv/onnv-gate/usr/src/uts/common/inet/tcp/tcp.c#18342 > > > > what are their relationship? what is nwam? kernel-sockets? this is the userspace implementation of nwamd, manager of the network interface. > > Those are project gates, based on some version of onnv-gate. The only one > you need to be concerned with is onnv-gate. > > > another questions is how is STREAM and tcp related? is tcp built on > > top of stream? > > Yes and no -- and the answer to this has changed over time. As of now, > TCP is part of the IP module, which is indeed a STREAMS module, but the > datapath generally does not use STREAMS. I say "generally" because there > are compatibility modes that will lead to STREAMS being used on the > datapath (see e.g., so_tpi_fallback()). I'd recommend using DTrace is to > understand the common TCP codepaths. > from the stacktrace as given in the crashdump in the URL above, putnext() is called, so we can deduced that tcp is STREAM implementation based. correct? thank you for the info...i will look into so_tpi_fallback() and dtrace. > -- > meem > -- Regards, Peter Teoh _______________________________________________ opensolaris-code mailing list opensolaris-code@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/opensolaris-code
