interface controllers (of varying intelligence levels).
Contributed by Wasabi Systems, Inc. Primarily written by Steve Woodford,
with some modification by me.
throughput significantly in a wide variety of test cases, including
local gigabit ethernet with both jumbo and standard frames,
transcontinental (U.S.) connections with e2e bandwidths ranging from
10Mbit/sec to 155Mbit/sec, and on a variety of test connections
between the NetBSD Project public servers and machines in Australia.
The impact of this change is less dramatic for high-delay connections
when Path MTU is in use but still measurable.
For optimal performance on local gigabit networks, a higher socket
buffer size (at least 64K) will still yield a substantial improvement
in performance, but 32K gets us most of the way there in my test
cases, with only a cost of _doubling_ memory use per socket rather
than _quadrupling_ it.
N.B. Windows NT, at least since Win2k SP2, uses a default socket buffer
size (or their analogue thereof) of 64K, which is a useful data
point.
This fixes a bug introduced in revision 1.120 of ffs_vfsops dated 2003/09/13
which results in fs_flags having a value of 0x7fffff00 when a superblock
is updated to use the new layout.
Discussed in http://mail-index.netbsd.org/tech-kern/2003/09/28/0003.html
to pci_attach_args. The latter is allocated on the stack during auto-
configuration and so will not be valid after that time.
It's amazing how the old code worked for so long. I guess pci_attach_args
is allocated deep in pid#0's kernel stack on most platforms.
* return NULL to indicate an error if a NULL name is passed
* fix a crash if description is NULL
Thanks to Julian Coleman for finding and fixing these.
fit in the number of bits used for the port still does something.
This fixes PR pkg/18741 for ac97-based hardware. Other audio drivers
might need a similar fix.
thorpej