| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
(see http://www.freebsd.org/cgi/query-pr.cgi?pr=kern/32827).
itojun@ ok
|
|
|
|
|
|
Make insertion of data into socket buffers O(C):
* Keep pointers to the first and last mbufs of the last record in the
socket buffer.
* Use the sb_lastrecord pointer in the sbappend*() family of functions
to avoid traversing the packet chain to find the last record.
* Add a new sbappend_stream() function for stream protocols which
guarantee that there will never be more than one record in the
socket buffer. This function uses the sb_mbtail pointer to perform
the data insertion. Make TCP use sbappend_stream(). On a profiling
run, this makes sbappend of a TCP transmission using
a 1M socket buffer go from 50% of the time to .02% of the time. Thanks
to Bill Sommerfeld and YAMAMOTO Takashi for their debugging
assistance!
|
|
|
|
|
|
|
|
well (not at all) with shortages of the vm_map where the pages are mapped
(usually kmem_map).
Try to deal with it:
- group all information the backend allocator for a pool in a separate
struct. The pool will only have a pointer to that struct.
- change the pool_init API to reflect that.
- link all pools allocating from the same allocator on a linked list.
- Since an allocator is responsible to wait for physical memory it will
only fail (waitok) when it runs out of its backing vm_map, carefully
drain pools using the same allocator so that va space is freed.
(see comments in code for caveats and details).
- change pool_reclaim to return if it actually succeeded to free some
memory, use that information to make draining easier and more efficient.
- get rid of PR_URGENT, noone uses it.
|
|
art@ ok
|
|
|
|
verified with both netbsd and freebsd.
from netbsd:
Tue Jun 8 02:39:57 1999 UTC by thorpej
In sosend(), if so_error is set, clear it before returning the error to
the process (i.e. pre-Reno behavior). The 4.4BSD behavior (introduced
in Reno) caused transient errors to stick incorrectly.
This is from PR #7640 (Havard Eidnes), cross-checked w/ FreeBSD, where
Bill Fenner committed the same fix (as described in a comment in the
Vat sources, by Van Jacobsen).
|
|
from NetBSD:
Wed Jan 7 23:47:08 1998 UTC by thorpej
Make insertion and removal of sockets from the partial and incoming
connections queues O(C) rather than O(N).
|
|
|
|
it is to be friendly with postfix daemon-to-daemon communication
(not 100% sure if which behavior is correct, specwise). patch similar to netbsd.
|
|
|
|
|
|
okay deraadt@, millert@
from jlemon@freebsd.org:
extend kqueue down to the device layer, backwards compatible approach
suggested by peter@freebsd.org
|
|
is disconnected by RST right before accept(2). fixes NetBSD PR 10698/12027.
checked with SUSv2, XNET 5.2, and Stevens (unix network programming
vol 1 2nd ed) section 5.11.
|
|
do not return junk data in mbuf (= sockaddr on accept(2)'s 2nd arg).
set the length to zero.
behavior checked with bsdi and freebsd.
partial solution to NetBSD PR 12027 and 10698 (need more investigation).
|
|
okay art@, millert@
|
|
This fix is taken from BSD/OS (the file in question being BSD licensed).
It continues to remove a datagram from a socket receive buffer even if there is
an error on the copy-out, so as to leave the buffer in a reasonable state.
Before, the kernel would stop in mid-receive if the copy-out failed, and the
buffer's structural requirements would be violated (since the start of a
datagram must be an address iff ).
Note that if the user provides any invalid addresses as arguments to a
recvmsg(), the datagram at the front of the buffer will be discarded. The more
correct behavior would be not to remove this datagram if the arguments are
invalid. Implementing this behavior requires a lot of significant changes, and
socket receives are a critical path.
Also included are two simple and fairly obvious fixes from the same source.
If non-blocking I/O is set, it makes sure the receieve is non-blocking. It also
fixes a slightly over-aggressive optimization.
|
|
|
|
|
|
|
|
|
|
(size_t) and don't return EINVAL if it is < 0 in sys_{read,write}. Remove check for uio_resid < 0 uiomove() now that uio_resid is unsigned and brack remaining panics with #ifdef DIAGNOSTIC. vn_rdwr() must now take a size_t * as its 9th argument so change that and clean up uses of vn_rdwr(). Fixes 549 + more
|
|
fast.. Sigh. I will change this again later
|
|
|
|
|
|
|
|
and euid as well, then deliver them using new csignal() interface
which ensures that pgid setting process is permitted to signal the
pgid process(es). Thanks to newsham@aloha.net for extensive help and
discussion.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
SOMAXCONN (kern.somaxconn), SOMINCONN (kern.sominconn), and TCPTV_KEEP_INIT
(net.inet.tcp.keepinittime). when this is not enough (ie. overfull), start
doing tail drop, but slightly prefer the same port.
|
|
|
|
explorer@flame.org
|
|
|
|
|
|
|
|
|
