david/ipxe
david/ipxe
Archived
1
0
Fork 0
Code
2,749 Commits 1 Branch 0 Tags 14 MiB
7d36a1b7b0
Commit Graph

71 Commits

Author SHA1 Message Date
Michael Brown
b59e0cc56e [i386] Change [u]int32_t to [unsigned] int, rather than [unsigned] long 
This brings us in to line with Linux definitions, and also simplifies
adding x86_64 support since both platforms have 2-byte shorts, 4-byte
ints and 8-byte long longs.
 2008-11-19 19:15:44 +00:00
Michael Brown
1a68d3fef3 [TCP] Avoid shrinking TCP window 
Maintain state for the advertised window length, and only ever increase
it (instead of calculating it afresh on each transmit).  This avoids
triggering "treason uncloaked" messages on Linux peers.

Respond to zero-length TCP keepalives (i.e. empty data packets
transmitted outside the window).  Even if the peer wouldn't otherwise
expect an ACK (because its packet consumed no sequence space), force an
ACK if it was outside the window.

We don't yet generate TCP keepalives.  It could be done, but it's unclear
what benefit this would have.  (Linux, for example, doesn't start sending
keepalives until the connection has been idle for two hours.)
 2008-06-05 00:28:17 +01:00
Alexey Zaytsev
a1572e0ab0 Modify gPXE core and drivers to work with the new timer subsystem 
Signed-off-by: Alexey Zaytsev <alexey.zaytsev@gmail.com>
 2008-03-02 03:41:10 +03:00
Michael Brown
f6a8158eed Make seek information part of the xfer metadata, rather than an entirely 
separate xfer method.

Add missing .alloc_iob entries to several xfer_interface_operations
structures.
 2008-01-08 16:46:55 +00:00
Michael Brown
df868476e7 Various warnings fixups for OpenBSD with gcc-3.3.5. 2007-12-07 00:11:43 +00:00
Michael Brown
2ff1b1245b Use start_timer_nodelay() in protocols which rely on the retry timer 
to generate the initial transmission; this cuts off around 0.3s per
instantiated connection.
 2007-08-13 11:03:33 -07:00
Michael Brown
9aa61ad5a2 Add per-file error identifiers 2007-07-24 17:11:31 +01:00
Michael Brown
096fa94f0c Add support for TCP timestamps 2007-07-13 11:32:53 +01:00
Michael Brown
eb530845d4 Adjust received length to take into account any already-received data 
in tcp_rx_data().

Clarify comments on discarding duplicate or out-of-order data.
 2007-07-13 11:31:58 +01:00
Michael Brown
d5735c631c Avoid reusing auto-allocated ports after connection close. 2007-07-13 11:25:00 +01:00
Michael Brown
edded7546e Limit xmit window to one MTU. (Path MTU discovery not yet 
implemented; should be done at some point.)
 2007-07-08 14:33:53 +01:00
Michael Brown
35afb379af TCP limits advertised TCP window to size of application window 
obtained via xfer_window().
 2007-07-08 14:14:59 +01:00
Michael Brown
b34d4d0449 Separate the "is data ready" function of xfer_seek() into an 
xfer_window() function, which can return a scalar rather than a
boolean.
 2007-07-08 14:11:07 +01:00
Michael Brown
ca4c6f9eee Kill off unused request() method in data-xfer interface. 2007-07-08 02:10:54 +01:00
Holger Lubitz
dc6c1c294d convert to zalloc 2007-07-06 20:47:15 +02:00
Michael Brown
5de837cc67 Mildly ugly hack to force correct linkage. 2007-06-28 18:06:25 +01:00
Michael Brown
95adce0ce1 Add concept of transfer metadata, to be used by UDP in order to 
implement sendto()/recvfrom() equivalents.
 2007-06-11 15:04:39 +01:00
Michael Brown
3acd51597a Add sanity check and extra debug message 2007-05-28 20:06:41 +00:00
Michael Brown
1311b4194c Eliminate PF_INET; just use the AF_INET from the struct sockaddr instead. 2007-05-26 20:41:23 +00:00
Michael Brown
b8618d76db Modify data-xfer semantics: it is no longer necessary to call one of 
request(), seek() or deliver_xxx() in order to start the data flow.
Autonomous generators must be genuinely autonomous (having their own
process), or otherwise arrange to be called.  TCP does this by
starting the retry timer immediately.

Add some debugging statements.
 2007-05-26 15:04:36 +00:00
Michael Brown
1136c19fa4 Updated tcp.c to provide a standardised data transfer interface. 2007-05-25 15:58:42 +00:00
Michael Brown
3e2c6b6736 pkbuff->iobuf changeover 
Achieved via Perl using:

perl -pi -e 's/pk_buff/io_buffer/g; s/Packet buffer/I\/O buffer/ig; ' \
	-e 's/pkbuff\.h/iobuf.h/g; s/pkb_/iob_/g; s/_pkb/_iob/g; ' \
	-e 's/pkb/iobuf/g; s/PKB/IOB/g;'
 2007-05-19 18:39:40 +00:00
Michael Brown
3fd10074d8 SSL needs quite a lot of spare memory 2007-02-01 09:38:16 +00:00
Michael Brown
6d32f0e6e2 Changed to use the generic stream API. 2007-01-31 02:09:13 +00:00
Michael Brown
73b09ecba6 Use stdio.h instead of vsprintf.h 2007-01-19 01:13:12 +00:00
Michael Brown
c014f607a8 Use total free memory as advertised window. This seems to be sufficient 
to avoid drops even on slow NICs.
 2007-01-18 20:39:17 +00:00
Michael Brown
06630a3036 Switch from calloc() to malloc()+memset() to match the practices used 
almost everywhere else.
 2007-01-18 12:45:58 +00:00
Michael Brown
76aa9ad07d Make TCP give up immediately when it receives -ENETUNREACH from 
tcpip_tx().  This avoids the irritating wait when you accidentally type
"kernel pxelinux.0" before bringing up the network interface.

Add ENETUNREACH to strerror()'s list.
 2007-01-14 16:47:03 +00:00
Michael Brown
526d314266 Advertise a larger MSS to improve TCP performance. 2007-01-13 17:36:17 +00:00
Michael Brown
f008b77ba2 Allow an explicit network device to be specified for IP-layer 
transmissions.
 2007-01-10 02:25:11 +00:00
Michael Brown
18e5353bed Rename pkb_available() to pkb_tailroom() for consistency with Linux's 
skb_tailroom().  Add pkb_headroom().
 2007-01-09 20:56:31 +00:00
Michael Brown
2eeb7c4fe7 Define "connected" as 
"when SYN is ACKed and we have already received SYN", or
  "when SYN is received and we have already had SYN ACKed"

rather than just

  "when SYN is ACKed"

This avoids spuriously calling the connected() method when we receive
a RST,ACK in response to a SYN.
 2007-01-09 05:01:22 +00:00
Michael Brown
b0daa99dce Send RST packets when we get a non-matching connection, or receive an 
out-of-range ACK.
 2007-01-03 22:21:59 +00:00
Michael Brown
eea1167a85 Accept RST on received packets 2007-01-03 21:51:36 +00:00
Michael Brown
94fc68895d Improve consistency between TCP and UDP RX datapaths 2007-01-03 21:13:11 +00:00
Michael Brown
a0525a4ed3 Verify checksums on the RX datapath. 
Simplify checksum generation on the TX datapath.
 2007-01-03 20:48:52 +00:00
Michael Brown
feb4f65d1e Use auto-colourisation for debug messages. 
Truncate TX length to TCP window at time of transmission rather than at
time of adding to TX packet; this is conceptually cleaner and also allows
the application to call tcp_send() multiple times to build up a single
packet.
 2006-12-29 14:03:03 +00:00
Michael Brown
ce9819af27 Redefine TCP state to include "flags that have been sent" rather than 
"flags that are currently being sent".  This allows at least one special
case (checking that we haven't already sent a FIN in tcp_rx_fin()) to be
collapsed.
 2006-12-29 00:44:31 +00:00
Michael Brown
be31970e36 Don't send a second FIN if we receive a FIN while in FIN_WAIT_2 2006-12-27 23:44:05 +00:00
Michael Brown
61ed298bc7 Merge changes from mcb-tcp-fixes branch. 2006-12-27 23:09:46 +00:00
Michael Brown
5edb85c8b4 Work around another confused-by-RST bug 2006-12-06 18:09:33 +00:00
Michael Brown
b9f9230661 Work around another instance of the port re-use bug. 2006-12-05 23:15:35 +00:00
Michael Brown
cc0b92652f Call closed() method only when the connection is genuinely in the 
TCP_CLOSED state (i.e. after the final FIN/ACK exchange), and has been
removed from the list of TCP connections.
 2006-12-05 22:53:28 +00:00
Michael Brown
fcc70c9d60 Ensure that tcp_rx() always frees its received packet buffer. 2006-12-05 21:23:23 +00:00
Michael Brown
28c711b91d Killed off spurious <stdio.h> inclusion 2006-12-05 21:16:52 +00:00
Michael Brown
3fbfbad0f7 Avoid calling NULL methods! 2006-12-05 21:09:37 +00:00
Marty Connor
6ac78f6aff added stdio.h to includes for DBG compilation 2006-09-27 05:57:06 +00:00
Nikhil Chandru Rao
dccb8358bd Fixed a bug in the TCP state machine. Added a transition from ESTABLISHED to LAST_ACK on receiving a FIN,ACK 2006-08-19 16:14:53 +00:00
Michael Brown
f0718d562f Don't call stop_timer() from within the timer expiry callback; it's 
already stopped.

Don't call start_timer() when sending a dataless ACK.  This may or may
not be the right thing to do; I can't tell.

Back out broken "send ACK only if required to" logic temporarily.
 2006-08-09 15:50:20 +00:00
Nikhil Chandru Rao
3aff701001 Changed the TCP state machines behaviour in the ESTABLISHED state 2006-08-09 15:04:26 +00:00