#
da976f9d |
| 05-Dec-2014 |
Enji Cooper <ngie@FreeBSD.org> |
MFhead @ r275496
|
#
24ce0a02 |
| 04-Dec-2014 |
Dimitry Andric <dim@FreeBSD.org> |
Merge ^/head r275387 through r275477.
|
#
4e88d37a |
| 02-Dec-2014 |
Michael Tuexen <tuexen@FreeBSD.org> |
Do the renaming of sb_cc to sb_ccc in a way with less code changes by using a macro. This is an alternate approach to https://svnweb.freebsd.org/changeset/base/275326 which is easier to handle upstre
Do the renaming of sb_cc to sb_ccc in a way with less code changes by using a macro. This is an alternate approach to https://svnweb.freebsd.org/changeset/base/275326 which is easier to handle upstream.
Discussed with: rrs, glebius
show more ...
|
#
60c317af |
| 01-Dec-2014 |
Dimitry Andric <dim@FreeBSD.org> |
Merge ^/head r275262 through r275363.
|
#
0f9d0a73 |
| 30-Nov-2014 |
Gleb Smirnoff <glebius@FreeBSD.org> |
Merge from projects/sendfile:
o Introduce a notion of "not ready" mbufs in socket buffers. These mbufs are now being populated by some I/O in background and are referenced outside. This forces fol
Merge from projects/sendfile:
o Introduce a notion of "not ready" mbufs in socket buffers. These mbufs are now being populated by some I/O in background and are referenced outside. This forces following implications: - An mbuf which is "not ready" can't be taken out of the buffer. - An mbuf that is behind a "not ready" in the queue neither. - If sockbet buffer is flushed, then "not ready" mbufs shouln't be freed.
o In struct sockbuf the sb_cc field is split into sb_ccc and sb_acc. The sb_ccc stands for ""claimed character count", or "committed character count". And the sb_acc is "available character count". Consumers of socket buffer API shouldn't already access them directly, but use sbused() and sbavail() respectively. o Not ready mbufs are marked with M_NOTREADY, and ready but blocked ones with M_BLOCKED. o New field sb_fnrdy points to the first not ready mbuf, to avoid linear search. o New function sbready() is provided to activate certain amount of mbufs in a socket buffer.
A special note on SCTP: SCTP has its own sockbufs. Unfortunately, FreeBSD stack doesn't yet allow protocol specific sockbufs. Thus, SCTP does some hacks to make itself compatible with FreeBSD: it manages sockbufs on its own, but keeps sb_cc updated to inform the stack of amount of data in them. The new notion of "not ready" data isn't supported by SCTP. Instead, only a mechanical substitute is done: s/sb_cc/sb_ccc/. A proper solution would be to take away struct sockbuf from struct socket and allow protocols to implement their own socket buffers, like SCTP already does. This was discussed with rrs@.
Sponsored by: Netflix Sponsored by: Nginx, Inc.
show more ...
|
#
9268022b |
| 19-Nov-2014 |
Simon J. Gerraty <sjg@FreeBSD.org> |
Merge from head@274682
|
Revision tags: release/10.1.0 |
|
#
8c3cfe0b |
| 04-Nov-2014 |
Alexander V. Chernikov <melifaro@FreeBSD.org> |
Hide 'struct rtentry' and all its macro inside new header: net/route_internal.h The goal is to make its opaque for all code except route/rtsock and proto domain _rmx.
|
#
257480b8 |
| 04-Nov-2014 |
Alexander V. Chernikov <melifaro@FreeBSD.org> |
Convert netinet6/ to use new routing API.
* Remove &ifpp from ip6_output() in favor of ri->ri_nh_info * Provide different wrappers to in6_selectsrc: Currently it is used by 2 differenct type of cu
Convert netinet6/ to use new routing API.
* Remove &ifpp from ip6_output() in favor of ri->ri_nh_info * Provide different wrappers to in6_selectsrc: Currently it is used by 2 differenct type of customers: - socket-based one, which all are unsure about provided address scope and - in-kernel ones (ND code mostly), which don't have any sockets, options, crededentials, etc. So, we provide two different wrappers to in6_selectsrc() returning select source. * Make different versions of selectroute(): Currenly selectroute() is used in two scenarios: - SAS, via in6_selecsrc() -> in6_selectif() -> selectroute() - output, via in6_output -> wrapper -> selectroute() Provide different versions for each customer: - fib6_lookup_nh_basic()-based in6_selectif() which is capable of returning interface only, without MTU/NHOP/L2 calculations - full-blown fib6_selectroute() with cached route/multipath/ MTU/L2 * Stop using routing table for link-local address lookups * Add in6_ifawithifp_lla() to make for-us check faster for link-local * Add in6_splitscope / in6_setllascope for faster embed/deembed scopes
show more ...
|
#
b4e8f808 |
| 19-Oct-2014 |
Alexander V. Chernikov <melifaro@FreeBSD.org> |
Switch IPv4 output path to use new routing api.
The goals of the new API is to provide consumers with minimal needed information, but as fast as possible. So we provide full nexthop info copied
Switch IPv4 output path to use new routing api.
The goals of the new API is to provide consumers with minimal needed information, but as fast as possible. So we provide full nexthop info copied into alighed on-cache structure instead of rte/ia pointers, their refcounts and locks. This does not provide solution for protecting from egress ifp destruction, but does not make it any worse.
Current changes:
nhops: Add fib4_lookup_prepend() function which stores either full L2+L3 prepend info (e.g. MAC header in case of plain IPv4) or L3 info with NH_FLAGS_L2_INCOMPLETE flag indicating that no valid L2 info exists and we have to take "slow" path.
ip_output: Currently ip[ 46]_output consumers use 'struct route' for the following purposes: 1) double lookup avoidance(route caching) 2) plain route caching 3) get path MTU to be able to notify source. The former pattern is mostly used by various tunnels (gif, gre, stf). (Actually, gre is the only remaining, others were already converted. Their locking model did not scale good enogh to benefit from such caching, so we have (temporarily) removed it without any performance loss). Plain route caching used by SCTP is simply wrong and should be removed. Temporary break it for now just to be able to compile. Optimize path mtu reporting by providing it in new 'route_info' stucture.
Minimize games with @ia locking/refcounting for route lookup: add special nhop[46]_extended structure to store more route attributes. Pointer to given structure can be passed to fib4_lookup_prepend() to indicate we want this info (we actually needs it for UDP and raw IP).
ether_output: Provide light-weight ether_output2() call to deal with transmitting L2 frame (e.g. properly handle broadcast/simloop/bridge/ other L2 hooks before actually transmitting frame by if_transmit()). Add a hack based on new RT_NHOP ro_flag to distinguish which version should we call. Better way is probably to add a new "if_output_frame" driver callbacks.
Next steps: * Convert ip_fastfwd part * Implement auto-growing array for per-radix nexthops * Implement LLE tracking for nexthop calculations to be able to immediately provide all necessary info in single route lookup for gateway routes * Switch radix locking scheme to runtime/cfg lock * Implement multipath support for rtsock * Implement "tracked nexthops" for tunnels (e.g. _proper_ nexthop caching) * Add IPv6 support for remaining parts (postponed not to interfere with user/ae/inet6 branch) * Consider adding "if_output_frame" driver call to ease logical frame pushing.
show more ...
|
#
1ce4b357 |
| 04-Oct-2014 |
Alexander V. Chernikov <melifaro@FreeBSD.org> |
Sync to HEAD@r272516.
|
#
4e27d36d |
| 17-Sep-2014 |
Neel Natu <neel@FreeBSD.org> |
IFC @r271694
|
#
43f9f175 |
| 16-Sep-2014 |
Michael Tuexen <tuexen@FreeBSD.org> |
The MTU is handled as a 32-bit entity within the SCTP stack. This was reported by Peter Kasting from Google.
MFC after: 3 days
|
#
f47f328d |
| 06-Sep-2014 |
Michael Tuexen <tuexen@FreeBSD.org> |
Fix the handling of sysctl variables when used with VIMAGE. While there do some cleanup of the code.
MFC after: 1 week
|
#
246e7a2b |
| 02-Sep-2014 |
Neel Natu <neel@FreeBSD.org> |
IFC @r269962
Submitted by: Anish Gupta (akgupt3@gmail.com)
|
#
ee7b0571 |
| 19-Aug-2014 |
Simon J. Gerraty <sjg@FreeBSD.org> |
Merge head from 7/28
|
#
1b833d53 |
| 13-Aug-2014 |
Alexander V. Chernikov <melifaro@FreeBSD.org> |
Sync to HEAD@r269943.
|
#
0c8682e8 |
| 11-Jul-2014 |
Michael Tuexen <tuexen@FreeBSD.org> |
Whitespace changes.
MFC after: 1 week
|
#
fcc34a23 |
| 11-Jul-2014 |
Gleb Smirnoff <glebius@FreeBSD.org> |
Fix style bug: rename the refcount field of m_ext to ext_cnt, to match other members.
Sponsored by: Nginx, Inc.
|
Revision tags: release/9.3.0 |
|
#
6cec9cad |
| 03-Jun-2014 |
Peter Grehan <grehan@FreeBSD.org> |
MFC @ r266724
An SVM update will follow this.
|
#
3b8f0845 |
| 28-Apr-2014 |
Simon J. Gerraty <sjg@FreeBSD.org> |
Merge head
|
#
84e51a1b |
| 23-Apr-2014 |
Alan Somers <asomers@FreeBSD.org> |
IFC @264767
|
#
1709ccf9 |
| 29-Mar-2014 |
Martin Matuska <mm@FreeBSD.org> |
Merge head up to r263906.
|
#
e3a7aa6f |
| 05-Mar-2014 |
Gleb Smirnoff <glebius@FreeBSD.org> |
- Remove rt_metrics_lite and simply put its members into rtentry. - Use counter(9) for rt_pksent (former rt_rmx.rmx_pksent). This removes another cache trashing ++ from packet forwarding path. - Cr
- Remove rt_metrics_lite and simply put its members into rtentry. - Use counter(9) for rt_pksent (former rt_rmx.rmx_pksent). This removes another cache trashing ++ from packet forwarding path. - Create zini/fini methods for the rtentry UMA zone. Via initialize mutex and counter in them. - Fix reporting of rmx_pksent to routing socket. - Fix netstat(1) to report "Use" both in kvm(3) and sysctl(3) mode.
The change is mostly targeted for stable/10 merge. For head, rt_pksent is expected to just disappear.
Discussed with: melifaro Sponsored by: Netflix Sponsored by: Nginx, Inc.
show more ...
|
Revision tags: release/10.0.0 |
|
#
0bfd163f |
| 18-Oct-2013 |
Gleb Smirnoff <glebius@FreeBSD.org> |
Merge head r233826 through r256722.
|
#
1ccca3b5 |
| 10-Oct-2013 |
Alan Somers <asomers@FreeBSD.org> |
IFC @256277
Approved by: ken (mentor)
|