Adrian Chadd 8e7393944d Push the actual TX processing into the ath taskqueue, rather than having ... it run out of multiple concurrent contexts. Right now the ath(4) TX processing is a bit hairy. Specifically: * It was running out of ath_start(), which could occur from multiple concurrent sending processes (as if_start() can be started from multiple sending threads nowdays.. sigh) * during RX if fast frames are enabled (so not really at the moment, not until I fix this particular feature again..) * during ath_reset() - so anything which calls that * during ath_tx_proc*() in the ath taskqueue - ie, TX is attempted again after TX completion, as there's now hopefully some ath_bufs available. * Then, the ic_raw_xmit() method can queue raw frames for transmission at any time, from any net80211 TX context. Ew. This has caused packet ordering issues in the past - specifically, there's absolutely no guarantee that preemption won't occuring _during_ ath_start() by the TX completion processing, which will call ath_start() again. It's a mess - 802.11 really, really wants things to be in sequence or things go all kinds of loopy. So: * create a new task struct for TX'ing; * make the if_start method simply queue the task on the ath taskqueue; * make ath_start() just be called by the new TX task; * make ath_tx_kick() just schedule the ath TX task, rather than directly calling ath_start(). Now yes, this means that I've taken a step backwards in terms of concurrency - TX -and- RX now occur in the same single-task taskqueue. But there's nothing stopping me from separating out the TX / TX completion code into a separate taskqueue which runs in parallel with the RX path, if that ends up being appropriate for some platforms. This fixes the CCMP/seqno concurrency issues that creep up when you transmit large amounts of uni-directional UDP traffic (>200MBit) on a FreeBSD STA -> AP, as now there's only one TX context no matter what's going on (TX completion->retry/software queue, userland->net80211->ath_start(), TX completion -> ath_start()); but it won't fix any concurrency issues between raw transmitted frames and non-raw transmitted frames (eg EAPOL frames on TID 16 and any other TID 16 multicast traffic that gets put on the CABQ.) That is going to require a bunch more re-architecture before it's feasible to fix. In any case, this is a big step towards making the majority of the TX path locking irrelevant, as now almost all TX activity occurs in the taskqueue. Phew.		2012-10-14 20:44:08 +00:00
..
amd64	Print the %rip value for uprintf_signal.	2012-10-14 17:08:46 +00:00
arm	Loop reading the RTC registers until the same values are obtained	2012-10-07 20:36:46 +00:00
boot	Since the introduction of the new advanced boot menu (r222417), options like	2012-10-14 06:52:49 +00:00
bsm
cam	Add explicit check for not set time inside cam_periph_freeze_after_event().	2012-10-14 08:50:05 +00:00
cddl	Revert previous commit...	2012-10-10 08:36:38 +00:00
compat	Revert previous commit...	2012-10-10 08:36:38 +00:00
conf	Import a FreeBSD port of the FUSE Linux module.	2012-10-13 23:54:26 +00:00
contrib	Fix defines in r241245. We actually don't define FreeBSD.	2012-10-14 15:03:06 +00:00
crypto	Revert previous commit...	2012-10-10 08:36:38 +00:00
ddb
dev	Push the actual TX processing into the ath taskqueue, rather than having	2012-10-14 20:44:08 +00:00
fs	Grammar fixes.	2012-10-14 18:13:33 +00:00
gdb
geom	NULL-ify last previously used pointer instead of last possible pointer.	2012-10-10 20:41:37 +00:00
gnu/fs	Add an unified macro to deny ability from the compiler to reorder	2012-10-09 14:32:30 +00:00
i386	MFamd64: add machdep.uprintf_signal.	2012-10-14 17:09:50 +00:00
ia64	Eliminate a stale comment. It describes another use case for the pmap in	2012-09-28 05:30:59 +00:00
isa
kern	Add a KPI to allow to reserve some amount of space in the numvnodes	2012-10-14 19:43:37 +00:00
kgssapi
libkern	s/ is is / is /g	2012-09-14 22:00:03 +00:00
mips	Correct an error in pmap_pv_reclaim(). It can legitimately encounter	2012-10-14 03:40:08 +00:00
modules	Import a FreeBSD port of the FUSE Linux module.	2012-10-13 23:54:26 +00:00
net	Revert previous commit...	2012-10-10 08:36:38 +00:00
net80211	Revert previous commit...	2012-10-10 08:36:38 +00:00
netatalk
netgraph	Add NG_NETFLOW_V9INFO_TYPE command to be able to request netflowv9-specific	2012-10-11 16:15:18 +00:00
netinet	Fix a miss from r241344: in ip_mloopback() we need to go to	2012-10-14 15:08:07 +00:00
netinet6	Cleanup documentation: cloning route support has been removed in r186119.	2012-10-13 09:31:01 +00:00
netipsec	Add missing break	2012-09-18 08:00:43 +00:00
netipx
netnatm
netncp
netpfil	Revert previous commit...	2012-10-10 08:36:38 +00:00
netsmb
nfs	Revert previous commit...	2012-10-10 08:36:38 +00:00
nfsclient
nfsserver	Revert previous commit...	2012-10-10 08:36:38 +00:00
nlm
ofed	The drbr(9) API appeared to be so unclear, that most drivers in	2012-09-28 18:28:27 +00:00
opencrypto	Revert previous commit...	2012-10-10 08:36:38 +00:00
pc98	Add an unified macro to deny ability from the compiler to reorder	2012-10-09 14:32:30 +00:00
pci
powerpc	Eliminate a stale comment. It describes another use case for the pmap in	2012-09-28 05:30:59 +00:00
rpc	rpc: convert all uid and gid variables to u_int.	2012-10-04 04:15:18 +00:00
security
sparc64	Add an unified macro to deny ability from the compiler to reorder	2012-10-09 14:32:30 +00:00
sys	Add a KPI to allow to reserve some amount of space in the numvnodes	2012-10-14 19:43:37 +00:00
teken
tools	Revert previous commit...	2012-10-10 08:36:38 +00:00
ufs	Fix up kernel sources to be ready for a 64-bit ino_t.	2012-09-27 23:30:49 +00:00
vm	Move vm_page_requeue() to the only file that uses it.	2012-10-13 20:19:43 +00:00
x86	Add an unified macro to deny ability from the compiler to reorder	2012-10-09 14:32:30 +00:00
xdr
xen
Makefile