| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
As WMM is required for HT/VHT operation, treat bad WMM parameters
more gracefully by falling back to default parameters instead of
not using WMM assocation. This makes it possible to still use HT
or VHT, although potentially with reduced quality of service due
to unintended WMM parameters.
Signed-off-by: Johannes Berg <[email protected]>
|
|
Disabling WMM has a huge impact these days. It implies that
HT and VHT will be disabled which means that the throughput
will be drammatically reduced.
Since the AIFSN is a transmission parameter, we can play a
bit and fix it up to make it compliant with the 802.11
specification which requires it to be at least 2.
Increasing it from 1 to 2 will slightly reduce the
likelyhood to get a transmission opportunity compared to
other clients that would accept to set AIFSN=1, but at
least it will allow HT and VHT which is a huge gain.
Signed-off-by: Emmanuel Grumbach <[email protected]>
Signed-off-by: Johannes Berg <[email protected]>
|
|
The function currently determines this value, for use in bss_info.qos,
based on the interface type itself. Make it a parameter instead and
set it with the same logic for now.
Signed-off-by: Johannes Berg <[email protected]>
|
|
llid_in_use needs to be limited to stations of the same VIF, otherwise it
will cause a NULL deref as the sta_info of non-mesh-VIFs don't have
sta->mesh set.
Steps to reproduce:
modprobe mac80211_hwsim channels=2
iw phy phy0 interface add ibss0 type ibss
iw phy phy0 interface add mesh0 type mp
iw phy phy1 interface add ibss1 type ibss
iw phy phy1 interface add mesh1 type mp
ip link set ibss0 up
ip link set mesh0 up
ip link set ibss1 up
ip link set mesh1 up
iw dev ibss0 ibss join foo 2412
iw dev ibss1 ibss join foo 2412
# Ensure that ibss0 and ibss1 are actually associated; I often need to
# leave and join the cell on ibss1 a second time.
iw dev mesh0 mesh join bar
iw dev mesh1 mesh join bar # crash
Signed-off-by: Matthias Schiffer <[email protected]>
Signed-off-by: Johannes Berg <[email protected]>
|
|
When 11n peers performs a TDLS connection on a legacy BSS, the HT
operation IE must be specified according to IEEE802.11-2012 section
9.23.3.2. Otherwise HT-protection is compromised and the medium becomes
noisy for both the TDLS and the BSS links.
Signed-off-by: Arik Nemtsov <[email protected]>
Signed-off-by: Emmanuel Grumbach <[email protected]>
Signed-off-by: Johannes Berg <[email protected]>
|
|
Scheduled scan has to be reconfigured only if wowlan wasn't
configured, since otherwise it should continue to run (with
the 'any' trigger) or be aborted.
The current code will end up asking the driver to start a new
scheduled scan without stopping the previous one, and leaking
some memory (from the previous request.)
Fix this by doing the abort/restart under the proper conditions.
Signed-off-by: Eliad Peller <[email protected]>
Signed-off-by: Emmanuel Grumbach <[email protected]>
Signed-off-by: Johannes Berg <[email protected]>
|
|
If drv_start() fails during hw_restart, all the running
interfaces are being closed/stopped, which results in
drv_stop() being called, although the driver was never
started successfully.
This might cause drivers to perform operations on uninitialized
memory (as they assume it was initialized on drv_start)
Consider the local->started flag, and call the driver's stop()
op only if drv_start() succeeded before.
Move drv_start() and drv_stop() to driver-ops.c, as they are no
longer simple wrappers.
Signed-off-by: Eliad Peller <[email protected]>
Signed-off-by: Emmanuel Grumbach <[email protected]>
Signed-off-by: Johannes Berg <[email protected]>
|
|
The recalc_smps work can run after the station disassociates.
At this stage we already released the channel, but the work
will be cancelled only when the interface stops.
In this scenario we can hit the warning in ieee80211_recalc_smps, so
just remove it.
Signed-off-by: Andrei Otcheretianski <[email protected]>
Signed-off-by: Emmanuel Grumbach <[email protected]>
Signed-off-by: Johannes Berg <[email protected]>
|
|
Requesting hw restart during suspend might result
in the restart work being executed after mac80211
and the hw are suspended.
Solve the race by simply scheduling the restart
work on a freezable workqueue.
Note that there can be some cases of reconfiguration
on resume (besides the hardware restart):
* wowlan is not configured -
All the interfaces removed were removed on suspend,
and drv_stop() was called. At this point the driver
shouldn't expect for hw_restart anyway, so we can
simply cancel it (on resume).
* wowlan is configured, drv_resume() == 1
There is no definitive expected behavior in this case,
as each driver might have different expectations (e.g.
setting some flags on suspend/restart vs. not handling
spurious recovery).
For now, simply let the hw_restart work run again after
resume, and hope the driver will handle it well (or at
least initiate another hw restart).
Signed-off-by: Eliad Peller <[email protected]>
Signed-off-by: Emmanuel Grumbach <[email protected]>
Signed-off-by: Johannes Berg <[email protected]>
|
|
Local request to deauthenticate wasn't handled while associating, thus
the association could continue even when the user space required to
disconnect.
Cc: [email protected]
Signed-off-by: Andrei Otcheretianski <[email protected]>
Signed-off-by: Emmanuel Grumbach <[email protected]>
Signed-off-by: Johannes Berg <[email protected]>
|
|
In TDLS channel-switch operations the chandef can sometimes be NULL.
Avoid an oops in the trace code for these cases and just print a
chandef full of zeros.
Cc: [email protected]
Fixes: a7a6bdd0670fe ("mac80211: introduce TDLS channel switch ops")
Signed-off-by: Arik Nemtsov <[email protected]>
Signed-off-by: Emmanuel Grumbach <[email protected]>
Signed-off-by: Johannes Berg <[email protected]>
|
|
If parse_acl_data succeeds but the subsequent parsing of smps
attributes fails, there will be a memory leak due to early returns.
Fix that by moving the ACL parsing later.
Cc: [email protected]
Fixes: 18998c381b19b ("cfg80211: allow requesting SMPS mode on ap start")
Signed-off-by: Ola Olsson <[email protected]>
Signed-off-by: Johannes Berg <[email protected]>
|
|
In case of one shot NOA the interval can be 0, catch that
instead of potentially (depending on the driver) crashing
like this:
divide error: 0000 [#1] SMP
[...]
Call Trace:
<IRQ>
[<ffffffffc08e891c>] ieee80211_extend_absent_time+0x6c/0xb0 [mac80211]
[<ffffffffc08e8a17>] ieee80211_update_p2p_noa+0xb7/0xe0 [mac80211]
[<ffffffffc069cc30>] ath9k_p2p_ps_timer+0x170/0x190 [ath9k]
[<ffffffffc070adf8>] ath_gen_timer_isr+0xc8/0xf0 [ath9k_hw]
[<ffffffffc0691156>] ath9k_tasklet+0x296/0x2f0 [ath9k]
[<ffffffff8107ad65>] tasklet_action+0xe5/0xf0
[...]
Cc: [email protected] [3.16+, due to d463af4a1c34 using it]
Signed-off-by: Janusz Dziedzic <[email protected]>
Signed-off-by: Johannes Berg <[email protected]>
|
|
There are some netdev features, which when disabled on an upper device,
such as a bonding master or a bridge, must be disabled and cannot be
re-enabled on underlying devices.
This is a rework of an earlier more heavy-handed appraoch, which simply
disables and prevents re-enabling of netdev features listed in a new
define in include/net/netdev_features.h, NETIF_F_UPPER_DISABLES. Any upper
device that disables a flag in that feature mask, the disabling will
propagate down the stack, and any lower device that has any upper device
with one of those flags disabled should not be able to enable said flag.
Initially, only LRO is included for proof of concept, and because this
code effectively does the same thing as dev_disable_lro(), though it will
also activate from the ethtool path, which was one of the goals here.
[root@dell-per730-01 ~]# ethtool -k bond0 |grep large
large-receive-offload: on
[root@dell-per730-01 ~]# ethtool -k p5p1 |grep large
large-receive-offload: on
[root@dell-per730-01 ~]# ethtool -K bond0 lro off
[root@dell-per730-01 ~]# ethtool -k bond0 |grep large
large-receive-offload: off
[root@dell-per730-01 ~]# ethtool -k p5p1 |grep large
large-receive-offload: off
dmesg dump:
[ 1033.277986] bond0: Disabling feature 0x0000000000008000 on lower dev p5p2.
[ 1034.067949] bnx2x 0000:06:00.1 p5p2: using MSI-X IRQs: sp 74 fp[0] 76 ... fp[7] 83
[ 1034.753612] bond0: Disabling feature 0x0000000000008000 on lower dev p5p1.
[ 1035.591019] bnx2x 0000:06:00.0 p5p1: using MSI-X IRQs: sp 62 fp[0] 64 ... fp[7] 71
This has been successfully tested with bnx2x, qlcnic and netxen network
cards as slaves in a bond interface. Turning LRO on or off on the master
also turns it on or off on each of the slaves, new slaves are added with
LRO in the same state as the master, and LRO can't be toggled on the
slaves.
Also, this should largely remove the need for dev_disable_lro(), and most,
if not all, of its call sites can be replaced by simply making sure
NETIF_F_LRO isn't included in the relevant device's feature flags.
Note that this patch is driven by bug reports from users saying it was
confusing that bonds and slaves had different settings for the same
features, and while it won't be 100% in sync if a lower device doesn't
support a feature like LRO, I think this is a good step in the right
direction.
CC: "David S. Miller" <[email protected]>
CC: Eric Dumazet <[email protected]>
CC: Jay Vosburgh <[email protected]>
CC: Veaceslav Falico <[email protected]>
CC: Andy Gospodarek <[email protected]>
CC: Jiri Pirko <[email protected]>
CC: Nikolay Aleksandrov <[email protected]>
CC: Michal Kubecek <[email protected]>
CC: Alexander Duyck <[email protected]>
CC: [email protected]
Signed-off-by: Jarod Wilson <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
sit0 device allocates its percpu storage twice :
- One time in ipip6_tunnel_init()
- One time in ipip6_fb_tunnel_init()
Thus we leak 48 bytes per possible cpu per network namespace dismantle.
ipip6_fb_tunnel_init() can be much simpler and does not
return an error, and should be called after register_netdev()
Note that ipip6_tunnel_clone_6rd() also needs to be called
after register_netdev() (calling ipip6_tunnel_init())
Fixes: ebe084aafb7e ("sit: Use ipip6_tunnel_init as the ndo_init function.")
Signed-off-by: Eric Dumazet <[email protected]>
Reported-by: Dmitry Vyukov <[email protected]>
Cc: Steffen Klassert <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
This patch fixes following problems :
1) percpu_counter_init() can return an error, therefore
init_frag_mem_limit() must propagate this error so that
inet_frags_init_net() can do the same up to its callers.
2) If ip[46]_frags_ns_ctl_register() fail, we must unwind
properly and free the percpu_counter.
Without this fix, we leave freed object in percpu_counters
global list (if CONFIG_HOTPLUG_CPU) leading to crashes.
This bug was detected by KASAN and syzkaller tool
(http://github.com/google/syzkaller)
Fixes: 6d7b857d541e ("net: use lib/percpu_counter API for fragmentation mem accounting")
Signed-off-by: Eric Dumazet <[email protected]>
Reported-by: Dmitry Vyukov <[email protected]>
Cc: Hannes Frederic Sowa <[email protected]>
Cc: Jesper Dangaard Brouer <[email protected]>
Acked-by: Hannes Frederic Sowa <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
The following bit in ceph_msg_revoke_incoming() is unsafe:
struct ceph_connection *con = msg->con;
if (!con)
return;
mutex_lock(&con->mutex);
<more msg->con use>
There is nothing preventing con from getting destroyed right after
msg->con test. One easy way to reproduce this is to disable message
signing only on the server side and try to map an image. The system
will go into a
libceph: read_partial_message ffff880073f0ab68 signature check failed
libceph: osd0 192.168.255.155:6801 bad crc/signature
libceph: read_partial_message ffff880073f0ab68 signature check failed
libceph: osd0 192.168.255.155:6801 bad crc/signature
loop which has to be interrupted with Ctrl-C. Hit Ctrl-C and you are
likely to end up with a random GP fault if the reset handler executes
"within" ceph_msg_revoke_incoming():
<yet another reply w/o a signature>
...
<Ctrl-C>
rbd_obj_request_end
ceph_osdc_cancel_request
__unregister_request
ceph_osdc_put_request
ceph_msg_revoke_incoming
...
osd_reset
__kick_osd_requests
__reset_osd
remove_osd
ceph_con_close
reset_connection
<clear con->in_msg->con>
<put con ref>
put_osd
<free osd/con>
<msg->con use> <-- !!!
If ceph_msg_revoke_incoming() executes "before" the reset handler,
osd/con will be leaked because ceph_msg_revoke_incoming() clears
con->in_msg but doesn't put con ref, while reset_connection() only puts
con ref if con->in_msg != NULL.
The current msg->con scheme was introduced by commits 38941f8031bf
("libceph: have messages point to their connection") and 92ce034b5a74
("libceph: have messages take a connection reference"), which defined
when messages get associated with a connection and when that
association goes away. Part of the problem is that this association is
supposed to go away in much too many places; closing this race entirely
requires either a rework of the existing or an addition of a new layer
of synchronization.
In lieu of that, we can make it *much* less likely to hit by
disassociating messages only on their destruction and resend through
a different connection. This makes the code simpler and is probably
a good thing to do regardless - this patch adds a msg_con_set() helper
which is is called from only three places: ceph_con_send() and
ceph_con_in_msg_alloc() to set msg->con and ceph_msg_release() to clear
it.
Signed-off-by: Ilya Dryomov <[email protected]>
|
|
Support for message signing was merged into 3.19, along with
nocephx_require_signatures option. But, all that option does is allow
the kernel client to talk to clusters that don't support MSG_AUTH
feature bit. That's pretty useless, given that it's been supported
since bobtail.
Meanwhile, if one disables message signing on the server side with
"cephx sign messages = false", it becomes impossible to use the kernel
client since it expects messages to be signed if MSG_AUTH was
negotiated. Add nocephx_sign_messages option to support this use case.
Signed-off-by: Ilya Dryomov <[email protected]>
|
|
supported_features and required_features serve no purpose at all, while
nocrc and tcp_nodelay belong to ceph_options::flags.
Signed-off-by: Ilya Dryomov <[email protected]>
|
|
I don't see a way for auth->authorizer to be NULL in
ceph_x_sign_message() or ceph_x_check_message_signature().
Signed-off-by: Ilya Dryomov <[email protected]>
|
|
We can use msg->con instead - at the point we sign an outgoing message
or check the signature on the incoming one, msg->con is always set. We
wouldn't know how to sign a message without an associated session (i.e.
msg->con == NULL) and being able to sign a message using an explicitly
provided authorizer is of no use.
Signed-off-by: Ilya Dryomov <[email protected]>
|
|
This patch changes the osd_req_op_data() macro to not evaluate
arguments more than once in order to follow the kernel coding style.
Signed-off-by: Ioana Ciornei <[email protected]>
Reviewed-by: Alex Elder <[email protected]>
[[email protected]: changelog, formatting]
Signed-off-by: Ilya Dryomov <[email protected]>
|
|
Commit ae385eaf24dc ("libceph: store session key in cephx authorizer")
introduced ceph_x_authorizer::session_key, but didn't update all the
exit/error paths. Introduce ceph_x_authorizer_cleanup() to encapsulate
ceph_x_authorizer cleanup and switch to it. This fixes ceph_x_destroy(),
which currently always leaks key and ceph_x_build_authorizer() error
paths.
Signed-off-by: Ilya Dryomov <[email protected]>
Reviewed-by: Yan, Zheng <[email protected]>
|
|
Use local variable cursor in place of &msg->cursor in
read_partial_msg_data() and write_partial_msg_data().
Signed-off-by: Shraddha Barke <[email protected]>
Signed-off-by: Ilya Dryomov <[email protected]>
|
|
Since handle_reply() does not use its con argument, remove it.
Signed-off-by: Shraddha Barke <[email protected]>
Signed-off-by: Ilya Dryomov <[email protected]>
|
|
NFS: NFSoRDMA Client Side Changes
In addition to a variety of bugfixes, these patches are mostly geared at
enabling both swap and backchannel support to the NFS over RDMA client.
Signed-off-by: Anna Schumake <[email protected]>
|
|
There are other error values besides ip6_null_entry that can be returned by
ip6_route_redirect(): fib6_rule_action() can also result in
ip6_blk_hole_entry and ip6_prohibit_entry if such ip rules are installed.
Only checking for ip6_null_entry in rt6_do_redirect() causes ip6_ins_rt()
to be called with rt->rt6i_table == NULL in these cases, making the kernel
crash.
Signed-off-by: Matthias Schiffer <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Forechannel transports get their own "bc_up" method to create an
endpoint for the backchannel service.
Signed-off-by: Chuck Lever <[email protected]>
[Anna Schumaker: Add forward declaration of struct net to xprt.h]
Signed-off-by: Anna Schumaker <[email protected]>
|
|
skb_set_owner_w() is called from various places that assume
skb->sk always point to a full blown socket (as it changes
sk->sk_wmem_alloc)
We'd like to attach skb to request sockets, and in the future
to timewait sockets as well. For these kind of pseudo sockets,
we need to take a traditional refcount and use sock_edemux()
as the destructor.
It is now time to un-inline skb_set_owner_w(), being too big.
Fixes: ca6fb0651883 ("tcp: attach SYNACK messages to request sockets instead of listener")
Signed-off-by: Eric Dumazet <[email protected]>
Bisected-by: Haiyang Zhang <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
br_should_learn() is protected by RCU and not by RTNL, so use correct
flavor of nbp_vlan_group().
Fixes: 907b1e6e83ed ("bridge: vlan: use proper rcu for the vlgrp
member")
Signed-off-by: Ido Schimmel <[email protected]>
Acked-by: Nikolay Aleksandrov <[email protected]>
Acked-by: Jiri Pirko <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
in preemptible context.
Fixes the following kernel BUG :
BUG: using __this_cpu_add() in preemptible [00000000] code: bash/2758
caller is __this_cpu_preempt_check+0x13/0x15
CPU: 0 PID: 2758 Comm: bash Tainted: P O 3.18.19 #2
ffffffff8170eaca ffff880110d1b788 ffffffff81482b2a 0000000000000000
0000000000000000 ffff880110d1b7b8 ffffffff812010ae ffff880007cab800
ffff88001a060800 ffff88013a899108 ffff880108b84240 ffff880110d1b7c8
Call Trace:
[<ffffffff81482b2a>] dump_stack+0x52/0x80
[<ffffffff812010ae>] check_preemption_disabled+0xce/0xe1
[<ffffffff812010d4>] __this_cpu_preempt_check+0x13/0x15
[<ffffffff81419d60>] ipmr_queue_xmit+0x647/0x70c
[<ffffffff8141a154>] ip_mr_forward+0x32f/0x34e
[<ffffffff8141af76>] ip_mroute_setsockopt+0xe03/0x108c
[<ffffffff810553fc>] ? get_parent_ip+0x11/0x42
[<ffffffff810e6974>] ? pollwake+0x4d/0x51
[<ffffffff81058ac0>] ? default_wake_function+0x0/0xf
[<ffffffff810553fc>] ? get_parent_ip+0x11/0x42
[<ffffffff810613d9>] ? __wake_up_common+0x45/0x77
[<ffffffff81486ea9>] ? _raw_spin_unlock_irqrestore+0x1d/0x32
[<ffffffff810618bc>] ? __wake_up_sync_key+0x4a/0x53
[<ffffffff8139a519>] ? sock_def_readable+0x71/0x75
[<ffffffff813dd226>] do_ip_setsockopt+0x9d/0xb55
[<ffffffff81429818>] ? unix_seqpacket_sendmsg+0x3f/0x41
[<ffffffff813963fe>] ? sock_sendmsg+0x6d/0x86
[<ffffffff813959d4>] ? sockfd_lookup_light+0x12/0x5d
[<ffffffff8139650a>] ? SyS_sendto+0xf3/0x11b
[<ffffffff810d5738>] ? new_sync_read+0x82/0xaa
[<ffffffff813ddd19>] compat_ip_setsockopt+0x3b/0x99
[<ffffffff813fb24a>] compat_raw_setsockopt+0x11/0x32
[<ffffffff81399052>] compat_sock_common_setsockopt+0x18/0x1f
[<ffffffff813c4d05>] compat_SyS_setsockopt+0x1a9/0x1cf
[<ffffffff813c4149>] compat_SyS_socketcall+0x180/0x1e3
[<ffffffff81488ea1>] cstar_dispatch+0x7/0x1e
Signed-off-by: Ani Sinha <[email protected]>
Acked-by: Eric Dumazet <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
The flag used to indicate if a VLAN should be used for filtering - as
opposed to context only - on the bridge itself (e.g. br0) is called
'brentry' and not 'brvlan'.
Signed-off-by: Ido Schimmel <[email protected]>
Acked-by: Nikolay Aleksandrov <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
When adding a port to a bridge we initialize VLAN filtering on it. We do
not bail out in case an error occurred in nbp_vlan_init, as it can be
used as a non VLAN filtering bridge.
However, if VLAN filtering is required and an error occurred in
nbp_vlan_init, we should set vlgrp to NULL, so that VLAN filtering
functions (e.g. br_vlan_find, br_get_pvid) will know the struct is
invalid and will not try to access it.
Signed-off-by: Ido Schimmel <[email protected]>
Signed-off-by: Nikolay Aleksandrov <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
IPv6 request sockets store a pointer to skb containing the SYN packet
to be able to transfer it to full blown socket when 3WHS is done
(ireq->pktopts -> np->pktoptions)
As explained in commit 5e0724d027f0 ("tcp/dccp: fix hashdance race for
passive sessions"), we must transfer the skb only if we won the
hashdance race, if multiple cpus receive the 'ack' packet completing
3WHS at the same time.
Fixes: e994b2f0fb92 ("tcp: do not lock listener to process SYN packets")
Fixes: 079096f103fa ("tcp/dccp: install syn_recv requests into ehash table")
Signed-off-by: Eric Dumazet <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
To further improve the RDS connection scalabilty on massive systems
where number of sockets grows into tens of thousands of sockets, there
is a need of larger bind hashtable. Pre-allocated 8K or 16K table is
not very flexible in terms of memory utilisation. The rhashtable
infrastructure gives us the flexibility to grow the hashtbable based
on use and also comes up with inbuilt efficient bucket(chain) handling.
Reviewed-by: David Miller <[email protected]>
Signed-off-by: Santosh Shilimkar <[email protected]>
Signed-off-by: Santosh Shilimkar <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
as result of function rds_iw_flush_mr_pool is nowhere checked,
changing its return type from int to void.
also removing the unused variable rc as there is nothing to return
Signed-off-by: Saurabh Sengar <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
This patch changes how the multipath hash is computed for locally
generated flows: now the hash comprises l4 information.
This allows better utilization of the available paths when the existing
flows have the same source IP and the same destination IP: with l3 hash,
even when multiple connections are in place simultaneously, a single path
will be used, while with l4 hash we can use all the available paths.
v2 changes:
- use get_hash_from_flowi4() instead of implementing just another l4 hash
function
Signed-off-by: Paolo Abeni <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Allow the use of other transport classes when handling a backward
direction RPC call.
Signed-off-by: Chuck Lever <[email protected]>
Reviewed-by: Sagi Grimberg <[email protected]>
Tested-By: Devesh Sharma <[email protected]>
Signed-off-by: Anna Schumaker <[email protected]>
|
|
On NFSv4.1 mount points, the Linux NFS client uses this transport
endpoint to receive backward direction calls and route replies back
to the NFSv4.1 server.
Signed-off-by: Chuck Lever <[email protected]>
Acked-by: "J. Bruce Fields" <[email protected]>
Reviewed-by: Sagi Grimberg <[email protected]>
Tested-By: Devesh Sharma <[email protected]>
Signed-off-by: Anna Schumaker <[email protected]>
|
|
Introduce a code path in the rpcrdma_reply_handler() to catch
incoming backward direction RPC calls and route them to the ULP's
backchannel server.
Signed-off-by: Chuck Lever <[email protected]>
Reviewed-by: Sagi Grimberg <[email protected]>
Tested-By: Devesh Sharma <[email protected]>
Signed-off-by: Anna Schumaker <[email protected]>
|
|
Backward direction RPC replies are sent via the client transport's
send_request method, the same way forward direction RPC calls are
sent.
Signed-off-by: Chuck Lever <[email protected]>
Reviewed-by: Sagi Grimberg <[email protected]>
Tested-By: Devesh Sharma <[email protected]>
Signed-off-by: Anna Schumaker <[email protected]>
|
|
Pre-allocate extra send and receive Work Requests needed to handle
backchannel receives and sends.
The transport doesn't know how many extra WRs to pre-allocate until
the xprt_setup_backchannel() call, but that's long after the WRs are
allocated during forechannel setup.
So, use a fixed value for now.
Signed-off-by: Chuck Lever <[email protected]>
Reviewed-by: Sagi Grimberg <[email protected]>
Tested-By: Devesh Sharma <[email protected]>
Signed-off-by: Anna Schumaker <[email protected]>
|
|
xprtrdma's backward direction send and receive buffers are the same
size as the forechannel's inline threshold, and must be pre-
registered.
The consumer has no control over which receive buffer the adapter
chooses to catch an incoming backwards-direction call. Any receive
buffer can be used for either a forward reply or a backward call.
Thus both types of RPC message must all be the same size.
Signed-off-by: Chuck Lever <[email protected]>
Reviewed-by: Sagi Grimberg <[email protected]>
Tested-By: Devesh Sharma <[email protected]>
Signed-off-by: Anna Schumaker <[email protected]>
|
|
xprt_{setup,destroy}_backchannel() won't be adequate for RPC/RMDA
bi-direction. In particular, receive buffers have to be pre-
registered and posted in order to receive incoming backchannel
requests.
Add a virtual function call to allow the insertion of appropriate
backchannel setup and destruction methods for each transport.
In addition, freeing a backchannel request is a little different
for RPC/RDMA. Introduce an rpc_xprt_op to handle the difference.
Signed-off-by: Chuck Lever <[email protected]>
Reviewed-by: Sagi Grimberg <[email protected]>
Tested-By: Devesh Sharma <[email protected]>
Signed-off-by: Anna Schumaker <[email protected]>
|
|
Now that RPC replies are processed in a workqueue, there's no need
to disable IRQs when managing send and receive buffers. This saves
noticeable overhead per RPC.
Signed-off-by: Chuck Lever <[email protected]>
Reviewed-by: Sagi Grimberg <[email protected]>
Tested-By: Devesh Sharma <[email protected]>
Signed-off-by: Anna Schumaker <[email protected]>
|
|
Clean up: The reply tasklet is no longer used.
Signed-off-by: Chuck Lever <[email protected]>
Reviewed-by: Sagi Grimberg <[email protected]>
Tested-By: Devesh Sharma <[email protected]>
Signed-off-by: Anna Schumaker <[email protected]>
|
|
The reply tasklet is fast, but it's single threaded. After reply
traffic saturates a single CPU, there's no more reply processing
capacity.
Replace the tasklet with a workqueue to spread reply handling across
all CPUs. This also moves RPC/RDMA reply handling out of the soft
IRQ context and into a context that allows sleeps.
Signed-off-by: Chuck Lever <[email protected]>
Reviewed-by: Sagi Grimberg <[email protected]>
Tested-By: Devesh Sharma <[email protected]>
Signed-off-by: Anna Schumaker <[email protected]>
|
|
The rb_send_bufs and rb_recv_bufs arrays are used to implement a
pair of stacks for keeping track of free rpcrdma_req and rpcrdma_rep
structs. Replace those arrays with free lists.
To allow more than 512 RPCs in-flight at once, each of these arrays
would be larger than a page (assuming 8-byte addresses and 4KB
pages). Allowing up to 64K in-flight RPCs (as TCP now does), each
buffer array would have to be 128 pages. That's an order-6
allocation. (Not that we're going there.)
A list is easier to expand dynamically. Instead of allocating a
larger array of pointers and copying the existing pointers to the
new array, simply append more buffers to each list.
This also makes it simpler to manage receive buffers that might
catch backwards-direction calls, or to post receive buffers in
bulk to amortize the overhead of ib_post_recv.
Signed-off-by: Chuck Lever <[email protected]>
Reviewed-by: Sagi Grimberg <[email protected]>
Reviewed-by: Devesh Sharma <[email protected]>
Tested-By: Devesh Sharma <[email protected]>
Signed-off-by: Anna Schumaker <[email protected]>
|
|
Clean up: The error cases in rpcrdma_reply_handler() almost never
execute. Ensure the compiler places them out of the hot path.
No behavior change expected.
Signed-off-by: Chuck Lever <[email protected]>
Reviewed-by: Sagi Grimberg <[email protected]>
Reviewed-by: Devesh Sharma <[email protected]>
Tested-By: Devesh Sharma <[email protected]>
Signed-off-by: Anna Schumaker <[email protected]>
|
|
Commit 8301a2c047cc ("xprtrdma: Limit work done by completion
handler") was supposed to prevent xprtrdma's upcall handlers from
starving other softIRQ work by letting them return to the provider
before all CQEs have been polled.
The logic assumes the provider will call the upcall handler again
immediately if the CQ is re-armed while there are still queued CQEs.
This assumption is invalid. The IBTA spec says that after a CQ is
armed, the hardware must interrupt only when a new CQE is inserted.
xprtrdma can't rely on the provider calling again, even though some
providers do.
Therefore, leaving CQEs on queue makes sense only when there is
another mechanism that ensures all remaining CQEs are consumed in a
timely fashion. xprtrdma does not have such a mechanism. If a CQE
remains queued, the transport can wait forever to send the next RPC.
Finally, move the wcs array back onto the stack to ensure that the
poll array is always local to the CPU where the completion upcall is
running.
Fixes: 8301a2c047cc ("xprtrdma: Limit work done by completion ...")
Signed-off-by: Chuck Lever <[email protected]>
Reviewed-by: Sagi Grimberg <[email protected]>
Reviewed-by: Devesh Sharma <[email protected]>
Tested-By: Devesh Sharma <[email protected]>
Signed-off-by: Anna Schumaker <[email protected]>
|