As a first step in moving from LRO to GRO, revert commit af40da894e
("IPoIB: add LRO support"). Also eliminate the ethtool set_flags
callback which isn't needed anymore. Finally, we need to include
<linux/sched.h> directly to get the declaration of restart_syscall()
(which used to be included implicitly through <linux/inet_lro.h>).
Cc: Ben Hutchings <bhutchings@solarflare.com>
Cc: Eric W. Biederman <ebiederm@xmission.com>
Cc: Vladimir Sokolovsky <vlad@mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband: (63 commits)
IB/qib: clean up properly if pci_set_consistent_dma_mask() fails
IB/qib: Allow driver to load if PCIe AER fails
IB/qib: Fix uninitialized pointer if CONFIG_PCI_MSI not set
IB/qib: Fix extra log level in qib_early_err()
RDMA/cxgb4: Remove unnecessary KERN_<level> use
RDMA/cxgb3: Remove unnecessary KERN_<level> use
IB/core: Add link layer type information to sysfs
IB/mlx4: Add VLAN support for IBoE
IB/core: Add VLAN support for IBoE
IB/mlx4: Add support for IBoE
mlx4_en: Change multicast promiscuous mode to support IBoE
mlx4_core: Update data structures and constants for IBoE
mlx4_core: Allow protocol drivers to find corresponding interfaces
IB/uverbs: Return link layer type to userspace for query port operation
IB/srp: Sync buffer before posting send
IB/srp: Use list_first_entry()
IB/srp: Reduce number of BUSY conditions
IB/srp: Eliminate two forward declarations
IB/mlx4: Signal node desc changes to SM by using FW to generate trap 144
IB: Replace EXTRA_CFLAGS with ccflags-y
...
Use the new {max,min}3 macros to save some cycles and bytes on the stack.
This patch substitutes trivial nested macros with their counterpart.
Signed-off-by: Hagen Paul Pfeifer <hagen@jauu.net>
Cc: Joe Perches <joe@perches.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Roland Dreier <rolandd@cisco.com>
Cc: Sean Hefty <sean.hefty@intel.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Use the net device's dev_id field to encode the port number of the pci
device. This can be used to to associate a net device with the pci
device's port. The encoding is: dev_id = port - 1.
Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
IPoIB is IP-over-Infiniband link layer. In the case of IBoE, the link
layer is Ethernet and IP can work directly over Ethernet, so disable
IPoIB for non-IB_LINK_LAYER_INFINIBAND ports.
Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Sumeet Lahorani <sumeet.lahorani@oracle.com> reported that the IPoIB
child entries are world-writable; however we don't want ordinary users
to be able to create and destroy child interfaces, so fix them to be
writable only by root.
Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
Cc: <stable@kernel.org>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
IPoIB can miss a change in destination GID under some conditions. The
problem is caused when ipoib_neigh->dgid contains a stale address.
The fix is to set ipoib_neigh->dgid to zero in ipoib_neigh_alloc().
This can happen when a system using bonding on its IPoIB interfaces
has switched its active interface from interface A to B and back to A.
The system that fails over will not correctly processes the 2nd
address change, as described below.
When an address has changed neighbor->ha is updated with the new
address. Each neighbor has an associated ipoib_neigh.
ipoib_neigh->dgid also holds a copy of the remote node's hardware
address. When an address changes neighbor->ha is updated by the
network layer (arp code) with the new address. IPoIB detects this
change in ipoib_start_xmit() by comparing neighbor->ha with
ipoib_neigh->dgid. The bug is that ipoib_neigh->dgid may already
contain the new address (A) thus the change from B to A is missed by
ipoib. Here is the sequence of events:
ipoib_neigh->dgid = A and neighbor->ha = A
The address is switched to B (the first switch)
neighbor->ha = B
The change is seen in ipoib_start_xmit() -- neighbor->ha !=
ipoib_neigh->dgid so ipoib_neigh is released, and a new one is
allocated.
The allocator may return the same chunk of memory that was just
released, therefore ipoib_neigh->dgid still contains A at this point.
ipoib_neigh->dgid should be updated in neigh_add_path(), but if the
following conditions are true dgid is not updated:
1) __path_find() returns a path
2) path->ah is NULL
The remote system now switches from address B to A, neighbor->ha is
updated to A.
Now we have again : ipoib_neigh->dgid = A and neighbor->ha = A
Since the addresses are the same ipoib won't process the change in
address. Fix this by zeroing out the dgid field when allocating a new
struct ipoib_neigh.
Signed-off-by: David Wilder <dwilder@us.ibm.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
IPoIB currently must use irqsave locking for priv->lock, since it is
taken from interrupt context in one path. However, ipoib_send() does
skb_orphan(), and the network stack locking is not IRQ-safe.
Therefore we need to make sure we don't hold priv->lock when calling
ipoib_send() to avoid lockdep warnings (the code was almost certainly
safe in practice, since the only code path that takes priv->lock from
interrupt context would never call into the network stack).
Addresses: http://bugzilla.kernel.org/show_bug.cgi?id=13757
Reported-by: Bart Van Assche <bart.vanassche@gmail.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Define three accessors to get/set dst attached to a skb
struct dst_entry *skb_dst(const struct sk_buff *skb)
void skb_dst_set(struct sk_buff *skb, struct dst_entry *dst)
void skb_dst_drop(struct sk_buff *skb)
This one should replace occurrences of :
dst_release(skb->dst)
skb->dst = NULL;
Delete skb->dst field
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Last two drivers that need skb->dst in their start_xmit() function
Tell dev_hard_start_xmit() to no release it by unsetting IFF_XMIT_DST_RELEASE
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
If NAPI is enabled while IPoIB's CQ is being drained, it creates a
race on priv->ibwc between ipoib_poll() and ipoib_drain_cq(), leading
to memory corruption.
The solution is to enable/disable NAPI in ipoib_ib_dev_{open/stop}()
instead of in ipoib_{open/stop}(), and sync NAPI on the INITIALIZED
flag instead on the ADMIN_UP flag. This way NAPI will be disabled when
ipoib_drain_cq() is called.
This fixes <https://bugs.openfabrics.org/show_bug.cgi?id=1587>.
Signed-off-by: Yossi Etigin <yosefe@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next-2.6: (1750 commits)
ixgbe: Allow Priority Flow Control settings to survive a device reset
net: core: remove unneeded include in net/core/utils.c.
e1000e: update version number
e1000e: fix close interrupt race
e1000e: fix loss of multicast packets
e1000e: commonize tx cleanup routine to match e1000 & igb
netfilter: fix nf_logger name in ebt_ulog.
netfilter: fix warning in ebt_ulog init function.
netfilter: fix warning about invalid const usage
e1000: fix close race with interrupt
e1000: cleanup clean_tx_irq routine so that it completely cleans ring
e1000: fix tx hang detect logic and address dma mapping issues
bridge: bad error handling when adding invalid ether address
bonding: select current active slave when enslaving device for mode tlb and alb
gianfar: reallocate skb when headroom is not enough for fcb
Bump release date to 25Mar2009 and version to 0.22
r6040: Fix second PHY address
qeth: fix wait_event_timeout handling
qeth: check for completion of a running recovery
qeth: unregister MAC addresses during recovery.
...
Manually fixed up conflicts in:
drivers/infiniband/hw/cxgb3/cxio_hal.h
drivers/infiniband/hw/nes/nes_nic.c
If path_rec_start() returns error, call path_free() only if the path
was newly-created. If we free an existing path whose valid flag was zero,
(but do not detach it from the list) we cause corruption of the
path list (of which it is a member), and get a kernel crash.
The simplest solution is to not free an existing path -- just leave it
in the list as-is (i.e., with its valid flag cleared).
Thanks to Yossi Etigin of Voltaire for identifying the problem flow
which caused the kernel crash.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Moni Shua <monis@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
After commit fe25c561 ("IPoIB: Don't enable NAPI when it's already
enabled"), if an interface is brought up but the corresponding P_Key
never appears, then ipoib_stop() will hang in napi_disable(), because
ipoib_open() returns before it does napi_enable().
Fix this by changing ipoib_open() to call napi_enable() even if the
P_Key isn't present.
Reported-by: Yossi Etigin <yosefe@Voltaire.COM>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Fix bonding failover in the case both peers failover and the
gratuitous ARP is lost. In that case, the sender side will create an
ipoib_neigh and issue a path request with the old GID first. When
skb->dst->neighbour->ha changes due to ARP refresh, this ipoib_neigh
will not be added to the path->list of the path of the new GID,
because the ipoib_neigh already exists. It will not have an AH
either, because of sender-side failover. Therefore, it will not get
an AH when the path is resolved.
The solution here is to compare GIDs in ipoib_start_xmit() even if
neigh->ah is invalid. Comparing with an uninitialized value of
neigh->dgid should be fine, since a spurious match is harmless (and
astronomically unlikely too).
Signed-off-by: Moni Shoua <monis@voltaire.com>
Signed-off-by: Yossi Etigin <yosefe@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Fix a crash in path_rec_completion() during an SM up/down loop. If
more than one path record request is issued, the first completion
releases path->done, allowing ipoib_flush_paths() to free the path,
and thus corrupting it for the second completion.
Commit ee1e2c82 ("IPoIB: Refresh paths instead of flushing them on SM
change events") added the field path->valid and changed the test "if
(!path)" to "if (!path || !path->valid)". This change made it
possible for a path with an outstanding query to pass the test and
issue another query on the same path. Having two queries on the same
path leads to a crash.
This fixes <https://bugs.openfabrics.org/show_bug.cgi?id=1325>.
Signed-off-by: Yossi Etigin <yosefe@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
ipoib_flush_paths() can hang during an SM up/down loop: if
path_rec_start() fails (for instance, because there is no sm_ah), the
path is still added to the path list by neigh_add_path(). Then,
ipoib_flush_paths() will wait for path->done, but it will never
complete because the request was not issued at all. Fix this by
completing path->done if issuing the query fails.
This fixes <https://bugs.openfabrics.org/show_bug.cgi?id=1329>.
Signed-off-by: Yossi Etigin <yosefe@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
If a P_Key is not present when an interface is created, ipoib_open()
will return after doing napi_enable(). ipoib_open() will be called
again from ipoib_pkey_poll() when the P_Key appears, after NAPI has
already been enabled, and try to enable it again. This triggers a
BUG_ON() in napi_enable().
Fix this by moving the call to napi_enable() to after the test for
P_Key presence.
Signed-off-by: Yossi Etigin <yosefe@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Replace all uses of IPOIB_GID_FMT, IPOIB_GID_RAW_ARG() and IPOIB_GID_ARG()
Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Child devices were created without any offload features set, fix this by
moving the code that computes the features into generic function which is
now called through non-child and child device creation.
Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
-- v1 has a bug where the 'result' flag in ipoib_vlan_add may be used uninitialized
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Currently, IPoIB is an LLTX driver that uses its own IRQ-disabling
tx_lock. Not only do we want to get rid of LLTX, this actually causes
problems because of the skb_orphan() done with this tx_lock held: some
skb destructors expect to be run with interrupts enabled.
The simplest fix for this is to get rid of the driver-private tx_lock
and stop using LLTX. We kill off priv->tx_lock and use
netif_tx_lock[_bh]() instead; the patch to do this is a tiny bit
tricky because we need to update places that take priv->lock inside
the tx_lock to disable IRQs, rather than relying on tx_lock having
already disabled IRQs.
Also, there are a couple of places where we need to disable BHs to
make sure we have a consistent context to call netif_tx_lock() (since
we no longer can use _irqsave() variants), and we also have to change
ipoib_send_comp_handler() to call drain_tx_cq() through a timer rather
than directly, because ipoib_send_comp_handler() runs in interrupt
context and drain_tx_cq() must run in BH context so it can call
netif_tx_lock().
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Commit ee1e2c82 ("IPoIB: Refresh paths instead of flushing them on SM
change events") changed how paths are flushed on an SM event. This
change introduces a problem if the path record query triggered by
fails, causing path->ah to become NULL. A later successful path query
will then trigger WARN_ON() in path_rec_completion(), and crash
because path->ah has already been freed, so the ipoib_put_ah() inside
the lock in path_rec_completion() may actually drop the last reference
(contrary to the comment that claims this is safe).
Fix this by updating path->ah and freeing old_ah only when the path
record query is successful. This prevents the neighbour AH and that
path AH from getting out of sync.
This fixes <https://bugs.openfabrics.org/show_bug.cgi?id=1194>
Reported-by: Rabah Salem <ravah@mellanox.com>
Debugged-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Taking rtnl_lock in ipoib_mcast_join_complete() causes a deadlock with
ipoib_stop(). We avoid it by scheduling the piece of code that takes
the lock on ipoib_workqueue instead of executing it directly. This
works because we only flush the ipoib_workqueue with the RTNL not held.
The deadlock happens because ipoib_stop() calls ipoib_ib_dev_down()
which calls ipoib_mcast_dev_flush(), which calls ipoib_mcast_free(),
which calls ipoib_mcast_leave(). The latter calls
ib_sa_free_multicast(), and this waits until the multicast completion
handler finishes. This handler is ipoib_mcast_join_complete(), which
waits for the rtnl_lock(), which was already taken by ipoib_stop().
This bug was introduced in commit a77a57a1 ("IPoIB: Fix deadlock on
RTNL in ipoib_stop()").
Signed-off-by: Yossi Etigin <yosefe@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Commit c8c2afe3 ("IPoIB: Use rtnl lock/unlock when changing device
flags") added a call to rtnl_lock() in ipoib_mcast_join_task(), which
is run from the ipoib_workqueue. However, ipoib_stop() (which is run
inside rtnl_lock()) flushes this workqueue, which leads to a deadlock
if the join task is pending.
Fix this by simply not flushing the workqueue from ipoib_stop(). It
turns out that we really don't care about workqueue tasks running
during or after ipoib_stop(), as long as we make sure to flush the
workqueue before unregistering a netdev.
This fixes <https://bugs.openfabrics.org/show_bug.cgi?id=1114>.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Print the return code of ib_sa_path_rec_get() if it fails to help
debug errors.
Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
No need for a mutex around calls to ib_attach_mcast/ib_detach_mcast
since these operations are synchronized at the HW driver layer.
Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
The patch tries to solve the problem of device going down and paths being
flushed on an SM change event. The method is to mark the paths as candidates for
refresh (by setting the new valid flag to 0), and wait for an ARP
probe a new path record query.
The solution requires a different and less intrusive handling of SM
change event. For that, the second argument of the flush function
changes its meaning from a boolean flag to a level. In most cases, SM
failover doesn't cause LID change so traffic won't stop. In the rare
cases of LID change, the remote host (the one that hadn't changed its
LID) will lose connectivity until paths are refreshed. This is no
worse than the current state. In fact, preventing the device from
going down saves packets that otherwise would be lost.
Signed-off-by: Moni Levy <monil@voltaire.com>
Signed-off-by: Moni Shoua <monis@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Add "ipoib_use_lro" module parameter to enable LRO and an
"ipoib_lro_max_aggr" module parameter to set the max number of packets
to be aggregated. Make LRO controllable and LRO statistics accessible
through ethtool.
Signed-off-by: Vladimir Sokolovsky <vlad@mellanox.co.il>
Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
The connected mode implementation in the IPoIB driver has a large
overhead in the way SKBs are handled in the receive flow. It usually
allocates an SKB with as big as was used in the currently received SKB
and moves unused fragments from the old SKB to the new one. This
involves a loop on all the remaining fragments and incurs overhead on
the CPU. This patch, for small SKBs, allocates an SKB just large
enough to contain the received data and copies to it the data from the
received SKB. The newly allocated SKB is passed to the stack and the
old SKB is reposted.
When running netperf, UDP small messages, without this pach I get:
UDP UNIDIRECTIONAL SEND TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to
14.4.3.178 (14.4.3.178) port 0 AF_INET
Socket Message Elapsed Messages
Size Size Time Okay Errors Throughput
bytes bytes secs # # 10^6bits/sec
114688 128 10.00 5142034 0 526.31
114688 10.00 1130489 115.71
With this patch I get both send and receive at ~315 mbps.
The reason that send performance actually slows down is as follows:
When using this patch, the overhead of the CPU for handling RX packets
is dramatically reduced. As a result, we do not experience RNR NAK
messages from the receiver which cause the connection to be closed and
reopened again; when the patch is not used, the receiver cannot handle
the packets fast enough so there is less time to post new buffers and
hence the mentioned RNR NACKs. So what happens is that the
application *thinks* it posted a certain number of packets for
transmission but these packets are flushed and do not really get
transmitted. Since the connection gets opened and closed many times,
each time netperf gets the CPU time that otherwise would have been
given to IPoIB to actually transmit the packets. This can be verified
when looking at the port counters -- the output of ifconfig and the
oputput of netperf (this is for the case without the patch):
tx packets
==========
port counter: 1,543,996
ifconfig: 1,581,426
netperf: 5,142,034
rx packets
==========
netperf 1,1304,089
Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Use a dedicated CQ for UD send completions. Also, do not arm the UD
send CQ, which reduces the number of interrupts generated. This patch
farther reduces overhead by not calling poll CQ for every posted send
WR -- it does polls only when there 16 or more outstanding work requests.
Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
This patch enables IPoIB to use 4K UD messages (when the underlying
device and fabrics support a 4K MTU) by using two scatter buffers when
PAGE_SIZE is less than or equal to thhe HCA IB MTU size. The first
buffer is for IPoIB header + GRH header, and the second buffer is the
IPoIB payload, which is 4K-4.
Signed-off-by: Shirley Ma <xma@us.ibm.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Just add the infrastructure so we can add functionality later.
Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
For HCAs that support TCP segmentation offload (IB_DEVICE_UD_TSO), set
NETIF_F_TSO and use HW LSO to offload TCP segmentation.
Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Convert list_splice() + INIT_LIST_HEAD() to the equivalent list_splice_init()
Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
For HCAs that support checksum offload (ie that set IB_DEVICE_UD_IP_CSUM
in the device capabilities flags), have IPoIB set NETIF_F_IP_CSUM and
use the HCA to generate and verify IP checksums.
Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Commit 7143740d ("IPoIB: Add send gather support") made struct
ipoib_tx_buf significantly larger, since the mapping member changed
from a single u64 to an array with MAX_SKB_FRAGS + 1 entries. This
means that allocating tx_rings with kzalloc() may fail because there
is not enough contiguous memory for the new, much bigger size. Fix
this regression by allocating the rings with vmalloc() instead.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
All current InfiniBand devices can handle all DMA addresses, and it's
hard to imagine anyone would be silly enough to build a new device
that couldn't. Therefore, enable the NETIF_F_HIGHDMA feature for IPoIB.
This has no effect for no, but is needed when we enable gather/scatter
support and checksum stateless offloads.
Signed-off-by: Eli Cohen <eli@mellnaox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Commit 732a2170 ("IB/ipoib: Bound the net device to the ipoib_neigh
structue") left a misleading debug print (n->dev would be a bond
device only if boding is used). Clean it up.
Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Move up the code that checks for a situation where the remote GID
stored in the ipoib_neigh is different than the one present in the
neighbour (handle gratuitous ARP) or that a bonding fail over has
happened but the neighbour still has a pointer to an ipoib_neigh
created by a different device than the current slave. This will cause
the driver to apply the check also for connected mode neighbours.
Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
qdisc_run() now tests for queue_stopped() before calling
__qdisc_run(), and the same check is done in every iteration of
__qdisc_run(), so another check is not required in the driver xmit.
This means that ipoib_start_xmit() no longer needs to test
netif_queue_stopped(); the test was added to fix earlier kernels,
where the networking stack did not guarantee that the xmit method of
an LLTX driver would not be called after the queue was stopped, but
current kernels do provide this guarantee.
To validate, I put a debug in the TX_BUSY path which never hit with 64
threads running overnight exercising this code a few 100 million
times.
Signed-off-by: Krishna Kumar <krkumar2@in.ibm.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Some HCAs (such as ehca2) support SRQ, but only support fewer than 16 SG
entries for SRQs. Currently IPoIB/CM implicitly assumes all HCAs will
support 16 SG entries for SRQs (to handle a 64K MTU with 4K pages). This
patch removes that restriction by limiting the maximum MTU in connected
mode to what the maximum number of SRQ SG entries allows.
This patch addresses <https://bugs.openfabrics.org/show_bug.cgi?id=728>
Signed-off-by: Pradeep Satyanarayana <pradeeps@linux.vnet.ibm.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Some IB adapters (notably IBM's eHCA) do not implement SRQs (shared
receive queues). The current IPoIB connected mode support only works
on devices that support SRQs.
Fix this by adding support for using the receive queue of each
connected mode receive QP. The disadvantage of this compared to using
an SRQ is that it means a full queue of receives must be posted for
each remote connected mode peer, which means that total memory usage
is potentially much higher than when using SRQs. To manage this, add
a new module parameter "max_nonsrq_conn_qp" that limits the number of
connections allowed per interface.
The rest of the changes are fairly straightforward: we use a table of
struct ipoib_cm_rx to hold all the active connections, and put the
table index of the connection in the high bits of receive WR IDs.
This is needed because we cannot rely on the struct ib_wc.qp field for
non-SRQ receive completions. Most of the rest of the changes just
test whether or not an SRQ is available, and post receives or find
received packets in the right place depending on the answer.
Cleaning up dead connections actually becomes simpler, because we do
not have to do the "last WQE reached" dance that is required to
destroy QPs attached to an SRQ. We just move the QP to the error
state and wait for all pending receives to be flushed.
Signed-off-by: Pradeep Satyanarayana <pradeeps@linux.vnet.ibm.com>
[ Completely rewritten and split up, based on Pradeep's work. Several
bugs fixed and no doubt several bugs introduced. - Roland ]
Signed-off-by: Roland Dreier <rolandd@cisco.com>
