Zach Brown [Tue, 18 May 2010 22:44:50 +0000 (15:44 -0700)]
RDS/IB: get the xmit max_sge from the RDS IB device on the connection
rds_ib_xmit_rdma() was calling ib_get_client_data() to get at the rds_ibdevice
just to get the max_sge for the transmit. This patch instead has it get it
directly off the rds_ibdev which is stored on the connection.
The current code won't free the rds_ibdev until all the IB connections that use
it are freed. So it's safe to reference the rds_ibdev this way. In the future
it also makes it easier to support proper reference counting of the rds_ibdev
struct.
As an additional bonus, this gets rid of the performance hit of calling in to
the IB stack to look up the rds_ibdev. The current implementation in the IB
stack acquires an interrupt blocking spinlock to protect the registration of
client callback data.
Zach Brown [Mon, 24 May 2010 20:14:59 +0000 (13:14 -0700)]
RDS/IB: rds_ib_cm_handle_connect() forgot to unlock c_cm_lock
rds_ib_cm_handle_connect() could return without unlocking the c_conn_lock if
rds_setup_qp() failed. Rather than adding another imbalanced mutex_unlock() to
this error path we only unlock the mutex once as we exit the function, reducing
the likelyhood of making this same mistake in the future. We remove the
previous mulitple return sites, leaving one unambigious return path.
Chris Mason [Tue, 11 May 2010 23:15:35 +0000 (16:15 -0700)]
rds: Fix reference counting on the for xmit_atomic and xmit_rdma
This makes sure we have the proper number of references in
rds_ib_xmit_atomic and rds_ib_xmit_rdma. We also consistently
drop references the same way for all message types as the IOs end.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Chris Mason [Tue, 11 May 2010 22:14:16 +0000 (15:14 -0700)]
rds: Fix RDMA message reference counting
The RDS send_xmit code was trying to get fancy with message
counting and was dropping the final reference on the RDMA messages
too early. This resulted in memory corruption and oopsen.
The fix here is to always add a ref as the parts of the message passes
through rds_send_xmit, and always drop a ref as the parts of the message
go through completion handling.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Chris Mason [Tue, 11 May 2010 22:09:45 +0000 (15:09 -0700)]
rds: Use RCU for the bind lookup searches
The RDS bind lookups are somewhat expensive in terms of CPU
time and locking overhead. This commit changes them into a
faster RCU based hash tree instead of the rbtrees they were using
before.
On large NUMA systems it is a significant improvement.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Chris Mason [Fri, 23 Apr 2010 01:59:15 +0000 (21:59 -0400)]
rds: rcu-ize rds_ib_get_device()
rds_ib_get_device is called very often as we turn an
ip address into a corresponding device structure. It currently
take a global spinlock as it walks different lists to find active
devices.
This commit changes the lists over to RCU, which isn't very complex
because they are not updated very often at all.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Chris Mason [Thu, 15 Apr 2010 20:38:14 +0000 (16:38 -0400)]
RDS: Use a generation counter to avoid rds_send_xmit loop
rds_send_xmit is required to loop around after it releases the lock
because someone else could done a trylock, found someone working on the
list and backed off.
But, once we drop our lock, it is possible that someone else does come
in and make progress on the list. We should detect this and not loop
around if another process is actually working on the list.
This patch adds a generation counter that is bumped every time we
get the lock and do some send work. If the retry notices someone else
has bumped the generation counter, it does not need to loop around and
continue working.
Signed-off-by: Chris Mason <chris.mason@oracle.com> Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Tue, 30 Mar 2010 00:08:49 +0000 (17:08 -0700)]
RDS: Remove send_quota from send_xmit()
The purpose of the send quota was really to give fairness
when different connections were all using the same
workq thread to send backlogged msgs -- they could only send
so many before another connection could make progress.
Now that each connection is pushing the backlog from its
completion handler, they are all guaranteed to make progress
and the quota isn't needed any longer.
A thread *will* have to send all previously queued data, as well
as any further msgs placed on the queue while while c_send_lock
was held. In a pathological case a single process can get
roped into doing this for long periods while other threads
get off free. But, since it can only do this until the transport
reports full, this is a bounded scenario.
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Mon, 29 Mar 2010 23:20:18 +0000 (16:20 -0700)]
RDS: Call rds_send_xmit() directly from sendmsg()
rds_sendmsg() is calling the send worker function to
send the just-queued datagrams, presumably because it wants
the behavior where anything not sent will re-call the send
worker. We now ensure all queued datagrams are sent by retrying
from the send completion handler, so this isn't needed any more.
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Wed, 24 Mar 2010 00:39:07 +0000 (17:39 -0700)]
RDS: Change send lock from a mutex to a spinlock
This change allows us to call rds_send_xmit() from a tasklet,
which is crucial to our new operating model.
* Change c_send_lock to a spinlock
* Update stats fields "sem_" to "_lock"
* Remove unneeded rds_conn_is_sending()
About locking between shutdown and send -- send checks if the
connection is up. Shutdown puts the connection into
DISCONNECTING. After this, all threads entering send will exit
immediately. However, a thread could be *in* send_xmit(), so
shutdown acquires the c_send_lock to ensure everyone is out
before proceeding with connection shutdown.
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Fri, 19 Mar 2010 00:19:52 +0000 (17:19 -0700)]
RDS: Refill recv ring directly from tasklet
Performance is better if we use allocations that don't block
to refill the receive ring. Since the whole reason we were
kicking out to the worker thread was so we could do blocking
allocs, we no longer need to do this.
Remove gfp params from rds_ib_recv_refill(); we always use
GFP_NOWAIT.
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Mon, 22 Mar 2010 22:22:04 +0000 (15:22 -0700)]
RDS: Stop supporting old cong map sending method
We now ask the transport to give us a rm for the congestion
map, and then we handle it normally. Previously, the
transport defined a function that we would call to send
a congestion map.
Convert TCP and loop transports to new cong map method.
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Thu, 4 Mar 2010 03:25:21 +0000 (19:25 -0800)]
RDS/IB: Do not wait for send ring to be empty on conn shutdown
Now that we are signaling send completions much less, we are likely
to have dirty entries in the send queue when the connection is
shut down (on rmmod, for example.) These are cleaned up a little
further down in conn_shutdown, but if we wait on the ring_empty_wait
for them, it'll never happen, and we hand on unload.
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Mon, 1 Mar 2010 22:03:09 +0000 (14:03 -0800)]
RDS: Perform unmapping ops in stages
Previously, RDS would wait until the final send WR had completed
and then handle cleanup. With silent ops, we do not know
if an atomic, rdma, or data op will be last. This patch
handles any of these cases by keeping a pointer to the last
op in the message in m_last_op.
When the TX completion event fires, rds dispatches to per-op-type
cleanup functions, and then does whole-message cleanup, if the
last op equalled m_last_op.
This patch also moves towards having op-specific functions take
the op struct, instead of the overall rm struct.
rds_ib_connection has a pointer to keep track of a a partially-
completed data send operation. This patch changes it from an
rds_message pointer to the narrower rm_data_op pointer, and
modifies places that use this pointer as needed.
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Tue, 13 Apr 2010 19:00:35 +0000 (12:00 -0700)]
RDS: Make sure cmsgs aren't used in improper ways
It hasn't cropped up in the field, but this code ensures it is
impossible to issue operations that pass an rdma cookie (DEST, MAP)
in the same sendmsg call that's actually initiating rdma or atomic
ops.
Disallowing this perverse-but-technically-allowed usage makes silent
RDMA heuristics slightly easier.
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Tue, 2 Mar 2010 00:10:40 +0000 (16:10 -0800)]
RDS: Add flag for silent ops. Do atomic op before RDMA
Add a flag to the API so users can indicate they want
silent operations. This is needed because silent ops
cannot be used with USE_ONCE MRs, so we can't just
assume silent.
Also, change send_xmit to do atomic op before rdma op if
both are present, and centralize the hairy logic to determine if
we want to attempt silent, or not.
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Sat, 20 Feb 2010 02:04:58 +0000 (18:04 -0800)]
RDS: queue failure notifications for dropped atomic ops
When dropping ops in the send queue, we notify the client
of failed rdma ops they asked for notifications on, but not
atomic ops. It should be for both.
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Mon, 1 Mar 2010 22:11:53 +0000 (14:11 -0800)]
RDS: Remove struct rds_rdma_op
A big changeset, but it's all pretty dumb.
struct rds_rdma_op was already embedded in struct rm_rdma_op.
Remove rds_rdma_op and put its members in rm_rdma_op. Rename
members with "op_" prefix instead of "r_", for consistency.
Of course this breaks a lot, so fixup the code accordingly.
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Thu, 28 Jan 2010 00:07:30 +0000 (16:07 -0800)]
RDS: Inline rdma_prepare into cmsg_rdma_args
cmsg_rdma_args just calls rdma_prepare and does a little
arg checking -- not quite enough to justify its existence.
Plus, it is the only caller of rdma_prepare().
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Fri, 15 Jan 2010 23:55:26 +0000 (15:55 -0800)]
RDS/IB: Make all flow control code conditional on i_flowctl
Maybe things worked fine with the flow control code running
even in the non-flow-control case, but making it explicitly
conditional helps the non-fc case be easier to read.
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Thu, 14 Jan 2010 20:18:11 +0000 (12:18 -0800)]
RDS: rewrite rds_ib_xmit
Now that the header always goes first, it is possible to
simplify rds_ib_xmit. Instead of having a path to handle 0-byte
dgrams and another path to handle >0, these can both be handled
in one path. This lets us eliminate xmit_populate_wr().
Rename sent to bytes_sent, to differentiate better from other
variable named "send".
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Thu, 14 Jan 2010 00:32:24 +0000 (16:32 -0800)]
RDS/IB: Remove ib_[header/data]_sge() functions
These functions were to cope with differently ordered
sg entries depending on RDS 3.0 or 3.1+. Now that
we've dropped 3.0 compatibility we no longer need them.
Also, modify usage sites for these to refer to sge[0] or [1]
directly. Reorder code to initialize header sgs first.
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Wed, 13 Jan 2010 23:50:09 +0000 (15:50 -0800)]
RDS/IB: Disallow connections less than RDS 3.1
RDS 3.0 connections (in OFED 1.3 and earlier) put the
header at the end. 3.1 connections put it at the head.
The code has significant added complexity in order to
handle both configurations. In OFED 1.6 we can
drop this and simplify the code by only supporting
"header-first" configuration.
This patch checks the protocol version, and if prior
to 3.1, does not complete the connection.
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Tue, 12 Jan 2010 22:43:06 +0000 (14:43 -0800)]
RDS/IB: eliminate duplicate code
both atomics and rdmas need to convert ib-specific completion codes
into RDS status codes. Rename rds_ib_rdma_send_complete to
rds_ib_send_complete, and have it take a pointer to the function to
call with the new error code.
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Tue, 12 Jan 2010 18:50:48 +0000 (10:50 -0800)]
RDS: Base init_depth and responder_resources on hw values
Instead of using a constant for initiator_depth and
responder_resources, read the per-QP values when the
device is enumerated, and then use these values when creating
the connection.
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Tue, 12 Jan 2010 22:19:32 +0000 (14:19 -0800)]
RDS: Clear up some confusing code in send_remove_from_sock
The previous code was correct, but made the assumption that
if r_notifier was non-NULL then either r_recverr or r_notify
was true. Valid, but fragile. Changed to explicitly check
r_recverr (shows up in greps for recverr now, too.)
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Tue, 12 Jan 2010 22:13:15 +0000 (14:13 -0800)]
RDS: make m_rdma_op a member of rds_message
This eliminates a separate memory alloc, although
it is now necessary to add an "r_active" flag, since
it is no longer to use the m_rdma_op pointer as an
indicator of if an rdma op is present.
rdma SGs allocated from rm sg pool.
rds_rm_size also gets bigger. It's a little inefficient to
run through CMSGs twice, but it makes later steps a lot smoother.
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Tue, 12 Jan 2010 20:56:06 +0000 (12:56 -0800)]
RDS: Explicitly allocate rm in sendmsg()
r_m_copy_from_user used to allocate the rm as well as kernel
buffers for the data, and then copy the data in. Now, sendmsg()
allocates the rm, although the data buffer alloc still happens
in r_m_copy_from_user.
SGs are still allocated with rm, but now r_m_alloc_sgs() is
used to reserve them. This allows multiple SG lists to be
allocated from the one rm -- this is important once we also
want to alloc our rdma sgl from this pool.
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Tue, 12 Jan 2010 20:37:17 +0000 (12:37 -0800)]
RDS: cleanup/fix rds_rdma_unuse
First, it looks to me like the atomic_inc is wrong.
We should be decrementing refcount only once here, no? It's
already being done by the mr_put() at the end.
Second, simplify the logic a bit by bailing early (with a warning)
if !mr.
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Andy Grover [Sat, 20 Feb 2010 02:01:41 +0000 (18:01 -0800)]
RDS: Rewrite rds_send_drop_to() for clarity
This function has been the source of numerous bugs; it's just
too complicated. Simplified to nest spinlocks cleanly within
the second loop body, and kick out early if there are no
rms to drop.
This will be a little slower because conn lock is grabbed for
each entry instead of "caching" the lock across rms, but this
should be entirely irrelevant to fastpath performance.
Signed-off-by: Andy Grover <andy.grover@oracle.com>
Tina Yang [Thu, 1 Apr 2010 21:09:00 +0000 (14:09 -0700)]
RDS: Fix corrupted rds_mrs
On second look at this bug (OFED #2002), it seems that the
collision is not with the retransmission queue (packet acked
by the peer), but with the local send completion. A theoretical
sequence of events (from time t0 to t3) is thought to be as
follows,
The problem with the rds_rdma_drop_keys() is it could
remove a mr's refcount more than its due (i.e. repeatedly
as long as it still remains in the tree (mr->r_refcount > 0)).
Theoretically it should remove only one reference - reference
by the tree.
/* Release any MRs associated with this socket */
while ((node = rb_first(&rs->rs_rdma_keys))) {
mr = container_of(node, struct rds_mr, r_rb_node);
if (mr->r_trans == rs->rs_transport)
mr->r_invalidate = 0;
rds_mr_put(mr);
}
I think the correct way of doing it is to remove the mr from
the tree and rds_destroy_mr it first, then a rds_mr_put()
to decrement its reference count by one. Whichever thread
holds the last reference will free the mr via rds_mr_put().
Signed-off-by: Tina Yang <tina.yang@oracle.com> Signed-off-by: Andy Grover <andy.grover@oracle.com>
If failed to get skb frags using napi_get_frags(),
the packet is dropped.
Reported-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il> Signed-off-by: David S. Miller <davem@davemloft.net>
Joe Perches [Sun, 5 Sep 2010 21:31:11 +0000 (21:31 +0000)]
net/caif: Use pr_fmt
This patch standardizes caif message logging prefixes.
Add #define pr_fmt(fmt) KBUILD_MODNAME ":%s(): " fmt, __func__
Add missing "\n"s to some logging messages
Convert pr_warning to pr_warn
This changes the logging message prefix from CAIF: to caif:
for all uses but caif_socket.c and chnl_net.c. Those now use
their filename without extension.
Signed-off-by: Joe Perches <joe@perches.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Julia Lawall [Sun, 5 Sep 2010 09:00:23 +0000 (09:00 +0000)]
drivers/atm/firestream.c: Fix unsigned return type
The function has an unsigned return type, but returns a negative constant
to indicate an error condition. The result of calling the function is
always stored in a variable of type (signed) int, and thus unsigned can be
dropped from the return type.
A sematic match that finds this problem is as follows:
(http://coccinelle.lip6.fr/)
Julia Lawall [Sun, 5 Sep 2010 09:00:21 +0000 (09:00 +0000)]
net/9p/trans_fd.c: Fix unsigned return type
The function has an unsigned return type, but returns a negative constant
to indicate an error condition. The result of calling the function is
always stored in a variable of type (signed) int, and thus unsigned can be
dropped from the return type.
A sematic match that finds this problem is as follows:
(http://coccinelle.lip6.fr/)
Joe Perches [Sat, 4 Sep 2010 22:21:05 +0000 (22:21 +0000)]
drivers/net/jme: Use pr_<level>
Add #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
Convert jeprintk to pr_err
Remove jeprintk macro define
Remove periods from end of logging messages
Coalesce format strings
Signed-off-by: Joe Perches <joe@perches.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Dan Carpenter [Sat, 4 Sep 2010 08:33:03 +0000 (08:33 +0000)]
isdn: cleanup: make buffer smaller
This showed up in my audit because we use strcpy() to copy "ds" into a
32 character buffer inside the isdn_tty_dial() function. But it turns
out that we only ever use the first 32 characters so it's OK. I have
changed the declaration to make the static checkers happy.
Signed-off-by: Dan Carpenter <error27@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Dan Carpenter [Sat, 4 Sep 2010 08:38:59 +0000 (08:38 +0000)]
isdn: potential buffer overflows
cs->ics.parm.setup.phone is a 32 character array. In each of these
cases we're copying from a 35 character array into a 32 character array
so we should use strlcpy() instead of strcpy().
Signed-off-by: Dan Carpenter <error27@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
bnad_set_rx_mode omit to unlock bna_lock on one fail path. Fix that.
Signed-off-by: Jiri Slaby <jslaby@suse.cz> Cc: Debashis Dutt <ddutt@brocade.com> Cc: Rasesh Mody <rmody@brocade.com> Cc: David S. Miller <davem@davemloft.net> Signed-off-by: David S. Miller <davem@davemloft.net>
Julia Lawall [Sat, 4 Sep 2010 00:12:43 +0000 (00:12 +0000)]
drivers/net/fs_enet/fs_enet-main.c: Add of_node_put to avoid memory leak
In this case, a device_node structure is stored in another structure that
is then freed without first decrementing the reference count of the
device_node structure.
The semantic match that finds this problem is as follows:
(http://coccinelle.lip6.fr/)
Santiago Leon [Fri, 3 Sep 2010 18:29:53 +0000 (18:29 +0000)]
ibmveth: Update module information and version
Add an entry to the MAINTAINERS file for ibmveth, clean up the copyright
and add all authors. Change the name of the module to reflect the product name
over the last number of years.
Considering all the changes we have made, bump the driver version.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Santiago Leon <santil@linux.vnet.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Santiago Leon [Fri, 3 Sep 2010 18:29:47 +0000 (18:29 +0000)]
ibmveth: Remove some unnecessary include files
These files probably came across from the skeleton driver. Remove them.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Santiago Leon <santil@linux.vnet.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Santiago Leon [Fri, 3 Sep 2010 18:29:41 +0000 (18:29 +0000)]
ibmveth: Convert driver specific assert to BUG_ON
We had a driver specific assert function which wasn't enabled most of the
time. Convert them to BUG_ON and enable them all the time.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Santiago Leon <santil@linux.vnet.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Santiago Leon [Fri, 3 Sep 2010 18:29:36 +0000 (18:29 +0000)]
ibmveth: Return -EINVAL on all ->probe errors
We had a few cases where we returned success on error.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Santiago Leon <santil@linux.vnet.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Santiago Leon [Fri, 3 Sep 2010 18:29:30 +0000 (18:29 +0000)]
ibmveth: Coding style fixes
Fix most of the kernel coding style issues in ibmveth.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Santiago Leon <santil@linux.vnet.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Santiago Leon [Fri, 3 Sep 2010 18:29:25 +0000 (18:29 +0000)]
ibmveth: Some formatting fixes
IbmVethNumBufferPools -> IBMVETH_NUM_BUFF_POOLS
Also change IBMVETH_MAX_MTU -> IBMVETH_MIN_MTU, it refers to the minimum
size not the maximum.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Santiago Leon <santil@linux.vnet.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Santiago Leon [Fri, 3 Sep 2010 18:29:19 +0000 (18:29 +0000)]
ibmveth: Convert driver specific error functions to netdev_err
Use netdev_err to standardise the error output.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Santiago Leon <santil@linux.vnet.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Santiago Leon [Fri, 3 Sep 2010 18:29:14 +0000 (18:29 +0000)]
ibmveth: Convert driver specific debug to netdev_dbg
Use netdev_dbg to standardise the debug output.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Santiago Leon <santil@linux.vnet.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Santiago Leon [Fri, 3 Sep 2010 18:29:08 +0000 (18:29 +0000)]
ibmveth: Remove redundant function prototypes
These functions appear before their use, so we can remove the redundant
prototypes.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Santiago Leon <santil@linux.vnet.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Santiago Leon [Fri, 3 Sep 2010 18:29:03 +0000 (18:29 +0000)]
ibmveth: Convert to netdev_alloc_skb
We were using alloc_skb which doesn't create any headroom. Change it to
use netdev_alloc_skb to match most other drivers.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Santiago Leon <santil@linux.vnet.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Santiago Leon [Fri, 3 Sep 2010 18:28:58 +0000 (18:28 +0000)]
ibmveth: remove procfs code
We export all the driver specific statistics via ethtool, so there is no need
to duplicate this in procfs.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Santiago Leon <santil@linux.vnet.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Santiago Leon [Fri, 3 Sep 2010 18:28:52 +0000 (18:28 +0000)]
ibmveth: Enable IPv6 checksum offload
This patch enables TCP checksum offload support for IPv6 on ibmveth.
This completely eliminates the generation and checking of the checksum
for IPv6 packets that are completely virtual and never touch a physical
network. A basic TCPIPV6_STREAM netperf run showed a ~30% throughput
improvement when an MTU of 64000 was used.
This featured is enabled by default, as is the case for IPv4 checksum
offload. When checksum offload is enabled the driver will negotiate
IPv4 and IPv6 offload with the firmware separately and enable what
is available. As long as either IPv4 or IPv6 offload is supported
and enabled the device will report that checksum offload is enabled.
The device stats, available through ethtool, will display which
checksum offload features are supported/enabled by firmware.
Performance testing against a stock kernel shows no regression for IPv4
or IPv6 in terms of throughput or processor utilization with checksum
disabled or enabled.
Signed-off-by: Robert Jennings <rcj@linux.vnet.ibm.com> Signed-off-by: Santiago Leon <santil@linux.vnet.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Remove code in the device probe function where we set up the checksum
offload feature and replace it with a call to an existing function that
is doing the same. This is done to clean up the driver in preparation
of adding IPv6 checksum offload support.
Signed-off-by: Robert Jennings <rcj@linux.vnet.ibm.com> Signed-off-by: Santiago Leon <santil@linux.vnet.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Santiago Leon [Fri, 3 Sep 2010 18:28:41 +0000 (18:28 +0000)]
ibmveth: Add optional flush of rx buffer
On some machines we can improve the bandwidth by ensuring rx buffers are
not in the cache. Add a module option that is disabled by default that flushes
rx buffers on insertion.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Santiago Leon <santil@linux.vnet.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Santiago Leon [Fri, 3 Sep 2010 18:28:36 +0000 (18:28 +0000)]
ibmveth: Add scatter-gather support
ibmveth can scatter gather up to 6 segments. If we go over this then
we have no option but to call skb_linearize, like other drivers with
similar limitations do.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Santiago Leon <santil@linux.vnet.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Anton Blanchard [Tue, 7 Sep 2010 01:21:41 +0000 (18:21 -0700)]
ibmveth: Use lighter weight read memory barrier in ibmveth_poll
We want to order the read in ibmveth_rxq_pending_buffer and the read of
ibmveth_rxq_buffer_valid which are both cacheable memory. smp_rmb() is good
enough for this.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Santiago Leon <santil@linux.vnet.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Santiago Leon [Fri, 3 Sep 2010 18:28:25 +0000 (18:28 +0000)]
ibmveth: Add rx_copybreak
For small packets, create a new skb and copy the packet into it so we
avoid tearing down and creating a TCE entry.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Santiago Leon <santil@linux.vnet.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Santiago Leon [Fri, 3 Sep 2010 18:28:20 +0000 (18:28 +0000)]
ibmveth: Add tx_copybreak
Use the existing bounce buffer if we send a buffer under a certain size.
This saves the overhead of a TCE map/unmap.
I can't see any reason for the wmb() in the bounce buffer case, if we need
a barrier it will be before we call h_send_logical_lan but we have
nothing in the common case. Remove it.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Santiago Leon <santil@linux.vnet.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
projects / linux-beck.git / log