Eliminate ucm.h.  Replace ucm_dbg with direct call to printk KERN_ERR.
Signed-off-by: NSean Hefty <sean.hefty@intel.com>

Add kernel support for userspace calling poll CQ, request CQ
notification, post send, post receive, post SRQ receive, create AH and
destroy AH commands.  These commands allow us to support userspace
verbs for devices that can't perform these operations directly from
userspace (eg the PathScale HCA).
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Give each device a uverbs_cmd_mask, so that a low-level driver can
control which methods may be called on behalf of userspace.
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

If the SA query module's initialization fails for a device, then that
device won't have a struct ib_sa_device associated.  We should fail SA
queries in that case, rather than blindly dereferencing the NULL
pointer we get back from ib_get_client_data().
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

A couple of functions were missing spin_unlock calls in error paths.
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

ipoib_create_qp() no longer creates IPoIB's QP, so it shouldn't
destroy the QP on failure -- that unwinding happens elsewhere, so the
current code can cause a double free.  While we're at it, the
function's name should match what it actually does, so rename it to
ipoib_init_qp().
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Check the sizes of CQs, QPs and SRQs when creating objects, and fail
instead of creating too-big queues.  Also return real limits instead
of just plausible-sounding values from mthca_query_device().
Signed-off-by: NJack Morgenstein <jackm@mellanox.co.il>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

There is a bug in ib_mad_init_device(): if ib_agent_port_open() fails
for a given port, then the current code doesn't call ib_mad_port_close()
for that port.
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

The hardware relies on us keeping one extra work request that never
gets used in SRQs.  Add checks to the SRQ work request posting
functions so that they fail when someone is about to use up that extra
work request, rather than when someone uses the very last work request.
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Our hardware supports generating an event when the number of receives
posted to a shared receive queue (SRQ) falls below a user-specified
limit.  Implement mthca_modify_srq() to arm the limit, and add code to
handle dispatching SRQ events when they occur.
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Check port number before passing query_port or modify_port operations
on to device driver.
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Reject userspace memory registrations with invalid permission flags:
"local write" is required if "remote write" or "remote atomic" is also
requested.

Pointed out by Jack Morgenstein <jackm@mellanox.co.il>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Add code to fill in the bad_pkey_cntr, max_mtu, active_mtu and
subnet_timeout fields in mthca_query_port().
Signed-off-by: NJack Morgenstein <jackm@mellanox.co.il>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Add abi_version attribute to uverbs class devices to allow for
ABI versioning of device-specific interfaces.
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

By waiting to add resources to our lists until after the last
operation that can fail, we don't have to remove them from their lists
in the error path.  Also, we should hold the idr mutex until we know
whether resource creation has succeed or failed, to avoid someone
finding a resource in our table before we're ready.

Loosely based on work by Robert Walsh <rjwalsh@pathscale.com>.
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Introduce new userspace verbs ABI version 3.  This eliminates some
unneeded commands, and adds support for user-created completion
channels.  This cleans up problems with file leaks on error paths, and
also makes sure that file descriptors are always installed into the
correct process.
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Return correct atomic capability flag from mthca query function.
Signed-off-by: NJack Morgenstein <jackm@mellanox.co.il>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Add checks so that we only allow multicast attach/detach with
a valid multicast GID and the correct QP type.
Signed-off-by: NJack Morgenstein <jackm@mellanox.co.il>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

 - added typedef unsigned int __nocast gfp_t;

 - replaced __nocast uses for gfp flags with gfp_t - it gives exactly
   the same warnings as far as sparse is concerned, doesn't change
   generated code (from gcc point of view we replaced unsigned int with
   typedef) and documents what's going on far better.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Remember to free the multicast group context memory table.
Signed-off-by: NMichael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Rename IPoIB driver's path_lookup() to ipoib_path_lookup() to avoid a
clashes with the kernel global path_lookup().  We don't hit this with
the current kernel source, but some external patches seem to trigger
this, and it's cleaner to avoid clashing with global names anyway.
Signed-off-by: NRoland Dreier <rolandd@cisco.com>
refs/heads/for-linus

The IB spec defines the field to be 32 bits, not 16 bits.
Signed-off-by: NJack Morgenstein <jackm@mellanox.co.il>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

When allocating a table for mem-free HCA context, don't assume that
obj_size * nobj is an even multiple of MTHCA_TABLE_CHUNK_SIZE.  In
particular, make sure we allocate at least one slot even if the table
is smaller than MTHCA_TABLE_CHUNK_SIZE.
Signed-off-by: NMichael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Al Viro pointed out that the current IB userspace verbs interface
allows userspace to cause mischief by closing file descriptors before
we're ready, or issuing the same command twice at the same time.  This
patch closes those races, and fixes other obvious problems such as a
module reference leak.

Some other interface bogosities will require an ABI change to fix
properly, so I'm deferring those fixes until 2.6.15.
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

The loop in mthca_map_cmd() would fill one entry past the end of the
mailbox buffer before calling the firmware command.
Signed-off-by: NMichael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

We should use the first word of the clear interrupt register if
the bit we're after is < 32, not < 31.
Signed-off-by: NMichael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

If we allocate a bunch of doorbell records and then free them, we'll
end up with completely empty pages, which we then free.  However, when
we come back to allocate more doorbell pages, we have to reallocate
those empty pages rather than always trying to take a slot that we've
never used.  If we don't, we eventually use up every slot and fail to
allocate a doorbell record, even though we have plenty of free space.
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Based on simplification idea from Sean Hefty <sean.hefty@intel.com>
Signed-off-by: NHal Rosenstock <halr@voltaire.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Clean up code by using enums instead of hard-coded magic numbers.
Signed-off-by: NSean Hefty <sean.hefty@intel.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Clean up QP table array on device removal.
Signed-off-by: NMichael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

ipoib_mcast_restart_task() is always called from within the
single-threaded IPoIB workqueue, so flushing the workqueue from within
the function can lead to a recursion overflow.  But since we're
running in a single-threaded workqueue, we're already synchronized
against other items in the workqueue, so just get rid of the flush in
ipoib_mcast_restart_task().
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

We need to subtract off the header length from our payload
length when sending multi-packet SA messages.
Signed-off-by: NHal Rosenstock <halr@voltaire.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Userspace SRQs don't have a buffer allocated for them in the kernel, so
it doesn't make sense to set srq->last during initialization.  In fact,
this can crash trying to follow a nonexistent buffer pointer.
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

The error handling paths in mthca_tavor_post_srq_recv() and
mthca_arbel_post_srq_recv() are quite bogus, the result of a
screwed up merge.  Fix them so they work as intended.

Pointed out by Michael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

In mthca_create_eq(), we call get_eqe() before setting eq->nent.  This
is wrong, because get_eqe() uses eq->nent.  Fix this, and clean up the
code a little while we're at it.  (We got lucky with the current code,
because eq->nent was cleared to 0, which get_eqe() made happen to do
the right thing)

Pointed out by Michael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

We got a little mixed up with what the backoff member holds in the
IPoIB multicast group structure: sometimes it was used as a number of
seconds, and sometimes it was used as a number of jiffies.  Fix the
code so that backoff is always in seconds.
Signed-off-by: NHal Rosenstock <halr@voltaire.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Fix posting first WQE for mem-free HCAs: we need to link to previous
WQE even in that case.  While we're at it, simplify code for
Tavor-mode HCAs.  We don't really need the conditional test there
either; we can similarly always link to the previous WQE.

Based on Michael S. Tsirkin's analogous fix for userspace libmthca.
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

The hardware reads the ACK timeout field from the most significant 5
bits of struct mthca_qp_path's ackto field, not the least significant
bits.  This fix has the driver put the timeout in the right place.
Without this, we get a timeout that is 2^8 times too small.
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Since ipoib uses queue_delayed_work to run flush task on port state events,
it must flush scheduled work after unregistering the event handler.
Signed-off-by: NMichael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

If the CM REQ handling function gets to error2, then it frees
cm_id_priv->timewait_info.  But the next line goes through
ib_destroy_cm_id() -> ib_send_cm_rej() -> cm_reset_to_idle(),
which ends up calling cm_cleanup_timewait(), which dereferences the
pointer we just freed.  Make sure we clear cm_id_priv->timewait_info
after freeing it, so that doesn't happen.
Signed-off-by: NRoland Dreier <rolandd@cisco.com>