This patch adds support for the IB "base memory management extension"
(BMME) and the equivalent iWARP operations (which the iWARP verbs
mandates all devices must implement). The new operations are:
- Allocate an ib_mr for use in fast register work requests.
- Allocate/free a physical buffer lists for use in fast register work
requests. This allows device drivers to allocate this memory as
needed for use in posting send requests (eg via dma_alloc_coherent).
- New send queue work requests:
* send with remote invalidate
* fast register memory region
* local invalidate memory region
* RDMA read with invalidate local memory region (iWARP only)
Consumer interface details:
- A new device capability flag IB_DEVICE_MEM_MGT_EXTENSIONS is added
to indicate device support for these features.
- New send work request opcodes IB_WR_FAST_REG_MR, IB_WR_LOCAL_INV,
IB_WR_RDMA_READ_WITH_INV are added.
- A new consumer API function, ib_alloc_mr() is added to allocate
fast register memory regions.
- New consumer API functions, ib_alloc_fast_reg_page_list() and
ib_free_fast_reg_page_list() are added to allocate and free
device-specific memory for fast registration page lists.
- A new consumer API function, ib_update_fast_reg_key(), is added to
allow the key portion of the R_Key and L_Key of a fast registration
MR to be updated. Consumers call this if desired before posting
a IB_WR_FAST_REG_MR work request.
Consumers can use this as follows:
- MR is allocated with ib_alloc_mr().
- Page list memory is allocated with ib_alloc_fast_reg_page_list().
- MR R_Key/L_Key "key" field is updated with ib_update_fast_reg_key().
- MR made VALID and bound to a specific page list via
ib_post_send(IB_WR_FAST_REG_MR)
- MR made INVALID via ib_post_send(IB_WR_LOCAL_INV),
ib_post_send(IB_WR_RDMA_READ_WITH_INV) or an incoming send with
invalidate operation.
- MR is deallocated with ib_dereg_mr()
- page lists dealloced via ib_free_fast_reg_page_list().
Applications can allocate a fast register MR once, and then can
repeatedly bind the MR to different physical block lists (PBLs) via
posting work requests to a send queue (SQ). For each outstanding
MR-to-PBL binding in the SQ pipe, a fast_reg_page_list needs to be
allocated (the fast_reg_page_list is owned by the low-level driver
from the consumer posting a work request until the request completes).
Thus pipelining can be achieved while still allowing device-specific
page_list processing.
The 32-bit fast register memory key/STag is composed of a 24-bit index
and an 8-bit key. The application can change the key each time it
fast registers thus allowing more control over the peer's use of the
key/STag (ie it can effectively be changed each time the rkey is
rebound to a page list).
Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Add a new IB_WR_SEND_WITH_INV send opcode that can be used to mark a
"send with invalidate" work request as defined in the iWARP verbs and
the InfiniBand base memory management extensions. Also put "imm_data"
and a new "invalidate_rkey" member in a new "ex" union in struct
ib_send_wr. The invalidate_rkey member can be used to pass in an
R_Key/STag to be invalidated. Add this new union to struct
ib_uverbs_send_wr. Add code to copy the invalidate_rkey field in
ib_uverbs_post_send().
Fix up low-level drivers to deal with the change to struct ib_send_wr,
and just remove the imm_data initialization from net/sunrpc/xprtrdma/,
since that code never does any send with immediate operations.
Also, move the existing IB_DEVICE_SEND_W_INV flag to a new bit, since
the iWARP drivers currently in the tree set the bit. The amso1100
driver at least will silently fail to honor the IB_SEND_INVALIDATE bit
if passed in as part of userspace send requests (since it does not
implement kernel bypass work request queueing). Remove the flag from
all existing drivers that set it until we know which ones are OK.
The values chosen for the new flag is not consecutive to avoid clashing
with flags defined in the XRC patches, which are not merged yet but
which are already in use and are likely to be merged soon.
This resurrects a patch sent long ago by Mikkel Hagen <mhagen@iol.unh.edu>.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Add a create_flags member to struct ib_qp_init_attr that will allow a
kernel verbs consumer to create a pass special flags when creating a QP.
Add a flag value for telling low-level drivers that a QP will be used
for IPoIB UD LSO. The create_flags member will also be useful for XRC
and ehca low-latency QP support.
Since no create_flags handling is implemented yet, add code to all
low-level drivers to return -EINVAL if create_flags is non-zero.
Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Commit 9ead190b ("IB/uverbs: Don't serialize with ib_uverbs_idr_mutex")
rewrote how userspace objects are looked up in the uverbs module's
idrs, and introduced a severe bug in the process: there is no checking
that an operation is being performed by the right process any more.
Fix this by adding the missing check of uobj->context in __idr_get_uobj().
Apparently everyone is being very careful to only touch their own
objects, because this bug was introduced in June 2006 in 2.6.18, and
has gone undetected until now.
Cc: stable <stable@kernel.org>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Export ib_umem_get()/ib_umem_release() and put low-level drivers in
control of when to call ib_umem_get() to pin and DMA map userspace,
rather than always calling it in ib_uverbs_reg_mr() before calling the
low-level driver's reg_user_mr method.
Also move these functions to be in the ib_core module instead of
ib_uverbs, so that driver modules using them do not depend on
ib_uverbs.
This has a number of advantages:
- It is better design from the standpoint of making generic code a
library that can be used or overridden by device-specific code as
the details of specific devices dictate.
- Drivers that do not need to pin userspace memory regions do not
need to take the performance hit of calling ib_mem_get(). For
example, although I have not tried to implement it in this patch,
the ipath driver should be able to avoid pinning memory and just
use copy_{to,from}_user() to access userspace memory regions.
- Buffers that need special mapping treatment can be identified by
the low-level driver. For example, it may be possible to solve
some Altix-specific memory ordering issues with mthca CQs in
userspace by mapping CQ buffers with extra flags.
- Drivers that need to pin and DMA map userspace memory for things
other than memory regions can use ib_umem_get() directly, instead
of hacks using extra parameters to their reg_phys_mr method. For
example, the mlx4 driver that is pending being merged needs to pin
and DMA map QP and CQ buffers, but it does not need to create a
memory key for these buffers. So the cleanest solution is for mlx4
to call ib_umem_get() in the create_qp and create_cq methods.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Add a num_comp_vectors member to struct ib_device and extend
ib_create_cq() to pass in a comp_vector parameter -- this parallels
the userspace libibverbs API. Update all hardware drivers to set
num_comp_vectors to 1 and have all ULPs pass 0 for the comp_vector
value. Pass the value of num_comp_vectors to userspace rather than
hard-coding a value of 1.
We want multiple CQ event vector support (via MSI-X or similar for
adapters that can generate multiple interrupts), but it's not clear
how many vectors we want, or how we want to deal with policy issues
such as how to decide which vector to use or how to set up interrupt
affinity. This patch is useful for experimenting, since no core
changes will be necessary when updating a driver to support multiple
vectors, and we know that we want to make at least these changes
anyway.
Signed-off-by: Michael S. Tsirkin <mst@dev.mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
If no matching PD is found in ib_uverbs_reg_mr(), then the function
jumps to err_release without setting the return value ret. This means
that ret will hold the return value of the call to ib_umem_get() a few
lines earlier; if the function reaches the point where it looks for
the PD, we know that ib_umem_get() must have returned 0, so
ib_uverbs_reg_mr() ends up return 0 for a bad PD ID. Fix this by
setting ret to -EINVAL before jumping to the exit path when no PD is
found.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
struct ib_wc currently only includes the local QP number: this matches
the IB spec, but seems mostly useless. The following patch replaces
this with the pointer to qp itself, and updates all low level drivers
and all users.
This has the following advantages:
- Ability to get a per-qp context through wc->qp->qp_context
- Existing drivers already have the qp pointer ready in poll cq, so
this change actually saves a tiny bit (extra memory read) on data path
(for ehca it would actually be expensive to find the QP pointer when
polling a CQ, but ehca does not support SRQ so we can leave wc->qp as
NULL for ehca)
- Users that need the QP number can still get it through wc->qp->qp_num
Use case:
In IPoIB connected mode code, I have a common CQ shared by multiple
QPs. To track connection usage, I need a way to get at some per-QP
context upon the completion, and I would like to avoid allocating
context object per work request just to stick a QP pointer into it.
With this code, I can just use wc->qp->qp_context.
Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Return the sq_draining value back to user space for query_qp instead
of the en_sqd_async notify value, which is valid only for
modify_qp. For query_qp, the draining status should returned.
Signed-off-by: Jack Morgenstein <jackm@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Remove some trailing whitespace that has snuck in despite the best
efforts of whitespace=error-all. Also fix a few other whitespace
bogosities.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Pass a struct ib_udata to the low-level driver's ->modify_srq() and
->modify_qp() methods, so that it can get to the device-specific data
passed in by the userspace driver.
Signed-off-by: Ralph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Add a ib_uverbs_resize_cq_resp.driver_data field so that low-level
drivers can return data from a resize CQ operation to userspace. Have
ib_uverbs_resize_cq() only copy the cqe field, to avoid having to bump
the userspace ABI.
Signed-off-by: Ralph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Lockdep warns when userspace creates a QP that uses different CQs for
send completions and receive completions, because both CQs are locked
and their mutexes belong to the same lock class. However, we know
that the mutexes are distinct and the nesting is safe (there is no
possibility of AB-BA deadlock because the mutexes are locked with
down_read()), so annotate the situation with SINGLE_DEPTH_NESTING to
get rid of the lockdep warning.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
There were two functions that open-coded idr_read_cq() in terms of
idr_read_uobj() rather than using the helper.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Lockdep warns because uverbs is trying to take uobj->mutex when it
already holds that lock. This is because there are really multiple
types of uobjs even though all of their locks are initialized in
common code.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
ib_uverbs_create_ah() and ib_uverbs_create_srq() did not release the
PD's read lock in their error paths, which lead to deadlock when
destroying the PD.
Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Store away the user handle passed in from userspace when creating an
SRQ, so that the kernel can return the correct handle when an SRQ
asynchronous event occurs. (A 0 was incorrectly stored as the user
handle as part of the changes in 9ead190b, "IB/uverbs: Don't serialize
with ib_uverbs_idr_mutex")
Signed-off-by: Roland Dreier <rolandd@cisco.com>
In ib_uverbs_post_send(), move the "out:" label after the loop that
frees the list of work requests, since the only place that jumps there
is before any work requests could possibly be added to the list.
This removes a compile warning: "is_ud might be used uninitialized in
this function".
Signed-off-by: Krishna Kumar <krkumar2@in.ibm.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Currently, all userspace verbs operations that call into the kernel
are serialized by ib_uverbs_idr_mutex. This can be a scalability
issue for some workloads, especially for devices driven by the ipath
driver, which needs to call into the kernel even for datapath
operations.
Fix this by adding reference counts to the userspace objects, and then
converting ib_uverbs_idr_mutex into a spinlock that only protects the
idrs long enough to take a reference on the object being looked up.
Because remove operations may fail, we have to do a slightly funky
two-step deletion, which is described in the comments at the top of
uverbs_cmd.c.
This also still leaves ib_uverbs_idr_lock as a single lock that is
possibly subject to contention. However, the lock hold time will only
be a single idr operation, so multiple threads should still be able to
make progress, even if ib_uverbs_idr_lock is being ping-ponged.
Surprisingly, these changes even shrink the object code:
add/remove: 23/5 grow/shrink: 4/21 up/down: 633/-693 (-60)
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Factor out common code for adding a userspace object to an idr into a
function idr_add_uobj(). This shrinks both the source and object code:
add/remove: 1/0 grow/shrink: 0/6 up/down: 57/-220 (-163)
function old new delta
idr_add_uobj - 57 +57
ib_uverbs_create_ah 543 512 -31
ib_uverbs_create_srq 662 630 -32
ib_uverbs_reg_mr 737 699 -38
ib_uverbs_create_cq 639 600 -39
ib_uverbs_alloc_pd 485 446 -39
ib_uverbs_create_qp 1020 979 -41
Signed-off-by: Roland Dreier <rolandd@cisco.com>
In error paths when destroying an object, uverbs should not decrement
associated objects' usecnt, since ib_dereg_mr(), ib_destroy_qp(),
etc. already do that.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
If ibdev->alloc_ucontext() fails then ib_uverbs_get_context() does not
unlock file->mutex before returning error.
Signed-off by: Ganapathi CH <cganapathi@novell.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
The old code incorrectly used the primary P_Key index as the alternate
index too.
Signed-off-by: Ami Perlmutter <amip@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
The old code didn't convert from the kernel's enum correctly.
Signed-off-by: Dotan Barak <dotanb@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Pass actual capacity of created SRQ back to userspace, so that
userspace can report accurate capacities. This requires an ABI bump,
to change struct ib_uverbs_create_srq_resp.
Signed-off-by: Dotan Barak <dotanb@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Add support to uverbs to handle querying userspace SRQs (shared
receive queues), including adding an ABI for marshalling requests and
responses. The kernel midlayer already has the underlying
ib_query_srq() function.
Signed-off-by: Dotan Barak <dotanb@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Add support to uverbs to handle querying userspace QPs (queue pairs),
including adding an ABI for marshalling requests and responses. The
kernel midlayer already has the underlying ib_query_qp() function.
Signed-off-by: Dotan Barak <dotanb@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Add support to uverbs to handle resizing userspace CQs (completion
queues), including adding an ABI for marshalling requests and
responses. The kernel midlayer already has ib_resize_cq().
Signed-off-by: Roland Dreier <rolandd@cisco.com>
semaphore to mutex conversion by Ingo and Arjan's script.
Signed-off-by: Ingo Molnar <mingo@elte.hu>
[ Sanity-checked on real IB hardware ]
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Add a node_guid field to struct ib_device. It is the responsibility
of the low-level driver to initialize this field before registering a
device with the midlayer. Convert everyone to looking at this field
instead of calling ib_query_device() when all they want is the node
GUID, and remove the node_guid field from struct ib_device_attr.
Signed-off-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
ib_uverbs_create_cq() should release the completion channel event file
if an error occurs after it looks it up. Also, if userspace asks for
a completion channel and we don't find it, an error should be returned
instead of silently creating a CQ without a completion channel.
Signed-off-by: Jack Morgenstein <jackm@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
If an operation fails after incrementing an object's reference count,
then it should decrement the reference count on the error path.
Signed-off-by: Jack Morgenstein <jackm@mellanox.co.il>
Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
uverbs needs to track which multicast groups is each qp
attached to, in order to properly detach when cleanup
is performed on device file close.
Signed-off-by: Jack Morgenstein <jackm@mellanox.co.il>
Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Move the computation of QP capabilities (max scatter/gather entries,
max inline data, etc) into the kernel, and have the uverbs module
return the values as part of the create QP response. This keeps
precise knowledge of device limits in the low-level kernel driver.
This requires an ABI bump, so while we're making changes, get rid of
the max_sge parameter for the modify SRQ command -- it's not used and
shouldn't be there.
Signed-off-by: Jack Morgenstein <jackm@mellanox.co.il>
Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Userspace CQs that have no completion event channel attached end up
with their cq_context set to NULL. However, asynchronous events like
"CQ overrun" can still occur on such CQs, so add a uverbs_file member
to struct ib_ucq_object that we can follow to deliver these events.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Move ib_uverbs module to using cdev_alloc() and class_device_create()
so that we can handle device lifetime properly. Now we can make sure
we keep all of our data structures around until the last way to reach
them is gone.
Signed-off-by: Roland Dreier <rolandd@cisco.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: Roland 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: Roland 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: Roland 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: Roland 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: Roland Dreier <rolandd@cisco.com>
completion events after destroying a CQ, QP or SRQ. We do this by
sweeping the event lists before returning from a destroy calls, and
then return the number of events already reported before the destroy
call. This allows userspace wait until it has processed all events
for an object returned from the kernel before it frees its context for
the object.
The ABI of the destroy CQ, destroy QP and destroy SRQ commands has to
change to return the event count, so bump the ABI version from 1 to 2.
The userspace libibverbs library has already been updated to handle
both the old and new ABI versions.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Add SRQ support to userspace verbs module. This adds several commands
and associated structures, but it's OK to do this without bumping the
ABI version because the commands are added at the end of the list so
they don't change the existing numbering. There are two cases to
worry about:
1. New kernel, old userspace. This is OK because old userspace simply
won't try to use the new SRQ commands. None of the old commands are
changed.
2. Old kernel, new userspace. This works perfectly as long as
userspace doesn't try to use SRQ commands. If userspace tries to
use SRQ commands, it will get EINVAL, which is perfectly
reasonable: the kernel doesn't support SRQs, so we couldn't do any
better.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Add the core of the InfiniBand userspace verbs implementation, including
creating character device nodes, dispatching requests from userspace, and
passing event notifications back up to userspace.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Steve Wise