The hfi1 user SDMA pinned-page cache will leave a stale cache entry when
the cache-entry's virtual address range is invalidated but that cache
entry is in-use by an outstanding SDMA request.
Subsequent user SDMA requests with buffers in or spanning the virtual
address range of the stale cache entry will result in packets constructed
from the wrong memory, the physical pages pointed to by the stale cache
entry.
To fix this, remove mmu_rb_node cache entries from the mmu_rb_handler
cache independent of the cache entry's refcount. Add 'struct kref
refcount' to struct mmu_rb_node and manage mmu_rb_node lifetime with
kref_get() and kref_put().
mmu_rb_node.refcount makes sdma_mmu_node.refcount redundant. Remove
'atomic_t refcount' from struct sdma_mmu_node and change sdma_mmu_node
code to use mmu_rb_node.refcount.
Move the mmu_rb_handler destructor call after a
wait-for-SDMA-request-completion call so mmu_rb_nodes that need
mmu_rb_handler's workqueue to queue themselves up for destruction from an
interrupt context may do so.
Fixes: f48ad614c1 ("IB/hfi1: Move driver out of staging")
Fixes: 00cbce5cbf ("IB/hfi1: Fix bugs with non-PAGE_SIZE-end multi-iovec user SDMA requests")
Link: https://lore.kernel.org/r/168451393605.3700681.13493776139032178861.stgit@awfm-02.cornelisnetworks.com
Reviewed-by: Dean Luick <dean.luick@cornelisnetworks.com>
Signed-off-by: Brendan Cunningham <bcunningham@cornelisnetworks.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@cornelisnetworks.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
hfi1 user SDMA request processing has two bugs that can cause data
corruption for user SDMA requests that have multiple payload iovecs
where an iovec other than the tail iovec does not run up to the page
boundary for the buffer pointed to by that iovec.a
Here are the specific bugs:
1. user_sdma_txadd() does not use struct user_sdma_iovec->iov.iov_len.
Rather, user_sdma_txadd() will add up to PAGE_SIZE bytes from iovec
to the packet, even if some of those bytes are past
iovec->iov.iov_len and are thus not intended to be in the packet.
2. user_sdma_txadd() and user_sdma_send_pkts() fail to advance to the
next iovec in user_sdma_request->iovs when the current iovec
is not PAGE_SIZE and does not contain enough data to complete the
packet. The transmitted packet will contain the wrong data from the
iovec pages.
This has not been an issue with SDMA packets from hfi1 Verbs or PSM2
because they only produce iovecs that end short of PAGE_SIZE as the tail
iovec of an SDMA request.
Fixing these bugs exposes other bugs with the SDMA pin cache
(struct mmu_rb_handler) that get in way of supporting user SDMA requests
with multiple payload iovecs whose buffers do not end at PAGE_SIZE. So
this commit fixes those issues as well.
Here are the mmu_rb_handler bugs that non-PAGE_SIZE-end multi-iovec
payload user SDMA requests can hit:
1. Overlapping memory ranges in mmu_rb_handler will result in duplicate
pinnings.
2. When extending an existing mmu_rb_handler entry (struct mmu_rb_node),
the mmu_rb code (1) removes the existing entry under a lock, (2)
releases that lock, pins the new pages, (3) then reacquires the lock
to insert the extended mmu_rb_node.
If someone else comes in and inserts an overlapping entry between (2)
and (3), insert in (3) will fail.
The failure path code in this case unpins _all_ pages in either the
original mmu_rb_node or the new mmu_rb_node that was inserted between
(2) and (3).
3. In hfi1_mmu_rb_remove_unless_exact(), mmu_rb_node->refcount is
incremented outside of mmu_rb_handler->lock. As a result, mmu_rb_node
could be evicted by another thread that gets mmu_rb_handler->lock and
checks mmu_rb_node->refcount before mmu_rb_node->refcount is
incremented.
4. Related to #2 above, SDMA request submission failure path does not
check mmu_rb_node->refcount before freeing mmu_rb_node object.
If there are other SDMA requests in progress whose iovecs have
pointers to the now-freed mmu_rb_node(s), those pointers to the
now-freed mmu_rb nodes will be dereferenced when those SDMA requests
complete.
Fixes: 7be85676f1 ("IB/hfi1: Don't remove RB entry when not needed.")
Fixes: 7724105686 ("IB/hfi1: add driver files")
Signed-off-by: Brendan Cunningham <bcunningham@cornelisnetworks.com>
Signed-off-by: Patrick Kelsey <pat.kelsey@cornelisnetworks.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@cornelisnetworks.com>
Link: https://lore.kernel.org/r/168088636445.3027109.10054635277810177889.stgit@252.162.96.66.static.eigbox.net
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Fix three sources of error involving struct sdma_txreq.num_descs.
When _extend_sdma_tx_descs() extends the descriptor array, it uses the
value of tx->num_descs to determine how many existing entries from the
tx's original, internal descriptor array to copy to the newly allocated
one. As this value was incremented before the call, the copy loop will
access one entry past the internal descriptor array, copying its contents
into the corresponding slot in the new array.
If the call to _extend_sdma_tx_descs() fails, _pad_smda_tx_descs() then
invokes __sdma_tx_clean() which uses the value of tx->num_desc to drive a
loop that unmaps all descriptor entries in use. As this value was
incremented before the call, the unmap loop will invoke sdma_unmap_desc()
on a descriptor entry whose contents consist of whatever random data was
copied into it during (1), leading to cascading further calls into the
kernel and driver using arbitrary data.
_sdma_close_tx() was using tx->num_descs instead of tx->num_descs - 1.
Fix all of the above by:
- Only increment .num_descs after .descp is extended.
- Use .num_descs - 1 instead of .num_descs for last .descp entry.
Fixes: f4d26d81ad ("staging/rdma/hfi1: Add coalescing support for SDMA TX descriptors")
Link: https://lore.kernel.org/r/167656658879.2223096.10026561343022570690.stgit@awfm-02.cornelisnetworks.com
Signed-off-by: Brendan Cunningham <bcunningham@cornelisnetworks.com>
Signed-off-by: Patrick Kelsey <pat.kelsey@cornelisnetworks.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@cornelisnetworks.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
If there is a failure during probe of hfi1 before the sdma_map_lock is
initialized, the call to hfi1_free_devdata() will attempt to use a lock
that has not been initialized. If the locking correctness validator is on
then an INFO message and stack trace resembling the following may be seen:
INFO: trying to register non-static key.
The code is fine but needs lockdep annotation, or maybe
you didn't initialize this object before use?
turning off the locking correctness validator.
Call Trace:
register_lock_class+0x11b/0x880
__lock_acquire+0xf3/0x7930
lock_acquire+0xff/0x2d0
_raw_spin_lock_irq+0x46/0x60
sdma_clean+0x42a/0x660 [hfi1]
hfi1_free_devdata+0x3a7/0x420 [hfi1]
init_one+0x867/0x11a0 [hfi1]
pci_device_probe+0x40e/0x8d0
The use of sdma_map_lock in sdma_clean() is for freeing the sdma_map
memory, and sdma_map is not allocated/initialized until after
sdma_map_lock has been initialized. This code only needs to be run if
sdma_map is not NULL, and so checking for that condition will avoid trying
to use the lock before it is initialized.
Fixes: 473291b3ea ("IB/hfi1: Fix for early release of sdma context")
Fixes: 7724105686 ("IB/hfi1: add driver files")
Link: https://lore.kernel.org/r/20220520183701.48973.72434.stgit@awfm-01.cornelisnetworks.com
Reported-by: Zheyu Ma <zheyuma97@gmail.com>
Signed-off-by: Douglas Miller <doug.miller@cornelisnetworks.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@cornelisnetworks.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
- Various cleanup and small features for rtrs
- kmap_local_page() conversions
- Driver updates and fixes for: efa, rxe, mlx5, hfi1, qed, hns
- Cache the IB subnet prefix
- Rework how CRC is calcuated in rxe
- Clean reference counting in iwpm's netlink
- Pull object allocation and lifecycle for user QPs to the uverbs core
code
- Several small hns features and continued general code cleanups
- Fix the scatterlist confusion of orig_nents/nents introduced in an
earlier patch creating the append operation
-----BEGIN PGP SIGNATURE-----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=ZwOe
-----END PGP SIGNATURE-----
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma
Pull rdma updates from Jason Gunthorpe:
"This is quite a small cycle, no major series stands out. The HNS and
rxe drivers saw the most activity this cycle, with rxe being broken
for a good chunk of time. The significant deleted line count is due to
a SPDX cleanup series.
Summary:
- Various cleanup and small features for rtrs
- kmap_local_page() conversions
- Driver updates and fixes for: efa, rxe, mlx5, hfi1, qed, hns
- Cache the IB subnet prefix
- Rework how CRC is calcuated in rxe
- Clean reference counting in iwpm's netlink
- Pull object allocation and lifecycle for user QPs to the uverbs
core code
- Several small hns features and continued general code cleanups
- Fix the scatterlist confusion of orig_nents/nents introduced in an
earlier patch creating the append operation"
* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma: (90 commits)
RDMA/mlx5: Relax DCS QP creation checks
RDMA/hns: Delete unnecessary blank lines.
RDMA/hns: Encapsulate the qp db as a function
RDMA/hns: Adjust the order in which irq are requested and enabled
RDMA/hns: Remove RST2RST error prints for hw v1
RDMA/hns: Remove dqpn filling when modify qp from Init to Init
RDMA/hns: Fix QP's resp incomplete assignment
RDMA/hns: Fix query destination qpn
RDMA/hfi1: Convert to SPDX identifier
IB/rdmavt: Convert to SPDX identifier
RDMA/hns: Bugfix for incorrect association between dip_idx and dgid
RDMA/hns: Bugfix for the missing assignment for dip_idx
RDMA/hns: Bugfix for data type of dip_idx
RDMA/hns: Fix incorrect lsn field
RDMA/irdma: Remove the repeated declaration
RDMA/core/sa_query: Retry SA queries
RDMA: Use the sg_table directly and remove the opencoded version from umem
lib/scatterlist: Fix wrong update of orig_nents
lib/scatterlist: Provide a dedicated function to support table append
RDMA/hns: Delete unused hns bitmap interface
...
kmalloc_array() is called to allocate memory for tx->descp. If it fails,
the function __sdma_txclean() is called:
__sdma_txclean(dd, tx);
However, in the function __sdma_txclean(), tx-descp is dereferenced if
tx->num_desc is not zero:
sdma_unmap_desc(dd, &tx->descp[0]);
To fix this possible null-pointer dereference, assign the return value of
kmalloc_array() to a local variable descp, and then assign it to tx->descp
if it is not NULL. Otherwise, go to enomem.
Fixes: 7724105686 ("IB/hfi1: add driver files")
Link: https://lore.kernel.org/r/20210806133029.194964-1-islituo@gmail.com
Reported-by: TOTE Robot <oslab@tsinghua.edu.cn>
Signed-off-by: Tuo Li <islituo@gmail.com>
Tested-by: Mike Marciniszyn <mike.marciniszyn@cornelisnetworks.com>
Acked-by: Mike Marciniszyn <mike.marciniszyn@cornelisnetworks.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
This patch fixes bunch of kernel-doc compilation warnings like below:
drivers/infiniband/hw/i40iw/i40iw_cm.c:4372: warning: expecting prototype for i40iw_ifdown_notify(). Prototype was for i40iw_if_notify() instead
Link: https://lore.kernel.org/r/20210314133908.291945-2-leon@kernel.org
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Fixes the following W=1 kernel build warning(s):
drivers/infiniband/hw/hfi1/sdma.c:2476: warning: Function parameter or member 'count_out' not described in 'sdma_send_txlist'
drivers/infiniband/hw/hfi1/sdma.c:2476: warning: Excess function parameter 'count' description in 'sdma_send_txlist'
Link: https://lore.kernel.org/r/20210126124732.3320971-15-lee.jones@linaro.org
Cc: Mike Marciniszyn <mike.marciniszyn@cornelisnetworks.com>
Cc: Dennis Dalessandro <dennis.dalessandro@cornelisnetworks.com>
Cc: Doug Ledford <dledford@redhat.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: linux-rdma@vger.kernel.org
Signed-off-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
See Documentation/core-api/printk-formats.rst. h should no longer be used
in the format specifier for printk.
Link: https://lore.kernel.org/r/20201215183509.2072517-1-trix@redhat.com
Signed-off-by: Tom Rix <trix@redhat.com>
Acked-by: Mike Marciniszyn <mike.marciniszyn@cornelisnetworks.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
In preparation for unconditionally passing the struct tasklet_struct
pointer to all tasklet callbacks, switch to using the new tasklet_setup()
and from_tasklet() to pass the tasklet pointer explicitly.
Link: https://lore.kernel.org/r/20200903060637.424458-3-allen.lkml@gmail.com
Signed-off-by: Romain Perier <romain.perier@gmail.com>
Signed-off-by: Allen Pais <allen.lkml@gmail.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
The current codebase makes use of the zero-length array language
extension to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:
struct foo {
int stuff;
struct boo array[];
};
By making use of the mechanism above, we will get a compiler warning
in case the flexible array does not occur last in the structure, which
will help us prevent some kind of undefined behavior bugs from being
inadvertently introduced[3] to the codebase from now on.
Also, notice that, dynamic memory allocations won't be affected by
this change:
"Flexible array members have incomplete type, and so the sizeof operator
may not be applied. As a quirk of the original implementation of
zero-length arrays, sizeof evaluates to zero."[1]
This issue was found with the help of Coccinelle.
[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 7649773293 ("cxgb3/l2t: Fix undefined behaviour")
Link: https://lore.kernel.org/r/20200213010425.GA13068@embeddedor.com
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> # added a few more
Replace all the occurrences of FIELD_SIZEOF() with sizeof_field() except
at places where these are defined. Later patches will remove the unused
definition of FIELD_SIZEOF().
This patch is generated using following script:
EXCLUDE_FILES="include/linux/stddef.h|include/linux/kernel.h"
git grep -l -e "\bFIELD_SIZEOF\b" | while read file;
do
if [[ "$file" =~ $EXCLUDE_FILES ]]; then
continue
fi
sed -i -e 's/\bFIELD_SIZEOF\b/sizeof_field/g' $file;
done
Signed-off-by: Pankaj Bharadiya <pankaj.laxminarayan.bharadiya@intel.com>
Link: https://lore.kernel.org/r/20190924105839.110713-3-pankaj.laxminarayan.bharadiya@intel.com
Co-developed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Kees Cook <keescook@chromium.org>
Acked-by: David Miller <davem@davemloft.net> # for net
rhashtable_lookup_fast() internally calls rcu_read_lock() then,
calls rhashtable_lookup(). So if rcu_read_lock() is already held,
rhashtable_lookup() is enough.
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
There is no reason for a different pad buffer for the two
packet types.
Expand the current buffer allocation to allow for both
packet types.
Fixes: f8195f3b14 ("IB/hfi1: Eliminate allocation while atomic")
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Kaike Wan <kaike.wan@intel.com>
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Link: https://lore.kernel.org/r/20191004204934.26838.13099.stgit@awfm-01.aw.intel.com
Signed-off-by: Doug Ledford <dledford@redhat.com>
In sdma_init if rhashtable_init fails the allocated memory for
tmp_sdma_rht should be released.
Fixes: 5a52a7acf7 ("IB/hfi1: NULL pointer dereference when freeing rhashtable")
Link: https://lore.kernel.org/r/20190925144543.10141-1-navid.emamdoost@gmail.com
Signed-off-by: Navid Emamdoost <navid.emamdoost@gmail.com>
Acked-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Once an SDMA engine is taken down due to a link failure, any waiting QPs
that do not have outstanding descriptors in the ring will stay
on the dmawait list as long as the port is down.
Since there is no timer running, they will stay there for a long time.
The fix is to wake up all iowaits linked to dmawait. The send engine
will build and post packets that get flushed back.
Fixes: 7724105686 ("IB/hfi1: add driver files")
Reviewed-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Heavy contention of the sde flushlist_lock can cause hard lockups at
extreme scale when the flushing logic is under stress.
Mitigate by replacing the item at a time copy to the local list with
an O(1) list_splice_init() and using the high priority work queue to
do the flushes.
Fixes: 7724105686 ("IB/hfi1: add driver files")
Cc: <stable@vger.kernel.org>
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
In commit:
4b53a3412d ("sched/core: Remove the tsk_nr_cpus_allowed() wrapper")
the tsk_nr_cpus_allowed() wrapper was removed. There was not
much difference in !RT but in RT we used this to implement
migrate_disable(). Within a migrate_disable() section the CPU mask is
restricted to single CPU while the "normal" CPU mask remains untouched.
As an alternative implementation Ingo suggested to use:
struct task_struct {
const cpumask_t *cpus_ptr;
cpumask_t cpus_mask;
};
with
t->cpus_ptr = &t->cpus_mask;
In -RT we then can switch the cpus_ptr to:
t->cpus_ptr = &cpumask_of(task_cpu(p));
in a migration disabled region. The rules are simple:
- Code that 'uses' ->cpus_allowed would use the pointer.
- Code that 'modifies' ->cpus_allowed would use the direct mask.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20190423142636.14347-1-bigeasy@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
ACK packets are generally associated with request completion and resource
release and therefore should be sent first. This patch optimizes the
send engine by using the following policies:
(1) QPs with RVT_S_ACK_PENDING bit set in qp->s_flags or qpriv->s_flags
should have their priority incremented;
(2) QPs with ACK or TID-ACK packet queued should have their priority
incremented;
(3) When a QP is queued to the wait list due to resource constraints, it
will be queued to the head if it has ACK packet to send;
(4) When selecting qps to run from the wait list, the one with the highest
priority and starve_cnt will be selected; each priority will be equivalent
to a fixed number of starve_cnt (16).
Reviewed-by: Mitko Haralanov <mitko.haralanov@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
We already need to zero out memory for dma_alloc_coherent(), as such
using dma_zalloc_coherent() is superflous. Phase it out.
This change was generated with the following Coccinelle SmPL patch:
@ replace_dma_zalloc_coherent @
expression dev, size, data, handle, flags;
@@
-dma_zalloc_coherent(dev, size, handle, flags)
+dma_alloc_coherent(dev, size, handle, flags)
Suggested-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Luis Chamberlain <mcgrof@kernel.org>
[hch: re-ran the script on the latest tree]
Signed-off-by: Christoph Hellwig <hch@lst.de>
Commit 4e045572e2 ("IB/hfi1: Add unique txwait_lock for txreq events")
laid the ground work to support per resource waiting locking.
This patch adds that with a lock unique to each sdma engine and pio
sendcontext and makes necessary changes for verbs, PSM, and vnic to use
the new locks.
This is particularly beneficial for smaller messages that will exhaust
resources at a faster rate.
Fixes: 7724105686 ("IB/hfi1: add driver files")
Reviewed-by: Gary Leshner <Gary.S.Leshner@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Current implementation allows each qp to have only one send engine. As
such, each qp has only one list to queue prebuilt packets when send engine
resources are not available. To improve performance, it is desired to
support multiple send engines for each qp.
This patch creates the framework to support two send engines
(two legs) for each qp for the TID RDMA protocol, which can be easily
extended to support more send engines. It achieves the goal by creating a
leg specific struct, iowait_work in the iowait struct, to hold the
work_struct and the tx_list as well as a pointer to the parent iowait
struct.
The hfi1_pkt_state now has an additional field to record the current legs
work structure and that is now passed to all egress waiters to determine
the leg that needs to wait via a new iowait helper. The APIs are adjusted
to use the new leg specific struct as required.
Many new and modified helpers are added to support this change.
Reviewed-by: Mitko Haralanov <mitko.haralanov@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Hardware limits the maximum number of packets to u16 packets.
Match that size for all relevant sequence numbers in the user_sdma
engine.
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Remove the sizeable cache of the chip sizing CSRs and replace with CSR
reads as needed.
Reviewed-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Improve the safety of the code and ensure the array cannot be indexed
out of bounds when picking the CPU for a given SDMA engine.
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
- Clean up some function signatures in rxe for clarity
- Tidy the RDMA netlink header to remove unimplemented constants
- bnxt_re driver fixes, one is a regression this window.
- Minor hns driver fixes
- Various fixes from Dan Carpenter and his tool
- Fix IRQ cleanup race in HFI1
- HF1 performance optimizations and a fix to report counters in the right units
- Fix for an IPoIB startup sequence race with the external manager
- Oops fix for the new kabi path
- Endian cleanups for hns
- Fix for mlx5 related to the new automatic affinity support
-----BEGIN PGP SIGNATURE-----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=xwZz
-----END PGP SIGNATURE-----
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma
Pull more rdma updates from Doug Ledford:
"Items of note:
- two patches fix a regression in the 4.15 kernel. The 4.14 kernel
worked fine with NVMe over Fabrics and mlx5 adapters. That broke in
4.15. The fix is here.
- one of the patches (the endian notation patch from Lijun) looks
like a lot of lines of change, but it's mostly mechanical in
nature. It amounts to the biggest chunk of change in it (it's about
2/3rds of the overall pull request).
Summary:
- Clean up some function signatures in rxe for clarity
- Tidy the RDMA netlink header to remove unimplemented constants
- bnxt_re driver fixes, one is a regression this window.
- Minor hns driver fixes
- Various fixes from Dan Carpenter and his tool
- Fix IRQ cleanup race in HFI1
- HF1 performance optimizations and a fix to report counters in the right units
- Fix for an IPoIB startup sequence race with the external manager
- Oops fix for the new kabi path
- Endian cleanups for hns
- Fix for mlx5 related to the new automatic affinity support"
* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma: (38 commits)
net/mlx5: increase async EQ to avoid EQ overrun
mlx5: fix mlx5_get_vector_affinity to start from completion vector 0
RDMA/hns: Fix the endian problem for hns
IB/uverbs: Use the standard kConfig format for experimental
IB: Update references to libibverbs
IB/hfi1: Add 16B rcvhdr trace support
IB/hfi1: Convert kzalloc_node and kcalloc to use kcalloc_node
IB/core: Avoid a potential OOPs for an unused optional parameter
IB/core: Map iWarp AH type to undefined in rdma_ah_find_type
IB/ipoib: Fix for potential no-carrier state
IB/hfi1: Show fault stats in both TX and RX directions
IB/hfi1: Remove blind constants from 16B update
IB/hfi1: Convert PortXmitWait/PortVLXmitWait counters to flit times
IB/hfi1: Do not override given pcie_pset value
IB/hfi1: Optimize process_receive_ib()
IB/hfi1: Remove unnecessary fecn and becn fields
IB/hfi1: Look up ibport using a pointer in receive path
IB/hfi1: Optimize packet type comparison using 9B and bypass code paths
IB/hfi1: Compute BTH only for RDMA_WRITE_LAST/SEND_LAST packet
IB/hfi1: Remove dependence on qp->s_hdrwords
...
Kzalloc_node API doesn't check for overflows in size multiplication.
While kcalloc API check for overflows in size multiplication
but these implementations are not NUMA-aware.
This conversion allowed for correcting an allocation used in the hot
path to be on the local NUMA and ensure us overflow free multiplication
for the size of a memory allocation.
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kamenee Arumugam <kamenee.arumugam@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
The smp_read_barrier_depends() does nothing at all except on DEC Alpha,
and no current DEC Alpha systems use Infiniband:
lkml.kernel.org/r/20171023085921.jwbntptn6ictbnvj@tower
This commit therefore makes Infiniband depend on !ALPHA and removes
the now-ineffective invocations of smp_read_barrier_depends() from
the InfiniBand driver.
Please note that this patch should not be construed as my saying that
InfiniBand's memory ordering is correct, but rather that this patch does
not in any way affect InfiniBand's correctness. In other words, the
result of applying this patch is bug-for-bug compatible with the original.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Doug Ledford <dledford@redhat.com>
Cc: Richard Henderson <rth@twiddle.net>
Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Cc: Matt Turner <mattst88@gmail.com>
Cc: Michael Cree <mcree@orcon.net.nz>
Cc: Andrea Parri <parri.andrea@gmail.com>
Cc: <linux-rdma@vger.kernel.org>
Cc: <linux-alpha@vger.kernel.org>
[ paulmck: Removed drivers/dma/ioat/dma.c per Jason Gunthorpe's feedback. ]
Acked-by: Jason Gunthorpe <jgg@mellanox.com>
- Add iWARP support to qedr driver
- Lots of misc fixes across subsystem
- Multiple update series to hns roce driver
- Multiple update series to hfi1 driver
- Updates to vnic driver
- Add kref to wait struct in cxgb4 driver
- Updates to i40iw driver
- Mellanox shared pull request
- timer_setup changes
- massive cleanup series from Bart Van Assche
- Two series of SRP/SRPT changes from Bart Van Assche
- Core updates from Mellanox
- i40iw updates
- IPoIB updates
- mlx5 updates
- mlx4 updates
- hns updates
- bnxt_re fixes
- PCI write padding support
- Sparse/Smatch/warning cleanups/fixes
- CQ moderation support
- SRQ support in vmw_pvrdma
-----BEGIN PGP SIGNATURE-----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=A1oL
-----END PGP SIGNATURE-----
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dledford/rdma
Pull rdma updates from Doug Ledford:
"This is a fairly plain pull request. Lots of driver updates across the
stack, a huge number of static analysis cleanups including a close to
50 patch series from Bart Van Assche, and a number of new features
inside the stack such as general CQ moderation support.
Nothing really stands out, but there might be a few conflicts as you
take things in. In particular, the cleanups touched some of the same
lines as the new timer_setup changes.
Everything in this pull request has been through 0day and at least two
days of linux-next (since Stephen doesn't necessarily flag new
errors/warnings until day2). A few more items (about 30 patches) from
Intel and Mellanox showed up on the list on Tuesday. I've excluded
those from this pull request, and I'm sure some of them qualify as
fixes suitable to send any time, but I still have to review them
fully. If they contain mostly fixes and little or no new development,
then I will probably send them through by the end of the week just to
get them out of the way.
There was a break in my acceptance of patches which coincides with the
computer problems I had, and then when I got things mostly back under
control I had a backlog of patches to process, which I did mostly last
Friday and Monday. So there is a larger number of patches processed in
that timeframe than I was striving for.
Summary:
- Add iWARP support to qedr driver
- Lots of misc fixes across subsystem
- Multiple update series to hns roce driver
- Multiple update series to hfi1 driver
- Updates to vnic driver
- Add kref to wait struct in cxgb4 driver
- Updates to i40iw driver
- Mellanox shared pull request
- timer_setup changes
- massive cleanup series from Bart Van Assche
- Two series of SRP/SRPT changes from Bart Van Assche
- Core updates from Mellanox
- i40iw updates
- IPoIB updates
- mlx5 updates
- mlx4 updates
- hns updates
- bnxt_re fixes
- PCI write padding support
- Sparse/Smatch/warning cleanups/fixes
- CQ moderation support
- SRQ support in vmw_pvrdma"
* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dledford/rdma: (296 commits)
RDMA/core: Rename kernel modify_cq to better describe its usage
IB/mlx5: Add CQ moderation capability to query_device
IB/mlx4: Add CQ moderation capability to query_device
IB/uverbs: Add CQ moderation capability to query_device
IB/mlx5: Exposing modify CQ callback to uverbs layer
IB/mlx4: Exposing modify CQ callback to uverbs layer
IB/uverbs: Allow CQ moderation with modify CQ
iw_cxgb4: atomically flush the qp
iw_cxgb4: only call the cq comp_handler when the cq is armed
iw_cxgb4: Fix possible circular dependency locking warning
RDMA/bnxt_re: report vlan_id and sl in qp1 recv completion
IB/core: Only maintain real QPs in the security lists
IB/ocrdma_hw: remove unnecessary code in ocrdma_mbx_dealloc_lkey
RDMA/core: Make function rdma_copy_addr return void
RDMA/vmw_pvrdma: Add shared receive queue support
RDMA/core: avoid uninitialized variable warning in create_udata
RDMA/bnxt_re: synchronize poll_cq and req_notify_cq verbs
RDMA/bnxt_re: Flush CQ notification Work Queue before destroying QP
RDMA/bnxt_re: Set QP state in case of response completion errors
RDMA/bnxt_re: Add memory barriers when processing CQ/EQ entries
...
The code that allocates the tx ring in the sdma code fails to take
advantage of kvzalloc variations.
Fix by converting to use kvzalloc_node.
Reported-by: Leon Romanovsky <leon@kernel.org>
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Please do not apply this to mainline directly, instead please re-run the
coccinelle script shown below and apply its output.
For several reasons, it is desirable to use {READ,WRITE}_ONCE() in
preference to ACCESS_ONCE(), and new code is expected to use one of the
former. So far, there's been no reason to change most existing uses of
ACCESS_ONCE(), as these aren't harmful, and changing them results in
churn.
However, for some features, the read/write distinction is critical to
correct operation. To distinguish these cases, separate read/write
accessors must be used. This patch migrates (most) remaining
ACCESS_ONCE() instances to {READ,WRITE}_ONCE(), using the following
coccinelle script:
----
// Convert trivial ACCESS_ONCE() uses to equivalent READ_ONCE() and
// WRITE_ONCE()
// $ make coccicheck COCCI=/home/mark/once.cocci SPFLAGS="--include-headers" MODE=patch
virtual patch
@ depends on patch @
expression E1, E2;
@@
- ACCESS_ONCE(E1) = E2
+ WRITE_ONCE(E1, E2)
@ depends on patch @
expression E;
@@
- ACCESS_ONCE(E)
+ READ_ONCE(E)
----
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: davem@davemloft.net
Cc: linux-arch@vger.kernel.org
Cc: mpe@ellerman.id.au
Cc: shuah@kernel.org
Cc: snitzer@redhat.com
Cc: thor.thayer@linux.intel.com
Cc: tj@kernel.org
Cc: viro@zeniv.linux.org.uk
Cc: will.deacon@arm.com
Link: http://lkml.kernel.org/r/1508792849-3115-19-git-send-email-paulmck@linux.vnet.ibm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Conflicts:
drivers/infiniband/hw/cxgb4/cm.c
drivers/infiniband/hw/qib/qib_driver.c
drivers/infiniband/hw/qib/qib_mad.c
There were minor fixups needed in these files. Just minor context diffs
due to patches from independent sources touching the same basic area.
Signed-off-by: Doug Ledford <dledford@redhat.com>
In preparation for unconditionally passing the struct timer_list pointer to
all timer callbacks, switch to using the new timer_setup() and from_timer()
to pass the timer pointer explicitly. Switches test of .data field to
.function, since .data will be going away.
Cc: Mike Marciniszyn <mike.marciniszyn@intel.com>
Cc: Dennis Dalessandro <dennis.dalessandro@intel.com>
Cc: Doug Ledford <dledford@redhat.com>
Cc: Sean Hefty <sean.hefty@intel.com>
Cc: Hal Rosenstock <hal.rosenstock@gmail.com>
Cc: linux-rdma@vger.kernel.org
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Avoid that gcc 7 reports the following warning when building with W=1:
warning: this statement may fall through [-Wimplicit-fallthrough=]
Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com>
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Cc: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
There is no reason to set the default descriptor flag on every SDMA
engine initialization.
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Currently, link down interrupts queue link entries
on a workqueue intended for sending events only.
Create a workqueue for queuing link events.
Reviewed-by: Dean Luick <dean.luick@intel.com>
Signed-off-by: Sebastian Sanchez <sebastian.sanchez@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
When an egress resource(SDMA descriptors, pio credits) is not available,
a sending thread will be put on the resource's wait queue. When the
resource becomes available again, up to a fixed number of sending threads
can be awakened sequentially and removed from the wait queue, depending
on the number of waiting threads and the number of free resources. Since
each awakened sending thread will send as many packets as possible, it
is highly likely that the first sending thread will consume all the
egress resources. Subsequently, it will be put back to the end of the wait
queue. Depending on the timing when the later sending threads wake up,
they may not be able to send any packet and be again put back to the end
of the wait queue sequentially, right behind the first sending thread.
This starvation cycle continues until some sending threads exceed their
retry limit and consequently fail.
This patch fixes the issue by two simple approaches:
(1) Any starved sending thread will be put to the head of the wait queue
while a served sending thread will be put to the tail;
(2) The most starved sending thread will be served first.
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
sdma_init does not take a number of sdma engine parameters,
rather it initializes all of the sdma engines.
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Improve the safety of the code and ensure the array cannot be indexed
out of bounds when picking the CPU for a given SDMA engine.
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
A NULL pointer dereference occurs when the driver
is unloaded, and the SDMA rhashtable is freed if
the rhashtable_init() function has not been called.
Prevent this by changing sdma_rht to be a pointer
to a dynamically allocated hash table. The NULL-ness
of the pointer serves as an indication that the hash
table was initialized and that it needs to be
destroyed.
Fixes: 0cb2aa690c ("IB/hfi1: Add sysfs interface for affinity setup")
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Sebastian Sanchez <sebastian.sanchez@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
So the original intention of tsk_cpus_allowed() was to 'future-proof'
the field - but it's pretty ineffectual at that, because half of
the code uses ->cpus_allowed directly ...
Also, the wrapper makes the code longer than the original expression!
So just get rid of it. This also shrinks <linux/sched.h> a bit.
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
