- 29 Sep, 2017 30 commits
-
-
Lijun Ou authored
We replace usleep_range that was excessively long anyway with udelay to avoid using usleep_range function in spin_lock_bh spin region, thereby avoiding this calltrace: BUG: scheduling while atomic: insmod/1428/0x00000002 Modules linked in: hns-roce-hw-v2(+) hns_roce rdma_ucm rdma_cm iw_cm ib_uverbs ib_cm ib_core CPU: 0 PID: 1428 Comm: insmod Not tainted 4.12.0-rc1-00677-g252e8fd-dirty #43 Hardware name: (null) (DT) Call trace: [<ffff000008089d20>] dump_backtrace+0x0/0x274 [<ffff00000808a068>] show_stack+0x20/0x28 [<ffff00000844ea58>] dump_stack+0x94/0xb4 [<ffff0000080f975c>] __schedule_bug+0x68/0x84 [<ffff000008a988d4>] __schedule+0x5fc/0x70c [<ffff000008a98a24>] schedule+0x40/0xa4 [<ffff000008a9c6f0>] schedule_hrtimeout_range_clock+0x98/0xfc [<ffff000008a9c788>] schedule_hrtimeout_range+0x34/0x40 [<ffff000008a9c098>] usleep_range+0x6c/0x80 [<ffff000000b9ae68>] hns_roce_cmd_send+0xe4/0x264 [hns-roce-hw-v2] [<ffff000000b9b748>] hns_roce_cmd_query_hw_info+0x40/0x60 [hns-roce-hw-v2] [<ffff000000b9b790>] hns_roce_v2_profile+0x28/0x668 [hns-roce-hw-v2] [<ffff000000b6b1f4>] hns_roce_init+0x6c/0x948 [hns-roce-hw-v2] Signed-off-by:
Lijun Ou <oulijun@huawei.com> Signed-off-by:
Wei Hu (Xavier) <xavier.huwei@huawei.com> Signed-off-by:
Shaobo Xu <xushaobo2@huawei.com> Signed-off-by:
Doug Ledford <dledford@redhat.com>
-
Lijun Ou authored
The driver doesn't actually register an inetaddr notifier function, so there is no need to unregister it on shutdown. Signed-off-by:
-
Wei Hu(Xavier) authored
After the loop in hns_roce_v1_mr_free_work_fn function, it is possible that all qps will have been freed (in which case ne will be 0). If that happens, then later in the function when we dereference hr_qp we will get an exception. Check ne is not 0 to make sure we actually have an hr_qp left to work on. This patch fixes the smatch error as below: drivers/infiniband/hw/hns/hns_roce_hw_v1.c:1009 hns_roce_v1_mr_free_work_fn() error: we previously assumed 'hr_qp' could be null Signed-off-by:
-
Lijun Ou authored
When querying qp, It needs to return RoCE device ah_attr type that may be specific to RoCE devices. Signed-off-by:
-
Lijun Ou authored
Only when the IB_QP_DEST_QPN flag of attr_mask is set is it valid to assign the dest_qp_num into the dest_qp field of qp context. Signed-off-by:
-
Wei Hu(Xavier) authored
When lp_qp_work is NULL, we should return ENOMEM. In order to do so, we had to make some upper layer functions return a value instead of being void type so we can propagate the error up the stack. This patch fixes the smatch error as below: drivers/infiniband/hw/hns/hns_roce_hw_v1.c:918 hns_roce_v1_recreate_lp_qp() error: potential null dereference 'lp_qp_work'. (kzalloc returns null) Signed-off-by:
-
Lijun Ou authored
Put the code for checking the send doorbell status into a separate function and call it from check_qp_db_process_status to improve indenting and readability. It fixes the warning from static checker: drivers/infiniband/hw/hns/hns_roce_hw_v1.c:3562 check_qp_db_process_status() warn: inconsistent indenting. Fixes: 5f110ac4 ("IB/hns: Fix for checkpatch.pl comment style) Signed-off-by:
-
Lijun Ou authored
The value of max_rd_atomic and max_dest_rd_atomic in query_qp are incorrect. It should be assigned by left shifting of the bit in hip06 SoC. Signed-off-by:
-
Steve Wise authored
For messages sent from the host to fw that solicit a reply from fw, the c4iw_wr_wait struct pointer is passed in the host->fw message, and included in the fw->host fw6_msg reply. This allows the sender to wait until the reply is received, and the code processing the ingress reply to wake up the sender. If c4iw_wait_for_reply() times out, however, we need to keep the c4iw_wr_wait object around in case the reply eventually does arrive. Otherwise we have touch-after-free bugs in the wake_up paths. This was hit due to a bad kernel driver that blocked ingress processing of cxgb4 for a long time, causing iw_cxgb4 timeouts, but eventually resuming ingress processing and thus hitting the touch-after-free bug. So I want to fix iw_cxgb4 such that we'll at least keep the wait object around until the reply comes. If it never comes we leak a small amount of memory, but if it does come late, we won't potentially crash the system. So add a kref struct in the c4iw_wr_wait struct, and take a reference before sending a message to FW that will generate a FW6 reply. And remove the reference (and potentially free the wait object) when the reply is processed. The ep code also uses the wr_wait for non FW6 CPL messages and doesn't embed the c4iw_wr_wait object in the message sent to firmware. So for those cases we add c4iw_wake_up_noref(). The mr/mw, cq, and qp object create/destroy paths do need this reference logic. For these paths, c4iw_ref_send_wait() is introduced to take the wr_wait reference, send the msg to fw, and then wait for the reply. So going forward, iw_cxgb4 either uses c4iw_ofld_send(), c4iw_wait_for_reply() and c4iw_wake_up_noref() like is done in the some of the endpoint logic, or c4iw_ref_send_wait() and c4iw_wake_up_deref() (formerly c4iw_wake_up()) when sending messages with the c4iw_wr_wait object pointer embedded in the message and resulting FW6 reply. Signed-off-by:
Steve Wise <swise@opengridcomputing.com> Signed-off-by:
-
Steve Wise authored
Remove the embedded c4iw_wr_wait object in preparation for correctly handling timeouts. Signed-off-by:
-
Steve Wise authored
Remove the local stack allocated c4iw_wr_wait object in preparation for correctly handling timeouts. Also cleaned up some error path unwind logic to make it more readable. Signed-off-by:
-
Steve Wise authored
Remove the local stack allocated c4iw_wr_wait object in preparation for correctly handling timeouts. Also cleaned up some error path unwind logic to make it more readable. Signed-off-by:
-
Steve Wise authored
Remove the local stack allocated c4iw_wr_wait object in preparation for correctly handling timeouts. Also refactored some code to simplify it and make errpath unwinding more readable. Signed-off-by:
-
Colin Ian King authored
Variable j is being assigned to loop_j and then later being assigned to a new value in for loops. The first initialization is therefore redundant and can be removed. Cleans up clang warning: warning: Value stored to 'j' is never read Signed-off-by:
Colin Ian King <colin.king@canonical.com> Signed-off-by:
-
Colin Ian King authored
The functions hns_roce_table_mhop_get, hns_roce_table_mhop_put, hns_roce_cleanup_mhop_hem_table, hns_roce_v1_post_mbox, hns_roce_cmq_setup_basic_desc, hns_roce_cmq_send, hns_roce_cmq_query_hw_info are all local to the source and do not need to be in global scope, so make them static. Cleans up sparse warnings: symbol 'hns_roce_table_mhop_get' was not declared. Should it be static? symbol 'hns_roce_table_mhop_put' was not declared. Should it be static? symbol 'hns_roce_cleanup_mhop_hem_table' was not declared. Should it be static? symbol 'hns_roce_v1_post_mbox' was not declared. Should it be static? symbol 'hns_roce_cmq_setup_basic_desc' was not declared. Should it be static? symbol 'hns_roce_cmq_send' was not declared. Should it be static? symbol 'hns_roce_cmq_query_hw_info' was not declared. Should it be static? Signed-off-by:
-
Dan Carpenter authored
These lines were indented too far by mistake. Signed-off-by:
Dan Carpenter <dan.carpenter@oracle.com> Reviewed-by:
Leon Romanovsky <leonro@mellanox.com> Acked-by:
Shiraz Saleem <shiraz.saleem@intel.com> Signed-off-by:
-
Ajaykumar Hotchandani authored
Vendor error print should be consistent across protocols to avoid any confusion. This patch corrects that. Suggested-by:
Santosh Shilimkar <santosh.shilimkar@oracle.com> Signed-off-by:
Ajaykumar Hotchandani <ajaykumar.hotchandani@oracle.com> Reviewed-by:
Wengang Wang <wen.gang.wang@oracle.com> Reviewed-by:
Yuval Shaia <yuval.shaia@oracle.com> Signed-off-by:
Sagi Grimberg <sagi@grimberg.me> Acked-by:
Roi Dayan <roid@mellanox.com> Signed-off-by:
-
Himanshu Jha authored
Use setup_timer and mod_timer API instead of structure assignments. This is done using Coccinelle and semantic patch used for this as follows: @@ expression x,y,z,a,b; @@ -init_timer (&x); +setup_timer (&x, y, z); +mod_timer (&a, b); -x.function = y; -x.data = z; -x.expires = b; -add_timer(&a); Signed-off-by:
Himanshu Jha <himanshujha199640@gmail.com> Signed-off-by:
-
Kalderon, Michal authored
Initialize the rdma_type (iWARP or RoCE) which is set according to device configuration in qed. Fixes: e6a38c54 ("RDMA/qedr: Add support for registering an iWARP device") Signed-off-by:
Michal Kalderon <Michal.Kalderon@cavium.com> Signed-off-by:
Ariel Elior <Ariel.Elior@cavium.com> Signed-off-by:
-
Doug Ledford authored
Signed-off-by: -
Doug Ledford authored
Signed-off-by: -
Michael J. Ruhl authored
The IOCTL is a bit unwieldy. Refactor reset_ctxt() to be a bit more manageable. Reviewed-by:
Ira Weiny <ira.weiny@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:
-
Michael J. Ruhl authored
The IOCTL is a bit unwieldy. Refactor to a common pattern. Refactor _RECV_CTRL, _POLL_TYPE, _ACK_EVENT and _SET_PKEY IOCTLs to a common pattern. Reviewed-by:
-
Michael J. Ruhl authored
The IOCTL is a bit unwieldy. Refactor to a common pattern. Refactor _TID_INVAL_READ IOCTLs. Reviewed-by:
-
Michael J. Ruhl authored
The IOCTL is a bit unwieldy. Refactor to a common pattern. Refactor the _TID_FREE IOCTL. Reviewed-by:
-
Michael J. Ruhl authored
The IOCTL is a bit unwieldy. Refactor to a common pattern. Refactor the _TID_UPDATE IOCTL. Reviewed-by:
-
Michael J. Ruhl authored
The IOCTL is a bit unwieldy. Refactor to a common pattern. Reviewed-by:
-
Michael J. Ruhl authored
In preparation to refactoring get_base_info(), cleanup some checkpatch issues. Reviewed-by:
-
Michael J. Ruhl authored
The IOCTL is a bit unwieldy. Refactor to a common pattern. Reviewed-by:
-
Michael J. Ruhl authored
The IOCTL is a bit unwieldy. Refactor to a common pattern. Refactor the assign_ctxt() IOCTL. Reviewed-by:
-
- 27 Sep, 2017 10 commits
-
-
Michael J. Ruhl authored
hfi1_rcd_get_by_index assumes that the given index is in the correct range. In most cases this is correct because the index is bounded by a loop. For these cases, adding a range check to the function is redundant. For the use case that is not bounded by the loop range, a _safe wrapper function is needed to validate the index before accessing the rcd array. Add a _safe wrapper to _get_by_index to validate the index range. Update appropriate call sites with the new _safe function. Reviewed-by:
-
Mike Marciniszyn authored
The etype field from the rhf can have more values than just IB and BYPASS. Extend the current tracing to report a symbolic for the etype field for non-bypass packets. Bypass packets will continue to report the l2. As part of this fix the etype and the l2 are added to the tracing struct and are available for trigger and filter operations. Fixes: Commit 863cf89d ("IB/hfi1: Add 16B trace support") Reviewed-by:
Don Hiatt <don.hiatt@intel.com> Signed-off-by:
Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by:
-
Jan Sokolowski authored
Following variables: hfi1_cpulist and hfi1_cpulist_count are unused. Remove them. Reviewed-by:
Harish Chegondi <harish.chegondi@intel.com> Reviewed-by:
Jakub Byczkowski <jakub.byczkowski@intel.com> Signed-off-by:
Jan Sokolowski <jan.sokolowski@intel.com> Signed-off-by:
-
Michael J. Ruhl authored
Calculating the offset to a context is done several times throughout the code. Create a common inlined function for doing this calculation. Reviewed-by:
-
Jan Sokolowski authored
Per-cpu variables int_counter, rcv_limit, and send_schedule print unnecessary error messages on failed allocations. Remove the error messages. Reviewed-by:
-
Ira Weiny authored
devdata->link_default is no longer variable Maintain number of holes by moving dc_shutdown Reviewed-by:
Sebastian Sanchez <sebastian.sanchez@intel.com> Signed-off-by:
-
Ira Weiny authored
There is no reason to set the default descriptor flag on every SDMA engine initialization. Reviewed-by:
-
Michael J. Ruhl authored
The HFI PCI IRQ code uses an obsolete PCI API. Update the code to use the new PCI IRQ API and any necessary changes because of the new API. Reviewed-by:
-
Michael J. Ruhl authored
The QIB PCI IRQ code uses an obsolete PCI API. Updating the code to use the new PCI IRQ API and any necessary changes because of the new API. Reviewed-by:
-
Harish Chegondi authored
Remove the debug trace statement in pin_sdma_pages() that gets executed when there is a memory allocation failure as the trace message doesn't help with debugging the memory allocation failure. Cc: Leon Romanovsky <leon@kernel.org> Reviewed-by:
-
