aboutsummaryrefslogtreecommitdiff
AgeCommit message (Collapse)AuthorFilesLines
2016-05-12IB/srp: Fix a commentBart Van Assche1-1/+1
The free request list was removed through patch "IB/srp: Use block layer tags". Hence update a comment that refers to that free request list. Signed-off-by: Bart Van Assche <[email protected]> Cc: Christoph Hellwig <[email protected]> Cc: Sagi Grimberg <[email protected]> Cc: Laurence Oberman <[email protected]> Reviewed-by: Sagi Grimberg <[email protected]> Reviewed-by: Christoph Hellwig <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-05-12IB/srp: Fix a spelling error in a source code commentBart Van Assche1-1/+1
Change one occurrence of "boundries" into "boundaries". Signed-off-by: Bart Van Assche <[email protected]> Reviewed-by: Sagi Grimberg <[email protected]> Cc: Christoph Hellwig <[email protected]> Cc: Laurence Oberman <[email protected]> Reviewed-by: Christoph Hellwig <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-05-05Merge branches 'hfi1' and 'iw_cxgb4' into k.o/for-4.7Doug Ledford40-546/+1138
2016-05-05RDMA/iw_cxgb4: remove abort_connection() usage from ep_timeout()Hariprasad S1-10/+2
Use c4iw_ep_disconnect() instead. This is part of getting rid of abort_connection() altogether so we properly clean up on send_abort() failures. This is the last user of abort_connection(), so remove it too. Signed-off-by: Steve Wise <[email protected]> Signed-off-by: Hariprasad Shenai <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-05-05RDMA/iw_cxgb4: move QP -> ERROR on fatal disconnect errorsHariprasad S1-0/+13
In c4iw_ep_disconnect(), if we fail to initiate a close operation, then move the qp to ERROR to disassociate the ep from the qp. Failure to do this will leak the ep resources. Signed-off-by: Steve Wise <[email protected]> Signed-off-by: Hariprasad Shenai <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-05-05RDMA/iw_cxgb4: don't use abort_connection in process_mpa_request()Hariprasad S1-29/+35
Instead return whether the caller needs to disconnect. This is part of getting rid of abort_connection() altogether so we properly clean up on send_abort() failures. Signed-off-by: Steve Wise <[email protected]> Signed-off-by: Hariprasad Shenai <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-05-05RDMA/iw_cxgb4: remove abort_connection() usage from accept/rejectHariprasad S1-14/+15
Use c4iw_ep_disconnect() instead. This is part of getting rid of abort_connection() altogether so we properly clean up on send_abort() failures. Signed-off-by: Steve Wise <[email protected]> Signed-off-by: Hariprasad Shenai <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-05-05RDMA/iw_cxgb4: free resources when send_flowc() failsHariprasad S1-5/+18
Signed-off-by: Steve Wise <[email protected]> Signed-off-by: Hariprasad Shenai <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-05-05RDMA/iw_cxgb4: remove connection abort from process_mpa_replyHariprasad S1-2/+13
Instead, have the caller, rx_data() handle the close/abort like it does for process_mpa_request(). This is part of getting rid of abort_connection() altogether so we properly clean up on send_abort() failures. Signed-off-by: Steve Wise <[email protected]> Signed-off-by: Hariprasad Shenai <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-05-05RDMA/iw_cxgb4: ensure eps don't get freed while the mutex is heldHariprasad S1-0/+7
In rx_data(), with the ep in FPDU_MODE, refcnt=2, if we get unexpected streaming data, we call c4iw_modify_rc_qp() and move the qp from RTS -> TERMINATE. In c4iw_modify_rc_qp(), if rdma_fini() returns an error, the ep will be dereferenced (refcnt=1). Then rx_data() calls c4iw_ep_disconnect() which starts the close operation. But if send_halfclose() fails in c4iw_ep_disconnect(), we will call release_ep_resources() derefing the ep which reduces the refcnt to 0 and and frees the ep. However we still has the ep mutex at that point, so we have a touch-after-free bug. There is a similar issue where peer_close() calls c4iw_ep_disconnect(). The solution is to add a reference to the ep in c4iw_ep_disconnect() after acquiring the mutex, and release it after releasing the mutex. Signed-off-by: Steve Wise <[email protected]> Signed-off-by: Hariprasad Shenai <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-05-05RDMA/iw_cxgb4: stop ep timer on close failureHariprasad S1-1/+6
In c4iw_ep_disconnect(), if we start the ep timer to begin a close, but send_halfclose() fails, we need to stop the timer and send a CLOSE event up to the IWCM before releasing the resources. Otherwise, we can crash when the ep timer fires if the ep is referencing a previous instance of the device. This can happen as part of adapter reset/recovery, for instance. Signed-off-by: Steve Wise <[email protected]> Signed-off-by: Hariprasad Shenai <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-05-05RDMA/iw_cxgb4: release ep resources on accept arp failureHariprasad S1-16/+66
If ARP fails before the CPL_PASS_ACCEPT_RPL is seen by hardware, the tid will be stuck in SYN_PEND and never released. So create an arp failure handler specifically for this message to release the endpoint resources. In pass_accept_rpl_arp_failure(), put the parent endpoint so it will be freed when destroyed. Also we don't need to call release_tid() here because _c4iw_free_ep() calls cxgb4_remove_tid() which releases the hwtid. If we get an ABORT_REQ_RSS instead of a PASS_ESTABLISH (because the peer's ACK to our SYN is never received), then put the parent as well in peer_abort(). Treat accept_cr() failures just like arp failures: put the parent ep and release the ep resources destroying the tid The ARP failure handlers are called in an atomic context, so we need to schedule some of the processing which might block. Namely _c4iw_free_ep() which needs a mutex. So create a "special" CPL opcode and handler and schedule it via sched() to be run by process_work() in a blockable context. Also rework the active open arp failure handler to make use of release_ep_resources(). This allows both the active and passive arp failure handlers to use the same deferred cleanup function. Signed-off-by: Steve Wise <[email protected]> Signed-off-by: Hariprasad Shenai <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-05-05IB/iser: Fix max_sectors calculationChristoph Hellwig1-4/+10
iSER currently has a couple places that set max_sectors in either the host template or SCSI host, and all of them get it wrong. This patch instead uses a single assignment that (hopefully) gets it right: the max_sectors value must be derived from the number of segments in the FR or FMR structure, but actually be one lower than the page size multiplied by the number of sectors, as it has to handle the case of non-aligned I/O. Without this I get trivial to reproduce hangs when running xfstests (on XFS) over iSER to Linux targets. Signed-off-by: Christoph Hellwig <[email protected]> Reviewed-by: Max Gurtovoy <[email protected]> Acked-by: Sagi Grimberg <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/nes: don't leak skb if carrier downFlorian Westphal1-3/+0
Alternatively one could free the skb, OTOH I don't think this test is useful so just remove it. Cc: <[email protected]> Signed-off-by: Florian Westphal <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Fix for removing quad hash entriesTatyana Nikolova1-1/+3
Fix for removing a quad hash entry when the corresponding quad hash entry hasn't been added, which is the case in loopback connections Signed-off-by: Tatyana Nikolova <[email protected]> Signed-off-by: Faisal Latif <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Fix for checking if the QP is destroyedTatyana Nikolova3-0/+8
Fix for checking if the QP associated with a completion has been destroyed while processing CQ elements. If that is the case, move the CQ head to the next element and continue completion processing. Signed-off-by: Tatyana Nikolova <[email protected]> Signed-off-by: Faisal Latif <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Fix for using one sge for RDMA READShiraz Saleem1-1/+5
A check is added to validate the requested sge number. iWARP doesn't support multiple sg elements for RDMA READ work requests. Signed-off-by: Tatyana Nikolova <[email protected]> Signed-off-by: Faisal Latif <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Fix for the size of kernel mode SQShiraz Saleem1-2/+0
Fix to calculate the SQ size based on the max frag_count, requested by the application instead of overwriting it with the max supported frag_count Signed-off-by: Tatyana Nikolova <[email protected]> Signed-off-by: Faisal Latif <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Fix for a NOP WQE sizeMohammad Khan1-0/+3
Fix for filling in the WQE size for NOP Signed-off-by: Tatyana Nikolova <[email protected]> Signed-off-by: Faisal Latif <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Correct STag mask to min of 14 bitsChien Tin Tung1-1/+2
STag index mask is calculated incorrectly, missing the 14 bits minimum requirement. Add max macro to use either # of MRs or 14 bits in the mask size calculation. Signed-off-by: Tatyana Nikolova <[email protected]> Signed-off-by: Faisal Latif <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Fixes for WQE alignmentShiraz Saleem2-2/+18
Invalidation after every WQE write is changed to invalidate only if required. NOPs are padded so that WQE writes are aligned to 64B boundary. Signed-off-by: Tatyana Nikolova <[email protected]> Signed-off-by: Faisal Latif <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Adding queue drain functionsIsmail, Mustafa2-0/+40
Adding sq and rq drain functions, which block until all previously posted wr-s in the specified queue have completed. A completion object is signaled to unblock the thread, when the last cqe for the corresponding queue is processed. Signed-off-by: Mustafa Ismail <[email protected]> Signed-off-by: Faisal Latif <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Fix SD calculation for initial HMC creationIsmail, Mustafa2-33/+62
Correct SD calculation by using base address returned from commit FPM. This alleviates any assumptions on resource ordering and alignment requirement. Also consolidate SD estimation code into i40iw_est_sd(). Signed-off-by: Mustafa Ismail <[email protected]> Signed-off-by: Faisal Latif <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Fix endian issues and warningsIsmail, Mustafa8-56/+54
Fix endian warnings and errors due to u32 stored to u16. Reported-by: Dan Carpenter <[email protected]> Signed-off-by: Mustafa Ismail <[email protected]> Signed-off-by: Faisal Latif <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Add base memory management extensionsIsmail, Mustafa4-19/+271
Implement fast register mr, Local invalidate, send with invalidate and RDMA read with invalidate. Signed-off-by: Mustafa Ismail <[email protected]> Signed-off-by: Faisal Latif <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Initialize max enabled vfs variableIsmail, Mustafa1-0/+1
Initialize max enabled vfs to max rdma vfs instead of 0. Signed-off-by: Mustafa Ismail <[email protected]> Signed-off-by: Faisal Latif <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Correct return code check in add_pble_poolIsmail, Mustafa1-4/+5
Move return code check to immediately after i40iw_hmc_sd_one call where it is set instead of outside the then statement. Signed-off-by: Mustafa Ismail <[email protected]> Signed-off-by: Faisal Latif <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Add virtual channel message queueIsmail, Mustafa6-63/+103
Queue users of virtual channel on a waitqueue until the channel is clear instead of failing the call when the channel is occupied. Signed-off-by: Mustafa Ismail <[email protected]> Signed-off-by: Faisal Latif <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Remove unused code and fix warningIsmail, Mustafa4-16/+3
Remove unused code and fix warning. Signed-off-by: Mustafa Ismail <[email protected]> Signed-off-by: Faisal Latif <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Populate vendor_id and vendor_part_id fieldsIsmail, Mustafa1-2/+2
Populate PCI info fields from PCI device structure. Signed-off-by: Mustafa Ismail <[email protected]> Signed-off-by: Faisal Latif <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Set vendor_err only if there is an actual errorIsmail, Mustafa1-5/+5
Add a check for cq_poll_info.error before setting vendor_err instead of always setting it. Signed-off-by: Mustafa Ismail <[email protected]> Signed-off-by: Faisal Latif <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Add qp table lock around AE processingIsmail, Mustafa3-2/+15
QP may be freed during Async Event processing. Add a lock around QP table to prevent it. Signed-off-by: Mustafa Ismail <[email protected]> Signed-off-by: Faisal Latif <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Do not set self-referencing pointer to NULL after freeIsmail, Mustafa1-1/+0
iwqp->allocated_buffer is a self-referencing pointer to iwqp. Do not set iwqp->allocated_buffer to NULL after freeing it. Signed-off-by: Mustafa Ismail <[email protected]> Signed-off-by: Faisal Latif <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Correct max message size in query portIsmail, Mustafa1-1/+1
Fix to correct max reported message size in query port. Signed-off-by: Mustafa Ismail <[email protected]> Signed-off-by: Faisal Latif <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Fix refused connectionsIsmail, Mustafa1-41/+44
Make sure cm_node is setup before sending SYN packet and ORD/IRD negotiation. Signed-off-by: Mustafa Ismail <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Correct QP size calculationIsmail, Mustafa4-66/+58
Include inline data size as part of SQ size calculation. RQ size calculation uses only number of SGEs and does not support 96 byte WQE size. Signed-off-by: Mustafa Ismail <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28RDMA/i40iw: Fix overflow of region lengthIsmail, Mustafa2-3/+7
Change region_length to u64 as a region can be > 4GB. Signed-off-by: Mustafa Ismail <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28IB/hfi1: Serialize hrtimer function callsJubin John2-15/+12
hrtimer functions do not guarantee serialization, so we extend the cca_timer_lock to cover the hrtimer_forward_now() in the hrtimer callback handler and the hrtimer_start() in process_becn(). This prevents races between these 2 functions to update the hrtimer state leading to problems such as: kernel BUG at kernel/hrtimer.c:1282! encountered during validation of the CCA feature. Reviewed-by: Mike Marciniszyn <[email protected]> Signed-off-by: Jubin John <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28IB/hfi1: Fix MAD port poll for active cablesDean Luick1-1/+15
A MAD directive to start polling must go through the normal link tuning and start steps in order to correctly handle active cables. Reviewed-by: Easwar Hariharan <[email protected]> Signed-off-by: Dean Luick <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28IB/hfi1: Correctly report neighbor link down reasonDean Luick1-14/+9
The code to save the link down reason for reporting to the SMA was in a location before the actual reason was read. Move the SMA link down reason assignment to a better location. Reviewed-by: Easwar Hariharan <[email protected]> Signed-off-by: Dean Luick <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28IB/hfi1: Use the neighbor link down reason only when validDean Luick2-9/+130
The 8051 uses a link down reason to inform the driver why the link went down. The neighbor planned link down reason code is only valid when a link down idle message is received by the 8051. Enhance the explanation on why the link went down. Reviewed-by: Easwar Hariharan <[email protected]> Signed-off-by: Dean Luick <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28IB/hfi1: Ignore link downgrade with 0 lanesDean Luick1-1/+5
Versions of the 8051 firmware < 0.38 may report a link failure as a link downgrade with a width of 0 followed by a link down notification. Ignore the zero width downgrade notification - the driver should follow the link down path. Reviewed-by: Easwar Hariharan <[email protected]> Signed-off-by: Dean Luick <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28IB/hfi1: Add RSM rule for user FECN handlingDean Luick1-0/+84
Add a receive side mapping rule to extract expected user packets with the FECN bit set and place them in an eager buffer. This will allow user libraries to recognize that a FECN was sent when using header suppression and respond appropriately. Reviewed-by: Dennis Dalessandro <[email protected]> Signed-off-by: Dean Luick <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28IB/hfi1: Create a routine to set a receive side mapping ruleDean Luick1-18/+57
Move the rule setting code into its own routine for improved searchability and reuse. Reviewed-by: Dennis Dalessandro <[email protected]> Signed-off-by: Dean Luick <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28IB/hfi1: Move QOS decision logic into its own functionDean Luick1-17/+50
The decision to use QOS affects other resource allocation. Move the QOS decision logic into its own function so it can be called by other interested parties. Reviewed-by: Dennis Dalessandro <[email protected]> Signed-off-by: Dean Luick <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28IB/hfi1: Extract RSM map table init from QOSDean Luick2-21/+71
Refactor the allocation, tracking, and writing of the RSM map table into its own set of routines. This will allow the map table to be passed to multiple users to fill in as needed. Start with the original user, QOS. Reviewed-by: Dennis Dalessandro <[email protected]> Signed-off-by: Dean Luick <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28IB/hfi1: Reduce kernel context pio buffer allocationJianxin Xiong4-25/+61
The pio buffers were pooled evenly among all kernel contexts and user contexts. However, the demand from kernel contexts is much lower than user contexts. This patch reduces the allocation for kernel contexts and thus makes more credits available for PSM, helping performance. This is especially useful on high core-count systems where large numbers of contexts are used. A new context type SC_VL15 is added to distinguish the context used for VL15 from other kernel contexts. The reason is that VL15 needs to support 2KB sized packet while other kernel contexts need only support packets up to the size determined by "piothreshold", which has a default value of 256. The new allocation method allows triple buffering of largest pio packets configured for these contexts. This is sufficient to maintain verbs performance. The largest pio packet size is 2048B for VL15 and "piothreshold" for other kernel contexts. A cap is applied to "piothreshold" to avoid excessive buffer allocation. The special case that SDMA is disable is handled differently. In that case, the original pooling allocation is used to better support the much higher pio traffic. Notice that if adaptive pio is disabled (piothreshold==0), the pio buffer size doesn't matter for non-VL15 kernel send contexts when SDMA is enabled because pio is not used at all on these contexts and thus the new allocation is still valid. If SDMA is disabled then pooling allocation is used as mentioned in previous paragraph. Adjustment is also made to the calculation of the credit return threshold for the kernel contexts. Instead of purely based on the MTU size, a percentage based threshold is also considered and the smaller one of the two is chosen. This is necessary to ensure that with the reduced buffer allocation credits are returned in time to avoid unnecessary stall in the send path. Reviewed-by: Mike Marciniszyn <[email protected]> Reviewed-by: Dean Luick <[email protected]> Reviewed-by: Dennis Dalessandro <[email protected]> Reviewed-by: Mark Debbage <[email protected]> Reviewed-by: Jubin John <[email protected]> Signed-off-by: Jianxin Xiong <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28IB/hfi1: Change default number of user contextsJubin John3-54/+84
Change the default number of user contexts to the number of real (non-HT) cpu cores in order to reduce the division of hfi1 hardware contexts in the case of high core counts with hyper-threading enabled. Reviewed-by: Dean Luick <[email protected]> Reviewed-by: Dennis Dalessandro <[email protected]> Reviewed-by: Mitko Haralanov <[email protected]> Signed-off-by: Jubin John <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28IB/hfi1: Use global defines for upper bits in opcodeMike Marciniszyn3-7/+4
The awkward coding for setting the allowed_ops field was tripping an smatch warning. This patch uses the more appropriate defines from include/rdma to avoid the issue. As part of the patch remove a mask that was duplicated in rdmavt include files and use that mask as appropriate. Fixes: 8bea6b1cfe6f ("IB/rdmavt: Add create queue pair functionality") Reported-by: Dan Carpenter <[email protected]> Reviewed-by: Dennis Dalessandro <[email protected]> Signed-off-by: Mike Marciniszyn <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
2016-04-28IB/hfi1: Remove unreachable codeMike Marciniszyn1-1/+1
Remove unreachable code from RC ack handling to fix an smatch error. Fixes: 633d27399514 ("staging/rdma/hfi1: use mod_timer when appropriate") Reported-by: Dan Carpenter <[email protected]> Reviewed-by: Dennis Dalessandro <[email protected]> Signed-off-by: Mike Marciniszyn <[email protected]> Signed-off-by: Doug Ledford <[email protected]>
Powered by cgit, Gruvboxed by Blaster4385.