kernel/drivers/infiniband/ulp/iser/iscsi_iser.h, branch linux-rolling-stable

IB/iser: Remove unused declaration in header file

2024-09-10T13:41:03Z

The definition of iser_finalize_rdma_unaligned_sg() has been removed since commit dd0107a08996 ("IB/iser: set block queue_virt_boundary"). Let's remove the unused declaration in header file. Signed-off-by: Zhang Zekun Link: https://patch.msgid.link/20240909121408.80079-2-zhangzekun11@huawei.com Reviewed-by: Kalesh AP Reviewed-by: Sagi Grimberg Acked-by: Max Gurtovoy Signed-off-by: Leon Romanovsky

IB/iser: Prevent invalidating wrong MR

2024-01-04T20:37:03Z

The iser_reg_resources structure has two pointers to MR but only one mr_valid field. The implementation assumes that we use only *sig_mr when pi_enable is true. Otherwise, we use only *mr. However, it is only sometimes correct. Read commands without protection information occur even when pi_enble is true. For example, the following SCSI commands have a Data-In buffer but never have protection information: READ CAPACITY (16), INQUIRY, MODE SENSE(6), MAINTENANCE IN. So, we use *sig_mr for some SCSI commands and *mr for the other SCSI commands. In most cases, it works fine because the remote invalidation is applied. However, there are two cases when the remote invalidation is not applicable. 1. Small write commands when all data is sent as an immediate. 2. The target does not support the remote invalidation feature. The lazy invalidation is used if the remote invalidation is impossible. Since, at the lazy invalidation, we always invalidate the MR we want to use, the wrong MR may be invalidated. To fix the issue, we need a field per MR that indicates the MR needs invalidation. Since the ib_mr structure already has such a field, let's use ib_mr.need_inval instead of iser_reg_resources.mr_valid. Fixes: b76a439982f8 ("IB/iser: Use IB_WR_REG_MR_INTEGRITY for PI handover") Link: https://lore.kernel.org/r/20231219072311.40989-1-sergeygo@nvidia.com Acked-by: Max Gurtovoy Signed-off-by: Sergey Gorenko Signed-off-by: Jason Gunthorpe

IB/iser: iscsi_iser.h: fix kernel-doc warning and spellos

2023-12-26T08:40:46Z

Drop one kernel-doc comment to prevent a warning: iscsi_iser.h:313: warning: Excess struct member 'mr' description in 'iser_device' and spell 2 words correctly (buffer and deferred). Signed-off-by: Randy Dunlap Cc: Sagi Grimberg Cc: Max Gurtovoy Cc: Jason Gunthorpe Cc: Leon Romanovsky Cc: linux-rdma@vger.kernel.org Link: https://lore.kernel.org/r/20231222234623.25231-1-rdunlap@infradead.org Reviewed-by: Sagi Grimberg Acked-by: Max Gurtovoy Signed-off-by: Leon Romanovsky

IB/iser: Fix typo in comment

2022-05-24T14:24:57Z

Spelling mistake (triple letters) in comment. Detected with the help of Coccinelle. Link: https://lore.kernel.org/r/20220521111145.81697-4-Julia.Lawall@inria.fr Signed-off-by: Julia Lawall Reviewed-by: Sagi Grimberg Acked-by: Max Gurtovoy Signed-off-by: Jason Gunthorpe

IB/iser: Generalize map/unmap dma tasks

2022-03-18T17:37:49Z

Avoid code duplication and add the mapping/unmapping of the protection buffers to the iser_dma_map_task_data/iser_dma_unmap_task_data functions. Link: https://lore.kernel.org/r/20220308145546.8372-4-mgurtovoy@nvidia.com Reviewed-by: Sergey Gorenko Signed-off-by: Max Gurtovoy Acked-by: Sagi Grimberg Signed-off-by: Jason Gunthorpe

IB/iser: Use iser_fr_desc as registration context

2022-03-18T17:37:49Z

After removing the FMR support in iSER, there is only one type of registration context. Replace the void pointer with the explicit structure for registration (struct iser_fr_desc). Link: https://lore.kernel.org/r/20220308145546.8372-3-mgurtovoy@nvidia.com Reviewed-by: Sergey Gorenko Signed-off-by: Max Gurtovoy Acked-by: Sagi Grimberg Signed-off-by: Jason Gunthorpe

IB/iser: Don't suppress send completions

2022-01-05T23:36:20Z

In order to complete a scsi command and guarantee that the HCA will never perform an access violation when retrying a send operation we must complete a scsi request only when both send and receive completions has arrived. This is a preparation commit that remove the send completions suppression. Next step will be taking care of the local invalidation mechanism and adding a reference counter for commands. Currently, we don't do anything upon getting the send completion and just "consume" it. Link: https://lore.kernel.org/r/20211215135721.3662-5-mgurtovoy@nvidia.com Signed-off-by: Max Gurtovoy Reviewed-by: Israel Rukshin Signed-off-by: Jason Gunthorpe

IB/iser: Fix RNR errors

2022-01-05T23:36:20Z

Some users complain about RNR errors on the target, when heavy high-priority tasks run on the initiator. After the investigation, we found out that the receive WRs were exhausted, because the initiator could not post them on time. Receive work reqeusts are posted in chunks to reduce the number of hits to the HCA. The WRs are posted in the receive completion handler when the number of free receive buffers reaches the threshold. But on a high-loaded host, receive CQEs processing can be delayed and all receive WRs will be exhausted. In this case, the target will get an RNR error. To avoid this, we post receive WR, as soon as possible and not in a batch. This increases the number of hits to the HCA, but also the common implementation in most of Linux ULPs (e.g. NVMe-oF/RDMA). As a rule of thumb, performance improvements and heuristics are being added to the RDMA core layer or vendors low level drivers and it's about time to align iSER as well. Link: https://lore.kernel.org/r/20211215135721.3662-3-mgurtovoy@nvidia.com Signed-off-by: Sergey Gorenko Signed-off-by: Max Gurtovoy Reviewed-by: Israel Rukshin Signed-off-by: Jason Gunthorpe

IB/iser: Remove deprecated pi_guard module param

2022-01-05T23:36:20Z

No need for this dead code. This commit doesn't change any functionality since one can still run "modprobe ib_iser pi_guard=". Link: https://lore.kernel.org/r/20211215135721.3662-2-mgurtovoy@nvidia.com Signed-off-by: Max Gurtovoy Signed-off-by: Jason Gunthorpe

RDMA/iser: struct iscsi_iser_task is declared twice

2021-03-30T22:53:08Z

struct iscsi_iser_task has been declared at 201st line. Remove the duplicate. Link: https://lore.kernel.org/r/20210326113347.903976-1-wanjiabing@vivo.com Signed-off-by: Wan Jiabing Signed-off-by: Jason Gunthorpe