- 13 Mar, 2018 9 commits
-
-
James Smart authored
The driver fails to allocate command buffers in the routine lpfc_new_scsi_buf_s4 There is an inconsistency between lpfc_mem_alloc(), where the phba->lpfc_sg_dma_buf_pool is created, and lpfc_new_scsi_buf_s4(), when we allocate a buffer from the pool and check the alignment. The alignment should be on a page boundary, based on LPFC_SLI3_BG_ENABLED in sli3_options, for both cases. Fix by explicitly tracking sli4 vs sli3 and BG options. The result is that phba->cfg_sg_dma_buf_size is now set correctly for SLI-4. Signed-off-by:
Dick Kennedy <dick.kennedy@broadcom.com> Signed-off-by:
James Smart <james.smart@broadcom.com> Signed-off-by:
Martin K. Petersen <martin.petersen@oracle.com>
-
James Smart authored
POST_SGL_PAGES mailbox command failed with status (timeout). wait_event_interruptible_timeout when called from mailbox wait interface, gets interrupted, and will randomly fail. Behavior seems very specific to 1 particular server type. Fix by changing from wait_event_interruptible_timeout to wait_for_completion_timeout. Signed-off-by:
-
James Smart authored
When a port is configured for NVME and SCSI Initiator support and it probes a target supporting both SCSI and NVME, NVME devices are discovered, but SCSI devices are not. The nlp_fc4_type for all NPorts should be cleared on Link Up or just before GID_FTs get issued, as opposed to just during GID_FT cmpl. RSCN activity as well as Link Up can trigger GID_FT. One GID_FT may complete before the next one is issued. Fix by clearng nlp_fc4_type on link up and just before both GID_FTs are issued. During port swapping, copy nlp_fc4_type to the new ndlp Signed-off-by:
-
James Smart authored
To reduce latency when initializing WQE content, created templates for the most common wqes. This reduces the number of operations taken to set the content. It's not a lot of speed up, but every bit helps. This patch updates the NVME target path. [mkp: fixed typo] Signed-off-by:
-
James Smart authored
To reduce latency when initializing WQE content, create templates for the most common wqes. This reduces the number of operations taken to set the content. It's not a lot of speed up, but every bit helps. This patch updates the NVME initiator path. [mkp: fixed typo] Signed-off-by:
-
James Smart authored
The driver is very sloppy about the WQE structure passed between routines. The base struct type is a 64byte wqe. But in many routines they typecast and access 128byte wqes. There were a couple of cases in the past (corrected already) where the typecasts were incorrectly done and the 64byte buffer was accessed as a 128 byte buffer. Clean this up by properly declaring wqe's as 128byte wqe's and removing the typecasts. 64byte wqes are considered a subset of the 128byte wqes. Signed-off-by:
-
James Smart authored
First Burst support was not properly indicated in NVMe PRLI. Correct the bit position and the logic to check and set first burst support. Signed-off-by:
-
James Smart authored
Commit 6e8e1c14 ("scsi: lpfc: Add WQ Full Logic for NVME Target") fails the static checker. Checker correctly identified a missing unlock on a return path. Add the unlock. Fixes: 6e8e1c14 ("scsi: lpfc: Add WQ Full Logic for NVME Target") Signed-off-by:
-
Souptick Joarder authored
Use dma_pool_zalloc() instead of dma_pool_alloc + memset Signed-off-by:
Souptick Joarder <jrdr.linux@gmail.com> Acked-by:
Brian King <brking@linux.vnet.ibm.com> Signed-off-by:
-
- 07 Mar, 2018 4 commits
-
-
James Smart authored
Commit 1351e69f ("scsi: lpfc: Add push-to-adapter support to sli4") fails compilation on some 32-bit systems as writeq() is not supported on all architectures. Additionally, it was pointed out that as writeX() does byteswapping if necessary for pci vs the cpu endianness, the code was broken on BE PPC. After discussions with Arnd Bergmann, we've resolved the issue to the following: Instead of writeX(), use __raw_writeX() - which writes to io space while preserving byte order. To use this, the code was changed to use a different buffer that lpfc prepped via sli_pcimem_bcopy() that was set to the bytestream to be written. On platforms with __raw_writeq support, use the routine, otherwise use __raw_writel() [mkp: checkpatch] Fixes: 1351e69f ("scsi: lpfc: Add push-to-adapter support to sli4") Signed-off-by:
Arnd Bergmann <arnd@arndb.de> Signed-off-by:
-
Jianchao Wang authored
In scsi core, __scsi_queue_insert should just put request back on the queue and retry using the same command as before. However, for blk-mq, scsi_mq_requeue_cmd is employed here which will unprepare the request. To align with the semantics of __scsi_queue_insert, use blk_mq_requeue_request with kick_requeue_list == true and put the reference of scsi_device. Cc: Christoph Hellwig <hch@lst.de> Signed-off-by:
Jianchao Wang <jianchao.w.wang@oracle.com> Reviewed-by:
Bart Van Assche <bart.vanassche@wdc.com> Signed-off-by:
-
Souptick Joarder authored
Use dma_pool_zalloc() instead of dma_pool_alloc + memset Signed-off-by:
Sumit Saxena <sumit.saxena@broadcom.com> Signed-off-by:
-
Kevin Barnett authored
Reviewed-by:
Scott Benesh <scott.benesh@microsemi.com> Signed-off-by:
Kevin Barnett <kevin.barnett@microsemi.com> Signed-off-by:
Don Brace <don.brace@microsemi.com> Signed-off-by:
-
- 06 Mar, 2018 1 commit
-
-
Arnd Bergmann authored
The newly added code mixes up phys_addr_t/resource_size_t with dma_addr_t and void pointers, as seen from these compiler warning: drivers/scsi/mpt3sas/mpt3sas_base.c: In function '_base_get_chain_phys': drivers/scsi/mpt3sas/mpt3sas_base.c:235:21: error: cast to pointer from integer of different size [-Werror=int-to-pointer-cast] base_chain_phys = (void *)ioc->chip_phys + MPI_FRAME_START_OFFSET + ^ drivers/scsi/mpt3sas/mpt3sas_base.c: In function '_clone_sg_entries': drivers/scsi/mpt3sas/mpt3sas_base.c:427:20: error: cast from pointer to integer of different size [-Werror=pointer-to-int-cast] sgel->Address = (dma_addr_t)dst_addr_phys; ^ drivers/scsi/mpt3sas/mpt3sas_base.c:438:7: error: cast from pointer to integer of different size [-Werror=pointer-to-int-cast] (dma_addr_t)buff_ptr_phys; ^ drivers/scsi/mpt3sas/mpt3sas_base.c:444:10: error: cast from pointer to integer of different size [-Werror=pointer-to-int-cast] (dma_addr_t)buff_ptr_phys; Both dma_addr_t and phys_addr_t may be wider than a pointer, so we must avoid the conversion to pointer types. This also helps readability. A second problem is treating MMIO addresses from a 'struct resource' as addresses that can be used for DMA on that device. In almost all cases, those are the same, but on some of the more obscure architectures, PCI memory address 0 is mapped into the CPU address space at a nonzero offset. I don't have a good fix for that, so I'm adding a comment here, plus a WARN_ON() that triggers whenever the phys_addr_t number is outside of the low 32-bit address space and causes a straight overflow when assigned to the 32-bit sgel->Address. Fixes: 182ac784 ("scsi: mpt3sas: Introduce Base function for cloning.") Signed-off-by:Sreekanth Reddy <Sreekanth.Reddy@broadcom.com> Signed-off-by:
-
- 02 Mar, 2018 5 commits
-
-
Weiping Zhang authored
Add a file for documenting SCSI sd module parameters and describe the cache_type setting. [mkp: tweaked text a bit] Signed-off-by:
Weiping Zhang <zhangweiping@didichuxing.com> Signed-off-by:
-
Dan Carpenter authored
On 64 bit CPUs there is a memory corruption bug on probe(). It should be a u32 pointer instead of an unsigned long pointer or we write past the end of the setupdata[] array. Signed-off-by:
Dan Carpenter <dan.carpenter@oracle.com> Reviewed-by:
Hannes Reinecke <hare@suse.com> Signed-off-by:
-
Wilfried Weissmann authored
This patch fixes the byte order of the SGPIO api and brings it back in sync with ledmon v0.80 and above. [mkp: added missing SoB and fixed whitespace] Signed-off-by:
Wilfried Weissmann <wilfried.weissmann@gmx.at> Signed-off-by:
-
Bart Van Assche authored
Avoid that the kernel-doc tool complains about mismatches between kernel-doc headers and function definitions. Avoid that errors like the following are reported when building the UFS driver with W=1: drivers/scsi/ufs/tc-dwc-g210-pci.c:60: error: Cannot parse struct or union! drivers/scsi/ufs/tc-dwc-g210-pltfrm.c:26: warning: cannot understand function prototype: 'struct ufs_hba_variant_ops tc_dwc_g210_20bit_pltfm_hba_vops = ' Signed-off-by:
-
Bart Van Assche authored
Avoid that the kernel-doc tool complains about a mismatch between the kernel-doc header and the function argument list. Signed-off-by:
-
- 28 Feb, 2018 10 commits
-
-
Suganath Prabu S authored
Sending I/O through 32 bit descriptors to Ventura series of controller results in IO timeout on certain conditions. This error only occurs on systems with high I/O activity. Changes in this patch will prevent driver from using 32 bit descriptor and use 64 bit Descriptors Signed-off-by:
Suganath Prabu S <suganath-prabu.subramani@broadcom.com> Reviewed-by:
Tomas Henzl <thenzl@redhat.com> Reviewed-by:
-
Suganath Prabu Subramani authored
If the posted request has an error of any type, the IOC writes a Reply message into a host-based system reply message frame. This functions clone it in the BAR0 mapped region. Signed-off-by:
-
Suganath Prabu Subramani authored
1) Added function _base_clone_mpi_to_sys_mem to clone MPI request into system BAR0 mapped region. 2) Separate out MPI Endpoint IO submissions to function _base_put_smid_mpi_ep_scsi_io. 3) MPI EP requests are submitted in two 32 bit MMIO writes. from _base_mpi_ep_writeq. For 32 bit Arch,_base_writeq function is identical to _base_mpi_ep_writeq, Removed duplicate code as suggested. Signed-off-by:
-
Suganath Prabu Subramani authored
All scsi IO's and config request's data buffer and sgl are cloned to system memory in _clone_sg_entries before submitting it to firmware. Signed-off-by:
-
Suganath Prabu Subramani authored
For MPI Endpoint/Mcpu, driver should double buffer data buffer/SGLs. This is normally copied from host to internal memory of IOC by DMA engine of PCI device. Since the interface to DMA from host to mCPU is not present for Mcpu/MPI Endpoint device, driver does double copy of those buffers directly to the mCPU memory region via BAR0 region. Introduced API to calculate and return BAR0 mapped host buffer's physical and virtual address for the provided smid. Signed-off-by:
-
Suganath Prabu Subramani authored
This configures shost max sector to 128, single reply descriptor post queue, sgl table size to 16 and 32 bit DMA for MPI Endpoint and it supports 64K as max IO. Signed-off-by:
-
Suganath Prabu Subramani authored
Add device ID and flag for Andromeda/MPI Endpoint. [mkp: typo] Signed-off-by:
-
Jianchao Wang authored
No functional changes. Just fix two wrong indentation cases in scsi_finish_command and scsi_decide_disposition. Signed-off-by:
-
Arnd Bergmann authored
A bugfix I did caused a build regression in some other randconfig builds in a rare combination of options: In file included from drivers/scsi/qedi/qedi_fw.c:16: drivers/scsi/qedi/qedi_gbl.h:26:38: error: array type has incomplete element type 'struct qedi_debugfs_ops' extern const struct qedi_debugfs_ops qedi_debugfs_ops[]; ^~~~~~~~~~~~~~~~ This removes the useless #ifdef around the declarations in qedi_dbg.h to make it always build. Fixes: 779936fa ("scsi: qedi: fix building with LTO") Signed-off-by: -
Bart Van Assche authored
Avoid that building with W=1 causes the kernel-doc tool to complain about function arguments that have not been documented in the libsas kernel-doc headers. Avoid that the short description starts with a hyphen by changing "--" into "-" in the first line of the kernel-doc headers. Signed-off-by:
John Garry <john.garry@huawei.com> Signed-off-by:
-
- 23 Feb, 2018 11 commits
-
-
James Smart authored
Updated Copyright in files updated as part of 12.0.0.0 Signed-off-by:
Johannes Thumshirn <jthumshirn@suse.de> Signed-off-by:
-
James Smart authored
Update the driver version to 12.0.0.0 Signed-off-by:
-
James Smart authored
The hardware offload for NVME commands was created when the FC-NVME standard was setting SGL Descriptor Type to SGL Data Block Descriptor (0h) and SGL Descriptor Sub Type to Address (0h). A late change in NVMe-over-Fabrics obsoleted these values, creating a transport SGL descriptor type with new values to go into these fields. For initial hardware support, in order to be compliant to the spec, use host-supplied cmd IU buffers instead of the adapter generated values. Later hardware will correct this. Add a module parameter to override this offload disablement if looking for lowest latency. This is reasonable as nothing in FC-NVME uses the SQE SGL values. Signed-off-by:
-
James Smart authored
Newer hardware more strictly enforces buffer lenghts, causing an mis-set value to be identified. Older hardware won't catch it. The difference is benign on old hardware. Set the right embedded buffer length for nvme ios. Signed-off-by:
-
James Smart authored
The current driver isn't taking advantage of a performance hint whereby the initial data buffer descriptor can be placed in the WQE as well as the SGL. Add the logic to detect support for the feature and to use it when supported. Signed-off-by:
-
James Smart authored
Current code is very explicit in what it allows to be downloaded. The driver checking prevented G7 firmware download. The driver checking is unnecessary as the device will validate what it receives. Revise the firmware download interface checking. Added a little debug support in case there is still a failure. Signed-off-by:
-
James Smart authored
Traditional SLI4 required the driver to clear Valid bits on EQEs and CQEs after consuming them. The new if_type=6 hardware will cycle the value for what is valid on each queue itteration. The driver no longer has to touch the valid bits. This also means all the cpu cache dirtying and perhaps flush/refill's done by the hardware in accessing the EQ/CQ elements is eliminated. Signed-off-by:
-
James Smart authored
The G7 adapter supports 64G link speeds. Add support to the driver. In addition, a small cleanup to replace the odd bitmap logic with a switch case. Signed-off-by:
-
James Smart authored
Add PCI ids for the new G7 adapter Signed-off-by:
-
James Smart authored
New if_type=6 adapters support an additional BAR that provides apertures to allow direct WQE to adapter push support - termed Direct Packet Push (DPP). WQ creation differs slightly to ask for a WQ to be DPP-ized. When submitting a WQE to a DPP WQ, it is submitted to the host memory for the WQ normally, but is also written by the host cpu directly to a BAR aperture. Write buffer coalescing in hardware is (hopefully) turned on, enabling single pci write operation support. The doorbell is thing rung to indicate the WQE is available and was pushed to the aperture. This patch: - Updates the WQ Create commands for the DPP options - Adds the bar mapping for if_type=6 DPP bar - Adds the WQE pushing to the DDP aperture received from WQ create - Adds a new module parameter to disable DPP operation if desired. Default is enabled. Signed-off-by:
-
James Smart authored
New hardware supports a SLI-4 interface, but with a new if_type variant of 6. If_type=6 has a different PCI BAR map, separate EQ/CQ doorbells, and some changes in doorbell formats. Add the changes for the if_type into headers, adapter initialization and control flows. Add new eq and cq handlers. Signed-off-by:
-
