diff options
Diffstat (limited to 'drivers/infiniband/hw/bnxt_re')
-rw-r--r-- | drivers/infiniband/hw/bnxt_re/bnxt_re.h | 5 | ||||
-rw-r--r-- | drivers/infiniband/hw/bnxt_re/ib_verbs.c | 292 | ||||
-rw-r--r-- | drivers/infiniband/hw/bnxt_re/ib_verbs.h | 19 | ||||
-rw-r--r-- | drivers/infiniband/hw/bnxt_re/main.c | 192 | ||||
-rw-r--r-- | drivers/infiniband/hw/bnxt_re/qplib_fp.c | 104 | ||||
-rw-r--r-- | drivers/infiniband/hw/bnxt_re/qplib_fp.h | 2 | ||||
-rw-r--r-- | drivers/infiniband/hw/bnxt_re/qplib_rcfw.c | 664 | ||||
-rw-r--r-- | drivers/infiniband/hw/bnxt_re/qplib_rcfw.h | 52 | ||||
-rw-r--r-- | drivers/infiniband/hw/bnxt_re/qplib_res.c | 177 | ||||
-rw-r--r-- | drivers/infiniband/hw/bnxt_re/qplib_res.h | 34 | ||||
-rw-r--r-- | drivers/infiniband/hw/bnxt_re/qplib_sp.c | 19 | ||||
-rw-r--r-- | drivers/infiniband/hw/bnxt_re/qplib_sp.h | 5 |
12 files changed, 1189 insertions, 376 deletions
diff --git a/drivers/infiniband/hw/bnxt_re/bnxt_re.h b/drivers/infiniband/hw/bnxt_re/bnxt_re.h index 2c95e6f3d47a..ea81b2497511 100644 --- a/drivers/infiniband/hw/bnxt_re/bnxt_re.h +++ b/drivers/infiniband/hw/bnxt_re/bnxt_re.h @@ -39,6 +39,7 @@ #ifndef __BNXT_RE_H__ #define __BNXT_RE_H__ +#include <rdma/uverbs_ioctl.h> #include "hw_counters.h" #define ROCE_DRV_MODULE_NAME "bnxt_re" @@ -179,10 +180,14 @@ struct bnxt_re_dev { #define BNXT_RE_ROCEV2_IPV4_PACKET 2 #define BNXT_RE_ROCEV2_IPV6_PACKET 3 +#define BNXT_RE_CHECK_RC(x) ((x) && ((x) != -ETIMEDOUT)) + static inline struct device *rdev_to_dev(struct bnxt_re_dev *rdev) { if (rdev) return &rdev->ibdev.dev; return NULL; } + +extern const struct uapi_definition bnxt_re_uapi_defs[]; #endif diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c index 952811c40c54..abef0b8baa7c 100644 --- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c +++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c @@ -61,6 +61,15 @@ #include "bnxt_re.h" #include "ib_verbs.h" + +#include <rdma/uverbs_types.h> +#include <rdma/uverbs_std_types.h> + +#include <rdma/ib_user_ioctl_cmds.h> + +#define UVERBS_MODULE_NAME bnxt_re +#include <rdma/uverbs_named_ioctl.h> + #include <rdma/bnxt_re-abi.h> static int __from_ib_access_flags(int iflags) @@ -534,12 +543,57 @@ fail: return rc; } +static struct bnxt_re_user_mmap_entry* +bnxt_re_mmap_entry_insert(struct bnxt_re_ucontext *uctx, u64 mem_offset, + enum bnxt_re_mmap_flag mmap_flag, u64 *offset) +{ + struct bnxt_re_user_mmap_entry *entry; + int ret; + + entry = kzalloc(sizeof(*entry), GFP_KERNEL); + if (!entry) + return NULL; + + entry->mem_offset = mem_offset; + entry->mmap_flag = mmap_flag; + entry->uctx = uctx; + + switch (mmap_flag) { + case BNXT_RE_MMAP_SH_PAGE: + ret = rdma_user_mmap_entry_insert_exact(&uctx->ib_uctx, + &entry->rdma_entry, PAGE_SIZE, 0); + break; + case BNXT_RE_MMAP_UC_DB: + case BNXT_RE_MMAP_WC_DB: + ret = rdma_user_mmap_entry_insert(&uctx->ib_uctx, + &entry->rdma_entry, PAGE_SIZE); + break; + default: + ret = -EINVAL; + break; + } + + if (ret) { + kfree(entry); + return NULL; + } + if (offset) + *offset = rdma_user_mmap_get_offset(&entry->rdma_entry); + + return entry; +} + /* Protection Domains */ int bnxt_re_dealloc_pd(struct ib_pd *ib_pd, struct ib_udata *udata) { struct bnxt_re_pd *pd = container_of(ib_pd, struct bnxt_re_pd, ib_pd); struct bnxt_re_dev *rdev = pd->rdev; + if (udata) { + rdma_user_mmap_entry_remove(pd->pd_db_mmap); + pd->pd_db_mmap = NULL; + } + bnxt_re_destroy_fence_mr(pd); if (pd->qplib_pd.id) { @@ -558,7 +612,8 @@ int bnxt_re_alloc_pd(struct ib_pd *ibpd, struct ib_udata *udata) struct bnxt_re_ucontext *ucntx = rdma_udata_to_drv_context( udata, struct bnxt_re_ucontext, ib_uctx); struct bnxt_re_pd *pd = container_of(ibpd, struct bnxt_re_pd, ib_pd); - int rc; + struct bnxt_re_user_mmap_entry *entry = NULL; + int rc = 0; pd->rdev = rdev; if (bnxt_qplib_alloc_pd(&rdev->qplib_res.pd_tbl, &pd->qplib_pd)) { @@ -568,15 +623,15 @@ int bnxt_re_alloc_pd(struct ib_pd *ibpd, struct ib_udata *udata) } if (udata) { - struct bnxt_re_pd_resp resp; + struct bnxt_re_pd_resp resp = {}; if (!ucntx->dpi.dbr) { /* Allocate DPI in alloc_pd to avoid failing of * ibv_devinfo and family of application when DPIs * are depleted. */ - if (bnxt_qplib_alloc_dpi(&rdev->qplib_res.dpi_tbl, - &ucntx->dpi, ucntx)) { + if (bnxt_qplib_alloc_dpi(&rdev->qplib_res, + &ucntx->dpi, ucntx, BNXT_QPLIB_DPI_TYPE_UC)) { rc = -ENOMEM; goto dbfail; } @@ -585,12 +640,21 @@ int bnxt_re_alloc_pd(struct ib_pd *ibpd, struct ib_udata *udata) resp.pdid = pd->qplib_pd.id; /* Still allow mapping this DBR to the new user PD. */ resp.dpi = ucntx->dpi.dpi; - resp.dbr = (u64)ucntx->dpi.umdbr; - rc = ib_copy_to_udata(udata, &resp, sizeof(resp)); + entry = bnxt_re_mmap_entry_insert(ucntx, (u64)ucntx->dpi.umdbr, + BNXT_RE_MMAP_UC_DB, &resp.dbr); + + if (!entry) { + rc = -ENOMEM; + goto dbfail; + } + + pd->pd_db_mmap = &entry->rdma_entry; + + rc = ib_copy_to_udata(udata, &resp, min(sizeof(resp), udata->outlen)); if (rc) { - ibdev_err(&rdev->ibdev, - "Failed to copy user response\n"); + rdma_user_mmap_entry_remove(pd->pd_db_mmap); + rc = -EFAULT; goto dbfail; } } @@ -614,12 +678,20 @@ int bnxt_re_destroy_ah(struct ib_ah *ib_ah, u32 flags) { struct bnxt_re_ah *ah = container_of(ib_ah, struct bnxt_re_ah, ib_ah); struct bnxt_re_dev *rdev = ah->rdev; + bool block = true; + int rc = 0; - bnxt_qplib_destroy_ah(&rdev->qplib_res, &ah->qplib_ah, - !(flags & RDMA_DESTROY_AH_SLEEPABLE)); + block = !(flags & RDMA_DESTROY_AH_SLEEPABLE); + rc = bnxt_qplib_destroy_ah(&rdev->qplib_res, &ah->qplib_ah, block); + if (BNXT_RE_CHECK_RC(rc)) { + if (rc == -ETIMEDOUT) + rc = 0; + else + goto fail; + } atomic_dec(&rdev->ah_count); - - return 0; +fail: + return rc; } static u8 bnxt_re_stack_to_dev_nw_type(enum rdma_network_type ntype) @@ -3955,6 +4027,7 @@ int bnxt_re_alloc_ucontext(struct ib_ucontext *ctx, struct ib_udata *udata) container_of(ctx, struct bnxt_re_ucontext, ib_uctx); struct bnxt_re_dev *rdev = to_bnxt_re_dev(ibdev, ibdev); struct bnxt_qplib_dev_attr *dev_attr = &rdev->dev_attr; + struct bnxt_re_user_mmap_entry *entry; struct bnxt_re_uctx_resp resp = {}; u32 chip_met_rev_num = 0; int rc; @@ -3993,6 +4066,16 @@ int bnxt_re_alloc_ucontext(struct ib_ucontext *ctx, struct ib_udata *udata) resp.comp_mask |= BNXT_RE_UCNTX_CMASK_HAVE_MODE; resp.mode = rdev->chip_ctx->modes.wqe_mode; + if (rdev->chip_ctx->modes.db_push) + resp.comp_mask |= BNXT_RE_UCNTX_CMASK_WC_DPI_ENABLED; + + entry = bnxt_re_mmap_entry_insert(uctx, 0, BNXT_RE_MMAP_SH_PAGE, NULL); + if (!entry) { + rc = -ENOMEM; + goto cfail; + } + uctx->shpage_mmap = &entry->rdma_entry; + rc = ib_copy_to_udata(udata, &resp, min(udata->outlen, sizeof(resp))); if (rc) { ibdev_err(ibdev, "Failed to copy user context"); @@ -4016,6 +4099,8 @@ void bnxt_re_dealloc_ucontext(struct ib_ucontext *ib_uctx) struct bnxt_re_dev *rdev = uctx->rdev; + rdma_user_mmap_entry_remove(uctx->shpage_mmap); + uctx->shpage_mmap = NULL; if (uctx->shpg) free_page((unsigned long)uctx->shpg); @@ -4023,8 +4108,7 @@ void bnxt_re_dealloc_ucontext(struct ib_ucontext *ib_uctx) /* Free DPI only if this is the first PD allocated by the * application and mark the context dpi as NULL */ - bnxt_qplib_dealloc_dpi(&rdev->qplib_res, - &rdev->qplib_res.dpi_tbl, &uctx->dpi); + bnxt_qplib_dealloc_dpi(&rdev->qplib_res, &uctx->dpi); uctx->dpi.dbr = NULL; } } @@ -4035,27 +4119,177 @@ int bnxt_re_mmap(struct ib_ucontext *ib_uctx, struct vm_area_struct *vma) struct bnxt_re_ucontext *uctx = container_of(ib_uctx, struct bnxt_re_ucontext, ib_uctx); - struct bnxt_re_dev *rdev = uctx->rdev; + struct bnxt_re_user_mmap_entry *bnxt_entry; + struct rdma_user_mmap_entry *rdma_entry; + int ret = 0; u64 pfn; - if (vma->vm_end - vma->vm_start != PAGE_SIZE) + rdma_entry = rdma_user_mmap_entry_get(&uctx->ib_uctx, vma); + if (!rdma_entry) return -EINVAL; - if (vma->vm_pgoff) { - vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot); - if (io_remap_pfn_range(vma, vma->vm_start, vma->vm_pgoff, - PAGE_SIZE, vma->vm_page_prot)) { - ibdev_err(&rdev->ibdev, "Failed to map DPI"); - return -EAGAIN; - } - } else { - pfn = virt_to_phys(uctx->shpg) >> PAGE_SHIFT; - if (remap_pfn_range(vma, vma->vm_start, - pfn, PAGE_SIZE, vma->vm_page_prot)) { - ibdev_err(&rdev->ibdev, "Failed to map shared page"); - return -EAGAIN; + bnxt_entry = container_of(rdma_entry, struct bnxt_re_user_mmap_entry, + rdma_entry); + + switch (bnxt_entry->mmap_flag) { + case BNXT_RE_MMAP_WC_DB: + pfn = bnxt_entry->mem_offset >> PAGE_SHIFT; + ret = rdma_user_mmap_io(ib_uctx, vma, pfn, PAGE_SIZE, + pgprot_writecombine(vma->vm_page_prot), + rdma_entry); + break; + case BNXT_RE_MMAP_UC_DB: + pfn = bnxt_entry->mem_offset >> PAGE_SHIFT; + ret = rdma_user_mmap_io(ib_uctx, vma, pfn, PAGE_SIZE, + pgprot_noncached(vma->vm_page_prot), + rdma_entry); + break; + case BNXT_RE_MMAP_SH_PAGE: + ret = vm_insert_page(vma, vma->vm_start, virt_to_page(uctx->shpg)); + break; + default: + ret = -EINVAL; + break; + } + + rdma_user_mmap_entry_put(rdma_entry); + return ret; +} + +void bnxt_re_mmap_free(struct rdma_user_mmap_entry *rdma_entry) +{ + struct bnxt_re_user_mmap_entry *bnxt_entry; + + bnxt_entry = container_of(rdma_entry, struct bnxt_re_user_mmap_entry, + rdma_entry); + + kfree(bnxt_entry); +} + +static int UVERBS_HANDLER(BNXT_RE_METHOD_ALLOC_PAGE)(struct uverbs_attr_bundle *attrs) +{ + struct ib_uobject *uobj = uverbs_attr_get_uobject(attrs, BNXT_RE_ALLOC_PAGE_HANDLE); + enum bnxt_re_alloc_page_type alloc_type; + struct bnxt_re_user_mmap_entry *entry; + enum bnxt_re_mmap_flag mmap_flag; + struct bnxt_qplib_chip_ctx *cctx; + struct bnxt_re_ucontext *uctx; + struct bnxt_re_dev *rdev; + u64 mmap_offset; + u32 length; + u32 dpi; + u64 dbr; + int err; + + uctx = container_of(ib_uverbs_get_ucontext(attrs), struct bnxt_re_ucontext, ib_uctx); + if (IS_ERR(uctx)) + return PTR_ERR(uctx); + + err = uverbs_get_const(&alloc_type, attrs, BNXT_RE_ALLOC_PAGE_TYPE); + if (err) + return err; + + rdev = uctx->rdev; + cctx = rdev->chip_ctx; + + switch (alloc_type) { + case BNXT_RE_ALLOC_WC_PAGE: + if (cctx->modes.db_push) { + if (bnxt_qplib_alloc_dpi(&rdev->qplib_res, &uctx->wcdpi, + uctx, BNXT_QPLIB_DPI_TYPE_WC)) + return -ENOMEM; + length = PAGE_SIZE; + dpi = uctx->wcdpi.dpi; + dbr = (u64)uctx->wcdpi.umdbr; + mmap_flag = BNXT_RE_MMAP_WC_DB; + } else { + return -EINVAL; } + + break; + + default: + return -EOPNOTSUPP; } + entry = bnxt_re_mmap_entry_insert(uctx, dbr, mmap_flag, &mmap_offset); + if (!entry) + return -ENOMEM; + + uobj->object = entry; + uverbs_finalize_uobj_create(attrs, BNXT_RE_ALLOC_PAGE_HANDLE); + err = uverbs_copy_to(attrs, BNXT_RE_ALLOC_PAGE_MMAP_OFFSET, + &mmap_offset, sizeof(mmap_offset)); + if (err) + return err; + + err = uverbs_copy_to(attrs, BNXT_RE_ALLOC_PAGE_MMAP_LENGTH, + &length, sizeof(length)); + if (err) + return err; + + err = uverbs_copy_to(attrs, BNXT_RE_ALLOC_PAGE_DPI, + &dpi, sizeof(length)); + if (err) + return err; + return 0; } + +static int alloc_page_obj_cleanup(struct ib_uobject *uobject, + enum rdma_remove_reason why, + struct uverbs_attr_bundle *attrs) +{ + struct bnxt_re_user_mmap_entry *entry = uobject->object; + struct bnxt_re_ucontext *uctx = entry->uctx; + + switch (entry->mmap_flag) { + case BNXT_RE_MMAP_WC_DB: + if (uctx && uctx->wcdpi.dbr) { + struct bnxt_re_dev *rdev = uctx->rdev; + + bnxt_qplib_dealloc_dpi(&rdev->qplib_res, &uctx->wcdpi); + uctx->wcdpi.dbr = NULL; + } + break; + default: + goto exit; + } + rdma_user_mmap_entry_remove(&entry->rdma_entry); +exit: + return 0; +} + +DECLARE_UVERBS_NAMED_METHOD(BNXT_RE_METHOD_ALLOC_PAGE, + UVERBS_ATTR_IDR(BNXT_RE_ALLOC_PAGE_HANDLE, + BNXT_RE_OBJECT_ALLOC_PAGE, + UVERBS_ACCESS_NEW, + UA_MANDATORY), + UVERBS_ATTR_CONST_IN(BNXT_RE_ALLOC_PAGE_TYPE, + enum bnxt_re_alloc_page_type, + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(BNXT_RE_ALLOC_PAGE_MMAP_OFFSET, + UVERBS_ATTR_TYPE(u64), + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(BNXT_RE_ALLOC_PAGE_MMAP_LENGTH, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(BNXT_RE_ALLOC_PAGE_DPI, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY)); + +DECLARE_UVERBS_NAMED_METHOD_DESTROY(BNXT_RE_METHOD_DESTROY_PAGE, + UVERBS_ATTR_IDR(BNXT_RE_DESTROY_PAGE_HANDLE, + BNXT_RE_OBJECT_ALLOC_PAGE, + UVERBS_ACCESS_DESTROY, + UA_MANDATORY)); + +DECLARE_UVERBS_NAMED_OBJECT(BNXT_RE_OBJECT_ALLOC_PAGE, + UVERBS_TYPE_ALLOC_IDR(alloc_page_obj_cleanup), + &UVERBS_METHOD(BNXT_RE_METHOD_ALLOC_PAGE), + &UVERBS_METHOD(BNXT_RE_METHOD_DESTROY_PAGE)); + +const struct uapi_definition bnxt_re_uapi_defs[] = { + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(BNXT_RE_OBJECT_ALLOC_PAGE), + {} +}; diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.h b/drivers/infiniband/hw/bnxt_re/ib_verbs.h index 31f7e34040f7..32d9e9d09791 100644 --- a/drivers/infiniband/hw/bnxt_re/ib_verbs.h +++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.h @@ -60,6 +60,8 @@ struct bnxt_re_pd { struct bnxt_re_dev *rdev; struct bnxt_qplib_pd qplib_pd; struct bnxt_re_fence_data fence; + struct rdma_user_mmap_entry *pd_db_mmap; + struct rdma_user_mmap_entry *pd_wcdb_mmap; }; struct bnxt_re_ah { @@ -134,8 +136,23 @@ struct bnxt_re_ucontext { struct ib_ucontext ib_uctx; struct bnxt_re_dev *rdev; struct bnxt_qplib_dpi dpi; + struct bnxt_qplib_dpi wcdpi; void *shpg; spinlock_t sh_lock; /* protect shpg */ + struct rdma_user_mmap_entry *shpage_mmap; +}; + +enum bnxt_re_mmap_flag { + BNXT_RE_MMAP_SH_PAGE, + BNXT_RE_MMAP_UC_DB, + BNXT_RE_MMAP_WC_DB, +}; + +struct bnxt_re_user_mmap_entry { + struct rdma_user_mmap_entry rdma_entry; + struct bnxt_re_ucontext *uctx; + u64 mem_offset; + u8 mmap_flag; }; static inline u16 bnxt_re_get_swqe_size(int nsge) @@ -213,6 +230,8 @@ struct ib_mr *bnxt_re_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, int bnxt_re_alloc_ucontext(struct ib_ucontext *ctx, struct ib_udata *udata); void bnxt_re_dealloc_ucontext(struct ib_ucontext *context); int bnxt_re_mmap(struct ib_ucontext *context, struct vm_area_struct *vma); +void bnxt_re_mmap_free(struct rdma_user_mmap_entry *rdma_entry); + unsigned long bnxt_re_lock_cqs(struct bnxt_re_qp *qp); void bnxt_re_unlock_cqs(struct bnxt_re_qp *qp, unsigned long flags); diff --git a/drivers/infiniband/hw/bnxt_re/main.c b/drivers/infiniband/hw/bnxt_re/main.c index 3073398a2183..b42166fe7454 100644 --- a/drivers/infiniband/hw/bnxt_re/main.c +++ b/drivers/infiniband/hw/bnxt_re/main.c @@ -83,6 +83,45 @@ static int bnxt_re_netdev_event(struct notifier_block *notifier, unsigned long event, void *ptr); static struct bnxt_re_dev *bnxt_re_from_netdev(struct net_device *netdev); static void bnxt_re_dev_uninit(struct bnxt_re_dev *rdev); +static int bnxt_re_hwrm_qcaps(struct bnxt_re_dev *rdev); + +static int bnxt_re_hwrm_qcfg(struct bnxt_re_dev *rdev, u32 *db_len, + u32 *offset); +static void bnxt_re_set_db_offset(struct bnxt_re_dev *rdev) +{ + struct bnxt_qplib_chip_ctx *cctx; + struct bnxt_en_dev *en_dev; + struct bnxt_qplib_res *res; + u32 l2db_len = 0; + u32 offset = 0; + u32 barlen; + int rc; + + res = &rdev->qplib_res; + en_dev = rdev->en_dev; + cctx = rdev->chip_ctx; + + /* Issue qcfg */ + rc = bnxt_re_hwrm_qcfg(rdev, &l2db_len, &offset); + if (rc) + dev_info(rdev_to_dev(rdev), + "Couldn't get DB bar size, Low latency framework is disabled\n"); + /* set register offsets for both UC and WC */ + res->dpi_tbl.ucreg.offset = res->is_vf ? BNXT_QPLIB_DBR_VF_DB_OFFSET : + BNXT_QPLIB_DBR_PF_DB_OFFSET; + res->dpi_tbl.wcreg.offset = res->dpi_tbl.ucreg.offset; + + /* If WC mapping is disabled by L2 driver then en_dev->l2_db_size + * is equal to the DB-Bar actual size. This indicates that L2 + * is mapping entire bar as UC-. RoCE driver can't enable WC mapping + * in such cases and DB-push will be disabled. + */ + barlen = pci_resource_len(res->pdev, RCFW_DBR_PCI_BAR_REGION); + if (cctx->modes.db_push && l2db_len && en_dev->l2_db_size != barlen) { + res->dpi_tbl.wcreg.offset = en_dev->l2_db_size; + dev_info(rdev_to_dev(rdev), "Low latency framework is enabled\n"); + } +} static void bnxt_re_set_drv_mode(struct bnxt_re_dev *rdev, u8 mode) { @@ -91,6 +130,9 @@ static void bnxt_re_set_drv_mode(struct bnxt_re_dev *rdev, u8 mode) cctx = rdev->chip_ctx; cctx->modes.wqe_mode = bnxt_qplib_is_chip_gen_p5(rdev->chip_ctx) ? mode : BNXT_QPLIB_WQE_MODE_STATIC; + if (bnxt_re_hwrm_qcaps(rdev)) + dev_err(rdev_to_dev(rdev), + "Failed to query hwrm qcaps\n"); } static void bnxt_re_destroy_chip_ctx(struct bnxt_re_dev *rdev) @@ -112,6 +154,7 @@ static int bnxt_re_setup_chip_ctx(struct bnxt_re_dev *rdev, u8 wqe_mode) { struct bnxt_qplib_chip_ctx *chip_ctx; struct bnxt_en_dev *en_dev; + int rc; en_dev = rdev->en_dev; @@ -130,6 +173,12 @@ static int bnxt_re_setup_chip_ctx(struct bnxt_re_dev *rdev, u8 wqe_mode) rdev->qplib_res.is_vf = BNXT_EN_VF(en_dev); bnxt_re_set_drv_mode(rdev, wqe_mode); + + bnxt_re_set_db_offset(rdev); + rc = bnxt_qplib_map_db_bar(&rdev->qplib_res); + if (rc) + return rc; + if (bnxt_qplib_determine_atomics(en_dev->pdev)) ibdev_info(&rdev->ibdev, "platform doesn't support global atomics."); @@ -283,15 +332,21 @@ static void bnxt_re_start_irq(void *handle, struct bnxt_msix_entry *ent) for (indx = 0; indx < rdev->num_msix; indx++) rdev->en_dev->msix_entries[indx].vector = ent[indx].vector; - bnxt_qplib_rcfw_start_irq(rcfw, msix_ent[BNXT_RE_AEQ_IDX].vector, - false); + rc = bnxt_qplib_rcfw_start_irq(rcfw, msix_ent[BNXT_RE_AEQ_IDX].vector, + false); + if (rc) { + ibdev_warn(&rdev->ibdev, "Failed to reinit CREQ\n"); + return; + } for (indx = BNXT_RE_NQ_IDX ; indx < rdev->num_msix; indx++) { nq = &rdev->nq[indx - 1]; rc = bnxt_qplib_nq_start_irq(nq, indx - 1, msix_ent[indx].vector, false); - if (rc) + if (rc) { ibdev_warn(&rdev->ibdev, "Failed to reinit NQ index %d\n", indx - 1); + return; + } } } @@ -315,12 +370,11 @@ static int bnxt_re_register_netdev(struct bnxt_re_dev *rdev) return rc; } -static void bnxt_re_init_hwrm_hdr(struct bnxt_re_dev *rdev, struct input *hdr, - u16 opcd, u16 crid, u16 trid) +static void bnxt_re_init_hwrm_hdr(struct input *hdr, u16 opcd) { hdr->req_type = cpu_to_le16(opcd); - hdr->cmpl_ring = cpu_to_le16(crid); - hdr->target_id = cpu_to_le16(trid); + hdr->cmpl_ring = cpu_to_le16(-1); + hdr->target_id = cpu_to_le16(-1); } static void bnxt_re_fill_fw_msg(struct bnxt_fw_msg *fw_msg, void *msg, @@ -334,13 +388,60 @@ static void bnxt_re_fill_fw_msg(struct bnxt_fw_msg *fw_msg, void *msg, fw_msg->timeout = timeout; } +/* Query device config using common hwrm */ +static int bnxt_re_hwrm_qcfg(struct bnxt_re_dev *rdev, u32 *db_len, + u32 *offset) +{ + struct bnxt_en_dev *en_dev = rdev->en_dev; + struct hwrm_func_qcfg_output resp = {0}; + struct hwrm_func_qcfg_input req = {0}; + struct bnxt_fw_msg fw_msg; + int rc; + + memset(&fw_msg, 0, sizeof(fw_msg)); + bnxt_re_init_hwrm_hdr((void *)&req, HWRM_FUNC_QCFG); + req.fid = cpu_to_le16(0xffff); + bnxt_re_fill_fw_msg(&fw_msg, (void *)&req, sizeof(req), (void *)&resp, + sizeof(resp), DFLT_HWRM_CMD_TIMEOUT); + rc = bnxt_send_msg(en_dev, &fw_msg); + if (!rc) { + *db_len = PAGE_ALIGN(le16_to_cpu(resp.l2_doorbell_bar_size_kb) * 1024); + *offset = PAGE_ALIGN(le16_to_cpu(resp.legacy_l2_db_size_kb) * 1024); + } + return rc; +} + +/* Query function capabilities using common hwrm */ +int bnxt_re_hwrm_qcaps(struct bnxt_re_dev *rdev) +{ + struct bnxt_en_dev *en_dev = rdev->en_dev; + struct hwrm_func_qcaps_output resp = {}; + struct hwrm_func_qcaps_input req = {}; + struct bnxt_qplib_chip_ctx *cctx; + struct bnxt_fw_msg fw_msg = {}; + int rc; + + cctx = rdev->chip_ctx; + bnxt_re_init_hwrm_hdr((void *)&req, HWRM_FUNC_QCAPS); + req.fid = cpu_to_le16(0xffff); + bnxt_re_fill_fw_msg(&fw_msg, (void *)&req, sizeof(req), (void *)&resp, + sizeof(resp), DFLT_HWRM_CMD_TIMEOUT); + + rc = bnxt_send_msg(en_dev, &fw_msg); + if (rc) + return rc; + cctx->modes.db_push = le32_to_cpu(resp.flags) & FUNC_QCAPS_RESP_FLAGS_WCB_PUSH_MODE; + + return 0; +} + static int bnxt_re_net_ring_free(struct bnxt_re_dev *rdev, u16 fw_ring_id, int type) { struct bnxt_en_dev *en_dev; - struct hwrm_ring_free_input req = {0}; + struct hwrm_ring_free_input req = {}; struct hwrm_ring_free_output resp; - struct bnxt_fw_msg fw_msg; + struct bnxt_fw_msg fw_msg = {}; int rc = -EINVAL; if (!rdev) @@ -354,9 +455,7 @@ static int bnxt_re_net_ring_free(struct bnxt_re_dev *rdev, if (test_bit(BNXT_RE_FLAG_ERR_DEVICE_DETACHED, &rdev->flags)) return 0; - memset(&fw_msg, 0, sizeof(fw_msg)); - - bnxt_re_init_hwrm_hdr(rdev, (void *)&req, HWRM_RING_FREE, -1, -1); + bnxt_re_init_hwrm_hdr((void *)&req, HWRM_RING_FREE); req.ring_type = type; req.ring_id = cpu_to_le16(fw_ring_id); bnxt_re_fill_fw_msg(&fw_msg, (void *)&req, sizeof(req), (void *)&resp, @@ -373,16 +472,15 @@ static int bnxt_re_net_ring_alloc(struct bnxt_re_dev *rdev, u16 *fw_ring_id) { struct bnxt_en_dev *en_dev = rdev->en_dev; - struct hwrm_ring_alloc_input req = {0}; + struct hwrm_ring_alloc_input req = {}; struct hwrm_ring_alloc_output resp; - struct bnxt_fw_msg fw_msg; + struct bnxt_fw_msg fw_msg = {}; int rc = -EINVAL; if (!en_dev) return rc; - memset(&fw_msg, 0, sizeof(fw_msg)); - bnxt_re_init_hwrm_hdr(rdev, (void *)&req, HWRM_RING_ALLOC, -1, -1); + bnxt_re_init_hwrm_hdr((void *)&req, HWRM_RING_ALLOC); req.enables = 0; req.page_tbl_addr = cpu_to_le64(ring_attr->dma_arr[0]); if (ring_attr->pages > 1) { @@ -411,7 +509,7 @@ static int bnxt_re_net_stats_ctx_free(struct bnxt_re_dev *rdev, struct bnxt_en_dev *en_dev = rdev->en_dev; struct hwrm_stat_ctx_free_input req = {}; struct hwrm_stat_ctx_free_output resp = {}; - struct bnxt_fw_msg fw_msg; + struct bnxt_fw_msg fw_msg = {}; int rc = -EINVAL; if (!en_dev) @@ -420,9 +518,7 @@ static int bnxt_re_net_stats_ctx_free(struct bnxt_re_dev *rdev, if (test_bit(BNXT_RE_FLAG_ERR_DEVICE_DETACHED, &rdev->flags)) return 0; - memset(&fw_msg, 0, sizeof(fw_msg)); - - bnxt_re_init_hwrm_hdr(rdev, (void *)&req, HWRM_STAT_CTX_FREE, -1, -1); + bnxt_re_init_hwrm_hdr((void *)&req, HWRM_STAT_CTX_FREE); req.stat_ctx_id = cpu_to_le32(fw_stats_ctx_id); bnxt_re_fill_fw_msg(&fw_msg, (void *)&req, sizeof(req), (void *)&resp, sizeof(resp), DFLT_HWRM_CMD_TIMEOUT); @@ -439,10 +535,10 @@ static int bnxt_re_net_stats_ctx_alloc(struct bnxt_re_dev *rdev, u32 *fw_stats_ctx_id) { struct bnxt_qplib_chip_ctx *chip_ctx = rdev->chip_ctx; - struct hwrm_stat_ctx_alloc_output resp = {0}; - struct hwrm_stat_ctx_alloc_input req = {0}; + struct hwrm_stat_ctx_alloc_output resp = {}; + struct hwrm_stat_ctx_alloc_input req = {}; struct bnxt_en_dev *en_dev = rdev->en_dev; - struct bnxt_fw_msg fw_msg; + struct bnxt_fw_msg fw_msg = {}; int rc = -EINVAL; *fw_stats_ctx_id = INVALID_STATS_CTX_ID; @@ -450,9 +546,7 @@ static int bnxt_re_net_stats_ctx_alloc(struct bnxt_re_dev *rdev, if (!en_dev) return rc; - memset(&fw_msg, 0, sizeof(fw_msg)); - - bnxt_re_init_hwrm_hdr(rdev, (void *)&req, HWRM_STAT_CTX_ALLOC, -1, -1); + bnxt_re_init_hwrm_hdr((void *)&req, HWRM_STAT_CTX_ALLOC); req.update_period_ms = cpu_to_le32(1000); req.stats_dma_addr = cpu_to_le64(dma_map); req.stats_dma_length = cpu_to_le16(chip_ctx->hw_stats_size); @@ -466,6 +560,10 @@ static int bnxt_re_net_stats_ctx_alloc(struct bnxt_re_dev *rdev, return rc; } +static void bnxt_re_disassociate_ucontext(struct ib_ucontext *ibcontext) +{ +} + /* Device */ static struct bnxt_re_dev *bnxt_re_from_netdev(struct net_device *netdev) @@ -532,6 +630,7 @@ static const struct ib_device_ops bnxt_re_dev_ops = { .destroy_qp = bnxt_re_destroy_qp, .destroy_srq = bnxt_re_destroy_srq, .device_group = &bnxt_re_dev_attr_group, + .disassociate_ucontext = bnxt_re_disassociate_ucontext, .get_dev_fw_str = bnxt_re_query_fw_str, .get_dma_mr = bnxt_re_get_dma_mr, .get_hw_stats = bnxt_re_ib_get_hw_stats, @@ -539,6 +638,7 @@ static const struct ib_device_ops bnxt_re_dev_ops = { .get_port_immutable = bnxt_re_get_port_immutable, .map_mr_sg = bnxt_re_map_mr_sg, .mmap = bnxt_re_mmap, + .mmap_free = bnxt_re_mmap_free, .modify_qp = bnxt_re_modify_qp, .modify_srq = bnxt_re_modify_srq, .poll_cq = bnxt_re_poll_cq, @@ -579,6 +679,9 @@ static int bnxt_re_register_ib(struct bnxt_re_dev *rdev) ibdev->dev.parent = &rdev->en_dev->pdev->dev; ibdev->local_dma_lkey = BNXT_QPLIB_RSVD_LKEY; + if (IS_ENABLED(CONFIG_INFINIBAND_USER_ACCESS)) + ibdev->driver_def = bnxt_re_uapi_defs; + ib_set_device_ops(ibdev, &bnxt_re_dev_ops); ret = ib_device_set_netdev(&rdev->ibdev, rdev->netdev, 1); if (ret) @@ -822,7 +925,6 @@ static void bnxt_re_free_res(struct bnxt_re_dev *rdev) if (rdev->qplib_res.dpi_tbl.max) { bnxt_qplib_dealloc_dpi(&rdev->qplib_res, - &rdev->qplib_res.dpi_tbl, &rdev->dpi_privileged); } if (rdev->qplib_res.rcfw) { @@ -850,9 +952,9 @@ static int bnxt_re_alloc_res(struct bnxt_re_dev *rdev) if (rc) goto fail; - rc = bnxt_qplib_alloc_dpi(&rdev->qplib_res.dpi_tbl, + rc = bnxt_qplib_alloc_dpi(&rdev->qplib_res, &rdev->dpi_privileged, - rdev); + rdev, BNXT_QPLIB_DPI_TYPE_KERNEL); if (rc) goto dealloc_res; @@ -892,7 +994,6 @@ free_nq: bnxt_qplib_free_nq(&rdev->nq[i]); } bnxt_qplib_dealloc_dpi(&rdev->qplib_res, - &rdev->qplib_res.dpi_tbl, &rdev->dpi_privileged); dealloc_res: bnxt_qplib_free_res(&rdev->qplib_res); @@ -963,12 +1064,6 @@ static int bnxt_re_update_gid(struct bnxt_re_dev *rdev) if (!ib_device_try_get(&rdev->ibdev)) return 0; - if (!sgid_tbl) { - ibdev_err(&rdev->ibdev, "QPLIB: SGID table not allocated"); - rc = -EINVAL; - goto out; - } - for (index = 0; index < sgid_tbl->active; index++) { gid_idx = sgid_tbl->hw_id[index]; @@ -986,7 +1081,7 @@ static int bnxt_re_update_gid(struct bnxt_re_dev *rdev) rc = bnxt_qplib_update_sgid(sgid_tbl, &gid, gid_idx, rdev->qplib_res.netdev->dev_addr); } -out: + ib_device_put(&rdev->ibdev); return rc; } @@ -1039,14 +1134,13 @@ static int bnxt_re_setup_qos(struct bnxt_re_dev *rdev) static void bnxt_re_query_hwrm_intf_version(struct bnxt_re_dev *rdev) { struct bnxt_en_dev *en_dev = rdev->en_dev; - struct hwrm_ver_get_output resp = {0}; - struct hwrm_ver_get_input req = {0}; - struct bnxt_fw_msg fw_msg; + struct hwrm_ver_get_output resp = {}; + struct hwrm_ver_get_input req = {}; + struct bnxt_qplib_chip_ctx *cctx; + struct bnxt_fw_msg fw_msg = {}; int rc = 0; - memset(&fw_msg, 0, sizeof(fw_msg)); - bnxt_re_init_hwrm_hdr(rdev, (void *)&req, - HWRM_VER_GET, -1, -1); + bnxt_re_init_hwrm_hdr((void *)&req, HWRM_VER_GET); req.hwrm_intf_maj = HWRM_VERSION_MAJOR; req.hwrm_intf_min = HWRM_VERSION_MINOR; req.hwrm_intf_upd = HWRM_VERSION_UPDATE; @@ -1058,11 +1152,18 @@ static void bnxt_re_query_hwrm_intf_version(struct bnxt_re_dev *rdev) rc); return; } - rdev->qplib_ctx.hwrm_intf_ver = + + cctx = rdev->chip_ctx; + cctx->hwrm_intf_ver = (u64)le16_to_cpu(resp.hwrm_intf_major) << 48 | (u64)le16_to_cpu(resp.hwrm_intf_minor) << 32 | (u64)le16_to_cpu(resp.hwrm_intf_build) << 16 | le16_to_cpu(resp.hwrm_intf_patch); + + cctx->hwrm_cmd_max_timeout = le16_to_cpu(resp.max_req_timeout); + + if (!cctx->hwrm_cmd_max_timeout) + cctx->hwrm_cmd_max_timeout = RCFW_FW_STALL_MAX_TIMEOUT; } static int bnxt_re_ib_init(struct bnxt_re_dev *rdev) @@ -1200,7 +1301,7 @@ static int bnxt_re_dev_init(struct bnxt_re_dev *rdev, u8 wqe_mode) db_offt = bnxt_re_get_nqdb_offset(rdev, BNXT_RE_AEQ_IDX); vid = rdev->en_dev->msix_entries[BNXT_RE_AEQ_IDX].vector; rc = bnxt_qplib_enable_rcfw_channel(&rdev->rcfw, - vid, db_offt, rdev->is_virtfn, + vid, db_offt, &bnxt_re_aeq_handler); if (rc) { ibdev_err(&rdev->ibdev, "Failed to enable RCFW channel: %#x\n", @@ -1497,6 +1598,7 @@ static int bnxt_re_suspend(struct auxiliary_device *adev, pm_message_t state) */ set_bit(BNXT_RE_FLAG_ERR_DEVICE_DETACHED, &rdev->flags); set_bit(ERR_DEVICE_DETACHED, &rdev->rcfw.cmdq.flags); + wake_up_all(&rdev->rcfw.cmdq.waitq); mutex_unlock(&bnxt_re_mutex); return 0; diff --git a/drivers/infiniband/hw/bnxt_re/qplib_fp.c b/drivers/infiniband/hw/bnxt_re/qplib_fp.c index 8974f6235cfa..91aed77ce40d 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_fp.c +++ b/drivers/infiniband/hw/bnxt_re/qplib_fp.c @@ -399,6 +399,9 @@ static irqreturn_t bnxt_qplib_nq_irq(int irq, void *dev_instance) void bnxt_qplib_nq_stop_irq(struct bnxt_qplib_nq *nq, bool kill) { + if (!nq->requested) + return; + tasklet_disable(&nq->nq_tasklet); /* Mask h/w interrupt */ bnxt_qplib_ring_nq_db(&nq->nq_db.dbinfo, nq->res->cctx, false); @@ -406,11 +409,12 @@ void bnxt_qplib_nq_stop_irq(struct bnxt_qplib_nq *nq, bool kill) synchronize_irq(nq->msix_vec); if (kill) tasklet_kill(&nq->nq_tasklet); - if (nq->requested) { - irq_set_affinity_hint(nq->msix_vec, NULL); - free_irq(nq->msix_vec, nq); - nq->requested = false; - } + + irq_set_affinity_hint(nq->msix_vec, NULL); + free_irq(nq->msix_vec, nq); + kfree(nq->name); + nq->name = NULL; + nq->requested = false; } void bnxt_qplib_disable_nq(struct bnxt_qplib_nq *nq) @@ -436,6 +440,7 @@ void bnxt_qplib_disable_nq(struct bnxt_qplib_nq *nq) int bnxt_qplib_nq_start_irq(struct bnxt_qplib_nq *nq, int nq_indx, int msix_vector, bool need_init) { + struct bnxt_qplib_res *res = nq->res; int rc; if (nq->requested) @@ -447,10 +452,17 @@ int bnxt_qplib_nq_start_irq(struct bnxt_qplib_nq *nq, int nq_indx, else tasklet_enable(&nq->nq_tasklet); - snprintf(nq->name, sizeof(nq->name), "bnxt_qplib_nq-%d", nq_indx); + nq->name = kasprintf(GFP_KERNEL, "bnxt_re-nq-%d@pci:%s", + nq_indx, pci_name(res->pdev)); + if (!nq->name) + return -ENOMEM; rc = request_irq(nq->msix_vec, bnxt_qplib_nq_irq, 0, nq->name, nq); - if (rc) + if (rc) { + kfree(nq->name); + nq->name = NULL; + tasklet_disable(&nq->nq_tasklet); return rc; + } cpumask_clear(&nq->mask); cpumask_set_cpu(nq_indx, &nq->mask); @@ -461,7 +473,7 @@ int bnxt_qplib_nq_start_irq(struct bnxt_qplib_nq *nq, int nq_indx, nq->msix_vec, nq_indx); } nq->requested = true; - bnxt_qplib_ring_nq_db(&nq->nq_db.dbinfo, nq->res->cctx, true); + bnxt_qplib_ring_nq_db(&nq->nq_db.dbinfo, res->cctx, true); return rc; } @@ -471,7 +483,6 @@ static int bnxt_qplib_map_nq_db(struct bnxt_qplib_nq *nq, u32 reg_offt) resource_size_t reg_base; struct bnxt_qplib_nq_db *nq_db; struct pci_dev *pdev; - int rc = 0; pdev = nq->pdev; nq_db = &nq->nq_db; @@ -481,8 +492,7 @@ static int bnxt_qplib_map_nq_db(struct bnxt_qplib_nq *nq, u32 reg_offt) if (!nq_db->reg.bar_base) { dev_err(&pdev->dev, "QPLIB: NQ BAR region %d resc start is 0!", nq_db->reg.bar_id); - rc = -ENOMEM; - goto fail; + return -ENOMEM; } reg_base = nq_db->reg.bar_base + reg_offt; @@ -492,15 +502,14 @@ static int bnxt_qplib_map_nq_db(struct bnxt_qplib_nq *nq, u32 reg_offt) if (!nq_db->reg.bar_reg) { dev_err(&pdev->dev, "QPLIB: NQ BAR region %d mapping failed", nq_db->reg.bar_id); - rc = -ENOMEM; - goto fail; + return -ENOMEM; } nq_db->dbinfo.db = nq_db->reg.bar_reg; nq_db->dbinfo.hwq = &nq->hwq; nq_db->dbinfo.xid = nq->ring_id; -fail: - return rc; + + return 0; } int bnxt_qplib_enable_nq(struct pci_dev *pdev, struct bnxt_qplib_nq *nq, @@ -614,7 +623,7 @@ int bnxt_qplib_create_srq(struct bnxt_qplib_res *res, hwq_attr.type = HWQ_TYPE_QUEUE; rc = bnxt_qplib_alloc_init_hwq(&srq->hwq, &hwq_attr); if (rc) - goto exit; + return rc; srq->swq = kcalloc(srq->hwq.max_elements, sizeof(*srq->swq), GFP_KERNEL); @@ -659,7 +668,7 @@ int bnxt_qplib_create_srq(struct bnxt_qplib_res *res, srq->dbinfo.xid = srq->id; srq->dbinfo.db = srq->dpi->dbr; srq->dbinfo.max_slot = 1; - srq->dbinfo.priv_db = res->dpi_tbl.dbr_bar_reg_iomem; + srq->dbinfo.priv_db = res->dpi_tbl.priv_db; if (srq->threshold) bnxt_qplib_armen_db(&srq->dbinfo, DBC_DBC_TYPE_SRQ_ARMENA); srq->arm_req = false; @@ -668,7 +677,7 @@ int bnxt_qplib_create_srq(struct bnxt_qplib_res *res, fail: bnxt_qplib_free_hwq(res, &srq->hwq); kfree(srq->swq); -exit: + return rc; } @@ -732,15 +741,14 @@ int bnxt_qplib_post_srq_recv(struct bnxt_qplib_srq *srq, struct rq_wqe *srqe; struct sq_sge *hw_sge; u32 sw_prod, sw_cons, count = 0; - int i, rc = 0, next; + int i, next; spin_lock(&srq_hwq->lock); if (srq->start_idx == srq->last_idx) { dev_err(&srq_hwq->pdev->dev, "FP: SRQ (0x%x) is full!\n", srq->id); - rc = -EINVAL; spin_unlock(&srq_hwq->lock); - goto done; + return -EINVAL; } next = srq->start_idx; srq->start_idx = srq->swq[next].next_idx; @@ -781,22 +789,19 @@ int bnxt_qplib_post_srq_recv(struct bnxt_qplib_srq *srq, srq->arm_req = false; bnxt_qplib_srq_arm_db(&srq->dbinfo, srq->threshold); } -done: - return rc; + + return 0; } /* QP */ static int bnxt_qplib_alloc_init_swq(struct bnxt_qplib_q *que) { - int rc = 0; int indx; que->swq = kcalloc(que->max_wqe, sizeof(*que->swq), GFP_KERNEL); - if (!que->swq) { - rc = -ENOMEM; - goto out; - } + if (!que->swq) + return -ENOMEM; que->swq_start = 0; que->swq_last = que->max_wqe - 1; @@ -804,8 +809,8 @@ static int bnxt_qplib_alloc_init_swq(struct bnxt_qplib_q *que) que->swq[indx].next_idx = indx + 1; que->swq[que->swq_last].next_idx = 0; /* Make it circular */ que->swq_last = 0; -out: - return rc; + + return 0; } int bnxt_qplib_create_qp1(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp) @@ -839,7 +844,7 @@ int bnxt_qplib_create_qp1(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp) hwq_attr.type = HWQ_TYPE_QUEUE; rc = bnxt_qplib_alloc_init_hwq(&sq->hwq, &hwq_attr); if (rc) - goto exit; + return rc; rc = bnxt_qplib_alloc_init_swq(sq); if (rc) @@ -927,7 +932,6 @@ sq_swq: kfree(sq->swq); fail_sq: bnxt_qplib_free_hwq(res, &sq->hwq); -exit: return rc; } @@ -992,7 +996,7 @@ int bnxt_qplib_create_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp) hwq_attr.type = HWQ_TYPE_QUEUE; rc = bnxt_qplib_alloc_init_hwq(&sq->hwq, &hwq_attr); if (rc) - goto exit; + return rc; rc = bnxt_qplib_alloc_init_swq(sq); if (rc) @@ -1140,7 +1144,6 @@ sq_swq: kfree(sq->swq); fail_sq: bnxt_qplib_free_hwq(res, &sq->hwq); -exit: return rc; } @@ -1614,7 +1617,7 @@ static int bnxt_qplib_put_inline(struct bnxt_qplib_qp *qp, il_src = (void *)wqe->sg_list[indx].addr; t_len += len; if (t_len > qp->max_inline_data) - goto bad; + return -ENOMEM; while (len) { if (pull_dst) { pull_dst = false; @@ -1638,8 +1641,6 @@ static int bnxt_qplib_put_inline(struct bnxt_qplib_qp *qp, } return t_len; -bad: - return -ENOMEM; } static u32 bnxt_qplib_put_sges(struct bnxt_qplib_hwq *hwq, @@ -2069,7 +2070,7 @@ int bnxt_qplib_create_cq(struct bnxt_qplib_res *res, struct bnxt_qplib_cq *cq) hwq_attr.sginfo = &cq->sg_info; rc = bnxt_qplib_alloc_init_hwq(&cq->hwq, &hwq_attr); if (rc) - goto exit; + return rc; bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req, CMDQ_BASE_OPCODE_CREATE_CQ, @@ -2104,7 +2105,7 @@ int bnxt_qplib_create_cq(struct bnxt_qplib_res *res, struct bnxt_qplib_cq *cq) cq->dbinfo.hwq = &cq->hwq; cq->dbinfo.xid = cq->id; cq->dbinfo.db = cq->dpi->dbr; - cq->dbinfo.priv_db = res->dpi_tbl.dbr_bar_reg_iomem; + cq->dbinfo.priv_db = res->dpi_tbl.priv_db; bnxt_qplib_armen_db(&cq->dbinfo, DBC_DBC_TYPE_CQ_ARMENA); @@ -2112,7 +2113,6 @@ int bnxt_qplib_create_cq(struct bnxt_qplib_res *res, struct bnxt_qplib_cq *cq) fail: bnxt_qplib_free_hwq(res, &cq->hwq); -exit: return rc; } @@ -2505,7 +2505,6 @@ static int bnxt_qplib_cq_process_res_rc(struct bnxt_qplib_cq *cq, struct bnxt_qplib_qp *qp; struct bnxt_qplib_q *rq; u32 wr_id_idx; - int rc = 0; qp = (struct bnxt_qplib_qp *)((unsigned long) le64_to_cpu(hwcqe->qp_handle)); @@ -2516,7 +2515,7 @@ static int bnxt_qplib_cq_process_res_rc(struct bnxt_qplib_cq *cq, if (qp->rq.flushed) { dev_dbg(&cq->hwq.pdev->dev, "%s: QP in Flush QP = %p\n", __func__, qp); - goto done; + return 0; } cqe = *pcqe; @@ -2572,8 +2571,7 @@ static int bnxt_qplib_cq_process_res_rc(struct bnxt_qplib_cq *cq, } } -done: - return rc; + return 0; } static int bnxt_qplib_cq_process_res_ud(struct bnxt_qplib_cq *cq, @@ -2586,7 +2584,6 @@ static int bnxt_qplib_cq_process_res_ud(struct bnxt_qplib_cq *cq, struct bnxt_qplib_qp *qp; struct bnxt_qplib_q *rq; u32 wr_id_idx; - int rc = 0; qp = (struct bnxt_qplib_qp *)((unsigned long) le64_to_cpu(hwcqe->qp_handle)); @@ -2597,7 +2594,7 @@ static int bnxt_qplib_cq_process_res_ud(struct bnxt_qplib_cq *cq, if (qp->rq.flushed) { dev_dbg(&cq->hwq.pdev->dev, "%s: QP in Flush QP = %p\n", __func__, qp); - goto done; + return 0; } cqe = *pcqe; cqe->opcode = hwcqe->cqe_type_toggle & CQ_BASE_CQE_TYPE_MASK; @@ -2659,8 +2656,8 @@ static int bnxt_qplib_cq_process_res_ud(struct bnxt_qplib_cq *cq, bnxt_qplib_add_flush_qp(qp); } } -done: - return rc; + + return 0; } bool bnxt_qplib_is_cq_empty(struct bnxt_qplib_cq *cq) @@ -2687,7 +2684,6 @@ static int bnxt_qplib_cq_process_res_raweth_qp1(struct bnxt_qplib_cq *cq, struct bnxt_qplib_srq *srq; struct bnxt_qplib_cqe *cqe; u32 wr_id_idx; - int rc = 0; qp = (struct bnxt_qplib_qp *)((unsigned long) le64_to_cpu(hwcqe->qp_handle)); @@ -2698,7 +2694,7 @@ static int bnxt_qplib_cq_process_res_raweth_qp1(struct bnxt_qplib_cq *cq, if (qp->rq.flushed) { dev_dbg(&cq->hwq.pdev->dev, "%s: QP in Flush QP = %p\n", __func__, qp); - goto done; + return 0; } cqe = *pcqe; cqe->opcode = hwcqe->cqe_type_toggle & CQ_BASE_CQE_TYPE_MASK; @@ -2767,8 +2763,7 @@ static int bnxt_qplib_cq_process_res_raweth_qp1(struct bnxt_qplib_cq *cq, } } -done: - return rc; + return 0; } static int bnxt_qplib_cq_process_terminal(struct bnxt_qplib_cq *cq, @@ -2790,11 +2785,8 @@ static int bnxt_qplib_cq_process_terminal(struct bnxt_qplib_cq *cq, qp = (struct bnxt_qplib_qp *)((unsigned long) le64_to_cpu(hwcqe->qp_handle)); - if (!qp) { - dev_err(&cq->hwq.pdev->dev, - "FP: CQ Process terminal qp is NULL\n"); + if (!qp) return -EINVAL; - } /* Must block new posting of SQ and RQ */ qp->state = CMDQ_MODIFY_QP_NEW_STATE_ERR; diff --git a/drivers/infiniband/hw/bnxt_re/qplib_fp.h b/drivers/infiniband/hw/bnxt_re/qplib_fp.h index d74d5ead2e32..a42820821c47 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_fp.h +++ b/drivers/infiniband/hw/bnxt_re/qplib_fp.h @@ -472,7 +472,7 @@ typedef int (*srqn_handler_t)(struct bnxt_qplib_nq *nq, struct bnxt_qplib_nq { struct pci_dev *pdev; struct bnxt_qplib_res *res; - char name[32]; + char *name; struct bnxt_qplib_hwq hwq; struct bnxt_qplib_nq_db nq_db; u16 ring_id; diff --git a/drivers/infiniband/hw/bnxt_re/qplib_rcfw.c b/drivers/infiniband/hw/bnxt_re/qplib_rcfw.c index de9069103177..b30e66b64827 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_rcfw.c +++ b/drivers/infiniband/hw/bnxt_re/qplib_rcfw.c @@ -53,112 +53,289 @@ static void bnxt_qplib_service_creq(struct tasklet_struct *t); -/* Hardware communication channel */ +/** + * bnxt_qplib_map_rc - map return type based on opcode + * @opcode - roce slow path opcode + * + * case #1 + * Firmware initiated error recovery is a safe state machine and + * driver can consider all the underlying rdma resources are free. + * In this state, it is safe to return success for opcodes related to + * destroying rdma resources (like destroy qp, destroy cq etc.). + * + * case #2 + * If driver detect potential firmware stall, it is not safe state machine + * and the driver can not consider all the underlying rdma resources are + * freed. + * In this state, it is not safe to return success for opcodes related to + * destroying rdma resources (like destroy qp, destroy cq etc.). + * + * Scope of this helper function is only for case #1. + * + * Returns: + * 0 to communicate success to caller. + * Non zero error code to communicate failure to caller. + */ +static int bnxt_qplib_map_rc(u8 opcode) +{ + switch (opcode) { + case CMDQ_BASE_OPCODE_DESTROY_QP: + case CMDQ_BASE_OPCODE_DESTROY_SRQ: + case CMDQ_BASE_OPCODE_DESTROY_CQ: + case CMDQ_BASE_OPCODE_DEALLOCATE_KEY: + case CMDQ_BASE_OPCODE_DEREGISTER_MR: + case CMDQ_BASE_OPCODE_DELETE_GID: + case CMDQ_BASE_OPCODE_DESTROY_QP1: + case CMDQ_BASE_OPCODE_DESTROY_AH: + case CMDQ_BASE_OPCODE_DEINITIALIZE_FW: + case CMDQ_BASE_OPCODE_MODIFY_ROCE_CC: + case CMDQ_BASE_OPCODE_SET_LINK_AGGR_MODE: + return 0; + default: + return -ETIMEDOUT; + } +} + +/** + * bnxt_re_is_fw_stalled - Check firmware health + * @rcfw - rcfw channel instance of rdev + * @cookie - cookie to track the command + * + * If firmware has not responded any rcfw command within + * rcfw->max_timeout, consider firmware as stalled. + * + * Returns: + * 0 if firmware is responding + * -ENODEV if firmware is not responding + */ +static int bnxt_re_is_fw_stalled(struct bnxt_qplib_rcfw *rcfw, + u16 cookie) +{ + struct bnxt_qplib_cmdq_ctx *cmdq; + struct bnxt_qplib_crsqe *crsqe; + + crsqe = &rcfw->crsqe_tbl[cookie]; + cmdq = &rcfw->cmdq; + + if (time_after(jiffies, cmdq->last_seen + + (rcfw->max_timeout * HZ))) { + dev_warn_ratelimited(&rcfw->pdev->dev, + "%s: FW STALL Detected. cmdq[%#x]=%#x waited (%d > %d) msec active %d ", + __func__, cookie, crsqe->opcode, + jiffies_to_msecs(jiffies - cmdq->last_seen), + rcfw->max_timeout * 1000, + crsqe->is_in_used); + return -ENODEV; + } + + return 0; +} + +/** + * __wait_for_resp - Don't hold the cpu context and wait for response + * @rcfw - rcfw channel instance of rdev + * @cookie - cookie to track the command + * + * Wait for command completion in sleepable context. + * + * Returns: + * 0 if command is completed by firmware. + * Non zero error code for rest of the case. + */ static int __wait_for_resp(struct bnxt_qplib_rcfw *rcfw, u16 cookie) { struct bnxt_qplib_cmdq_ctx *cmdq; - u16 cbit; - int rc; + struct bnxt_qplib_crsqe *crsqe; + int ret; cmdq = &rcfw->cmdq; - cbit = cookie % rcfw->cmdq_depth; - rc = wait_event_timeout(cmdq->waitq, - !test_bit(cbit, cmdq->cmdq_bitmap), - msecs_to_jiffies(RCFW_CMD_WAIT_TIME_MS)); - return rc ? 0 : -ETIMEDOUT; + crsqe = &rcfw->crsqe_tbl[cookie]; + + do { + if (test_bit(ERR_DEVICE_DETACHED, &cmdq->flags)) + return bnxt_qplib_map_rc(crsqe->opcode); + if (test_bit(FIRMWARE_STALL_DETECTED, &cmdq->flags)) + return -ETIMEDOUT; + + wait_event_timeout(cmdq->waitq, + !crsqe->is_in_used || + test_bit(ERR_DEVICE_DETACHED, &cmdq->flags), + msecs_to_jiffies(rcfw->max_timeout * 1000)); + + if (!crsqe->is_in_used) + return 0; + + bnxt_qplib_service_creq(&rcfw->creq.creq_tasklet); + + if (!crsqe->is_in_used) + return 0; + + ret = bnxt_re_is_fw_stalled(rcfw, cookie); + if (ret) + return ret; + + } while (true); }; +/** + * __block_for_resp - hold the cpu context and wait for response + * @rcfw - rcfw channel instance of rdev + * @cookie - cookie to track the command + * + * This function will hold the cpu (non-sleepable context) and + * wait for command completion. Maximum holding interval is 8 second. + * + * Returns: + * -ETIMEOUT if command is not completed in specific time interval. + * 0 if command is completed by firmware. + */ static int __block_for_resp(struct bnxt_qplib_rcfw *rcfw, u16 cookie) { - u32 count = RCFW_BLOCKED_CMD_WAIT_COUNT; - struct bnxt_qplib_cmdq_ctx *cmdq; - u16 cbit; + struct bnxt_qplib_cmdq_ctx *cmdq = &rcfw->cmdq; + struct bnxt_qplib_crsqe *crsqe; + unsigned long issue_time = 0; + + issue_time = jiffies; + crsqe = &rcfw->crsqe_tbl[cookie]; - cmdq = &rcfw->cmdq; - cbit = cookie % rcfw->cmdq_depth; - if (!test_bit(cbit, cmdq->cmdq_bitmap)) - goto done; do { + if (test_bit(ERR_DEVICE_DETACHED, &cmdq->flags)) + return bnxt_qplib_map_rc(crsqe->opcode); + if (test_bit(FIRMWARE_STALL_DETECTED, &cmdq->flags)) + return -ETIMEDOUT; + udelay(1); + bnxt_qplib_service_creq(&rcfw->creq.creq_tasklet); - } while (test_bit(cbit, cmdq->cmdq_bitmap) && --count); -done: - return count ? 0 : -ETIMEDOUT; + if (!crsqe->is_in_used) + return 0; + + } while (time_before(jiffies, issue_time + (8 * HZ))); + + return -ETIMEDOUT; }; -static int __send_message(struct bnxt_qplib_rcfw *rcfw, - struct bnxt_qplib_cmdqmsg *msg) +/* __send_message_no_waiter - get cookie and post the message. + * @rcfw - rcfw channel instance of rdev + * @msg - qplib message internal + * + * This function will just post and don't bother about completion. + * Current design of this function is - + * user must hold the completion queue hwq->lock. + * user must have used existing completion and free the resources. + * this function will not check queue full condition. + * this function will explicitly set is_waiter_alive=false. + * current use case is - send destroy_ah if create_ah is return + * after waiter of create_ah is lost. It can be extended for other + * use case as well. + * + * Returns: Nothing + * + */ +static void __send_message_no_waiter(struct bnxt_qplib_rcfw *rcfw, + struct bnxt_qplib_cmdqmsg *msg) { struct bnxt_qplib_cmdq_ctx *cmdq = &rcfw->cmdq; struct bnxt_qplib_hwq *hwq = &cmdq->hwq; struct bnxt_qplib_crsqe *crsqe; struct bnxt_qplib_cmdqe *cmdqe; u32 sw_prod, cmdq_prod; - struct pci_dev *pdev; - unsigned long flags; - u32 bsize, opcode; - u16 cookie, cbit; + u16 cookie; + u32 bsize; u8 *preq; - pdev = rcfw->pdev; + cookie = cmdq->seq_num & RCFW_MAX_COOKIE_VALUE; + __set_cmdq_base_cookie(msg->req, msg->req_sz, cpu_to_le16(cookie)); + crsqe = &rcfw->crsqe_tbl[cookie]; - opcode = __get_cmdq_base_opcode(msg->req, msg->req_sz); - if (!test_bit(FIRMWARE_INITIALIZED_FLAG, &cmdq->flags) && - (opcode != CMDQ_BASE_OPCODE_QUERY_FUNC && - opcode != CMDQ_BASE_OPCODE_INITIALIZE_FW && - opcode != CMDQ_BASE_OPCODE_QUERY_VERSION)) { - dev_err(&pdev->dev, - "RCFW not initialized, reject opcode 0x%x\n", opcode); - return -EINVAL; - } + /* Set cmd_size in terms of 16B slots in req. */ + bsize = bnxt_qplib_set_cmd_slots(msg->req); + /* GET_CMD_SIZE would return number of slots in either case of tlv + * and non-tlv commands after call to bnxt_qplib_set_cmd_slots() + */ + crsqe->is_internal_cmd = true; + crsqe->is_waiter_alive = false; + crsqe->is_in_used = true; + crsqe->req_size = __get_cmdq_base_cmd_size(msg->req, msg->req_sz); - if (test_bit(FIRMWARE_INITIALIZED_FLAG, &cmdq->flags) && - opcode == CMDQ_BASE_OPCODE_INITIALIZE_FW) { - dev_err(&pdev->dev, "RCFW already initialized!\n"); - return -EINVAL; - } + preq = (u8 *)msg->req; + do { + /* Locate the next cmdq slot */ + sw_prod = HWQ_CMP(hwq->prod, hwq); + cmdqe = bnxt_qplib_get_qe(hwq, sw_prod, NULL); + /* Copy a segment of the req cmd to the cmdq */ + memset(cmdqe, 0, sizeof(*cmdqe)); + memcpy(cmdqe, preq, min_t(u32, bsize, sizeof(*cmdqe))); + preq += min_t(u32, bsize, sizeof(*cmdqe)); + bsize -= min_t(u32, bsize, sizeof(*cmdqe)); + hwq->prod++; + } while (bsize > 0); + cmdq->seq_num++; - if (test_bit(FIRMWARE_TIMED_OUT, &cmdq->flags)) - return -ETIMEDOUT; + cmdq_prod = hwq->prod; + atomic_inc(&rcfw->timeout_send); + /* ring CMDQ DB */ + wmb(); + writel(cmdq_prod, cmdq->cmdq_mbox.prod); + writel(RCFW_CMDQ_TRIG_VAL, cmdq->cmdq_mbox.db); +} + +static int __send_message(struct bnxt_qplib_rcfw *rcfw, + struct bnxt_qplib_cmdqmsg *msg, u8 opcode) +{ + u32 bsize, free_slots, required_slots; + struct bnxt_qplib_cmdq_ctx *cmdq; + struct bnxt_qplib_crsqe *crsqe; + struct bnxt_qplib_cmdqe *cmdqe; + struct bnxt_qplib_hwq *hwq; + u32 sw_prod, cmdq_prod; + struct pci_dev *pdev; + unsigned long flags; + u16 cookie; + u8 *preq; + + cmdq = &rcfw->cmdq; + hwq = &cmdq->hwq; + pdev = rcfw->pdev; /* Cmdq are in 16-byte units, each request can consume 1 or more * cmdqe */ spin_lock_irqsave(&hwq->lock, flags); - if (msg->req->cmd_size >= HWQ_FREE_SLOTS(hwq)) { - dev_err(&pdev->dev, "RCFW: CMDQ is full!\n"); + required_slots = bnxt_qplib_get_cmd_slots(msg->req); + free_slots = HWQ_FREE_SLOTS(hwq); + cookie = cmdq->seq_num & RCFW_MAX_COOKIE_VALUE; + crsqe = &rcfw->crsqe_tbl[cookie]; + + if (required_slots >= free_slots) { + dev_info_ratelimited(&pdev->dev, + "CMDQ is full req/free %d/%d!", + required_slots, free_slots); spin_unlock_irqrestore(&hwq->lock, flags); return -EAGAIN; } - - - cookie = cmdq->seq_num & RCFW_MAX_COOKIE_VALUE; - cbit = cookie % rcfw->cmdq_depth; if (msg->block) cookie |= RCFW_CMD_IS_BLOCKING; - - set_bit(cbit, cmdq->cmdq_bitmap); __set_cmdq_base_cookie(msg->req, msg->req_sz, cpu_to_le16(cookie)); - crsqe = &rcfw->crsqe_tbl[cbit]; - if (crsqe->resp) { - spin_unlock_irqrestore(&hwq->lock, flags); - return -EBUSY; - } - /* change the cmd_size to the number of 16byte cmdq unit. - * req->cmd_size is modified here - */ bsize = bnxt_qplib_set_cmd_slots(msg->req); - - memset(msg->resp, 0, sizeof(*msg->resp)); + crsqe->free_slots = free_slots; crsqe->resp = (struct creq_qp_event *)msg->resp; crsqe->resp->cookie = cpu_to_le16(cookie); + crsqe->is_internal_cmd = false; + crsqe->is_waiter_alive = true; + crsqe->is_in_used = true; + crsqe->opcode = opcode; + crsqe->req_size = __get_cmdq_base_cmd_size(msg->req, msg->req_sz); if (__get_cmdq_base_resp_size(msg->req, msg->req_sz) && msg->sb) { struct bnxt_qplib_rcfw_sbuf *sbuf = msg->sb; - __set_cmdq_base_resp_addr(msg->req, msg->req_sz, cpu_to_le64(sbuf->dma_addr)); + + __set_cmdq_base_resp_addr(msg->req, msg->req_sz, + cpu_to_le64(sbuf->dma_addr)); __set_cmdq_base_resp_size(msg->req, msg->req_sz, - ALIGN(sbuf->size, BNXT_QPLIB_CMDQE_UNITS)); + ALIGN(sbuf->size, + BNXT_QPLIB_CMDQE_UNITS)); } preq = (u8 *)msg->req; @@ -166,11 +343,6 @@ static int __send_message(struct bnxt_qplib_rcfw *rcfw, /* Locate the next cmdq slot */ sw_prod = HWQ_CMP(hwq->prod, hwq); cmdqe = bnxt_qplib_get_qe(hwq, sw_prod, NULL); - if (!cmdqe) { - dev_err(&pdev->dev, - "RCFW request failed with no cmdqe!\n"); - goto done; - } /* Copy a segment of the req cmd to the cmdq */ memset(cmdqe, 0, sizeof(*cmdqe)); memcpy(cmdqe, preq, min_t(u32, bsize, sizeof(*cmdqe))); @@ -180,7 +352,7 @@ static int __send_message(struct bnxt_qplib_rcfw *rcfw, } while (bsize > 0); cmdq->seq_num++; - cmdq_prod = hwq->prod; + cmdq_prod = hwq->prod & 0xFFFF; if (test_bit(FIRMWARE_FIRST_FLAG, &cmdq->flags)) { /* The very first doorbell write * is required to set this flag @@ -194,51 +366,158 @@ static int __send_message(struct bnxt_qplib_rcfw *rcfw, wmb(); writel(cmdq_prod, cmdq->cmdq_mbox.prod); writel(RCFW_CMDQ_TRIG_VAL, cmdq->cmdq_mbox.db); -done: spin_unlock_irqrestore(&hwq->lock, flags); /* Return the CREQ response pointer */ return 0; } -int bnxt_qplib_rcfw_send_message(struct bnxt_qplib_rcfw *rcfw, - struct bnxt_qplib_cmdqmsg *msg) +/** + * __poll_for_resp - self poll completion for rcfw command + * @rcfw - rcfw channel instance of rdev + * @cookie - cookie to track the command + * + * It works same as __wait_for_resp except this function will + * do self polling in sort interval since interrupt is disabled. + * This function can not be called from non-sleepable context. + * + * Returns: + * -ETIMEOUT if command is not completed in specific time interval. + * 0 if command is completed by firmware. + */ +static int __poll_for_resp(struct bnxt_qplib_rcfw *rcfw, u16 cookie) +{ + struct bnxt_qplib_cmdq_ctx *cmdq = &rcfw->cmdq; + struct bnxt_qplib_crsqe *crsqe; + unsigned long issue_time; + int ret; + + issue_time = jiffies; + crsqe = &rcfw->crsqe_tbl[cookie]; + + do { + if (test_bit(ERR_DEVICE_DETACHED, &cmdq->flags)) + return bnxt_qplib_map_rc(crsqe->opcode); + if (test_bit(FIRMWARE_STALL_DETECTED, &cmdq->flags)) + return -ETIMEDOUT; + + usleep_range(1000, 1001); + + bnxt_qplib_service_creq(&rcfw->creq.creq_tasklet); + if (!crsqe->is_in_used) + return 0; + if (jiffies_to_msecs(jiffies - issue_time) > + (rcfw->max_timeout * 1000)) { + ret = bnxt_re_is_fw_stalled(rcfw, cookie); + if (ret) + return ret; + } + } while (true); +}; + +static int __send_message_basic_sanity(struct bnxt_qplib_rcfw *rcfw, + struct bnxt_qplib_cmdqmsg *msg, + u8 opcode) +{ + struct bnxt_qplib_cmdq_ctx *cmdq; + + cmdq = &rcfw->cmdq; + + /* Prevent posting if f/w is not in a state to process */ + if (test_bit(ERR_DEVICE_DETACHED, &rcfw->cmdq.flags)) + return bnxt_qplib_map_rc(opcode); + if (test_bit(FIRMWARE_STALL_DETECTED, &cmdq->flags)) + return -ETIMEDOUT; + + if (test_bit(FIRMWARE_INITIALIZED_FLAG, &cmdq->flags) && + opcode == CMDQ_BASE_OPCODE_INITIALIZE_FW) { + dev_err(&rcfw->pdev->dev, "QPLIB: RCFW already initialized!"); + return -EINVAL; + } + + if (!test_bit(FIRMWARE_INITIALIZED_FLAG, &cmdq->flags) && + (opcode != CMDQ_BASE_OPCODE_QUERY_FUNC && + opcode != CMDQ_BASE_OPCODE_INITIALIZE_FW && + opcode != CMDQ_BASE_OPCODE_QUERY_VERSION)) { + dev_err(&rcfw->pdev->dev, + "QPLIB: RCFW not initialized, reject opcode 0x%x", + opcode); + return -EOPNOTSUPP; + } + + return 0; +} + +/* This function will just post and do not bother about completion */ +static void __destroy_timedout_ah(struct bnxt_qplib_rcfw *rcfw, + struct creq_create_ah_resp *create_ah_resp) +{ + struct bnxt_qplib_cmdqmsg msg = {}; + struct cmdq_destroy_ah req = {}; + + bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req, + CMDQ_BASE_OPCODE_DESTROY_AH, + sizeof(req)); + req.ah_cid = create_ah_resp->xid; + msg.req = (struct cmdq_base *)&req; + msg.req_sz = sizeof(req); + __send_message_no_waiter(rcfw, &msg); + dev_info_ratelimited(&rcfw->pdev->dev, + "From %s: ah_cid = %d timeout_send %d\n", + __func__, req.ah_cid, + atomic_read(&rcfw->timeout_send)); +} + +/** + * __bnxt_qplib_rcfw_send_message - qplib interface to send + * and complete rcfw command. + * @rcfw - rcfw channel instance of rdev + * @msg - qplib message internal + * + * This function does not account shadow queue depth. It will send + * all the command unconditionally as long as send queue is not full. + * + * Returns: + * 0 if command completed by firmware. + * Non zero if the command is not completed by firmware. + */ +static int __bnxt_qplib_rcfw_send_message(struct bnxt_qplib_rcfw *rcfw, + struct bnxt_qplib_cmdqmsg *msg) { struct creq_qp_event *evnt = (struct creq_qp_event *)msg->resp; + struct bnxt_qplib_crsqe *crsqe; + unsigned long flags; u16 cookie; - u8 opcode, retry_cnt = 0xFF; int rc = 0; + u8 opcode; - /* Prevent posting if f/w is not in a state to process */ - if (test_bit(ERR_DEVICE_DETACHED, &rcfw->cmdq.flags)) - return 0; + opcode = __get_cmdq_base_opcode(msg->req, msg->req_sz); - do { - opcode = __get_cmdq_base_opcode(msg->req, msg->req_sz); - rc = __send_message(rcfw, msg); - cookie = le16_to_cpu(__get_cmdq_base_cookie(msg->req, msg->req_sz)) & - RCFW_MAX_COOKIE_VALUE; - if (!rc) - break; - if (!retry_cnt || (rc != -EAGAIN && rc != -EBUSY)) { - /* send failed */ - dev_err(&rcfw->pdev->dev, "cmdq[%#x]=%#x send failed\n", - cookie, opcode); - return rc; - } - msg->block ? mdelay(1) : usleep_range(500, 1000); + rc = __send_message_basic_sanity(rcfw, msg, opcode); + if (rc) + return rc; - } while (retry_cnt--); + rc = __send_message(rcfw, msg, opcode); + if (rc) + return rc; + + cookie = le16_to_cpu(__get_cmdq_base_cookie(msg->req, msg->req_sz)) + & RCFW_MAX_COOKIE_VALUE; if (msg->block) rc = __block_for_resp(rcfw, cookie); - else + else if (atomic_read(&rcfw->rcfw_intr_enabled)) rc = __wait_for_resp(rcfw, cookie); + else + rc = __poll_for_resp(rcfw, cookie); + if (rc) { - /* timed out */ - dev_err(&rcfw->pdev->dev, "cmdq[%#x]=%#x timedout (%d)msec\n", - cookie, opcode, RCFW_CMD_WAIT_TIME_MS); - set_bit(FIRMWARE_TIMED_OUT, &rcfw->cmdq.flags); - return rc; + spin_lock_irqsave(&rcfw->cmdq.hwq.lock, flags); + crsqe = &rcfw->crsqe_tbl[cookie]; + crsqe->is_waiter_alive = false; + if (rc == -ENODEV) + set_bit(FIRMWARE_STALL_DETECTED, &rcfw->cmdq.flags); + spin_unlock_irqrestore(&rcfw->cmdq.hwq.lock, flags); + return -ETIMEDOUT; } if (evnt->status) { @@ -250,6 +529,48 @@ int bnxt_qplib_rcfw_send_message(struct bnxt_qplib_rcfw *rcfw, return rc; } + +/** + * bnxt_qplib_rcfw_send_message - qplib interface to send + * and complete rcfw command. + * @rcfw - rcfw channel instance of rdev + * @msg - qplib message internal + * + * Driver interact with Firmware through rcfw channel/slow path in two ways. + * a. Blocking rcfw command send. In this path, driver cannot hold + * the context for longer period since it is holding cpu until + * command is not completed. + * b. Non-blocking rcfw command send. In this path, driver can hold the + * context for longer period. There may be many pending command waiting + * for completion because of non-blocking nature. + * + * Driver will use shadow queue depth. Current queue depth of 8K + * (due to size of rcfw message there can be actual ~4K rcfw outstanding) + * is not optimal for rcfw command processing in firmware. + * + * Restrict at max #RCFW_CMD_NON_BLOCKING_SHADOW_QD Non-Blocking rcfw commands. + * Allow all blocking commands until there is no queue full. + * + * Returns: + * 0 if command completed by firmware. + * Non zero if the command is not completed by firmware. + */ +int bnxt_qplib_rcfw_send_message(struct bnxt_qplib_rcfw *rcfw, + struct bnxt_qplib_cmdqmsg *msg) +{ + int ret; + + if (!msg->block) { + down(&rcfw->rcfw_inflight); + ret = __bnxt_qplib_rcfw_send_message(rcfw, msg); + up(&rcfw->rcfw_inflight); + } else { + ret = __bnxt_qplib_rcfw_send_message(rcfw, msg); + } + + return ret; +} + /* Completions */ static int bnxt_qplib_process_func_event(struct bnxt_qplib_rcfw *rcfw, struct creq_func_event *func_event) @@ -295,19 +616,20 @@ static int bnxt_qplib_process_func_event(struct bnxt_qplib_rcfw *rcfw, } static int bnxt_qplib_process_qp_event(struct bnxt_qplib_rcfw *rcfw, - struct creq_qp_event *qp_event) + struct creq_qp_event *qp_event, + u32 *num_wait) { struct creq_qp_error_notification *err_event; struct bnxt_qplib_hwq *hwq = &rcfw->cmdq.hwq; struct bnxt_qplib_crsqe *crsqe; + u32 qp_id, tbl_indx, req_size; struct bnxt_qplib_qp *qp; - u16 cbit, blocked = 0; + u16 cookie, blocked = 0; + bool is_waiter_alive; struct pci_dev *pdev; unsigned long flags; - __le16 mcookie; - u16 cookie; + u32 wait_cmds = 0; int rc = 0; - u32 qp_id, tbl_indx; pdev = rcfw->pdev; switch (qp_event->event) { @@ -339,33 +661,60 @@ static int bnxt_qplib_process_qp_event(struct bnxt_qplib_rcfw *rcfw, spin_lock_irqsave_nested(&hwq->lock, flags, SINGLE_DEPTH_NESTING); cookie = le16_to_cpu(qp_event->cookie); - mcookie = qp_event->cookie; blocked = cookie & RCFW_CMD_IS_BLOCKING; cookie &= RCFW_MAX_COOKIE_VALUE; - cbit = cookie % rcfw->cmdq_depth; - crsqe = &rcfw->crsqe_tbl[cbit]; - if (crsqe->resp && - crsqe->resp->cookie == mcookie) { - memcpy(crsqe->resp, qp_event, sizeof(*qp_event)); - crsqe->resp = NULL; - } else { - if (crsqe->resp && crsqe->resp->cookie) - dev_err(&pdev->dev, - "CMD %s cookie sent=%#x, recd=%#x\n", - crsqe->resp ? "mismatch" : "collision", - crsqe->resp ? crsqe->resp->cookie : 0, - mcookie); + crsqe = &rcfw->crsqe_tbl[cookie]; + crsqe->is_in_used = false; + + if (WARN_ONCE(test_bit(FIRMWARE_STALL_DETECTED, + &rcfw->cmdq.flags), + "QPLIB: Unreponsive rcfw channel detected.!!")) { + dev_info(&pdev->dev, + "rcfw timedout: cookie = %#x, free_slots = %d", + cookie, crsqe->free_slots); + spin_unlock_irqrestore(&hwq->lock, flags); + return rc; } - if (!test_and_clear_bit(cbit, rcfw->cmdq.cmdq_bitmap)) - dev_warn(&pdev->dev, - "CMD bit %d was not requested\n", cbit); - hwq->cons += crsqe->req_size; + + if (crsqe->is_internal_cmd && !qp_event->status) + atomic_dec(&rcfw->timeout_send); + + if (crsqe->is_waiter_alive) { + if (crsqe->resp) + memcpy(crsqe->resp, qp_event, sizeof(*qp_event)); + if (!blocked) + wait_cmds++; + } + + req_size = crsqe->req_size; + is_waiter_alive = crsqe->is_waiter_alive; + crsqe->req_size = 0; + if (!is_waiter_alive) + crsqe->resp = NULL; - if (!blocked) - wake_up(&rcfw->cmdq.waitq); + hwq->cons += req_size; + + /* This is a case to handle below scenario - + * Create AH is completed successfully by firmware, + * but completion took more time and driver already lost + * the context of create_ah from caller. + * We have already return failure for create_ah verbs, + * so let's destroy the same address vector since it is + * no more used in stack. We don't care about completion + * in __send_message_no_waiter. + * If destroy_ah is failued by firmware, there will be AH + * resource leak and relatively not critical + unlikely + * scenario. Current design is not to handle such case. + */ + if (!is_waiter_alive && !qp_event->status && + qp_event->event == CREQ_QP_EVENT_EVENT_CREATE_AH) + __destroy_timedout_ah(rcfw, + (struct creq_create_ah_resp *) + qp_event); spin_unlock_irqrestore(&hwq->lock, flags); } + *num_wait += wait_cmds; return rc; } @@ -379,6 +728,7 @@ static void bnxt_qplib_service_creq(struct tasklet_struct *t) struct creq_base *creqe; u32 sw_cons, raw_cons; unsigned long flags; + u32 num_wakeup = 0; /* Service the CREQ until budget is over */ spin_lock_irqsave(&hwq->lock, flags); @@ -392,12 +742,14 @@ static void bnxt_qplib_service_creq(struct tasklet_struct *t) * reading any further. */ dma_rmb(); + rcfw->cmdq.last_seen = jiffies; type = creqe->type & CREQ_BASE_TYPE_MASK; switch (type) { case CREQ_BASE_TYPE_QP_EVENT: bnxt_qplib_process_qp_event - (rcfw, (struct creq_qp_event *)creqe); + (rcfw, (struct creq_qp_event *)creqe, + &num_wakeup); creq->stats.creq_qp_event_processed++; break; case CREQ_BASE_TYPE_FUNC_EVENT: @@ -425,6 +777,8 @@ static void bnxt_qplib_service_creq(struct tasklet_struct *t) rcfw->res->cctx, true); } spin_unlock_irqrestore(&hwq->lock, flags); + if (num_wakeup) + wake_up_nr(&rcfw->cmdq.waitq, num_wakeup); } static irqreturn_t bnxt_qplib_creq_irq(int irq, void *dev_instance) @@ -556,7 +910,6 @@ skip_ctx_setup: void bnxt_qplib_free_rcfw_channel(struct bnxt_qplib_rcfw *rcfw) { - bitmap_free(rcfw->cmdq.cmdq_bitmap); kfree(rcfw->qp_tbl); kfree(rcfw->crsqe_tbl); bnxt_qplib_free_hwq(rcfw->res, &rcfw->cmdq.hwq); @@ -593,13 +946,11 @@ int bnxt_qplib_alloc_rcfw_channel(struct bnxt_qplib_res *res, "HW channel CREQ allocation failed\n"); goto fail; } - if (ctx->hwrm_intf_ver < HWRM_VERSION_RCFW_CMDQ_DEPTH_CHECK) - rcfw->cmdq_depth = BNXT_QPLIB_CMDQE_MAX_CNT_256; - else - rcfw->cmdq_depth = BNXT_QPLIB_CMDQE_MAX_CNT_8192; + + rcfw->cmdq_depth = BNXT_QPLIB_CMDQE_MAX_CNT; sginfo.pgsize = bnxt_qplib_cmdqe_page_size(rcfw->cmdq_depth); - hwq_attr.depth = rcfw->cmdq_depth; + hwq_attr.depth = rcfw->cmdq_depth & 0x7FFFFFFF; hwq_attr.stride = BNXT_QPLIB_CMDQE_UNITS; hwq_attr.type = HWQ_TYPE_CTX; if (bnxt_qplib_alloc_init_hwq(&cmdq->hwq, &hwq_attr)) { @@ -613,10 +964,6 @@ int bnxt_qplib_alloc_rcfw_channel(struct bnxt_qplib_res *res, if (!rcfw->crsqe_tbl) goto fail; - cmdq->cmdq_bitmap = bitmap_zalloc(rcfw->cmdq_depth, GFP_KERNEL); - if (!cmdq->cmdq_bitmap) - goto fail; - /* Allocate one extra to hold the QP1 entries */ rcfw->qp_tbl_size = qp_tbl_sz + 1; rcfw->qp_tbl = kcalloc(rcfw->qp_tbl_size, sizeof(struct bnxt_qplib_qp_node), @@ -624,6 +971,8 @@ int bnxt_qplib_alloc_rcfw_channel(struct bnxt_qplib_res *res, if (!rcfw->qp_tbl) goto fail; + rcfw->max_timeout = res->cctx->hwrm_cmd_max_timeout; + return 0; fail: @@ -636,6 +985,10 @@ void bnxt_qplib_rcfw_stop_irq(struct bnxt_qplib_rcfw *rcfw, bool kill) struct bnxt_qplib_creq_ctx *creq; creq = &rcfw->creq; + + if (!creq->requested) + return; + tasklet_disable(&creq->creq_tasklet); /* Mask h/w interrupts */ bnxt_qplib_ring_nq_db(&creq->creq_db.dbinfo, rcfw->res->cctx, false); @@ -644,17 +997,17 @@ void bnxt_qplib_rcfw_stop_irq(struct bnxt_qplib_rcfw *rcfw, bool kill) if (kill) tasklet_kill(&creq->creq_tasklet); - if (creq->requested) { - free_irq(creq->msix_vec, rcfw); - creq->requested = false; - } + free_irq(creq->msix_vec, rcfw); + kfree(creq->irq_name); + creq->irq_name = NULL; + creq->requested = false; + atomic_set(&rcfw->rcfw_intr_enabled, 0); } void bnxt_qplib_disable_rcfw_channel(struct bnxt_qplib_rcfw *rcfw) { struct bnxt_qplib_creq_ctx *creq; struct bnxt_qplib_cmdq_ctx *cmdq; - unsigned long indx; creq = &rcfw->creq; cmdq = &rcfw->cmdq; @@ -664,11 +1017,6 @@ void bnxt_qplib_disable_rcfw_channel(struct bnxt_qplib_rcfw *rcfw) iounmap(cmdq->cmdq_mbox.reg.bar_reg); iounmap(creq->creq_db.reg.bar_reg); - indx = find_first_bit(cmdq->cmdq_bitmap, rcfw->cmdq_depth); - if (indx != rcfw->cmdq_depth) - dev_err(&rcfw->pdev->dev, - "disabling RCFW with pending cmd-bit %lx\n", indx); - cmdq->cmdq_mbox.reg.bar_reg = NULL; creq->creq_db.reg.bar_reg = NULL; creq->aeq_handler = NULL; @@ -679,9 +1027,11 @@ int bnxt_qplib_rcfw_start_irq(struct bnxt_qplib_rcfw *rcfw, int msix_vector, bool need_init) { struct bnxt_qplib_creq_ctx *creq; + struct bnxt_qplib_res *res; int rc; creq = &rcfw->creq; + res = rcfw->res; if (creq->requested) return -EFAULT; @@ -691,24 +1041,32 @@ int bnxt_qplib_rcfw_start_irq(struct bnxt_qplib_rcfw *rcfw, int msix_vector, tasklet_setup(&creq->creq_tasklet, bnxt_qplib_service_creq); else tasklet_enable(&creq->creq_tasklet); + + creq->irq_name = kasprintf(GFP_KERNEL, "bnxt_re-creq@pci:%s", + pci_name(res->pdev)); + if (!creq->irq_name) + return -ENOMEM; rc = request_irq(creq->msix_vec, bnxt_qplib_creq_irq, 0, - "bnxt_qplib_creq", rcfw); - if (rc) + creq->irq_name, rcfw); + if (rc) { + kfree(creq->irq_name); + creq->irq_name = NULL; + tasklet_disable(&creq->creq_tasklet); return rc; + } creq->requested = true; - bnxt_qplib_ring_nq_db(&creq->creq_db.dbinfo, rcfw->res->cctx, true); + bnxt_qplib_ring_nq_db(&creq->creq_db.dbinfo, res->cctx, true); + atomic_inc(&rcfw->rcfw_intr_enabled); return 0; } -static int bnxt_qplib_map_cmdq_mbox(struct bnxt_qplib_rcfw *rcfw, bool is_vf) +static int bnxt_qplib_map_cmdq_mbox(struct bnxt_qplib_rcfw *rcfw) { struct bnxt_qplib_cmdq_mbox *mbox; resource_size_t bar_reg; struct pci_dev *pdev; - u16 prod_offt; - int rc = 0; pdev = rcfw->pdev; mbox = &rcfw->cmdq.cmdq_mbox; @@ -733,11 +1091,10 @@ static int bnxt_qplib_map_cmdq_mbox(struct bnxt_qplib_rcfw *rcfw, bool is_vf) return -ENOMEM; } - prod_offt = is_vf ? RCFW_VF_COMM_PROD_OFFSET : - RCFW_PF_COMM_PROD_OFFSET; - mbox->prod = (void __iomem *)(mbox->reg.bar_reg + prod_offt); + mbox->prod = (void __iomem *)(mbox->reg.bar_reg + + RCFW_PF_VF_COMM_PROD_OFFSET); mbox->db = (void __iomem *)(mbox->reg.bar_reg + RCFW_COMM_TRIG_OFFSET); - return rc; + return 0; } static int bnxt_qplib_map_creq_db(struct bnxt_qplib_rcfw *rcfw, u32 reg_offt) @@ -798,7 +1155,7 @@ static void bnxt_qplib_start_rcfw(struct bnxt_qplib_rcfw *rcfw) int bnxt_qplib_enable_rcfw_channel(struct bnxt_qplib_rcfw *rcfw, int msix_vector, - int cp_bar_reg_off, int virt_fn, + int cp_bar_reg_off, aeq_handler_t aeq_handler) { struct bnxt_qplib_cmdq_ctx *cmdq; @@ -818,7 +1175,7 @@ int bnxt_qplib_enable_rcfw_channel(struct bnxt_qplib_rcfw *rcfw, creq->stats.creq_func_event_processed = 0; creq->aeq_handler = aeq_handler; - rc = bnxt_qplib_map_cmdq_mbox(rcfw, virt_fn); + rc = bnxt_qplib_map_cmdq_mbox(rcfw); if (rc) return rc; @@ -834,6 +1191,7 @@ int bnxt_qplib_enable_rcfw_channel(struct bnxt_qplib_rcfw *rcfw, return rc; } + sema_init(&rcfw->rcfw_inflight, RCFW_CMD_NON_BLOCKING_SHADOW_QD); bnxt_qplib_start_rcfw(rcfw); return 0; diff --git a/drivers/infiniband/hw/bnxt_re/qplib_rcfw.h b/drivers/infiniband/hw/bnxt_re/qplib_rcfw.h index dd5651478bbb..7b31bee3e000 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_rcfw.h +++ b/drivers/infiniband/hw/bnxt_re/qplib_rcfw.h @@ -45,13 +45,13 @@ #define RCFW_COMM_PCI_BAR_REGION 0 #define RCFW_COMM_CONS_PCI_BAR_REGION 2 #define RCFW_COMM_BASE_OFFSET 0x600 -#define RCFW_PF_COMM_PROD_OFFSET 0xc -#define RCFW_VF_COMM_PROD_OFFSET 0xc +#define RCFW_PF_VF_COMM_PROD_OFFSET 0xc #define RCFW_COMM_TRIG_OFFSET 0x100 #define RCFW_COMM_SIZE 0x104 #define RCFW_DBR_PCI_BAR_REGION 2 #define RCFW_DBR_BASE_PAGE_SHIFT 12 +#define RCFW_FW_STALL_MAX_TIMEOUT 40 /* Cmdq contains a fix number of a 16-Byte slots */ struct bnxt_qplib_cmdqe { @@ -67,11 +67,12 @@ static inline void bnxt_qplib_rcfw_cmd_prep(struct cmdq_base *req, req->cmd_size = cmd_size; } +/* Shadow queue depth for non blocking command */ +#define RCFW_CMD_NON_BLOCKING_SHADOW_QD 64 #define RCFW_CMD_WAIT_TIME_MS 20000 /* 20 Seconds timeout */ /* CMDQ elements */ -#define BNXT_QPLIB_CMDQE_MAX_CNT_256 256 -#define BNXT_QPLIB_CMDQE_MAX_CNT_8192 8192 +#define BNXT_QPLIB_CMDQE_MAX_CNT 8192 #define BNXT_QPLIB_CMDQE_BYTES(depth) ((depth) * BNXT_QPLIB_CMDQE_UNITS) static inline u32 bnxt_qplib_cmdqe_npages(u32 depth) @@ -89,6 +90,26 @@ static inline u32 bnxt_qplib_cmdqe_page_size(u32 depth) return (bnxt_qplib_cmdqe_npages(depth) * PAGE_SIZE); } +/* Get the number of command units required for the req. The + * function returns correct value only if called before + * setting using bnxt_qplib_set_cmd_slots + */ +static inline u32 bnxt_qplib_get_cmd_slots(struct cmdq_base *req) +{ + u32 cmd_units = 0; + + if (HAS_TLV_HEADER(req)) { + struct roce_tlv *tlv_req = (struct roce_tlv *)req; + + cmd_units = tlv_req->total_size; + } else { + cmd_units = (req->cmd_size + BNXT_QPLIB_CMDQE_UNITS - 1) / + BNXT_QPLIB_CMDQE_UNITS; + } + + return cmd_units; +} + static inline u32 bnxt_qplib_set_cmd_slots(struct cmdq_base *req) { u32 cmd_byte = 0; @@ -106,11 +127,10 @@ static inline u32 bnxt_qplib_set_cmd_slots(struct cmdq_base *req) return cmd_byte; } -#define RCFW_MAX_COOKIE_VALUE 0x7FFF +#define RCFW_MAX_COOKIE_VALUE (BNXT_QPLIB_CMDQE_MAX_CNT - 1) #define RCFW_CMD_IS_BLOCKING 0x8000 -#define RCFW_BLOCKED_CMD_WAIT_COUNT 20000000UL /* 20 sec */ -#define HWRM_VERSION_RCFW_CMDQ_DEPTH_CHECK 0x1000900020011ULL +#define HWRM_VERSION_DEV_ATTR_MAX_DPI 0x1000A0000000DULL /* Crsq buf is 1024-Byte */ struct bnxt_qplib_crsbe { @@ -132,6 +152,12 @@ typedef int (*aeq_handler_t)(struct bnxt_qplib_rcfw *, void *, void *); struct bnxt_qplib_crsqe { struct creq_qp_event *resp; u32 req_size; + /* Free slots at the time of submission */ + u32 free_slots; + u8 opcode; + bool is_waiter_alive; + bool is_internal_cmd; + bool is_in_used; }; struct bnxt_qplib_rcfw_sbuf { @@ -149,7 +175,7 @@ struct bnxt_qplib_qp_node { #define FIRMWARE_INITIALIZED_FLAG (0) #define FIRMWARE_FIRST_FLAG (31) -#define FIRMWARE_TIMED_OUT (3) +#define FIRMWARE_STALL_DETECTED (3) #define ERR_DEVICE_DETACHED (4) struct bnxt_qplib_cmdq_mbox { @@ -163,7 +189,7 @@ struct bnxt_qplib_cmdq_ctx { struct bnxt_qplib_cmdq_mbox cmdq_mbox; wait_queue_head_t waitq; unsigned long flags; - unsigned long *cmdq_bitmap; + unsigned long last_seen; u32 seq_num; }; @@ -186,6 +212,7 @@ struct bnxt_qplib_creq_ctx { u16 ring_id; int msix_vec; bool requested; /*irq handler installed */ + char *irq_name; }; /* RCFW Communication Channels */ @@ -200,6 +227,11 @@ struct bnxt_qplib_rcfw { u64 oos_prev; u32 init_oos_stats; u32 cmdq_depth; + atomic_t rcfw_intr_enabled; + struct semaphore rcfw_inflight; + atomic_t timeout_send; + /* cached from chip cctx for quick reference in slow path */ + u16 max_timeout; }; struct bnxt_qplib_cmdqmsg { @@ -234,7 +266,7 @@ int bnxt_qplib_rcfw_start_irq(struct bnxt_qplib_rcfw *rcfw, int msix_vector, bool need_init); int bnxt_qplib_enable_rcfw_channel(struct bnxt_qplib_rcfw *rcfw, int msix_vector, - int cp_bar_reg_off, int virt_fn, + int cp_bar_reg_off, aeq_handler_t aeq_handler); struct bnxt_qplib_rcfw_sbuf *bnxt_qplib_rcfw_alloc_sbuf( diff --git a/drivers/infiniband/hw/bnxt_re/qplib_res.c b/drivers/infiniband/hw/bnxt_re/qplib_res.c index 81b0c5e879f9..5fd8f7c90bb0 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_res.c +++ b/drivers/infiniband/hw/bnxt_re/qplib_res.c @@ -696,44 +696,76 @@ static int bnxt_qplib_alloc_pd_tbl(struct bnxt_qplib_res *res, } /* DPIs */ -int bnxt_qplib_alloc_dpi(struct bnxt_qplib_dpi_tbl *dpit, - struct bnxt_qplib_dpi *dpi, - void *app) +int bnxt_qplib_alloc_dpi(struct bnxt_qplib_res *res, + struct bnxt_qplib_dpi *dpi, + void *app, u8 type) { + struct bnxt_qplib_dpi_tbl *dpit = &res->dpi_tbl; + struct bnxt_qplib_reg_desc *reg; u32 bit_num; + u64 umaddr; + + reg = &dpit->wcreg; + mutex_lock(&res->dpi_tbl_lock); bit_num = find_first_bit(dpit->tbl, dpit->max); - if (bit_num == dpit->max) + if (bit_num == dpit->max) { + mutex_unlock(&res->dpi_tbl_lock); return -ENOMEM; + } /* Found unused DPI */ clear_bit(bit_num, dpit->tbl); dpit->app_tbl[bit_num] = app; - dpi->dpi = bit_num; - dpi->dbr = dpit->dbr_bar_reg_iomem + (bit_num * PAGE_SIZE); - dpi->umdbr = dpit->unmapped_dbr + (bit_num * PAGE_SIZE); + dpi->bit = bit_num; + dpi->dpi = bit_num + (reg->offset - dpit->ucreg.offset) / PAGE_SIZE; + + umaddr = reg->bar_base + reg->offset + bit_num * PAGE_SIZE; + dpi->umdbr = umaddr; + + switch (type) { + case BNXT_QPLIB_DPI_TYPE_KERNEL: + /* privileged dbr was already mapped just initialize it. */ + dpi->umdbr = dpit->ucreg.bar_base + + dpit->ucreg.offset + bit_num * PAGE_SIZE; + dpi->dbr = dpit->priv_db; + dpi->dpi = dpi->bit; + break; + case BNXT_QPLIB_DPI_TYPE_WC: + dpi->dbr = ioremap_wc(umaddr, PAGE_SIZE); + break; + default: + dpi->dbr = ioremap(umaddr, PAGE_SIZE); + break; + } + dpi->type = type; + mutex_unlock(&res->dpi_tbl_lock); return 0; + } int bnxt_qplib_dealloc_dpi(struct bnxt_qplib_res *res, - struct bnxt_qplib_dpi_tbl *dpit, - struct bnxt_qplib_dpi *dpi) + struct bnxt_qplib_dpi *dpi) { - if (dpi->dpi >= dpit->max) { - dev_warn(&res->pdev->dev, "Invalid DPI? dpi = %d\n", dpi->dpi); - return -EINVAL; - } - if (test_and_set_bit(dpi->dpi, dpit->tbl)) { - dev_warn(&res->pdev->dev, "Freeing an unused DPI? dpi = %d\n", - dpi->dpi); + struct bnxt_qplib_dpi_tbl *dpit = &res->dpi_tbl; + + mutex_lock(&res->dpi_tbl_lock); + if (dpi->dpi && dpi->type != BNXT_QPLIB_DPI_TYPE_KERNEL) + pci_iounmap(res->pdev, dpi->dbr); + + if (test_and_set_bit(dpi->bit, dpit->tbl)) { + dev_warn(&res->pdev->dev, + "Freeing an unused DPI? dpi = %d, bit = %d\n", + dpi->dpi, dpi->bit); + mutex_unlock(&res->dpi_tbl_lock); return -EINVAL; } if (dpit->app_tbl) - dpit->app_tbl[dpi->dpi] = NULL; + dpit->app_tbl[dpi->bit] = NULL; memset(dpi, 0, sizeof(*dpi)); - + mutex_unlock(&res->dpi_tbl_lock); return 0; } @@ -742,52 +774,38 @@ static void bnxt_qplib_free_dpi_tbl(struct bnxt_qplib_res *res, { kfree(dpit->tbl); kfree(dpit->app_tbl); - if (dpit->dbr_bar_reg_iomem) - pci_iounmap(res->pdev, dpit->dbr_bar_reg_iomem); - memset(dpit, 0, sizeof(*dpit)); + dpit->tbl = NULL; + dpit->app_tbl = NULL; + dpit->max = 0; } -static int bnxt_qplib_alloc_dpi_tbl(struct bnxt_qplib_res *res, - struct bnxt_qplib_dpi_tbl *dpit, - u32 dbr_offset) +static int bnxt_qplib_alloc_dpi_tbl(struct bnxt_qplib_res *res, + struct bnxt_qplib_dev_attr *dev_attr) { - u32 dbr_bar_reg = RCFW_DBR_PCI_BAR_REGION; - resource_size_t bar_reg_base; - u32 dbr_len, bytes; - - if (dpit->dbr_bar_reg_iomem) { - dev_err(&res->pdev->dev, "DBR BAR region %d already mapped\n", - dbr_bar_reg); - return -EALREADY; - } - - bar_reg_base = pci_resource_start(res->pdev, dbr_bar_reg); - if (!bar_reg_base) { - dev_err(&res->pdev->dev, "BAR region %d resc start failed\n", - dbr_bar_reg); - return -ENOMEM; - } + struct bnxt_qplib_dpi_tbl *dpit; + struct bnxt_qplib_reg_desc *reg; + unsigned long bar_len; + u32 dbr_offset; + u32 bytes; - dbr_len = pci_resource_len(res->pdev, dbr_bar_reg) - dbr_offset; - if (!dbr_len || ((dbr_len & (PAGE_SIZE - 1)) != 0)) { - dev_err(&res->pdev->dev, "Invalid DBR length %d\n", dbr_len); - return -ENOMEM; - } + dpit = &res->dpi_tbl; + reg = &dpit->wcreg; - dpit->dbr_bar_reg_iomem = ioremap(bar_reg_base + dbr_offset, - dbr_len); - if (!dpit->dbr_bar_reg_iomem) { - dev_err(&res->pdev->dev, - "FP: DBR BAR region %d mapping failed\n", dbr_bar_reg); - return -ENOMEM; + if (!bnxt_qplib_is_chip_gen_p5(res->cctx)) { + /* Offest should come from L2 driver */ + dbr_offset = dev_attr->l2_db_size; + dpit->ucreg.offset = dbr_offset; + dpit->wcreg.offset = dbr_offset; } - dpit->unmapped_dbr = bar_reg_base + dbr_offset; - dpit->max = dbr_len / PAGE_SIZE; + bar_len = pci_resource_len(res->pdev, reg->bar_id); + dpit->max = (bar_len - reg->offset) / PAGE_SIZE; + if (dev_attr->max_dpi) + dpit->max = min_t(u32, dpit->max, dev_attr->max_dpi); - dpit->app_tbl = kcalloc(dpit->max, sizeof(void *), GFP_KERNEL); + dpit->app_tbl = kcalloc(dpit->max, sizeof(void *), GFP_KERNEL); if (!dpit->app_tbl) - goto unmap_io; + return -ENOMEM; bytes = dpit->max >> 3; if (!bytes) @@ -797,17 +815,14 @@ static int bnxt_qplib_alloc_dpi_tbl(struct bnxt_qplib_res *res, if (!dpit->tbl) { kfree(dpit->app_tbl); dpit->app_tbl = NULL; - goto unmap_io; + return -ENOMEM; } memset((u8 *)dpit->tbl, 0xFF, bytes); + dpit->priv_db = dpit->ucreg.bar_reg + dpit->ucreg.offset; return 0; -unmap_io: - pci_iounmap(res->pdev, dpit->dbr_bar_reg_iomem); - dpit->dbr_bar_reg_iomem = NULL; - return -ENOMEM; } /* Stats */ @@ -874,7 +889,7 @@ int bnxt_qplib_alloc_res(struct bnxt_qplib_res *res, struct pci_dev *pdev, if (rc) goto fail; - rc = bnxt_qplib_alloc_dpi_tbl(res, &res->dpi_tbl, dev_attr->l2_db_size); + rc = bnxt_qplib_alloc_dpi_tbl(res, dev_attr); if (rc) goto fail; @@ -884,6 +899,46 @@ fail: return rc; } +void bnxt_qplib_unmap_db_bar(struct bnxt_qplib_res *res) +{ + struct bnxt_qplib_reg_desc *reg; + + reg = &res->dpi_tbl.ucreg; + if (reg->bar_reg) + pci_iounmap(res->pdev, reg->bar_reg); + reg->bar_reg = NULL; + reg->bar_base = 0; + reg->len = 0; + reg->bar_id = 0; +} + +int bnxt_qplib_map_db_bar(struct bnxt_qplib_res *res) +{ + struct bnxt_qplib_reg_desc *ucreg; + struct bnxt_qplib_reg_desc *wcreg; + + wcreg = &res->dpi_tbl.wcreg; + wcreg->bar_id = RCFW_DBR_PCI_BAR_REGION; + wcreg->bar_base = pci_resource_start(res->pdev, wcreg->bar_id); + + ucreg = &res->dpi_tbl.ucreg; + ucreg->bar_id = RCFW_DBR_PCI_BAR_REGION; + ucreg->bar_base = pci_resource_start(res->pdev, ucreg->bar_id); + ucreg->len = ucreg->offset + PAGE_SIZE; + if (!ucreg->len || ((ucreg->len & (PAGE_SIZE - 1)) != 0)) { + dev_err(&res->pdev->dev, "QPLIB: invalid dbr length %d", + (int)ucreg->len); + return -EINVAL; + } + ucreg->bar_reg = ioremap(ucreg->bar_base, ucreg->len); + if (!ucreg->bar_reg) { + dev_err(&res->pdev->dev, "privileged dpi map failed!"); + return -ENOMEM; + } + + return 0; +} + int bnxt_qplib_determine_atomics(struct pci_dev *dev) { int comp; diff --git a/drivers/infiniband/hw/bnxt_re/qplib_res.h b/drivers/infiniband/hw/bnxt_re/qplib_res.h index 982e2c96dac2..d850a553821e 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_res.h +++ b/drivers/infiniband/hw/bnxt_re/qplib_res.h @@ -47,7 +47,7 @@ extern const struct bnxt_qplib_gid bnxt_qplib_gid_zero; struct bnxt_qplib_drv_modes { u8 wqe_mode; - /* Other modes to follow here */ + bool db_push; }; struct bnxt_qplib_chip_ctx { @@ -55,9 +55,14 @@ struct bnxt_qplib_chip_ctx { u8 chip_rev; u8 chip_metal; u16 hw_stats_size; + u16 hwrm_cmd_max_timeout; struct bnxt_qplib_drv_modes modes; + u64 hwrm_intf_ver; }; +#define BNXT_QPLIB_DBR_PF_DB_OFFSET 0x10000 +#define BNXT_QPLIB_DBR_VF_DB_OFFSET 0x4000 + #define PTR_CNT_PER_PG (PAGE_SIZE / sizeof(void *)) #define PTR_MAX_IDX_PER_PG (PTR_CNT_PER_PG - 1) #define PTR_PG(x) (((x) & ~PTR_MAX_IDX_PER_PG) / PTR_CNT_PER_PG) @@ -109,6 +114,7 @@ enum bnxt_qplib_hwrm_pg_size { struct bnxt_qplib_reg_desc { u8 bar_id; resource_size_t bar_base; + unsigned long offset; void __iomem *bar_reg; size_t len; }; @@ -185,18 +191,27 @@ struct bnxt_qplib_sgid_tbl { u8 *vlan; }; +enum { + BNXT_QPLIB_DPI_TYPE_KERNEL = 0, + BNXT_QPLIB_DPI_TYPE_UC = 1, + BNXT_QPLIB_DPI_TYPE_WC = 2 +}; + struct bnxt_qplib_dpi { u32 dpi; + u32 bit; void __iomem *dbr; u64 umdbr; + u8 type; }; struct bnxt_qplib_dpi_tbl { void **app_tbl; unsigned long *tbl; u16 max; - void __iomem *dbr_bar_reg_iomem; - u64 unmapped_dbr; + struct bnxt_qplib_reg_desc ucreg; /* Hold entire DB bar. */ + struct bnxt_qplib_reg_desc wcreg; + void __iomem *priv_db; }; struct bnxt_qplib_stats { @@ -241,7 +256,6 @@ struct bnxt_qplib_ctx { struct bnxt_qplib_tqm_ctx tqm_ctx; struct bnxt_qplib_stats stats; struct bnxt_qplib_vf_res vf_res; - u64 hwrm_intf_ver; }; struct bnxt_qplib_res { @@ -253,6 +267,8 @@ struct bnxt_qplib_res { struct bnxt_qplib_pd_tbl pd_tbl; struct bnxt_qplib_sgid_tbl sgid_tbl; struct bnxt_qplib_dpi_tbl dpi_tbl; + /* To protect the dpi table bit map */ + struct mutex dpi_tbl_lock; bool prio; bool is_vf; }; @@ -344,11 +360,10 @@ int bnxt_qplib_alloc_pd(struct bnxt_qplib_pd_tbl *pd_tbl, int bnxt_qplib_dealloc_pd(struct bnxt_qplib_res *res, struct bnxt_qplib_pd_tbl *pd_tbl, struct bnxt_qplib_pd *pd); -int bnxt_qplib_alloc_dpi(struct bnxt_qplib_dpi_tbl *dpit, - struct bnxt_qplib_dpi *dpi, - void *app); +int bnxt_qplib_alloc_dpi(struct bnxt_qplib_res *res, + struct bnxt_qplib_dpi *dpi, + void *app, u8 type); int bnxt_qplib_dealloc_dpi(struct bnxt_qplib_res *res, - struct bnxt_qplib_dpi_tbl *dpi_tbl, struct bnxt_qplib_dpi *dpi); void bnxt_qplib_cleanup_res(struct bnxt_qplib_res *res); int bnxt_qplib_init_res(struct bnxt_qplib_res *res); @@ -361,6 +376,9 @@ void bnxt_qplib_free_ctx(struct bnxt_qplib_res *res, int bnxt_qplib_alloc_ctx(struct bnxt_qplib_res *res, struct bnxt_qplib_ctx *ctx, bool virt_fn, bool is_p5); +int bnxt_qplib_map_db_bar(struct bnxt_qplib_res *res); +void bnxt_qplib_unmap_db_bar(struct bnxt_qplib_res *res); + int bnxt_qplib_determine_atomics(struct pci_dev *dev); static inline void bnxt_qplib_hwq_incr_prod(struct bnxt_qplib_hwq *hwq, u32 cnt) diff --git a/drivers/infiniband/hw/bnxt_re/qplib_sp.c b/drivers/infiniband/hw/bnxt_re/qplib_sp.c index b967a17a44be..ab45f9d4bb02 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_sp.c +++ b/drivers/infiniband/hw/bnxt_re/qplib_sp.c @@ -170,6 +170,9 @@ int bnxt_qplib_get_dev_attr(struct bnxt_qplib_rcfw *rcfw, attr->tqm_alloc_reqs[i * 4 + 3] = *(++tqm_alloc); } + if (rcfw->res->cctx->hwrm_intf_ver >= HWRM_VERSION_DEV_ATTR_MAX_DPI) + attr->max_dpi = le32_to_cpu(sb->max_dpi); + attr->is_atomic = bnxt_qplib_is_atomic_cap(rcfw); bail: bnxt_qplib_rcfw_free_sbuf(rcfw, sbuf); @@ -233,10 +236,6 @@ int bnxt_qplib_del_sgid(struct bnxt_qplib_sgid_tbl *sgid_tbl, struct bnxt_qplib_rcfw *rcfw = res->rcfw; int index; - if (!sgid_tbl) { - dev_err(&res->pdev->dev, "SGID table not allocated\n"); - return -EINVAL; - } /* Do we need a sgid_lock here? */ if (!sgid_tbl->active) { dev_err(&res->pdev->dev, "SGID table has no active entries\n"); @@ -297,10 +296,6 @@ int bnxt_qplib_add_sgid(struct bnxt_qplib_sgid_tbl *sgid_tbl, struct bnxt_qplib_rcfw *rcfw = res->rcfw; int i, free_idx; - if (!sgid_tbl) { - dev_err(&res->pdev->dev, "SGID table not allocated\n"); - return -EINVAL; - } /* Do we need a sgid_lock here? */ if (sgid_tbl->active == sgid_tbl->max) { dev_err(&res->pdev->dev, "SGID table is full\n"); @@ -468,13 +463,14 @@ int bnxt_qplib_create_ah(struct bnxt_qplib_res *res, struct bnxt_qplib_ah *ah, return 0; } -void bnxt_qplib_destroy_ah(struct bnxt_qplib_res *res, struct bnxt_qplib_ah *ah, - bool block) +int bnxt_qplib_destroy_ah(struct bnxt_qplib_res *res, struct bnxt_qplib_ah *ah, + bool block) { struct bnxt_qplib_rcfw *rcfw = res->rcfw; struct creq_destroy_ah_resp resp = {}; struct bnxt_qplib_cmdqmsg msg = {}; struct cmdq_destroy_ah req = {}; + int rc; /* Clean up the AH table in the device */ bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req, @@ -485,7 +481,8 @@ void bnxt_qplib_destroy_ah(struct bnxt_qplib_res *res, struct bnxt_qplib_ah *ah, bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req), sizeof(resp), block); - bnxt_qplib_rcfw_send_message(rcfw, &msg); + rc = bnxt_qplib_rcfw_send_message(rcfw, &msg); + return rc; } /* MRW */ diff --git a/drivers/infiniband/hw/bnxt_re/qplib_sp.h b/drivers/infiniband/hw/bnxt_re/qplib_sp.h index 5de874659cdf..264ef3cedc45 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_sp.h +++ b/drivers/infiniband/hw/bnxt_re/qplib_sp.h @@ -72,6 +72,7 @@ struct bnxt_qplib_dev_attr { u8 tqm_alloc_reqs[MAX_TQM_ALLOC_REQ]; bool is_atomic; u16 dev_cap_flags; + u32 max_dpi; }; struct bnxt_qplib_pd { @@ -327,8 +328,8 @@ int bnxt_qplib_set_func_resources(struct bnxt_qplib_res *res, struct bnxt_qplib_ctx *ctx); int bnxt_qplib_create_ah(struct bnxt_qplib_res *res, struct bnxt_qplib_ah *ah, bool block); -void bnxt_qplib_destroy_ah(struct bnxt_qplib_res *res, struct bnxt_qplib_ah *ah, - bool block); +int bnxt_qplib_destroy_ah(struct bnxt_qplib_res *res, struct bnxt_qplib_ah *ah, + bool block); int bnxt_qplib_alloc_mrw(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mrw); int bnxt_qplib_dereg_mrw(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mrw, |