| Message ID | 0-v1-b37437a73f35+49c-bnxt_re_dma_block_jgg@nvidia.com (mailing list archive) |
|---|---|
| State | Accepted |
| Delegated to: | Jason Gunthorpe |
| Headers | show |
| Series | RDMA/bnxt_re: Use rdma_umem_for_each_dma_block() | expand |
On Wed, Sep 30, 2020 at 5:54 AM Jason Gunthorpe <jgg@nvidia.com> wrote: > > This driver is taking the SGL out of the umem and passing it through a > struct bnxt_qplib_sg_info. Instead of passing the SGL pass the umem and > then use rdma_umem_for_each_dma_block() directly. > > Move the calls of ib_umem_num_dma_blocks() closer to their actual point of > use, npages is only set for non-umem pbl flows. > > Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Acked-by: Selvin Xavier <selvin.xavier@broadcom.com> Tested-by: Selvin Xavier <selvin.xavier@broadcom.com> Thanks > --- > drivers/infiniband/hw/bnxt_re/ib_verbs.c | 18 +++----------- > drivers/infiniband/hw/bnxt_re/qplib_res.c | 30 +++++++++++++---------- > drivers/infiniband/hw/bnxt_re/qplib_res.h | 3 +-- > 3 files changed, 22 insertions(+), 29 deletions(-) > > This is part of the umem cleanup. It is a bit complicated, would be good for > someone to check it. Thanks > > diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c > index a0e8d93595d8e8..e2707b27c9500c 100644 > --- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c > +++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c > @@ -940,9 +940,7 @@ static int bnxt_re_init_user_qp(struct bnxt_re_dev *rdev, struct bnxt_re_pd *pd, > return PTR_ERR(umem); > > qp->sumem = umem; > - qplib_qp->sq.sg_info.sghead = umem->sg_head.sgl; > - qplib_qp->sq.sg_info.npages = ib_umem_num_dma_blocks(umem, PAGE_SIZE); > - qplib_qp->sq.sg_info.nmap = umem->nmap; > + qplib_qp->sq.sg_info.umem = umem; > qplib_qp->sq.sg_info.pgsize = PAGE_SIZE; > qplib_qp->sq.sg_info.pgshft = PAGE_SHIFT; > qplib_qp->qp_handle = ureq.qp_handle; > @@ -955,10 +953,7 @@ static int bnxt_re_init_user_qp(struct bnxt_re_dev *rdev, struct bnxt_re_pd *pd, > if (IS_ERR(umem)) > goto rqfail; > qp->rumem = umem; > - qplib_qp->rq.sg_info.sghead = umem->sg_head.sgl; > - qplib_qp->rq.sg_info.npages = > - ib_umem_num_dma_blocks(umem, PAGE_SIZE); > - qplib_qp->rq.sg_info.nmap = umem->nmap; > + qplib_qp->rq.sg_info.umem = umem; > qplib_qp->rq.sg_info.pgsize = PAGE_SIZE; > qplib_qp->rq.sg_info.pgshft = PAGE_SHIFT; > } > @@ -1612,9 +1607,7 @@ static int bnxt_re_init_user_srq(struct bnxt_re_dev *rdev, > return PTR_ERR(umem); > > srq->umem = umem; > - qplib_srq->sg_info.sghead = umem->sg_head.sgl; > - qplib_srq->sg_info.npages = ib_umem_num_dma_blocks(umem, PAGE_SIZE); > - qplib_srq->sg_info.nmap = umem->nmap; > + qplib_srq->sg_info.umem = umem; > qplib_srq->sg_info.pgsize = PAGE_SIZE; > qplib_srq->sg_info.pgshft = PAGE_SHIFT; > qplib_srq->srq_handle = ureq.srq_handle; > @@ -2865,10 +2858,7 @@ int bnxt_re_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr, > rc = PTR_ERR(cq->umem); > goto fail; > } > - cq->qplib_cq.sg_info.sghead = cq->umem->sg_head.sgl; > - cq->qplib_cq.sg_info.npages = > - ib_umem_num_dma_blocks(cq->umem, PAGE_SIZE); > - cq->qplib_cq.sg_info.nmap = cq->umem->nmap; > + cq->qplib_cq.sg_info.umem = cq->umem; > cq->qplib_cq.dpi = &uctx->dpi; > } else { > cq->max_cql = min_t(u32, entries, MAX_CQL_PER_POLL); > diff --git a/drivers/infiniband/hw/bnxt_re/qplib_res.c b/drivers/infiniband/hw/bnxt_re/qplib_res.c > index 7efa6e5dce6282..fa7878336100ac 100644 > --- a/drivers/infiniband/hw/bnxt_re/qplib_res.c > +++ b/drivers/infiniband/hw/bnxt_re/qplib_res.c > @@ -45,6 +45,9 @@ > #include <linux/dma-mapping.h> > #include <linux/if_vlan.h> > #include <linux/vmalloc.h> > +#include <rdma/ib_verbs.h> > +#include <rdma/ib_umem.h> > + > #include "roce_hsi.h" > #include "qplib_res.h" > #include "qplib_sp.h" > @@ -87,12 +90,11 @@ static void __free_pbl(struct bnxt_qplib_res *res, struct bnxt_qplib_pbl *pbl, > static void bnxt_qplib_fill_user_dma_pages(struct bnxt_qplib_pbl *pbl, > struct bnxt_qplib_sg_info *sginfo) > { > - struct scatterlist *sghead = sginfo->sghead; > - struct sg_dma_page_iter sg_iter; > + struct ib_block_iter biter; > int i = 0; > > - for_each_sg_dma_page(sghead, &sg_iter, sginfo->nmap, 0) { > - pbl->pg_map_arr[i] = sg_page_iter_dma_address(&sg_iter); > + rdma_umem_for_each_dma_block(sginfo->umem, &biter, sginfo->pgsize) { > + pbl->pg_map_arr[i] = rdma_block_iter_dma_address(&biter); > pbl->pg_arr[i] = NULL; > pbl->pg_count++; > i++; > @@ -104,15 +106,16 @@ static int __alloc_pbl(struct bnxt_qplib_res *res, > struct bnxt_qplib_sg_info *sginfo) > { > struct pci_dev *pdev = res->pdev; > - struct scatterlist *sghead; > bool is_umem = false; > u32 pages; > int i; > > if (sginfo->nopte) > return 0; > - pages = sginfo->npages; > - sghead = sginfo->sghead; > + if (sginfo->umem) > + pages = ib_umem_num_dma_blocks(sginfo->umem, sginfo->pgsize); > + else > + pages = sginfo->npages; > /* page ptr arrays */ > pbl->pg_arr = vmalloc(pages * sizeof(void *)); > if (!pbl->pg_arr) > @@ -127,7 +130,7 @@ static int __alloc_pbl(struct bnxt_qplib_res *res, > pbl->pg_count = 0; > pbl->pg_size = sginfo->pgsize; > > - if (!sghead) { > + if (!sginfo->umem) { > for (i = 0; i < pages; i++) { > pbl->pg_arr[i] = dma_alloc_coherent(&pdev->dev, > pbl->pg_size, > @@ -183,14 +186,12 @@ int bnxt_qplib_alloc_init_hwq(struct bnxt_qplib_hwq *hwq, > struct bnxt_qplib_sg_info sginfo = {}; > u32 depth, stride, npbl, npde; > dma_addr_t *src_phys_ptr, **dst_virt_ptr; > - struct scatterlist *sghead = NULL; > struct bnxt_qplib_res *res; > struct pci_dev *pdev; > int i, rc, lvl; > > res = hwq_attr->res; > pdev = res->pdev; > - sghead = hwq_attr->sginfo->sghead; > pg_size = hwq_attr->sginfo->pgsize; > hwq->level = PBL_LVL_MAX; > > @@ -204,7 +205,7 @@ int bnxt_qplib_alloc_init_hwq(struct bnxt_qplib_hwq *hwq, > aux_pages++; > } > > - if (!sghead) { > + if (!hwq_attr->sginfo->umem) { > hwq->is_user = false; > npages = (depth * stride) / pg_size + aux_pages; > if ((depth * stride) % pg_size) > @@ -213,11 +214,14 @@ int bnxt_qplib_alloc_init_hwq(struct bnxt_qplib_hwq *hwq, > return -EINVAL; > hwq_attr->sginfo->npages = npages; > } else { > + unsigned long sginfo_num_pages = ib_umem_num_dma_blocks( > + hwq_attr->sginfo->umem, hwq_attr->sginfo->pgsize); > + > hwq->is_user = true; > - npages = hwq_attr->sginfo->npages; > + npages = sginfo_num_pages; > npages = (npages * PAGE_SIZE) / > BIT_ULL(hwq_attr->sginfo->pgshft); > - if ((hwq_attr->sginfo->npages * PAGE_SIZE) % > + if ((sginfo_num_pages * PAGE_SIZE) % > BIT_ULL(hwq_attr->sginfo->pgshft)) > if (!npages) > npages++; > diff --git a/drivers/infiniband/hw/bnxt_re/qplib_res.h b/drivers/infiniband/hw/bnxt_re/qplib_res.h > index 9da470d1e4a3c2..ceb94db20a786a 100644 > --- a/drivers/infiniband/hw/bnxt_re/qplib_res.h > +++ b/drivers/infiniband/hw/bnxt_re/qplib_res.h > @@ -126,8 +126,7 @@ struct bnxt_qplib_pbl { > }; > > struct bnxt_qplib_sg_info { > - struct scatterlist *sghead; > - u32 nmap; > + struct ib_umem *umem; > u32 npages; > u32 pgshft; > u32 pgsize; > -- > 2.28.0 >
On Tue, Sep 29, 2020 at 09:24:35PM -0300, Jason Gunthorpe wrote: > This driver is taking the SGL out of the umem and passing it through a > struct bnxt_qplib_sg_info. Instead of passing the SGL pass the umem and > then use rdma_umem_for_each_dma_block() directly. > > Move the calls of ib_umem_num_dma_blocks() closer to their actual point of > use, npages is only set for non-umem pbl flows. > > Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> > Acked-by: Selvin Xavier <selvin.xavier@broadcom.com> > Tested-by: Selvin Xavier <selvin.xavier@broadcom.com> > --- > drivers/infiniband/hw/bnxt_re/ib_verbs.c | 18 +++----------- > drivers/infiniband/hw/bnxt_re/qplib_res.c | 30 +++++++++++++---------- > drivers/infiniband/hw/bnxt_re/qplib_res.h | 3 +-- > 3 files changed, 22 insertions(+), 29 deletions(-) Applied to for-next, thanks for checking Jason
diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c index a0e8d93595d8e8..e2707b27c9500c 100644 --- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c +++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c @@ -940,9 +940,7 @@ static int bnxt_re_init_user_qp(struct bnxt_re_dev *rdev, struct bnxt_re_pd *pd, return PTR_ERR(umem); qp->sumem = umem; - qplib_qp->sq.sg_info.sghead = umem->sg_head.sgl; - qplib_qp->sq.sg_info.npages = ib_umem_num_dma_blocks(umem, PAGE_SIZE); - qplib_qp->sq.sg_info.nmap = umem->nmap; + qplib_qp->sq.sg_info.umem = umem; qplib_qp->sq.sg_info.pgsize = PAGE_SIZE; qplib_qp->sq.sg_info.pgshft = PAGE_SHIFT; qplib_qp->qp_handle = ureq.qp_handle; @@ -955,10 +953,7 @@ static int bnxt_re_init_user_qp(struct bnxt_re_dev *rdev, struct bnxt_re_pd *pd, if (IS_ERR(umem)) goto rqfail; qp->rumem = umem; - qplib_qp->rq.sg_info.sghead = umem->sg_head.sgl; - qplib_qp->rq.sg_info.npages = - ib_umem_num_dma_blocks(umem, PAGE_SIZE); - qplib_qp->rq.sg_info.nmap = umem->nmap; + qplib_qp->rq.sg_info.umem = umem; qplib_qp->rq.sg_info.pgsize = PAGE_SIZE; qplib_qp->rq.sg_info.pgshft = PAGE_SHIFT; } @@ -1612,9 +1607,7 @@ static int bnxt_re_init_user_srq(struct bnxt_re_dev *rdev, return PTR_ERR(umem); srq->umem = umem; - qplib_srq->sg_info.sghead = umem->sg_head.sgl; - qplib_srq->sg_info.npages = ib_umem_num_dma_blocks(umem, PAGE_SIZE); - qplib_srq->sg_info.nmap = umem->nmap; + qplib_srq->sg_info.umem = umem; qplib_srq->sg_info.pgsize = PAGE_SIZE; qplib_srq->sg_info.pgshft = PAGE_SHIFT; qplib_srq->srq_handle = ureq.srq_handle; @@ -2865,10 +2858,7 @@ int bnxt_re_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr, rc = PTR_ERR(cq->umem); goto fail; } - cq->qplib_cq.sg_info.sghead = cq->umem->sg_head.sgl; - cq->qplib_cq.sg_info.npages = - ib_umem_num_dma_blocks(cq->umem, PAGE_SIZE); - cq->qplib_cq.sg_info.nmap = cq->umem->nmap; + cq->qplib_cq.sg_info.umem = cq->umem; cq->qplib_cq.dpi = &uctx->dpi; } else { cq->max_cql = min_t(u32, entries, MAX_CQL_PER_POLL); diff --git a/drivers/infiniband/hw/bnxt_re/qplib_res.c b/drivers/infiniband/hw/bnxt_re/qplib_res.c index 7efa6e5dce6282..fa7878336100ac 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_res.c +++ b/drivers/infiniband/hw/bnxt_re/qplib_res.c @@ -45,6 +45,9 @@ #include <linux/dma-mapping.h> #include <linux/if_vlan.h> #include <linux/vmalloc.h> +#include <rdma/ib_verbs.h> +#include <rdma/ib_umem.h> + #include "roce_hsi.h" #include "qplib_res.h" #include "qplib_sp.h" @@ -87,12 +90,11 @@ static void __free_pbl(struct bnxt_qplib_res *res, struct bnxt_qplib_pbl *pbl, static void bnxt_qplib_fill_user_dma_pages(struct bnxt_qplib_pbl *pbl, struct bnxt_qplib_sg_info *sginfo) { - struct scatterlist *sghead = sginfo->sghead; - struct sg_dma_page_iter sg_iter; + struct ib_block_iter biter; int i = 0; - for_each_sg_dma_page(sghead, &sg_iter, sginfo->nmap, 0) { - pbl->pg_map_arr[i] = sg_page_iter_dma_address(&sg_iter); + rdma_umem_for_each_dma_block(sginfo->umem, &biter, sginfo->pgsize) { + pbl->pg_map_arr[i] = rdma_block_iter_dma_address(&biter); pbl->pg_arr[i] = NULL; pbl->pg_count++; i++; @@ -104,15 +106,16 @@ static int __alloc_pbl(struct bnxt_qplib_res *res, struct bnxt_qplib_sg_info *sginfo) { struct pci_dev *pdev = res->pdev; - struct scatterlist *sghead; bool is_umem = false; u32 pages; int i; if (sginfo->nopte) return 0; - pages = sginfo->npages; - sghead = sginfo->sghead; + if (sginfo->umem) + pages = ib_umem_num_dma_blocks(sginfo->umem, sginfo->pgsize); + else + pages = sginfo->npages; /* page ptr arrays */ pbl->pg_arr = vmalloc(pages * sizeof(void *)); if (!pbl->pg_arr) @@ -127,7 +130,7 @@ static int __alloc_pbl(struct bnxt_qplib_res *res, pbl->pg_count = 0; pbl->pg_size = sginfo->pgsize; - if (!sghead) { + if (!sginfo->umem) { for (i = 0; i < pages; i++) { pbl->pg_arr[i] = dma_alloc_coherent(&pdev->dev, pbl->pg_size, @@ -183,14 +186,12 @@ int bnxt_qplib_alloc_init_hwq(struct bnxt_qplib_hwq *hwq, struct bnxt_qplib_sg_info sginfo = {}; u32 depth, stride, npbl, npde; dma_addr_t *src_phys_ptr, **dst_virt_ptr; - struct scatterlist *sghead = NULL; struct bnxt_qplib_res *res; struct pci_dev *pdev; int i, rc, lvl; res = hwq_attr->res; pdev = res->pdev; - sghead = hwq_attr->sginfo->sghead; pg_size = hwq_attr->sginfo->pgsize; hwq->level = PBL_LVL_MAX; @@ -204,7 +205,7 @@ int bnxt_qplib_alloc_init_hwq(struct bnxt_qplib_hwq *hwq, aux_pages++; } - if (!sghead) { + if (!hwq_attr->sginfo->umem) { hwq->is_user = false; npages = (depth * stride) / pg_size + aux_pages; if ((depth * stride) % pg_size) @@ -213,11 +214,14 @@ int bnxt_qplib_alloc_init_hwq(struct bnxt_qplib_hwq *hwq, return -EINVAL; hwq_attr->sginfo->npages = npages; } else { + unsigned long sginfo_num_pages = ib_umem_num_dma_blocks( + hwq_attr->sginfo->umem, hwq_attr->sginfo->pgsize); + hwq->is_user = true; - npages = hwq_attr->sginfo->npages; + npages = sginfo_num_pages; npages = (npages * PAGE_SIZE) / BIT_ULL(hwq_attr->sginfo->pgshft); - if ((hwq_attr->sginfo->npages * PAGE_SIZE) % + if ((sginfo_num_pages * PAGE_SIZE) % BIT_ULL(hwq_attr->sginfo->pgshft)) if (!npages) npages++; diff --git a/drivers/infiniband/hw/bnxt_re/qplib_res.h b/drivers/infiniband/hw/bnxt_re/qplib_res.h index 9da470d1e4a3c2..ceb94db20a786a 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_res.h +++ b/drivers/infiniband/hw/bnxt_re/qplib_res.h @@ -126,8 +126,7 @@ struct bnxt_qplib_pbl { }; struct bnxt_qplib_sg_info { - struct scatterlist *sghead; - u32 nmap; + struct ib_umem *umem; u32 npages; u32 pgshft; u32 pgsize;
This driver is taking the SGL out of the umem and passing it through a struct bnxt_qplib_sg_info. Instead of passing the SGL pass the umem and then use rdma_umem_for_each_dma_block() directly. Move the calls of ib_umem_num_dma_blocks() closer to their actual point of use, npages is only set for non-umem pbl flows. Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> --- drivers/infiniband/hw/bnxt_re/ib_verbs.c | 18 +++----------- drivers/infiniband/hw/bnxt_re/qplib_res.c | 30 +++++++++++++---------- drivers/infiniband/hw/bnxt_re/qplib_res.h | 3 +-- 3 files changed, 22 insertions(+), 29 deletions(-) This is part of the umem cleanup. It is a bit complicated, would be good for someone to check it. Thanks