diff options
Diffstat (limited to 'io_uring')
| -rw-r--r-- | io_uring/kbuf.c | 2 | ||||
| -rw-r--r-- | io_uring/rsrc.c | 109 | ||||
| -rw-r--r-- | io_uring/zcrx.c | 56 | ||||
| -rw-r--r-- | io_uring/zcrx.h | 6 |
4 files changed, 107 insertions, 66 deletions
diff --git a/io_uring/kbuf.c b/io_uring/kbuf.c index 098109259671..953d5e742569 100644 --- a/io_uring/kbuf.c +++ b/io_uring/kbuf.c @@ -504,6 +504,8 @@ int io_provide_buffers_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe p->nbufs = tmp; p->addr = READ_ONCE(sqe->addr); p->len = READ_ONCE(sqe->len); + if (!p->len) + return -EINVAL; if (check_mul_overflow((unsigned long)p->len, (unsigned long)p->nbufs, &size)) diff --git a/io_uring/rsrc.c b/io_uring/rsrc.c index 5e64a8bb30a4..f80a77c4973f 100644 --- a/io_uring/rsrc.c +++ b/io_uring/rsrc.c @@ -175,6 +175,18 @@ void io_rsrc_cache_free(struct io_ring_ctx *ctx) io_alloc_cache_free(&ctx->imu_cache, kfree); } +static void io_clear_table_tags(struct io_rsrc_data *data) +{ + int i; + + for (i = 0; i < data->nr; i++) { + struct io_rsrc_node *node = data->nodes[i]; + + if (node) + node->tag = 0; + } +} + __cold void io_rsrc_data_free(struct io_ring_ctx *ctx, struct io_rsrc_data *data) { @@ -583,6 +595,7 @@ int io_sqe_files_register(struct io_ring_ctx *ctx, void __user *arg, io_file_table_set_alloc_range(ctx, 0, ctx->file_table.data.nr); return 0; fail: + io_clear_table_tags(&ctx->file_table.data); io_sqe_files_unregister(ctx); return ret; } @@ -902,8 +915,10 @@ int io_sqe_buffers_register(struct io_ring_ctx *ctx, void __user *arg, } ctx->buf_table = data; - if (ret) + if (ret) { + io_clear_table_tags(&ctx->buf_table); io_sqe_buffers_unregister(ctx); + } return ret; } @@ -1017,10 +1032,33 @@ static int validate_fixed_range(u64 buf_addr, size_t len, return 0; } +static int io_import_kbuf(int ddir, struct iov_iter *iter, + struct io_mapped_ubuf *imu, size_t len, size_t offset) +{ + size_t count = len + offset; + + iov_iter_bvec(iter, ddir, imu->bvec, imu->nr_bvecs, count); + iov_iter_advance(iter, offset); + + if (count < imu->len) { + const struct bio_vec *bvec = iter->bvec; + + while (len > bvec->bv_len) { + len -= bvec->bv_len; + bvec++; + } + iter->nr_segs = 1 + bvec - iter->bvec; + } + return 0; +} + static int io_import_fixed(int ddir, struct iov_iter *iter, struct io_mapped_ubuf *imu, u64 buf_addr, size_t len) { + const struct bio_vec *bvec; + size_t folio_mask; + unsigned nr_segs; size_t offset; int ret; @@ -1032,56 +1070,35 @@ static int io_import_fixed(int ddir, struct iov_iter *iter, if (!(imu->dir & (1 << ddir))) return -EFAULT; - /* - * Might not be a start of buffer, set size appropriately - * and advance us to the beginning. - */ offset = buf_addr - imu->ubuf; - iov_iter_bvec(iter, ddir, imu->bvec, imu->nr_bvecs, offset + len); - if (offset) { - /* - * Don't use iov_iter_advance() here, as it's really slow for - * using the latter parts of a big fixed buffer - it iterates - * over each segment manually. We can cheat a bit here for user - * registered nodes, because we know that: - * - * 1) it's a BVEC iter, we set it up - * 2) all bvecs are the same in size, except potentially the - * first and last bvec - * - * So just find our index, and adjust the iterator afterwards. - * If the offset is within the first bvec (or the whole first - * bvec, just use iov_iter_advance(). This makes it easier - * since we can just skip the first segment, which may not - * be folio_size aligned. - */ - const struct bio_vec *bvec = imu->bvec; - - /* - * Kernel buffer bvecs, on the other hand, don't necessarily - * have the size property of user registered ones, so we have - * to use the slow iter advance. - */ - if (offset < bvec->bv_len) { - iter->count -= offset; - iter->iov_offset = offset; - } else if (imu->is_kbuf) { - iov_iter_advance(iter, offset); - } else { - unsigned long seg_skip; + if (imu->is_kbuf) + return io_import_kbuf(ddir, iter, imu, len, offset); - /* skip first vec */ - offset -= bvec->bv_len; - seg_skip = 1 + (offset >> imu->folio_shift); + /* + * Don't use iov_iter_advance() here, as it's really slow for + * using the latter parts of a big fixed buffer - it iterates + * over each segment manually. We can cheat a bit here for user + * registered nodes, because we know that: + * + * 1) it's a BVEC iter, we set it up + * 2) all bvecs are the same in size, except potentially the + * first and last bvec + */ + folio_mask = (1UL << imu->folio_shift) - 1; + bvec = imu->bvec; + if (offset >= bvec->bv_len) { + unsigned long seg_skip; - iter->bvec += seg_skip; - iter->nr_segs -= seg_skip; - iter->count -= bvec->bv_len + offset; - iter->iov_offset = offset & ((1UL << imu->folio_shift) - 1); - } + /* skip first vec */ + offset -= bvec->bv_len; + seg_skip = 1 + (offset >> imu->folio_shift); + bvec += seg_skip; + offset &= folio_mask; } - + nr_segs = (offset + len + bvec->bv_offset + folio_mask) >> imu->folio_shift; + iov_iter_bvec(iter, ddir, bvec, nr_segs, len); + iter->iov_offset = offset; return 0; } diff --git a/io_uring/zcrx.c b/io_uring/zcrx.c index 80d4a6f71d29..fe86606b9f30 100644 --- a/io_uring/zcrx.c +++ b/io_uring/zcrx.c @@ -26,6 +26,11 @@ #include "zcrx.h" #include "rsrc.h" +static inline struct io_zcrx_ifq *io_pp_to_ifq(struct page_pool *pp) +{ + return pp->mp_priv; +} + #define IO_DMA_ATTR (DMA_ATTR_SKIP_CPU_SYNC | DMA_ATTR_WEAK_ORDERING) static void __io_zcrx_unmap_area(struct io_zcrx_ifq *ifq, @@ -46,14 +51,21 @@ static void __io_zcrx_unmap_area(struct io_zcrx_ifq *ifq, static void io_zcrx_unmap_area(struct io_zcrx_ifq *ifq, struct io_zcrx_area *area) { + guard(mutex)(&ifq->dma_lock); + if (area->is_mapped) __io_zcrx_unmap_area(ifq, area, area->nia.num_niovs); + area->is_mapped = false; } static int io_zcrx_map_area(struct io_zcrx_ifq *ifq, struct io_zcrx_area *area) { int i; + guard(mutex)(&ifq->dma_lock); + if (area->is_mapped) + return 0; + for (i = 0; i < area->nia.num_niovs; i++) { struct net_iov *niov = &area->nia.niovs[i]; dma_addr_t dma; @@ -181,7 +193,7 @@ static void io_zcrx_free_area(struct io_zcrx_area *area) kvfree(area->nia.niovs); kvfree(area->user_refs); if (area->pages) { - unpin_user_pages(area->pages, area->nia.num_niovs); + unpin_user_pages(area->pages, area->nr_folios); kvfree(area->pages); } kfree(area); @@ -192,7 +204,7 @@ static int io_zcrx_create_area(struct io_zcrx_ifq *ifq, struct io_uring_zcrx_area_reg *area_reg) { struct io_zcrx_area *area; - int i, ret, nr_pages; + int i, ret, nr_pages, nr_iovs; struct iovec iov; if (area_reg->flags || area_reg->rq_area_token) @@ -220,27 +232,28 @@ static int io_zcrx_create_area(struct io_zcrx_ifq *ifq, area->pages = NULL; goto err; } - area->nia.num_niovs = nr_pages; + area->nr_folios = nr_iovs = nr_pages; + area->nia.num_niovs = nr_iovs; - area->nia.niovs = kvmalloc_array(nr_pages, sizeof(area->nia.niovs[0]), + area->nia.niovs = kvmalloc_array(nr_iovs, sizeof(area->nia.niovs[0]), GFP_KERNEL | __GFP_ZERO); if (!area->nia.niovs) goto err; - area->freelist = kvmalloc_array(nr_pages, sizeof(area->freelist[0]), + area->freelist = kvmalloc_array(nr_iovs, sizeof(area->freelist[0]), GFP_KERNEL | __GFP_ZERO); if (!area->freelist) goto err; - for (i = 0; i < nr_pages; i++) + for (i = 0; i < nr_iovs; i++) area->freelist[i] = i; - area->user_refs = kvmalloc_array(nr_pages, sizeof(area->user_refs[0]), + area->user_refs = kvmalloc_array(nr_iovs, sizeof(area->user_refs[0]), GFP_KERNEL | __GFP_ZERO); if (!area->user_refs) goto err; - for (i = 0; i < nr_pages; i++) { + for (i = 0; i < nr_iovs; i++) { struct net_iov *niov = &area->nia.niovs[i]; niov->owner = &area->nia; @@ -248,7 +261,7 @@ static int io_zcrx_create_area(struct io_zcrx_ifq *ifq, atomic_set(&area->user_refs[i], 0); } - area->free_count = nr_pages; + area->free_count = nr_iovs; area->ifq = ifq; /* we're only supporting one area per ifq for now */ area->area_id = 0; @@ -274,6 +287,7 @@ static struct io_zcrx_ifq *io_zcrx_ifq_alloc(struct io_ring_ctx *ctx) ifq->ctx = ctx; spin_lock_init(&ifq->lock); spin_lock_init(&ifq->rq_lock); + mutex_init(&ifq->dma_lock); return ifq; } @@ -323,6 +337,7 @@ static void io_zcrx_ifq_free(struct io_zcrx_ifq *ifq) put_device(ifq->dev); io_free_rbuf_ring(ifq); + mutex_destroy(&ifq->dma_lock); kfree(ifq); } @@ -353,7 +368,8 @@ int io_register_zcrx_ifq(struct io_ring_ctx *ctx, return -EFAULT; if (copy_from_user(&rd, u64_to_user_ptr(reg.region_ptr), sizeof(rd))) return -EFAULT; - if (memchr_inv(®.__resv, 0, sizeof(reg.__resv))) + if (memchr_inv(®.__resv, 0, sizeof(reg.__resv)) || + reg.__resv2 || reg.zcrx_id) return -EINVAL; if (reg.if_rxq == -1 || !reg.rq_entries || reg.flags) return -EINVAL; @@ -393,10 +409,6 @@ int io_register_zcrx_ifq(struct io_ring_ctx *ctx, goto err; get_device(ifq->dev); - ret = io_zcrx_map_area(ifq, ifq->area); - if (ret) - goto err; - mp_param.mp_ops = &io_uring_pp_zc_ops; mp_param.mp_priv = ifq; ret = net_mp_open_rxq(ifq->netdev, reg.if_rxq, &mp_param); @@ -584,7 +596,7 @@ static void io_zcrx_refill_slow(struct page_pool *pp, struct io_zcrx_ifq *ifq) static netmem_ref io_pp_zc_alloc_netmems(struct page_pool *pp, gfp_t gfp) { - struct io_zcrx_ifq *ifq = pp->mp_priv; + struct io_zcrx_ifq *ifq = io_pp_to_ifq(pp); /* pp should already be ensuring that */ if (unlikely(pp->alloc.count)) @@ -616,7 +628,8 @@ static bool io_pp_zc_release_netmem(struct page_pool *pp, netmem_ref netmem) static int io_pp_zc_init(struct page_pool *pp) { - struct io_zcrx_ifq *ifq = pp->mp_priv; + struct io_zcrx_ifq *ifq = io_pp_to_ifq(pp); + int ret; if (WARN_ON_ONCE(!ifq)) return -EINVAL; @@ -629,13 +642,17 @@ static int io_pp_zc_init(struct page_pool *pp) if (pp->p.dma_dir != DMA_FROM_DEVICE) return -EOPNOTSUPP; + ret = io_zcrx_map_area(ifq, ifq->area); + if (ret) + return ret; + percpu_ref_get(&ifq->ctx->refs); return 0; } static void io_pp_zc_destroy(struct page_pool *pp) { - struct io_zcrx_ifq *ifq = pp->mp_priv; + struct io_zcrx_ifq *ifq = io_pp_to_ifq(pp); struct io_zcrx_area *area = ifq->area; if (WARN_ON_ONCE(area->free_count != area->nia.num_niovs)) @@ -664,6 +681,9 @@ static void io_pp_uninstall(void *mp_priv, struct netdev_rx_queue *rxq) struct io_zcrx_ifq *ifq = mp_priv; io_zcrx_drop_netdev(ifq); + if (ifq->area) + io_zcrx_unmap_area(ifq, ifq->area); + p->mp_ops = NULL; p->mp_priv = NULL; } @@ -790,7 +810,7 @@ static int io_zcrx_recv_frag(struct io_kiocb *req, struct io_zcrx_ifq *ifq, niov = netmem_to_net_iov(frag->netmem); if (niov->pp->mp_ops != &io_uring_pp_zc_ops || - niov->pp->mp_priv != ifq) + io_pp_to_ifq(niov->pp) != ifq) return -EFAULT; if (!io_zcrx_queue_cqe(req, niov, ifq, off + skb_frag_off(frag), len)) diff --git a/io_uring/zcrx.h b/io_uring/zcrx.h index 706cc7300780..f2bc811f022c 100644 --- a/io_uring/zcrx.h +++ b/io_uring/zcrx.h @@ -15,6 +15,7 @@ struct io_zcrx_area { bool is_mapped; u16 area_id; struct page **pages; + unsigned long nr_folios; /* freelist */ spinlock_t freelist_lock ____cacheline_aligned_in_smp; @@ -26,17 +27,18 @@ struct io_zcrx_ifq { struct io_ring_ctx *ctx; struct io_zcrx_area *area; + spinlock_t rq_lock ____cacheline_aligned_in_smp; struct io_uring *rq_ring; struct io_uring_zcrx_rqe *rqes; - u32 rq_entries; u32 cached_rq_head; - spinlock_t rq_lock; + u32 rq_entries; u32 if_rxq; struct device *dev; struct net_device *netdev; netdevice_tracker netdev_tracker; spinlock_t lock; + struct mutex dma_lock; }; #if defined(CONFIG_IO_URING_ZCRX) |