io_uring: add support for IORING_SETUP_SQE_MIXED

Normal rings support 64b SQEs for posting submissions, while certain
features require the ring to be configured with IORING_SETUP_SQE128, as
they need to convey more information per submission. This, in turn,
makes ALL the SQEs be 128b in size. This is somewhat wasteful and
inefficient, particularly when only certain SQEs need to be of the
bigger variant.

This adds support for setting up a ring with mixed SQE sizes, using
IORING_SETUP_SQE_MIXED. When setup in this mode, SQEs posted to the ring
may be either 64b or 128b in size. If a SQE is 128b in size, then opcode
will be set to a variante to indicate that this is the case. Any other
non-128b opcode will assume the SQ's default size.

SQEs on these types of mixed rings may also utilize NOP with skip
success set.  This can happen if the ring is one (small) SQE entry away
from wrapping, and an attempt is made to get a 128b SQE. As SQEs must be
contiguous in the SQ ring, a 128b SQE cannot wrap the ring. For this
case, a single NOP SQE should be inserted with the SKIP_SUCCESS flag
set. The kernel will process this as a normal NOP and without posting a
CQE.

Signed-off-by: Keith Busch <kbusch@kernel.org>
[axboe: {} style fix and assign sqe before opcode read]
Signed-off-by: Jens Axboe <axboe@kernel.dk>

diff --git a/include/uapi/linux/io_uring.h b/include/uapi/linux/io_uring.h
index 263bed13473ef175878716e95942cd2a7e136367..04797a9b76bc24dc4b553acbba3475c8d450c9ed 100644 (file)
--- a/include/uapi/linux/io_uring.h
+++ b/include/uapi/linux/io_uring.h
@@ -231,6 +231,12 @@ enum io_uring_sqe_flags_bit {
  */
 #define IORING_SETUP_CQE_MIXED         (1U << 18)
 
+/*
+ * Allow both 64b and 128b SQEs. If a 128b SQE is posted, it will have
+ * a 128b opcode.
+ */
+#define IORING_SETUP_SQE_MIXED         (1U << 19)
+
 enum io_uring_op {
        IORING_OP_NOP,
        IORING_OP_READV,
@@ -295,6 +301,8 @@ enum io_uring_op {
        IORING_OP_READV_FIXED,
        IORING_OP_WRITEV_FIXED,
        IORING_OP_PIPE,
+       IORING_OP_NOP128,
+       IORING_OP_URING_CMD128,
 
        /* this goes last, obviously */
        IORING_OP_LAST,

diff --git a/io_uring/fdinfo.c b/io_uring/fdinfo.c
index ff3364531c77be0925903671c373f521ecb1ab38..1a806ad168408c1bd797430d143a5ef93ab13a35 100644 (file)
--- a/io_uring/fdinfo.c
+++ b/io_uring/fdinfo.c
@@ -14,6 +14,7 @@
 #include "fdinfo.h"
 #include "cancel.h"
 #include "rsrc.h"
+#include "opdef.h"
 
 #ifdef CONFIG_NET_RX_BUSY_POLL
 static __cold void common_tracking_show_fdinfo(struct io_ring_ctx *ctx,
@@ -66,7 +67,6 @@ static void __io_uring_show_fdinfo(struct io_ring_ctx *ctx, struct seq_file *m)
        unsigned int cq_head = READ_ONCE(r->cq.head);
        unsigned int cq_tail = READ_ONCE(r->cq.tail);
        unsigned int sq_shift = 0;
-       unsigned int sq_entries;
        int sq_pid = -1, sq_cpu = -1;
        u64 sq_total_time = 0, sq_work_time = 0;
        unsigned int i;
@@ -89,26 +89,45 @@ static void __io_uring_show_fdinfo(struct io_ring_ctx *ctx, struct seq_file *m)
        seq_printf(m, "CqTail:\t%u\n", cq_tail);
        seq_printf(m, "CachedCqTail:\t%u\n", data_race(ctx->cached_cq_tail));
        seq_printf(m, "SQEs:\t%u\n", sq_tail - sq_head);
-       sq_entries = min(sq_tail - sq_head, ctx->sq_entries);
-       for (i = 0; i < sq_entries; i++) {
-               unsigned int entry = i + sq_head;
+       while (sq_head < sq_tail) {
                struct io_uring_sqe *sqe;
                unsigned int sq_idx;
+               bool sqe128 = false;
+               u8 opcode;
 
                if (ctx->flags & IORING_SETUP_NO_SQARRAY)
                        break;
-               sq_idx = READ_ONCE(ctx->sq_array[entry & sq_mask]);
+               sq_idx = READ_ONCE(ctx->sq_array[sq_head & sq_mask]);
                if (sq_idx > sq_mask)
                        continue;
+
                sqe = &ctx->sq_sqes[sq_idx << sq_shift];
+               opcode = READ_ONCE(sqe->opcode);
+               if (sq_shift) {
+                       sqe128 = true;
+               } else if (io_issue_defs[opcode].is_128) {
+                       if (!(ctx->flags & IORING_SETUP_SQE_MIXED)) {
+                               seq_printf(m,
+                                       "%5u: invalid sqe, 128B entry on non-mixed sq\n",
+                                       sq_idx);
+                               break;
+                       }
+                       if ((++sq_head & sq_mask) == 0) {
+                               seq_printf(m,
+                                       "%5u: corrupted sqe, wrapping 128B entry\n",
+                                       sq_idx);
+                               break;
+                       }
+                       sqe128 = true;
+               }
                seq_printf(m, "%5u: opcode:%s, fd:%d, flags:%x, off:%llu, "
                              "addr:0x%llx, rw_flags:0x%x, buf_index:%d "
                              "user_data:%llu",
-                          sq_idx, io_uring_get_opcode(sqe->opcode), sqe->fd,
+                          sq_idx, io_uring_get_opcode(opcode), sqe->fd,
                           sqe->flags, (unsigned long long) sqe->off,
                           (unsigned long long) sqe->addr, sqe->rw_flags,
                           sqe->buf_index, sqe->user_data);
-               if (sq_shift) {
+               if (sqe128) {
                        u64 *sqeb = (void *) (sqe + 1);
                        int size = sizeof(struct io_uring_sqe) / sizeof(u64);
                        int j;
@@ -120,6 +139,7 @@ static void __io_uring_show_fdinfo(struct io_ring_ctx *ctx, struct seq_file *m)
                        }
                }
                seq_printf(m, "\n");
+               sq_head++;
        }
        seq_printf(m, "CQEs:\t%u\n", cq_tail - cq_head);
        while (cq_head < cq_tail) {

diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c
index e4ede0bad36f08d103624621e44d896e535cb9f1..be44d636fe1f73709c9f67992e596040bcb38bb6 100644 (file)
--- a/io_uring/io_uring.c
+++ b/io_uring/io_uring.c
@@ -2164,7 +2164,7 @@ static __cold int io_init_fail_req(struct io_kiocb *req, int err)
 }
 
 static int io_init_req(struct io_ring_ctx *ctx, struct io_kiocb *req,
-                      const struct io_uring_sqe *sqe)
+                      const struct io_uring_sqe *sqe, unsigned int *left)
        __must_hold(&ctx->uring_lock)
 {
        const struct io_issue_def *def;
@@ -2190,6 +2190,24 @@ static int io_init_req(struct io_ring_ctx *ctx, struct io_kiocb *req,
        opcode = array_index_nospec(opcode, IORING_OP_LAST);
 
        def = &io_issue_defs[opcode];
+       if (def->is_128 && !(ctx->flags & IORING_SETUP_SQE128)) {
+               /*
+                * A 128b op on a non-128b SQ requires mixed SQE support as
+                * well as 2 contiguous entries.
+                */
+               if (!(ctx->flags & IORING_SETUP_SQE_MIXED) || *left < 2 ||
+                   !(ctx->cached_sq_head & (ctx->sq_entries - 1)))
+                       return io_init_fail_req(req, -EINVAL);
+               /*
+                * A 128b operation on a mixed SQ uses two entries, so we have
+                * to increment the head and cached refs, and decrement what's
+                * left.
+                */
+               current->io_uring->cached_refs++;
+               ctx->cached_sq_head++;
+               (*left)--;
+       }
+
        if (unlikely(sqe_flags & ~SQE_COMMON_FLAGS)) {
                /* enforce forwards compatibility on users */
                if (sqe_flags & ~SQE_VALID_FLAGS)
@@ -2299,13 +2317,13 @@ static __cold int io_submit_fail_init(const struct io_uring_sqe *sqe,
 }
 
 static inline int io_submit_sqe(struct io_ring_ctx *ctx, struct io_kiocb *req,
-                        const struct io_uring_sqe *sqe)
+                        const struct io_uring_sqe *sqe, unsigned int *left)
        __must_hold(&ctx->uring_lock)
 {
        struct io_submit_link *link = &ctx->submit_state.link;
        int ret;
 
-       ret = io_init_req(ctx, req, sqe);
+       ret = io_init_req(ctx, req, sqe, left);
        if (unlikely(ret))
                return io_submit_fail_init(sqe, req, ret);
 
@@ -2457,7 +2475,7 @@ int io_submit_sqes(struct io_ring_ctx *ctx, unsigned int nr)
                 * Continue submitting even for sqe failure if the
                 * ring was setup with IORING_SETUP_SUBMIT_ALL
                 */
-               if (unlikely(io_submit_sqe(ctx, req, sqe)) &&
+               if (unlikely(io_submit_sqe(ctx, req, sqe, &left)) &&
                    !(ctx->flags & IORING_SETUP_SUBMIT_ALL)) {
                        left--;
                        break;
@@ -2802,6 +2820,10 @@ unsigned long rings_size(unsigned int flags, unsigned int sq_entries,
                if (cq_entries < 2)
                        return SIZE_MAX;
        }
+       if (flags & IORING_SETUP_SQE_MIXED) {
+               if (sq_entries < 2)
+                       return SIZE_MAX;
+       }
 
 #ifdef CONFIG_SMP
        off = ALIGN(off, SMP_CACHE_BYTES);
@@ -3726,6 +3748,13 @@ static int io_uring_sanitise_params(struct io_uring_params *p)
        if ((flags & (IORING_SETUP_CQE32|IORING_SETUP_CQE_MIXED)) ==
            (IORING_SETUP_CQE32|IORING_SETUP_CQE_MIXED))
                return -EINVAL;
+       /*
+        * Nonsensical to ask for SQE128 and mixed SQE support, it's not
+        * supported to post 64b SQEs on a ring setup with SQE128.
+        */
+       if ((flags & (IORING_SETUP_SQE128|IORING_SETUP_SQE_MIXED)) ==
+           (IORING_SETUP_SQE128|IORING_SETUP_SQE_MIXED))
+               return -EINVAL;
 
        return 0;
 }

diff --git a/io_uring/io_uring.h b/io_uring/io_uring.h
index 78777bf1ea4b99f51452b97eccdbf5f213e4d27e..44b8091c7fcdbdcd0f245dce9f61cced44ddcbca 100644 (file)
--- a/io_uring/io_uring.h
+++ b/io_uring/io_uring.h
@@ -54,7 +54,8 @@
                        IORING_SETUP_REGISTERED_FD_ONLY |\
                        IORING_SETUP_NO_SQARRAY |\
                        IORING_SETUP_HYBRID_IOPOLL |\
-                       IORING_SETUP_CQE_MIXED)
+                       IORING_SETUP_CQE_MIXED |\
+                       IORING_SETUP_SQE_MIXED)
 
 #define IORING_ENTER_FLAGS (IORING_ENTER_GETEVENTS |\
                        IORING_ENTER_SQ_WAKEUP |\
@@ -565,17 +566,6 @@ static inline void io_req_queue_tw_complete(struct io_kiocb *req, s32 res)
        io_req_task_work_add(req);
 }
 
-/*
- * IORING_SETUP_SQE128 contexts allocate twice the normal SQE size for each
- * slot.
- */
-static inline size_t uring_sqe_size(struct io_ring_ctx *ctx)
-{
-       if (ctx->flags & IORING_SETUP_SQE128)
-               return 2 * sizeof(struct io_uring_sqe);
-       return sizeof(struct io_uring_sqe);
-}
-
 static inline bool io_file_can_poll(struct io_kiocb *req)
 {
        if (req->flags & REQ_F_CAN_POLL)

diff --git a/io_uring/opdef.c b/io_uring/opdef.c
index 932319633eac20791e1eef309ea061ed6e56e16d..df52d760240e475c2e9256b111892dfe9aa4e6d2 100644 (file)
--- a/io_uring/opdef.c
+++ b/io_uring/opdef.c
@@ -575,6 +575,24 @@ const struct io_issue_def io_issue_defs[] = {
                .prep                   = io_pipe_prep,
                .issue                  = io_pipe,
        },
+       [IORING_OP_NOP128] = {
+               .audit_skip             = 1,
+               .iopoll                 = 1,
+               .is_128                 = 1,
+               .prep                   = io_nop_prep,
+               .issue                  = io_nop,
+       },
+       [IORING_OP_URING_CMD128] = {
+               .buffer_select          = 1,
+               .needs_file             = 1,
+               .plug                   = 1,
+               .iopoll                 = 1,
+               .iopoll_queue           = 1,
+               .is_128                 = 1,
+               .async_size             = sizeof(struct io_async_cmd),
+               .prep                   = io_uring_cmd_prep,
+               .issue                  = io_uring_cmd,
+       },
 };
 
 const struct io_cold_def io_cold_defs[] = {
@@ -825,6 +843,14 @@ const struct io_cold_def io_cold_defs[] = {
        [IORING_OP_PIPE] = {
                .name                   = "PIPE",
        },
+       [IORING_OP_NOP128] = {
+               .name                   = "NOP128",
+       },
+       [IORING_OP_URING_CMD128] = {
+               .name                   = "URING_CMD128",
+               .sqe_copy               = io_uring_cmd_sqe_copy,
+               .cleanup                = io_uring_cmd_cleanup,
+       },
 };
 
 const char *io_uring_get_opcode(u8 opcode)

diff --git a/io_uring/opdef.h b/io_uring/opdef.h
index c2f0907ed78cc901497fa1bf73d86f5d029cfc38..aa37846880ffd247258f85937d521fed5a9713f4 100644 (file)
--- a/io_uring/opdef.h
+++ b/io_uring/opdef.h
@@ -27,6 +27,8 @@ struct io_issue_def {
        unsigned                iopoll_queue : 1;
        /* vectored opcode, set if 1) vectored, and 2) handler needs to know */
        unsigned                vectored : 1;
+       /* set to 1 if this opcode uses 128b sqes in a mixed sq */
+       unsigned                is_128 : 1;
 
        /* size of async data needed, if any */
        unsigned short          async_size;

diff --git a/io_uring/register.c b/io_uring/register.c
index 43eb0200482484d88830ed147558eea2f6844d58..1a3e05be6e7b2940d076697ff905d85088885fa0 100644 (file)
--- a/io_uring/register.c
+++ b/io_uring/register.c
@@ -394,7 +394,7 @@ static void io_register_free_rings(struct io_ring_ctx *ctx,
 #define RESIZE_FLAGS   (IORING_SETUP_CQSIZE | IORING_SETUP_CLAMP)
 #define COPY_FLAGS     (IORING_SETUP_NO_SQARRAY | IORING_SETUP_SQE128 | \
                         IORING_SETUP_CQE32 | IORING_SETUP_NO_MMAP | \
-                        IORING_SETUP_CQE_MIXED)
+                        IORING_SETUP_CQE_MIXED | IORING_SETUP_SQE_MIXED)
 
 static int io_register_resize_rings(struct io_ring_ctx *ctx, void __user *arg)
 {

diff --git a/io_uring/uring_cmd.c b/io_uring/uring_cmd.c
index 1225f8124e4bb2e5ad3d88bdfcd8ae7e4e282bbe..9d67a2a721aa3d034af2877d86ded1d45d727e79 100644 (file)
--- a/io_uring/uring_cmd.c
+++ b/io_uring/uring_cmd.c
@@ -216,6 +216,18 @@ int io_uring_cmd_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
        return 0;
 }
 
+/*
+ * IORING_SETUP_SQE128 contexts allocate twice the normal SQE size for each
+ * slot.
+ */
+static inline size_t uring_sqe_size(struct io_kiocb *req)
+{
+       if (req->ctx->flags & IORING_SETUP_SQE128 ||
+           req->opcode == IORING_OP_URING_CMD128)
+               return 2 * sizeof(struct io_uring_sqe);
+       return sizeof(struct io_uring_sqe);
+}
+
 void io_uring_cmd_sqe_copy(struct io_kiocb *req)
 {
        struct io_uring_cmd *ioucmd = io_kiocb_to_cmd(req, struct io_uring_cmd);
@@ -224,7 +236,7 @@ void io_uring_cmd_sqe_copy(struct io_kiocb *req)
        /* Should not happen, as REQ_F_SQE_COPIED covers this */
        if (WARN_ON_ONCE(ioucmd->sqe == ac->sqes))
                return;
-       memcpy(ac->sqes, ioucmd->sqe, uring_sqe_size(req->ctx));
+       memcpy(ac->sqes, ioucmd->sqe, uring_sqe_size(req));
        ioucmd->sqe = ac->sqes;
 }
 
@@ -242,7 +254,8 @@ int io_uring_cmd(struct io_kiocb *req, unsigned int issue_flags)
        if (ret)
                return ret;
 
-       if (ctx->flags & IORING_SETUP_SQE128)
+       if (ctx->flags & IORING_SETUP_SQE128 ||
+           req->opcode == IORING_OP_URING_CMD128)
                issue_flags |= IO_URING_F_SQE128;
        if (ctx->flags & (IORING_SETUP_CQE32 | IORING_SETUP_CQE_MIXED))
                issue_flags |= IO_URING_F_CQE32;