io_uring: user registered clockid for wait timeouts

Add a new registration opcode IORING_REGISTER_CLOCK, which allows the
user to select which clock id it wants to use with CQ waiting timeouts.
It only allows a subset of all posix clocks and currently supports
CLOCK_MONOTONIC and CLOCK_BOOTTIME.

Suggested-by: Lewis Baker <lewissbaker@gmail.com>
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/98f2bc8a3c36cdf8f0e6a275245e81e903459703.1723039801.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
This commit is contained in:
Pavel Begunkov 2024-08-07 15:18:14 +01:00 committed by Jens Axboe
parent d29cb3726f
commit 2b8e976b98
6 changed files with 56 additions and 3 deletions

View File

@ -239,6 +239,9 @@ struct io_ring_ctx {
struct io_rings *rings;
struct percpu_ref refs;
clockid_t clockid;
enum tk_offsets clock_offset;
enum task_work_notify_mode notify_method;
unsigned sq_thread_idle;
} ____cacheline_aligned_in_smp;

View File

@ -596,6 +596,8 @@ enum io_uring_register_op {
IORING_REGISTER_NAPI = 27,
IORING_UNREGISTER_NAPI = 28,
IORING_REGISTER_CLOCK = 29,
/* this goes last */
IORING_REGISTER_LAST,
@ -676,6 +678,11 @@ struct io_uring_restriction {
__u32 resv2[3];
};
struct io_uring_clock_register {
__u32 clockid;
__u32 __resv[3];
};
struct io_uring_buf {
__u64 addr;
__u32 len;

View File

@ -2377,7 +2377,8 @@ static inline int io_cqring_wait_schedule(struct io_ring_ctx *ctx,
ret = 0;
if (iowq->timeout == KTIME_MAX)
schedule();
else if (!schedule_hrtimeout(&iowq->timeout, HRTIMER_MODE_ABS))
else if (!schedule_hrtimeout_range_clock(&iowq->timeout, 0,
HRTIMER_MODE_ABS, ctx->clockid))
ret = -ETIME;
current->in_iowait = 0;
return ret;
@ -2422,7 +2423,7 @@ static int io_cqring_wait(struct io_ring_ctx *ctx, int min_events, u32 flags,
iowq.timeout = timespec64_to_ktime(ts);
if (!(flags & IORING_ENTER_ABS_TIMER))
iowq.timeout = ktime_add(iowq.timeout, ktime_get());
iowq.timeout = ktime_add(iowq.timeout, io_get_time(ctx));
}
if (sig) {
@ -3424,6 +3425,9 @@ static __cold int io_uring_create(unsigned entries, struct io_uring_params *p,
if (!ctx)
return -ENOMEM;
ctx->clockid = CLOCK_MONOTONIC;
ctx->clock_offset = 0;
if ((ctx->flags & IORING_SETUP_DEFER_TASKRUN) &&
!(ctx->flags & IORING_SETUP_IOPOLL) &&
!(ctx->flags & IORING_SETUP_SQPOLL))

View File

@ -437,6 +437,14 @@ static inline bool io_file_can_poll(struct io_kiocb *req)
return false;
}
static inline ktime_t io_get_time(struct io_ring_ctx *ctx)
{
if (ctx->clockid == CLOCK_MONOTONIC)
return ktime_get();
return ktime_get_with_offset(ctx->clock_offset);
}
enum {
IO_CHECK_CQ_OVERFLOW_BIT,
IO_CHECK_CQ_DROPPED_BIT,

View File

@ -283,7 +283,7 @@ void __io_napi_busy_loop(struct io_ring_ctx *ctx, struct io_wait_queue *iowq)
iowq->napi_busy_poll_dt = READ_ONCE(ctx->napi_busy_poll_dt);
if (iowq->timeout != KTIME_MAX) {
ktime_t dt = ktime_sub(iowq->timeout, ktime_get());
ktime_t dt = ktime_sub(iowq->timeout, io_get_time(ctx));
iowq->napi_busy_poll_dt = min_t(u64, iowq->napi_busy_poll_dt, dt);
}

View File

@ -335,6 +335,31 @@ err:
return ret;
}
static int io_register_clock(struct io_ring_ctx *ctx,
struct io_uring_clock_register __user *arg)
{
struct io_uring_clock_register reg;
if (copy_from_user(&reg, arg, sizeof(reg)))
return -EFAULT;
if (memchr_inv(&reg.__resv, 0, sizeof(reg.__resv)))
return -EINVAL;
switch (reg.clockid) {
case CLOCK_MONOTONIC:
ctx->clock_offset = 0;
break;
case CLOCK_BOOTTIME:
ctx->clock_offset = TK_OFFS_BOOT;
break;
default:
return -EINVAL;
}
ctx->clockid = reg.clockid;
return 0;
}
static int __io_uring_register(struct io_ring_ctx *ctx, unsigned opcode,
void __user *arg, unsigned nr_args)
__releases(ctx->uring_lock)
@ -511,6 +536,12 @@ static int __io_uring_register(struct io_ring_ctx *ctx, unsigned opcode,
break;
ret = io_unregister_napi(ctx, arg);
break;
case IORING_REGISTER_CLOCK:
ret = -EINVAL;
if (!arg || nr_args)
break;
ret = io_register_clock(ctx, arg);
break;
default:
ret = -EINVAL;
break;