Re: [PATCH v2 1/2] io/block: Refactoring the bdrv_drained_begin() function and implement a timeout mechanism.

[Stefan Hajnoczi <stefanha@redhat.com>]

[-- Attachment #1: Type: text/plain, Size: 12371 bytes --]

On Fri, Mar 21, 2025 at 03:09:16PM +0800, zoudongjie wrote:
> From: Zhu Yangyang <zhuyangyang14@huawei.com>
> 
> The bdrv_drained_begin() function is a blocking function. In scenarios where network storage
> is used and network links fail, it may block for a long time.
> Therefore, we add a timeout parameter to control the duration of the block.
> 
> Since bdrv_drained_begin() has been widely adopted, both bdrv_drained_begin()
> and bdrv_drained_begin_timeout() will be retained.
> 
> Signed-off-by: Zhu Yangyang <zhuyangyang14@huawei.com>
> ---
>  block/io.c               | 58 +++++++++++++++++++++++++++++++++-------
>  include/block/aio-wait.h | 49 +++++++++++++++++++++++++++++++++
>  include/block/block-io.h | 22 ++++++++++++++-
>  util/aio-wait.c          |  5 ++++
>  4 files changed, 123 insertions(+), 11 deletions(-)
> 
> diff --git a/block/io.c b/block/io.c
> index 1ba8d1aeea..912b76c4a4 100644
> --- a/block/io.c
> +++ b/block/io.c
> @@ -255,6 +255,8 @@ typedef struct {
>      bool begin;
>      bool poll;
>      BdrvChild *parent;
> +    uint64_t timeout_ns;
> +    int ret;
>  } BdrvCoDrainData;
>  
>  /* Returns true if BDRV_POLL_WHILE() should go into a blocking aio_poll() */
> @@ -283,6 +285,10 @@ static bool bdrv_drain_poll_top_level(BlockDriverState *bs,
>      return bdrv_drain_poll(bs, ignore_parent, false);
>  }
>  
> +static int bdrv_do_drained_begin_timeout(BlockDriverState *bs,
> +                                         BdrvChild *parent,
> +                                         bool poll,
> +                                         uint64_t timeout_ns);
>  static void bdrv_do_drained_begin(BlockDriverState *bs, BdrvChild *parent,
>                                    bool poll);
>  static void bdrv_do_drained_end(BlockDriverState *bs, BdrvChild *parent);
> @@ -296,7 +302,9 @@ static void bdrv_co_drain_bh_cb(void *opaque)
>      if (bs) {
>          bdrv_dec_in_flight(bs);
>          if (data->begin) {
> -            bdrv_do_drained_begin(bs, data->parent, data->poll);
> +            data->ret = bdrv_do_drained_begin_timeout(bs, data->parent,
> +                                                      data->poll,
> +                                                      data->timeout_ns);
>          } else {
>              assert(!data->poll);
>              bdrv_do_drained_end(bs, data->parent);
> @@ -310,10 +318,11 @@ static void bdrv_co_drain_bh_cb(void *opaque)
>      aio_co_wake(co);
>  }
>  
> -static void coroutine_fn bdrv_co_yield_to_drain(BlockDriverState *bs,
> -                                                bool begin,
> -                                                BdrvChild *parent,
> -                                                bool poll)
> +static int coroutine_fn bdrv_co_yield_to_drain_timeout(BlockDriverState *bs,
> +                                                       bool begin,
> +                                                       BdrvChild *parent,
> +                                                       bool poll,
> +                                                       uint64_t timeout_ns)
>  {
>      BdrvCoDrainData data;
>      Coroutine *self = qemu_coroutine_self();
> @@ -329,6 +338,8 @@ static void coroutine_fn bdrv_co_yield_to_drain(BlockDriverState *bs,
>          .begin = begin,
>          .parent = parent,
>          .poll = poll,
> +        .timeout_ns = timeout_ns,
> +        .ret = 0
>      };
>  
>      if (bs) {
> @@ -342,16 +353,27 @@ static void coroutine_fn bdrv_co_yield_to_drain(BlockDriverState *bs,
>      /* If we are resumed from some other event (such as an aio completion or a
>       * timer callback), it is a bug in the caller that should be fixed. */
>      assert(data.done);
> +    return data.ret;
>  }
>  
> -static void bdrv_do_drained_begin(BlockDriverState *bs, BdrvChild *parent,
> -                                  bool poll)
> +static void coroutine_fn bdrv_co_yield_to_drain(BlockDriverState *bs,
> +                                                bool begin,
> +                                                BdrvChild *parent,
> +                                                bool poll)
> +{
> +    bdrv_co_yield_to_drain_timeout(bs, begin, parent, poll, 0);
> +}
> +
> +static int bdrv_do_drained_begin_timeout(BlockDriverState *bs,
> +                                         BdrvChild *parent,
> +                                         bool poll,
> +                                         uint64_t timeout_ns)
>  {
>      IO_OR_GS_CODE();
>  
>      if (qemu_in_coroutine()) {
> -        bdrv_co_yield_to_drain(bs, true, parent, poll);
> -        return;
> +        return bdrv_co_yield_to_drain_timeout(bs, true, parent, poll,
> +                                              timeout_ns);
>      }
>  
>      GLOBAL_STATE_CODE();
> @@ -375,8 +397,17 @@ static void bdrv_do_drained_begin(BlockDriverState *bs, BdrvChild *parent,
>       * nodes.
>       */
>      if (poll) {
> -        BDRV_POLL_WHILE(bs, bdrv_drain_poll_top_level(bs, parent));
> +        return BDRV_POLL_WHILE_TIMEOUT(bs,
> +                                       bdrv_drain_poll_top_level(bs, parent),
> +                                       timeout_ns);
>      }
> +    return 0;
> +}
> +
> +static void bdrv_do_drained_begin(BlockDriverState *bs, BdrvChild *parent,
> +                                  bool poll)
> +{
> +    bdrv_do_drained_begin_timeout(bs, parent, poll, 0);
>  }
>  
>  void bdrv_do_drained_begin_quiesce(BlockDriverState *bs, BdrvChild *parent)
> @@ -391,6 +422,13 @@ bdrv_drained_begin(BlockDriverState *bs)
>      bdrv_do_drained_begin(bs, NULL, true);
>  }
>  
> +int coroutine_mixed_fn
> +bdrv_drained_begin_timeout(BlockDriverState *bs, uint64_t timeout_ns)
> +{
> +    IO_OR_GS_CODE();
> +    return bdrv_do_drained_begin_timeout(bs, NULL, true, timeout_ns);
> +}
> +
>  /**
>   * This function does not poll, nor must any of its recursively called
>   * functions.
> diff --git a/include/block/aio-wait.h b/include/block/aio-wait.h
> index cf5e8bde1c..24c31df620 100644
> --- a/include/block/aio-wait.h
> +++ b/include/block/aio-wait.h
> @@ -99,6 +99,53 @@ extern AioWait global_aio_wait;
>      qatomic_dec(&wait_->num_waiters);                              \
>      waited_; })
>  
> +/**
> + * AIO_WAIT_WHILE_TIMEOUT:
> + *
> + * Refer to the implementation of AIO_WAIT_WHILE_INTERNAL,
> + * the timeout parameter is added.
> + * @timeout_ns: maximum duration to wait, in nanoseconds, except the value
> + *              is unsigned, 0 means infinite.
> + */
> +#define AIO_WAIT_WHILE_TIMEOUT(ctx, cond, timeout_ns) ({                 \
> +    int ret_ = 0;                                                        \
> +    uint64_t timeout_ = (timeout_ns);                                    \
> +    AioWait *wait_ = &global_aio_wait;                                   \
> +    AioContext *ctx_ = (ctx);                                            \
> +    AioContext *current_ctx_ = NULL;                                     \
> +    QEMUTimer timer_;                                                    \
> +    /* Increment wait_->num_waiters before evaluating cond. */           \
> +    qatomic_inc(&wait_->num_waiters);                                    \
> +    /* Paired with smp_mb in aio_wait_kick(). */                         \
> +    smp_mb__after_rmw();                                                 \
> +    if (ctx_ && in_aio_context_home_thread(ctx_)) {                      \
> +        current_ctx_ = ctx_;                                             \
> +    } else {                                                             \
> +        assert(qemu_get_current_aio_context() ==                         \
> +               qemu_get_aio_context());                                  \
> +        current_ctx_ = qemu_get_aio_context();                           \
> +    }                                                                    \
> +    if (timeout_ > 0) {                                                  \
> +        timer_init_full(&timer_, &current_ctx_->tlg,                     \
> +                        QEMU_CLOCK_REALTIME,                             \
> +                        SCALE_NS, 0, aio_wait_timer_cb, NULL);           \
> +        timer_mod_ns(&timer_,                                            \
> +                     qemu_clock_get_ns(QEMU_CLOCK_REALTIME) +            \
> +                     timeout_);                                          \
> +    }                                                                    \
> +    while ((cond)) {                                                     \
> +        aio_poll(current_ctx_, true);                                    \
> +        if (timeout_ > 0 && !timer_pending(&timer_)) {                   \
> +            ret_ = -ETIMEDOUT;                                           \
> +            break;                                                       \
> +        }                                                                \
> +    }                                                                    \
> +    if (timeout_ > 0) {                                                  \
> +        timer_del(&timer_);                                              \
> +    }                                                                    \
> +    qatomic_dec(&wait_->num_waiters);                                    \
> +    ret_; })
> +
>  #define AIO_WAIT_WHILE(ctx, cond)                                  \
>      AIO_WAIT_WHILE_INTERNAL(ctx, cond)

It would be nice to unify AIO_WAIT_WHILE_INTERNAL() and
AIO_WAIT_WHILE_TIMEOUT(). I don't see any caller that uses the waited_
return value from AIO_WAIT_WHILE_INTERNAL(), so I think it's possible to
replace AIO_WAIT_WHILE_INTERNAL() with AIO_WAIT_WHILE_TIMEOUT(.., ..., 0).

>  
> @@ -149,4 +196,6 @@ static inline bool in_aio_context_home_thread(AioContext *ctx)
>      }
>  }
>  
> +void aio_wait_timer_cb(void *opaque);
> +
>  #endif /* QEMU_AIO_WAIT_H */
> diff --git a/include/block/block-io.h b/include/block/block-io.h
> index b49e0537dd..844e9cf350 100644
> --- a/include/block/block-io.h
> +++ b/include/block/block-io.h
> @@ -354,6 +354,11 @@ bdrv_co_copy_range(BdrvChild *src, int64_t src_offset,
>      AIO_WAIT_WHILE(bdrv_get_aio_context(bs_),              \
>                     cond); })
>  
> +#define BDRV_POLL_WHILE_TIMEOUT(bs, cond, timeout_ns) ({   \
> +    BlockDriverState *bs_ = (bs);                          \
> +    AIO_WAIT_WHILE_TIMEOUT(bdrv_get_aio_context(bs_),      \
> +                           cond, timeout_ns); })
> +
>  void bdrv_drain(BlockDriverState *bs);
>  
>  int co_wrapper_mixed_bdrv_rdlock
> @@ -432,7 +437,22 @@ bdrv_drain_poll(BlockDriverState *bs, BdrvChild *ignore_parent,
>  void bdrv_drained_begin(BlockDriverState *bs);
>  
>  /**
> - * bdrv_do_drained_begin_quiesce:
> + * bdrv_drained_begin_timeout:
> + *
> + * Added timeout parameter for bdrv_drained_begin() to make a time limited.
> + *
> + * @timeout_ns: maximum duration to wait; 0 means infinite, equal to call
> + *              bdrv_drained_begin().
> + *
> + * Returns: 0 if succeeded; -ETIMEDOUT when a timeout occurs.
> + *
> + * Note: when the timeout fails, we've already begin aquiesced section, so we
> + * still need to call bdrv_drained_end() to end the quiescent section.
> + */
> +int bdrv_drained_begin_timeout(BlockDriverState *bs, uint64_t timeout_ns);
> +
> +/**
> + * bdrv_do_drained_badegin_quiesce:
>   *
>   * Quiesces a BDS like bdrv_drained_begin(), but does not wait for already
>   * running requests to complete.
> diff --git a/util/aio-wait.c b/util/aio-wait.c
> index b5336cf5fd..64c3714fb8 100644
> --- a/util/aio-wait.c
> +++ b/util/aio-wait.c
> @@ -84,3 +84,8 @@ void aio_wait_bh_oneshot(AioContext *ctx, QEMUBHFunc *cb, void *opaque)
>      aio_bh_schedule_oneshot(ctx, aio_wait_bh, &data);
>      AIO_WAIT_WHILE_UNLOCKED(NULL, !data.done);
>  }
> +
> +void aio_wait_timer_cb(void *opaque)
> +{
> +    /* The point is to make AIO_WAIT_WHILE_TIMEOUT()'s aio_poll() return */
> +}
> -- 
> 2.33.0
> 

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]