Message ID | 20230412112606.80983-1-pbonzini@redhat.com |
---|---|
State | New |
Headers | show |
Series | [8.0,regression] block/nfs: do not poll within a coroutine | expand |
If you would like this in 8.0 then it needs to be reviewed and in a pullreq today... thanks -- PMM On Wed, 12 Apr 2023 at 12:27, Paolo Bonzini <pbonzini@redhat.com> wrote: > > Since the former nfs_get_allocated_file_size is now a coroutine > function, it must suspend rather than poll. Switch BDRV_POLL_WHILE() > to a qemu_coroutine_yield() loop and schedule nfs_co_generic_bh_cb() > in place of the call to bdrv_wakeup(). > > Fixes: 82618d7bc341 ("block: Convert bdrv_get_allocated_file_size() to co_wrapper", 2023-02-01) > Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> > --- > block/nfs.c | 24 +++++++++++++----------- > 1 file changed, 13 insertions(+), 11 deletions(-) > > diff --git a/block/nfs.c b/block/nfs.c > index 351dc6ec8d14..417defc0cfef 100644 > --- a/block/nfs.c > +++ b/block/nfs.c > @@ -726,10 +726,8 @@ nfs_get_allocated_file_size_cb(int ret, struct nfs_context *nfs, void *data, > if (task->ret < 0) { > error_report("NFS Error: %s", nfs_get_error(nfs)); > } > - > - /* Set task->complete before reading bs->wakeup. */ > - qatomic_mb_set(&task->complete, 1); > - bdrv_wakeup(task->bs); > + replay_bh_schedule_oneshot_event(task->client->aio_context, > + nfs_co_generic_bh_cb, task); > } > > static int64_t coroutine_fn nfs_co_get_allocated_file_size(BlockDriverState *bs) > @@ -743,15 +741,19 @@ static int64_t coroutine_fn nfs_co_get_allocated_file_size(BlockDriverState *bs) > return client->st_blocks * 512; > } > > - task.bs = bs; > + nfs_co_init_task(bs, &task); > task.st = &st; > - if (nfs_fstat_async(client->context, client->fh, nfs_get_allocated_file_size_cb, > - &task) != 0) { > - return -ENOMEM; > - } > + WITH_QEMU_LOCK_GUARD(&client->mutex) { > + if (nfs_fstat_async(client->context, client->fh, nfs_get_allocated_file_size_cb, > + &task) != 0) { > + return -ENOMEM; > + } > > - nfs_set_events(client); > - BDRV_POLL_WHILE(bs, !task.complete); > + nfs_set_events(client); > + } > + while (!task.complete) { > + qemu_coroutine_yield(); > + } > > return (task.ret < 0 ? task.ret : st.st_blocks * 512); > }
On Wed, Apr 12, 2023 at 01:26:06PM +0200, Paolo Bonzini wrote: > Since the former nfs_get_allocated_file_size is now a coroutine > function, it must suspend rather than poll. Switch BDRV_POLL_WHILE() > to a qemu_coroutine_yield() loop and schedule nfs_co_generic_bh_cb() > in place of the call to bdrv_wakeup(). > > Fixes: 82618d7bc341 ("block: Convert bdrv_get_allocated_file_size() to co_wrapper", 2023-02-01) > Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> > --- > block/nfs.c | 24 +++++++++++++----------- > 1 file changed, 13 insertions(+), 11 deletions(-) Reviewed-by: Eric Blake <eblake@redhat.com>
Am 12.04.2023 um 13:26 hat Paolo Bonzini geschrieben: > Since the former nfs_get_allocated_file_size is now a coroutine > function, it must suspend rather than poll. Switch BDRV_POLL_WHILE() > to a qemu_coroutine_yield() loop and schedule nfs_co_generic_bh_cb() > in place of the call to bdrv_wakeup(). > > Fixes: 82618d7bc341 ("block: Convert bdrv_get_allocated_file_size() to co_wrapper", 2023-02-01) > Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> > --- > block/nfs.c | 24 +++++++++++++----------- > 1 file changed, 13 insertions(+), 11 deletions(-) > > diff --git a/block/nfs.c b/block/nfs.c > index 351dc6ec8d14..417defc0cfef 100644 > --- a/block/nfs.c > +++ b/block/nfs.c > @@ -726,10 +726,8 @@ nfs_get_allocated_file_size_cb(int ret, struct nfs_context *nfs, void *data, > if (task->ret < 0) { > error_report("NFS Error: %s", nfs_get_error(nfs)); > } > - > - /* Set task->complete before reading bs->wakeup. */ > - qatomic_mb_set(&task->complete, 1); > - bdrv_wakeup(task->bs); > + replay_bh_schedule_oneshot_event(task->client->aio_context, > + nfs_co_generic_bh_cb, task); > } > > static int64_t coroutine_fn nfs_co_get_allocated_file_size(BlockDriverState *bs) > @@ -743,15 +741,19 @@ static int64_t coroutine_fn nfs_co_get_allocated_file_size(BlockDriverState *bs) > return client->st_blocks * 512; > } > > - task.bs = bs; > + nfs_co_init_task(bs, &task); > task.st = &st; > - if (nfs_fstat_async(client->context, client->fh, nfs_get_allocated_file_size_cb, > - &task) != 0) { > - return -ENOMEM; > - } > + WITH_QEMU_LOCK_GUARD(&client->mutex) { > + if (nfs_fstat_async(client->context, client->fh, nfs_get_allocated_file_size_cb, > + &task) != 0) { > + return -ENOMEM; > + } > > - nfs_set_events(client); > - BDRV_POLL_WHILE(bs, !task.complete); > + nfs_set_events(client); Tab damage in this line. > + } > + while (!task.complete) { > + qemu_coroutine_yield(); > + } > > return (task.ret < 0 ? task.ret : st.st_blocks * 512); > } With the indentation above fixed: Reviewed-by: Kevin Wolf <kwolf@redhat.com>
diff --git a/block/nfs.c b/block/nfs.c index 351dc6ec8d14..417defc0cfef 100644 --- a/block/nfs.c +++ b/block/nfs.c @@ -726,10 +726,8 @@ nfs_get_allocated_file_size_cb(int ret, struct nfs_context *nfs, void *data, if (task->ret < 0) { error_report("NFS Error: %s", nfs_get_error(nfs)); } - - /* Set task->complete before reading bs->wakeup. */ - qatomic_mb_set(&task->complete, 1); - bdrv_wakeup(task->bs); + replay_bh_schedule_oneshot_event(task->client->aio_context, + nfs_co_generic_bh_cb, task); } static int64_t coroutine_fn nfs_co_get_allocated_file_size(BlockDriverState *bs) @@ -743,15 +741,19 @@ static int64_t coroutine_fn nfs_co_get_allocated_file_size(BlockDriverState *bs) return client->st_blocks * 512; } - task.bs = bs; + nfs_co_init_task(bs, &task); task.st = &st; - if (nfs_fstat_async(client->context, client->fh, nfs_get_allocated_file_size_cb, - &task) != 0) { - return -ENOMEM; - } + WITH_QEMU_LOCK_GUARD(&client->mutex) { + if (nfs_fstat_async(client->context, client->fh, nfs_get_allocated_file_size_cb, + &task) != 0) { + return -ENOMEM; + } - nfs_set_events(client); - BDRV_POLL_WHILE(bs, !task.complete); + nfs_set_events(client); + } + while (!task.complete) { + qemu_coroutine_yield(); + } return (task.ret < 0 ? task.ret : st.st_blocks * 512); }
Since the former nfs_get_allocated_file_size is now a coroutine function, it must suspend rather than poll. Switch BDRV_POLL_WHILE() to a qemu_coroutine_yield() loop and schedule nfs_co_generic_bh_cb() in place of the call to bdrv_wakeup(). Fixes: 82618d7bc341 ("block: Convert bdrv_get_allocated_file_size() to co_wrapper", 2023-02-01) Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> --- block/nfs.c | 24 +++++++++++++----------- 1 file changed, 13 insertions(+), 11 deletions(-)