From: Paolo Bonzini <pbonzini@redhat.com>
To: Ming Lei <ming.lei@canonical.com>,
qemu-devel@nongnu.org, Peter Maydell <peter.maydell@linaro.org>,
Stefan Hajnoczi <stefanha@redhat.com>
Cc: Kevin Wolf <kwolf@redhat.com>, Fam Zheng <famz@redhat.com>,
"Michael S. Tsirkin" <mst@redhat.com>
Subject: Re: [Qemu-devel] [PATCH 01/15] qemu coroutine: support bypass mode
Date: Wed, 30 Jul 2014 15:45:20 +0200 [thread overview]
Message-ID: <53D8F6F0.7040106@redhat.com> (raw)
In-Reply-To: <1406720388-18671-2-git-send-email-ming.lei@canonical.com>
Il 30/07/2014 13:39, Ming Lei ha scritto:
> This patch introduces several APIs for supporting bypass qemu coroutine
> in case of being not necessary and for performance's sake.
No, this is wrong. Dataplane *must* use the same code as non-dataplane,
anything else is a step backwards.
If you want to bypass coroutines, bdrv_aio_readv/writev must detect the
conditions that allow doing that and call the bdrv_aio_readv/writev
directly.
To begin with, have you benchmarked QEMU and can you provide a trace of
*where* the coroutine overhead lies?
Paolo
> Signed-off-by: Ming Lei <ming.lei@canonical.com>
> ---
> include/block/coroutine.h | 8 ++++++++
> include/block/coroutine_int.h | 5 +++++
> qemu-coroutine-lock.c | 4 ++--
> qemu-coroutine.c | 33 +++++++++++++++++++++++++++++++++
> 4 files changed, 48 insertions(+), 2 deletions(-)
>
> diff --git a/include/block/coroutine.h b/include/block/coroutine.h
> index b408f96..46d2642 100644
> --- a/include/block/coroutine.h
> +++ b/include/block/coroutine.h
> @@ -223,4 +223,12 @@ void coroutine_fn co_aio_sleep_ns(AioContext *ctx, QEMUClockType type,
> * Note that this function clobbers the handlers for the file descriptor.
> */
> void coroutine_fn yield_until_fd_readable(int fd);
> +
> +/* qemu coroutine bypass APIs */
> +void qemu_coroutine_set_bypass(bool bypass);
> +bool qemu_coroutine_bypassed(Coroutine *self);
> +bool qemu_coroutine_self_bypassed(void);
> +void qemu_coroutine_set_var(void *var);
> +void *qemu_coroutine_get_var(void);
> +
> #endif /* QEMU_COROUTINE_H */
> diff --git a/include/block/coroutine_int.h b/include/block/coroutine_int.h
> index f133d65..106d0b2 100644
> --- a/include/block/coroutine_int.h
> +++ b/include/block/coroutine_int.h
> @@ -39,6 +39,11 @@ struct Coroutine {
> Coroutine *caller;
> QSLIST_ENTRY(Coroutine) pool_next;
>
> + bool bypass;
> +
> + /* only used in bypass mode */
> + void *opaque;
> +
> /* Coroutines that should be woken up when we yield or terminate */
> QTAILQ_HEAD(, Coroutine) co_queue_wakeup;
> QTAILQ_ENTRY(Coroutine) co_queue_next;
> diff --git a/qemu-coroutine-lock.c b/qemu-coroutine-lock.c
> index e4860ae..7c69ff6 100644
> --- a/qemu-coroutine-lock.c
> +++ b/qemu-coroutine-lock.c
> @@ -82,13 +82,13 @@ static bool qemu_co_queue_do_restart(CoQueue *queue, bool single)
>
> bool coroutine_fn qemu_co_queue_next(CoQueue *queue)
> {
> - assert(qemu_in_coroutine());
> + assert(qemu_in_coroutine() || qemu_coroutine_self_bypassed());
> return qemu_co_queue_do_restart(queue, true);
> }
>
> void coroutine_fn qemu_co_queue_restart_all(CoQueue *queue)
> {
> - assert(qemu_in_coroutine());
> + assert(qemu_in_coroutine() || qemu_coroutine_self_bypassed());
> qemu_co_queue_do_restart(queue, false);
> }
>
> diff --git a/qemu-coroutine.c b/qemu-coroutine.c
> index 4708521..0597ed9 100644
> --- a/qemu-coroutine.c
> +++ b/qemu-coroutine.c
> @@ -137,3 +137,36 @@ void coroutine_fn qemu_coroutine_yield(void)
> self->caller = NULL;
> coroutine_swap(self, to);
> }
> +
> +void qemu_coroutine_set_bypass(bool bypass)
> +{
> + Coroutine *self = qemu_coroutine_self();
> +
> + self->bypass = bypass;
> +}
> +
> +bool qemu_coroutine_bypassed(Coroutine *self)
> +{
> + return self->bypass;
> +}
> +
> +bool qemu_coroutine_self_bypassed(void)
> +{
> + Coroutine *self = qemu_coroutine_self();
> +
> + return qemu_coroutine_bypassed(self);
> +}
> +
> +void qemu_coroutine_set_var(void *var)
> +{
> + Coroutine *self = qemu_coroutine_self();
> +
> + self->opaque = var;
> +}
> +
> +void *qemu_coroutine_get_var(void)
> +{
> + Coroutine *self = qemu_coroutine_self();
> +
> + return self->opaque;
> +}
>
next prev parent reply other threads:[~2014-07-30 13:45 UTC|newest]
Thread overview: 71+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-07-30 11:39 [Qemu-devel] [PATCH 00/14] dataplane: optimization and multi virtqueue support Ming Lei
2014-07-30 11:39 ` [Qemu-devel] [PATCH 01/15] qemu coroutine: support bypass mode Ming Lei
2014-07-30 13:45 ` Paolo Bonzini [this message]
2014-07-30 17:15 ` Ming Lei
2014-07-30 23:37 ` Paolo Bonzini
2014-07-31 3:55 ` Ming Lei
2014-07-31 7:37 ` Benoît Canet
2014-07-31 9:47 ` Ming Lei
2014-07-31 10:45 ` Paolo Bonzini
2014-08-01 13:38 ` Ming Lei
2014-07-31 8:59 ` Ming Lei
2014-07-31 9:15 ` Paolo Bonzini
2014-07-31 10:06 ` Ming Lei
2014-07-31 16:13 ` Ming Lei
2014-07-31 16:30 ` Paolo Bonzini
2014-08-01 2:54 ` Ming Lei
2014-08-01 13:13 ` Stefan Hajnoczi
2014-08-01 13:48 ` Ming Lei
2014-08-01 14:17 ` Paolo Bonzini
2014-08-01 15:21 ` Ming Lei
2014-08-01 14:52 ` Ming Lei
2014-08-01 16:03 ` Stefan Hajnoczi
2014-08-02 2:42 ` Ming Lei
2014-07-30 11:39 ` [Qemu-devel] [PATCH 02/15] qemu aio: prepare for supporting selective bypass coroutine Ming Lei
2014-07-30 11:39 ` [Qemu-devel] [PATCH 03/15] block: support to bypass qemu coroutinue Ming Lei
2014-07-30 11:39 ` [Qemu-devel] [PATCH 04/15] Revert "raw-posix: drop raw_get_aio_fd() since it is no longer used" Ming Lei
2014-07-30 11:39 ` [Qemu-devel] [PATCH 05/15] dataplane: enable selective bypassing coroutine Ming Lei
2014-07-30 11:39 ` [Qemu-devel] [PATCH 06/15] qemu/obj_pool.h: introduce object allocation pool Ming Lei
2014-07-30 11:39 ` [Qemu-devel] [PATCH 07/15] dataplane: use object pool to speed up allocation for virtio blk request Ming Lei
2014-07-30 14:14 ` Paolo Bonzini
2014-07-30 15:09 ` Michael S. Tsirkin
2014-07-31 3:22 ` Ming Lei
2014-07-31 9:18 ` Paolo Bonzini
2014-08-01 7:42 ` Ming Lei
2014-08-04 10:21 ` Stefan Hajnoczi
2014-08-04 11:42 ` Ming Lei
2014-07-30 11:39 ` [Qemu-devel] [PATCH 08/15] virtio: decrease size of VirtQueueElement Ming Lei
2014-07-30 13:51 ` Paolo Bonzini
2014-07-30 14:40 ` Michael S. Tsirkin
2014-07-30 14:50 ` Paolo Bonzini
2014-07-31 2:11 ` Ming Lei
2014-07-31 2:07 ` Ming Lei
2014-07-31 9:38 ` Paolo Bonzini
2014-08-01 3:34 ` Ming Lei
2014-07-30 11:39 ` [Qemu-devel] [PATCH 09/15] linux-aio: fix submit aio as a batch Ming Lei
2014-07-30 13:59 ` Paolo Bonzini
2014-07-30 17:32 ` Ming Lei
2014-07-30 23:41 ` Paolo Bonzini
2014-07-30 11:39 ` [Qemu-devel] [PATCH 10/15] linux-aio: increase max event to 256 Ming Lei
2014-07-30 12:15 ` Eric Blake
2014-07-30 14:00 ` Paolo Bonzini
2014-07-30 17:20 ` Ming Lei
2014-08-04 10:26 ` Stefan Hajnoczi
2014-07-30 11:39 ` [Qemu-devel] [PATCH 11/15] linux-aio: remove 'node' from 'struct qemu_laiocb' Ming Lei
2014-07-30 11:39 ` [Qemu-devel] [PATCH 12/15] hw/virtio-pci: introduce num_queues property Ming Lei
2014-07-30 11:39 ` [Qemu-devel] [PATCH 13/15] hw/virtio/virtio-blk.h: introduce VIRTIO_BLK_F_MQ Ming Lei
2014-07-30 11:39 ` [Qemu-devel] [PATCH 14/15] hw/block/virtio-blk: create num_queues vqs if dataplane is enabled Ming Lei
2014-07-30 14:01 ` Paolo Bonzini
2014-07-30 15:12 ` Michael S. Tsirkin
2014-07-30 15:25 ` Paolo Bonzini
2014-07-31 3:47 ` Ming Lei
2014-07-31 8:52 ` Paolo Bonzini
2014-08-01 3:09 ` Ming Lei
2014-08-01 3:24 ` Ming Lei
2014-08-01 6:10 ` Paolo Bonzini
2014-08-01 7:35 ` Ming Lei
2014-08-01 7:46 ` Paolo Bonzini
2014-07-30 11:39 ` [Qemu-devel] [PATCH 15/15] dataplane: virtio-blk: support mutlti virtqueue Ming Lei
2014-07-30 12:42 ` [Qemu-devel] [PATCH 00/14] dataplane: optimization and multi virtqueue support Christian Borntraeger
2014-08-04 10:16 ` Stefan Hajnoczi
2014-08-04 10:45 ` Ming Lei
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=53D8F6F0.7040106@redhat.com \
--to=pbonzini@redhat.com \
--cc=famz@redhat.com \
--cc=kwolf@redhat.com \
--cc=ming.lei@canonical.com \
--cc=mst@redhat.com \
--cc=peter.maydell@linaro.org \
--cc=qemu-devel@nongnu.org \
--cc=stefanha@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).