When the AioContext changes, we need to associate a LinuxAioState with
the new AioContext. Use the bdrv_attach_aio_context callback and call
the new aio_setup_linux_aio(), which will allocate a new AioContext if
needed, and return errors on failures. If it fails for any reason,
fallback to threaded AIO with an error message, as the device is already
in-use by the guest.

Signed-off-by: Nishanth Aravamudan <naravamudan@digitalocean.com>
---
Note this patch didn't exist in v2, but is a result of feedback to that
posting.

 block/file-posix.c | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)

diff --git a/block/file-posix.c b/block/file-posix.c
index 6a1714d4a8..ce24950acf 100644
--- a/block/file-posix.c
+++ b/block/file-posix.c
@@ -1730,6 +1730,21 @@ static BlockAIOCB *raw_aio_flush(BlockDriverState *bs,
     return paio_submit(bs, s->fd, 0, NULL, 0, cb, opaque, QEMU_AIO_FLUSH);
 }
 
+static void raw_aio_attach_aio_context(BlockDriverState *bs,
+                                       AioContext *new_context)
+{
+#ifdef CONFIG_LINUX_AIO
+    BDRVRawState *s = bs->opaque;
+    if (s->use_linux_aio) {
+        if (aio_setup_linux_aio(new_context) != 0) {
+            error_report("Unable to use native AIO, falling back to "
+                         "thread pool");
+            s->use_linux_aio = false;
+        }
+    }
+#endif
+}
+
 static void raw_close(BlockDriverState *bs)
 {
     BDRVRawState *s = bs->opaque;
@@ -2608,6 +2623,7 @@ BlockDriver bdrv_file = {
     .bdrv_refresh_limits = raw_refresh_limits,
     .bdrv_io_plug = raw_aio_plug,
     .bdrv_io_unplug = raw_aio_unplug,
+    .bdrv_attach_aio_context = raw_aio_attach_aio_context,
 
     .bdrv_truncate = raw_truncate,
     .bdrv_getlength = raw_getlength,
-- 
2.17.1

On Thu, 06/21 15:21, Nishanth Aravamudan wrote:
> When the AioContext changes, we need to associate a LinuxAioState with
> the new AioContext. Use the bdrv_attach_aio_context callback and call
> the new aio_setup_linux_aio(), which will allocate a new AioContext if
> needed, and return errors on failures. If it fails for any reason,
> fallback to threaded AIO with an error message, as the device is already
> in-use by the guest.
> 
> Signed-off-by: Nishanth Aravamudan <naravamudan@digitalocean.com>
> ---
> Note this patch didn't exist in v2, but is a result of feedback to that
> posting.

This should be squashed into patch 1, no?

Fam

Am 22.06.2018 um 04:25 hat Fam Zheng geschrieben:
> On Thu, 06/21 15:21, Nishanth Aravamudan wrote:
> > When the AioContext changes, we need to associate a LinuxAioState with
> > the new AioContext. Use the bdrv_attach_aio_context callback and call
> > the new aio_setup_linux_aio(), which will allocate a new AioContext if
> > needed, and return errors on failures. If it fails for any reason,
> > fallback to threaded AIO with an error message, as the device is already
> > in-use by the guest.
> > 
> > Signed-off-by: Nishanth Aravamudan <naravamudan@digitalocean.com>
> > ---
> > Note this patch didn't exist in v2, but is a result of feedback to that
> > posting.
> 
> This should be squashed into patch 1, no?

Yes, without it, patch 1 is incorrect. Specifically, at least the
assertion in aio_get_linux_aio() won't hold true without it.

Kevin

On 22.06.2018 [11:02:06 +0200], Kevin Wolf wrote:
> Am 22.06.2018 um 04:25 hat Fam Zheng geschrieben:
> > On Thu, 06/21 15:21, Nishanth Aravamudan wrote:
> > > When the AioContext changes, we need to associate a LinuxAioState with
> > > the new AioContext. Use the bdrv_attach_aio_context callback and call
> > > the new aio_setup_linux_aio(), which will allocate a new AioContext if
> > > needed, and return errors on failures. If it fails for any reason,
> > > fallback to threaded AIO with an error message, as the device is already
> > > in-use by the guest.
> > > 
> > > Signed-off-by: Nishanth Aravamudan <naravamudan@digitalocean.com>
> > > ---
> > > Note this patch didn't exist in v2, but is a result of feedback to that
> > > posting.
> > 
> > This should be squashed into patch 1, no?
> 
> Yes, without it, patch 1 is incorrect. Specifically, at least the
> assertion in aio_get_linux_aio() won't hold true without it.

Yes, you are right! Sorry about that. I'll send a v4 today.

-Nish