On 6/1/23 10:49, Eelco Chaudron wrote:
On 1 Jun 2023, at 10:29, Maxime Coquelin wrote:
On 6/1/23 10:15, Eelco Chaudron wrote:
On 1 Jun 2023, at 4:18, Xia, Chenbo wrote:
-----Original Message-----
From: Maxime Coquelin <maxime.coque...@redhat.com>
Sent: Wednesday, May 31, 2023 5:29 PM
To: Xia, Chenbo <chenbo....@intel.com>; Thomas Monjalon
<tho...@monjalon.net>; Eelco Chaudron <echau...@redhat.com>;
david.march...@redhat.com
Cc: dev@dpdk.org
Subject: Re: [PATCH v3 4/4] vhost: add device op to offload the interrupt
kick
On 5/31/23 08:19, Xia, Chenbo wrote:
-----Original Message-----
From: Maxime Coquelin <maxime.coque...@redhat.com>
Sent: Tuesday, May 30, 2023 11:17 PM
To: Thomas Monjalon <tho...@monjalon.net>; Eelco Chaudron
<echau...@redhat.com>; Xia, Chenbo <chenbo....@intel.com>;
david.march...@redhat.com
Cc: dev@dpdk.org
Subject: Re: [PATCH v3 4/4] vhost: add device op to offload the
interrupt
kick
On 5/30/23 15:16, Thomas Monjalon wrote:
30/05/2023 15:02, Maxime Coquelin:
On 5/17/23 11:09, Eelco Chaudron wrote:
This patch adds an operation callback which gets called every time
the
library wants to call eventfd_write(). This eventfd_write() call
could
result in a system call, which could potentially block the PMD
thread.
The callback function can decide whether it's ok to handle the
eventfd_write() now or have the newly introduced function,
rte_vhost_notify_guest(), called at a later time.
This can be used by 3rd party applications, like OVS, to avoid
system
calls being called as part of the PMD threads.
Signed-off-by: Eelco Chaudron <echau...@redhat.com>
---
lib/vhost/meson.build | 2 ++
lib/vhost/rte_vhost.h | 23 +++++++++++++++++-
lib/vhost/socket.c | 63
++++++++++++++++++++++++++++++++++++++++++++++---
lib/vhost/version.map | 9 +++++++
lib/vhost/vhost.c | 38 ++++++++++++++++++++++++++++++
lib/vhost/vhost.h | 58 ++++++++++++++++++++++++++++++++---
---
-------
6 files changed, 171 insertions(+), 22 deletions(-)
The patch looks good to me, but that's the first time we use function
versioning in Vhost library, so I'd like another pair of eyes to be
sure
I don't miss anything.
Reviewed-by: Maxime Coquelin <maxime.coque...@redhat.com>
Thomas, do we need to mention it somewhere in the release note?
If compatibility is kept, I think we don't need to mention it.
Thanks Thomas for the information.
Maxime
About release note, except the versioning, there is also one new API
introduced
in this patch, so we still need to mention this in release note
Right, good catch.
Eelco, let me know what you would put, I'll add it while applying (No
need for a new revision).
Btw, the vhost_lib.rst also needs a new item..
What about the following?
diff --git a/doc/guides/prog_guide/vhost_lib.rst
b/doc/guides/prog_guide/vhost_lib.rst
index e8bb8c9b7b..0f964d7a4a 100644
--- a/doc/guides/prog_guide/vhost_lib.rst
+++ b/doc/guides/prog_guide/vhost_lib.rst
@@ -334,6 +334,11 @@ The following is an overview of some key Vhost API
functions:
Clean DMA vChannel finished to use. After this function is called,
the specified DMA vChannel should no longer be used by the Vhost library.
+* ``rte_vhost_notify_guest(int vid, uint16_t queue_id)``
+
+ Inject the offloaded interrupt received by the 'guest_notify' callback,
+ into the vhost device's queue.
+
Vhost-user Implementations
--------------------------
Maxime do you want to add it, or do you want a new rev?
That's fine, I just added it now :) Can you check the next-virtio main
branch and confirm all is OK for your series?
https://git.dpdk.org/next/dpdk-next-virtio/log/
One small spelling issue, the rest looks good:
+* **Added callback API support for interrupt handling in the vhost library.**
+
+ A new callback, guest_notify, is introduced that can be used to handle the
+ interrupt kick outside of the datapath fast path. In addition, a new API,
+ rte_vhost_notify_guest(), is added to raise the interrupt outside of the
+ past path.
Last sentence spells ‘past path’ instead of ‘fast path’.
I edited the line but it didn't caught my eye!
Fixing it now.
Thanks,
Maxime
Thanks,
Maxime
Thanks,
Chenbo
Thanks,
Maxime
Thanks,
Chenbo
