[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20170124080818.48c26a6c@xeon-e3>
Date: Tue, 24 Jan 2017 08:08:18 -0800
From: Stephen Hemminger <stephen@...workplumber.org>
To: Dexuan Cui <decui@...rosoft.com>
Cc: "gregkh@...uxfoundation.org" <gregkh@...uxfoundation.org>,
"driverdev-devel@...uxdriverproject.org"
<driverdev-devel@...uxdriverproject.org>,
KY Srinivasan <kys@...rosoft.com>,
Haiyang Zhang <haiyangz@...rosoft.com>,
Stephen Hemminger <sthemmin@...rosoft.com>,
"olaf@...fle.de" <olaf@...fle.de>,
Rolf Neugebauer <rolf.neugebauer@...ker.com>,
"jasowang@...hat.com" <jasowang@...hat.com>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
"apw@...onical.com" <apw@...onical.com>
Subject: Re: [PATCH] Drivers: hv: vmbus: finally fix
hv_need_to_signal_on_read()
On Tue, 24 Jan 2017 06:54:46 +0000
Dexuan Cui <decui@...rosoft.com> wrote:
> +static inline void
> +init_cached_read_index(struct vmbus_channel *channel)
> +{
> + struct hv_ring_buffer_info *rbi = &channel->inbound;
> +
> + rbi->cached_read_index = rbi->ring_buffer->read_index;
> +}
Looks good thanks. This should go in right away. Which versions are impacted?
Should it also go to stable?
In a future patch, the API function names for interacting with the ring buffer
should be changed to all have common prefix (hv_) and maybe do a little rethinking
about what needs to be in ring buffer and what could be local variables.
For example, the cached_read_index is only useful over the span of the loop
reading from the ring buffer. For me, it would be cleaner with a ring_buffer
iterator object which could abstract the API better.
struct vmbus_ringbuffer_iter iter;
vmbus_begin_read(&iter, channel);
while ((desc = vmbus_next_read(&iter), channel) {
...
}
vmbus_end_read(&iter, channel);
Powered by blists - more mailing lists