[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <MWHPR21MB1593F5A0494D7726E577E03DD7219@MWHPR21MB1593.namprd21.prod.outlook.com>
Date: Thu, 27 Jan 2022 21:22:34 +0000
From: "Michael Kelley (LINUX)" <mikelley@...rosoft.com>
To: Yanming Liu <yanminglr@...il.com>
CC: Wei Liu <wei.liu@...nel.org>,
Andrea Parri <parri.andrea@...il.com>,
"linux-hyperv@...r.kernel.org" <linux-hyperv@...r.kernel.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
"dri-devel@...ts.freedesktop.org" <dri-devel@...ts.freedesktop.org>,
"linux-fbdev@...r.kernel.org" <linux-fbdev@...r.kernel.org>,
KY Srinivasan <kys@...rosoft.com>,
Haiyang Zhang <haiyangz@...rosoft.com>,
Stephen Hemminger <sthemmin@...rosoft.com>,
Dexuan Cui <decui@...rosoft.com>,
"drawat.floss@...il.com" <drawat.floss@...il.com>,
"airlied@...ux.ie" <airlied@...ux.ie>,
"daniel@...ll.ch" <daniel@...ll.ch>,
"lkmlabelt@...il.com" <lkmlabelt@...il.com>
Subject: RE: [PATCH v2] hv: account for packet descriptor in maximum packet
size
From: Yanming Liu <yanminglr@...il.com> Sent: Wednesday, January 19, 2022 12:14 PM
>
> On Thu, Jan 20, 2022 at 2:12 AM Michael Kelley (LINUX)
> <mikelley@...rosoft.com> wrote:
> >
> > From: Wei Liu <wei.liu@...nel.org> Sent: Friday, January 14, 2022 11:13 AM
> > >
> > > On Mon, Jan 10, 2022 at 01:44:19AM +0100, Andrea Parri wrote:
> > > > (Extending Cc: list,)
> > > >
> > > > On Sun, Jan 09, 2022 at 05:55:16PM +0800, Yanming Liu wrote:
> >
> > The VSS driver in hv_snapshot.c allocates a receive buffer of 8 Kbytes
> > and sets max_pkt_size to 8 Kbytes. But the received messages are
> > all fixed size and small. I don't know why the driver uses an 8 Kbyte
> > receive buffer instead of 4 Kbytes, but the current settings are
> > more than sufficient.
> >
>
> Well, I'm not sure, on August 2021 there was a patch changing
> max_pkt_size to 8 KiB for VSS driver:
> https://lore.kernel.org/linux-hyperv/20210825190217.qh2c6yq5qr3ntum5@liuwe-devbox-debian-v2/T/
>
> The patch mentioned a 6304 bytes VSS message. Which is part of the
> reason I tried to address the more "general" problem of potentially
> mismatching buffer size.
>
This is certainly interesting. The Linux driver is not processing
all those bytes, so I'm not sure what Hyper-V is passing to the
guest. I'll check with the Hyper-V team to be sure.
Michael
Powered by blists - more mailing lists