On Tue, 2011-05-10 at 10:09 +0800, hayeswang wrote: > > Francois Romieu [mailto:romieu@fr.zoreil.com] > > Sent: Tuesday, May 10, 2011 3:23 AM > > To: David Woodhouse > > Cc: Hayeswang; netdev@vger.kernel.org > > Subject: Re: [PATCH 1/2] linux-firmware: update firmware for RTL8111E > > > > David Woodhouse : > > [...] > > > I'd be a lot happier if the WHENCE file contained version numbers. > > > Does the *driver* print a version string for the firmware > > after loading it? > > > > No. > > > > Hayes, do we agree that the firmware needs to embed a version > > identifier (and some checksum to identify the current > > firmware that don't embed one). > > > > I agree that. However, I have no idea about making it compatible with previous > parser and firmware. Let me think if there is a good way to do this. A PHY_DATA_AND at the beginning of a firmware blob instruction is a no-op, so you could use that to add version information in a compatible way. For RTL8168D, however, you are right that this would break compatibility with the earlier driver versions that only handled PHY_WRITE. Ben. -- Ben Hutchings Once a job is fouled up, anything done to improve it makes it worse.