Re: [Qemu-devel] [PATCH v4] fw_cfg: insert string blobs via qemu cmdline

[Gabriel L. Somlo]

On Thu, Oct 15, 2015 at 05:50:03PM +0200, Laszlo Ersek wrote:
> On 10/15/15 11:24, Gerd Hoffmann wrote:
> > On Mi, 2015-10-14 at 16:22 +0200, Laszlo Ersek wrote:
> >> Paolo,
> >>
> >> On 09/29/15 18:29, Gabriel L. Somlo wrote:
> >>> Allow users to provide custom fw_cfg blobs with ascii string
> >>> payloads specified directly on the qemu command line.
> >>>
> >>> Suggested-by: Jordan Justen <address@hidden>
> >>> Suggested-by: Laszlo Ersek <address@hidden>
> >>> Signed-off-by: Gabriel Somlo <address@hidden>
> >>> Reviewd-by: Laszlo Ersek <address@hidden>
> >>> ---
> >>>
> >>> New since v3:  s/content/string/g   for the option name, at Gerd's
> >>>           suggestion (to potentially allow for additional content
> >>>           types directly on the command line, should the need arise
> >>>           at a future date)
> >>>
> >>> Thanks,
> >>>   --Gabriel
> >>>
> >>>  docs/specs/fw_cfg.txt | 15 +++++++++++++++
> >>>  qemu-options.hx       |  7 ++++++-
> >>>  vl.c                  | 33 +++++++++++++++++++++++++++------
> >>>  3 files changed, 48 insertions(+), 7 deletions(-)
> >>
> >> I'm sure you're going to want to give me a hug for bringing this up, but
> >> can you please pick this up? O:-)
> >>
> >> Or maybe Peter can apply it directly?
> > 
> > I guess I'll go prepare a fw_cfg pull request, probably tomorrow.
> > There also is the fw_cfg_dma series from Marc ...
> 
> Great, thanks!
> 
> > 
> > While being at it: anything else which is ready?  The fw_cfg acpi
> > patches seem still to be on discussion still ...
> 
> That's my impression as well, yes. And, in any case, Marc's work and
> Gabriel's conflict (in context, not in purpose / implementation), so I
> guess Gabriel will have to rebase (purely because Marc's work has
> converged first).

That's right, and rebasing is what I had planned. Plus, I'll need to
figure out how to write a guest-side kernel driver which works with
both DT and ACPI, builds on all affected architectures, and hopefully
doesn't need to use *too* many ifdefs in the process :)

Cheers,
--Gabriel