[Spice-devel] [PATCH spice-gtk 0/5] RFC: Improve handling of large clipboard data

[Marc-André Lureau]

I'll hold it for now, since we are still discussing the protocol message.

On Thu, Nov 7, 2013 at 11:42 PM, Jonathon Jongsma <jjongsma at redhat.com> wrote:
> Other than the minor issues I pointed out separately, the rest of this
> looks OK to me.
>
> On Wed, 2013-11-06 at 22:25 +0100, Marc-André Lureau wrote:
>> From: Marc-André Lureau <marcandre.lureau at redhat.com>
>>
>> Hi,
>>
>> It's quite easy to crash a spice-gtk client when a clipboard operation
>> take large data buffers. There are several ways this can go wrong (2
>> ways client side, 2 ways guest side), from what I have found, gdk &
>> glib are mostly safe (at least on win32 backend, after gbo 711546 and
>> 711553 are fixed).
>>
>> However, it would be nice to somehow prevent large clipboard
>> operations, so I propose the following patches. They will also need a
>> few vdagent changes to deal with the proposed MAX_CLIPBOARD message.
>>
>> An extra would be to make dos2unix function OOM safe, by getting rid
>> of GString. I have not done that, since the max-clipboard limit should
>> be sufficient in reasonable cases, but this will be an easy
>> supplement.
>>
>> Marc-André Lureau (5):
>>   Block sending clipboard data > max-clipboard
>>   Add SPICE_MAX_CLIPBOARD environment variable
>>   main: use self for main channel variable
>>   main: simplify usage of VD_AGENT_HAS_CAPABILITY
>>   main: send max-clipboard to agent
>>
>>  gtk/channel-main.c      | 144 ++++++++++++++++++++++++++++++++----------------
>>  gtk/spice-gtk-session.c |   9 ++-
>>  2 files changed, 104 insertions(+), 49 deletions(-)
>>
>
>



-- 
Marc-André Lureau