On Mon, Oct 16, 2017 at 4:53 PM, Frediano Ziglio wrote:
>>
>> From: Jakub Janků
>>
>> Replace existing main while-loop with GMainLoop.
>>
>> Use GIOChannel with g_io_add_watch() to manage IO
>> from x11 connections in the main loop.
>>
>> udscs_connect() now internally creates GSources
>> by call
>
> From: Jakub Janků
>
> This is required for the following GMainLoop integration,
> which utilizes some of the new functions/definitions, namely:
>
> [definition][GLib version]
> g_clear_pointer() 2.34
> G_SOURCE_REMOVE 2.32
> G_SOURCE_CONTINUE 2.32
> g_unix_signal_add() 2.30
>
> GLi
>
> From: Jakub Janků
>
> Replace existing main while-loop with GMainLoop.
>
> Use GIOChannel with g_io_add_watch() to manage IO
> from x11 connections in the main loop.
>
> udscs_connect() now internally creates GSources
> by calling g_io_add_watch().
> This enables GMainLoop integration,
> c
From: Jakub Janků
Replace existing main while-loop with GMainLoop.
Use GIOChannel with g_io_add_watch() to manage IO
from x11 connections in the main loop.
udscs_connect() now internally creates GSources
by calling g_io_add_watch().
This enables GMainLoop integration,
clients no longer need to
From: Jakub Janků
Drop functions:
- udscs_client_fill_fds()
- udscs_client_handle_fds()
These became obsolete with GMainLoop integration.
Move required code to udscs_server_fill_fds()
and udscs_server_handle_fds(), respectively.
Acked-by: Frediano Ziglio
---
src/udscs.c | 48 ++
From: Victor Toso
This is mostly a preparatory patch to reduce the usage of globals and
the GMainLoop integration. The changes target one function by:
* Changing its name from client_setup() to client_setup_sync(): As it
important to clarify that the function block;
* Removing the 'reconnect'
From: Victor Toso
This patch creates vdagent_finalize_file_xfer() to finalize and stop
file xfer. Moving this code to a function removes some duplication.
Signed-off-by: Victor Toso
Acked-by: Frediano Ziglio
---
src/vdagent/vdagent.c | 22 --
1 file changed, 12 insertions(
From: Victor Toso
As we already depend on glib, let's remove code that glib can take
care of. In this case, we don't need to handle commandline parsing
ourselves.
In regard the global variables:
* static const char * -> static gchar * [only paths]
path variables: portdev, fx_dir, vdagentd_soc
Related to ongoing work to use GMainLoop and GTK integration
Based on work from Victor Toso
---
src/vdagent/vdagent.c | 121 ++
1 file changed, 72 insertions(+), 49 deletions(-)
diff --git a/src/vdagent/vdagent.c b/src/vdagent/vdagent.c
index 1f44
From: Jakub Janků
This is required for the following GMainLoop integration,
which utilizes some of the new functions/definitions, namely:
[definition][GLib version]
g_clear_pointer() 2.34
G_SOURCE_REMOVE 2.32
G_SOURCE_CONTINUE 2.32
g_unix_signal_add() 2.30
GLib version accross distribut
From: Victor Toso
This patch creates two functions:
- xfer_get_download_directory()
- vdagent_init_file_xfer()
The logic should be similar as it was before this patch, taking in
consideration the global variables fx_open_dir and fx_dir which are
set from command line.
Signed-off-by: Victor Toso
Changes since v3:
- remove GObject requirements, was not improving after
making udcsc independent.
Changes since v2:
- hide "-y" command line option in help
- remove 1 second delay introduced in previous versions,
when starting vdagent
- remove all GSources on VDAgent finalization
--
Major c
On Oct 16, 2017 4:20 PM, "Frediano Ziglio" wrote:
On Fri, Oct 13, 2017 at 3:49 PM Frediano Ziglio wrote:
>
> On Fri, Oct 13, 2017, 9:29 AM Frediano Ziglio wrote:
>
>> >
>> > On Wed, Oct 11, 2017 at 1:43 PM, Frediano Ziglio
>> wrote:
>> > >>
>> > >> Replace existing main while-loop with GMain
> On Fri, Oct 13, 2017 at 3:49 PM Frediano Ziglio < fzig...@redhat.com > wrote:
> > > > On Fri, Oct 13, 2017, 9:29 AM Frediano Ziglio < fzig...@redhat.com >
> > > > wrote:
> > >
> >
>
> > > > > >
> > > >
> > >
> >
>
> > > > > > On Wed, Oct 11, 2017 at 1:43 PM, Frediano Ziglio <
> > > > > >
On Mon, Oct 16, 2017 at 4:16 PM, Yedidyah Bar David wrote:
> On Mon, Oct 16, 2017 at 4:07 PM, Christophe Fergeau
> wrote:
>> My understanding is that the short log really should be saying "oVirt CI
>> support" rather than "standard CI support".
>
> Changed to 'standard oVirt CI support' and adde
On Mon, Oct 16, 2017 at 4:07 PM, Christophe Fergeau wrote:
> My understanding is that the short log really should be saying "oVirt CI
> support" rather than "standard CI support".
Changed to 'standard oVirt CI support' and added a link.
oVirt has also non-standard CI, which is currently used for
Looks good to me. Minor comment below. Adding also others.
On Mon, Oct 16, 2017 at 4:03 PM, Christophe Fergeau wrote:
> If these paths are unquoted, and the path contains spaces (C:\Program
> Files (x86)\...), this could be exploited by putting a binary with a
> crafted name (C:\Program.exe), lea
My understanding is that the short log really should be saying "oVirt CI
support" rather than "standard CI support".
Has gitlab built-in CI been considered for this work? In other words,
why use oVirt for CI rather than something more "standard"?
Where would the error messages go if CI fails? Are
From: Sandro Bonazzola
This patch was previously merged in ovirt gerrit with Change-Id
Ie16cc0902c4393e62b074661a86f2f5bb9a1efc1, but removed from master
branch there for review on spice-devel.
For reference:
https://www.ovirt.org/develop/dev-process/build-and-test-standards/
Change-Id: I042f3
If these paths are unquoted, and the path contains spaces (C:\Program
Files (x86)\...), this could be exploited by putting a binary with a
crafted name (C:\Program.exe), leading to priviledge escalation as this
is a service that is being started.
https://www.commonexploits.com/unquoted-service-pat
From: Sandro Bonazzola
This patch was previously merged in ovirt gerrit with Change-Id
Ie16cc0902c4393e62b074661a86f2f5bb9a1efc1, but removed from master
branch there for review on spice-devel.
Change-Id: I042f30d87b0730ae76e2c8b537e46a56cbbf65f7
Signed-off-by: Sandro Bonazzola
---
automation/
From: Sandro Bonazzola
Change-Id: Ie16cc0902c4393e62b074661a86f2f5bb9a1efc1
Signed-off-by: Sandro Bonazzola
---
automation/build-artifacts.packages | 15 +++
automation/build-artifacts.repos| 3 +++
automation/build-artifacts.sh | 23 +++
automation/ch
- Original Message -
> From: ge...@hostfission.com
> To: "Frediano Ziglio"
> Cc: "Christophe Fergeau" ,
> spice-devel@lists.freedesktop.org
> Sent: Saturday, October 14, 2017 8:34:42 AM
> Subject: Re: [Spice-devel] Spice Protocol Missing Headers
>
> On 2017-10-13 20:05, Frediano Ziglio
Hi,
> Why extending QXL protocol is not worth doing but writing workaround
> on workaround is?
Problem is extending the qxl protocol requires changes on the host
side. A fix which doesn't need that has the advantage that it'll work
on hosts with old spice-server and qemu versions.
We can look
On Fri, Oct 13, 2017 at 12:20:02PM -0400, Frediano Ziglio wrote:
> >
> > For the series,
> > Acked-by: Christophe Fergeau
> > though I would put the test after the fix, or squash both.
> >
> > Christophe
> >
>
> Some TDD methodologies suggest that test should first fail.
> But this does not wo
>
> On Thu, 2017-10-12 at 15:52 +0100, Frediano Ziglio wrote:
> > When guest close the device the host device has to be reset too.
> > This make easier to restart the guest device which can happen in case
> > of reboot, agent issues or if we want to update the agent.
> >
> > Signed-off-by: Fredia
>
> Hi,
>
> We need to figure a way to handle intel vgpu with spice, especially how
> to handle the cursor.
>
> What we'll get from the vgpu is two dma-bufs: One for the primary
> plane (aka guest display), one for the cursor plane. We also get the
> position of the cursor. We do *not* get th
27 matches
Mail list logo
