On Tue, Aug 14, 2018 at 3:01 PM Christophe Fergeau wrote:
>
> The simpletlv_* functions return lengths as an int, but the test is
> casting it to a size_t before comparing it against -1. This is not
> giving the expected result on i686, causing the test to fail. The
> variables used in the test do
Move code handling VDAGENTD_FILE_XFER_STATUS
from agent_read_complete() to a separate function
to make the switch statement less cluttered.
Improve formatting a bit.
---
src/vdagentd/vdagentd.c | 61 +
1 file changed, 37 insertions(+), 24 deletions(-)
diff
Some older parts of the code currently use
memory functions defined in stdlib.h
and usually handle allocation errors.
On the other hand, newer parts of the code
and GLib/GTK+ functions themselves commonly use
wrappers provided by GLib that terminate
the program when there isn't enough memory.
So
Move code handling VDAGENTD_GUEST_XORG_RESOLUTION
from agent_read_complete() to a separate function
to make the switch statement less cluttered.
Remove unnecessary res variable.
---
src/vdagentd/vdagentd.c | 61 +
1 file changed, 31 insertions(+), 30 deleti
Rewrite console-kit.c using GDBus instead of libdbus.
GDBus provides a more convenient way to access DBus objects.
All DBus method calls share a fair amount of common code here,
so add a helper function dbus_call_method_for_string().
---
src/vdagentd/console-kit.c | 597 --
Remove type_to_string, no_types arguments from
udscs_connect() and udscs_server_new().
udscs is used only in vdagent.c and vdagentd.c
and in both cases the args are the same
(vdagentd_messages, VDAGENTD_NO_MESSAGES).
Add debug_print_message_header().
---
src/udscs.c | 52 +
udscs_get_peer_cred() is currently used only in
vdagentd.c to obtain PID, so let's drop udscs_get_peer_cred()
and add udscs_get_peer_pid() instead.
---
src/udscs.c | 6 +++---
src/udscs.h | 4 ++--
src/vdagentd/vdagentd.c | 2 +-
3 files changed, 6 insertions(+), 6 deletion
Both console-kit.c and systemd-login.c
have been moved to GDBus.
---
Makefile.am | 2 --
configure.ac | 1 -
2 files changed, 3 deletions(-)
diff --git a/Makefile.am b/Makefile.am
index b291b19..64ed406 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -47,7 +47,6 @@ src_spice_vdagent_SOURCES =
Rewrite virtio-port.c using VDAgentConnection to integrate it
into GMainLoop and simplify the code.
virtio_port_destroy() does NOT close the underlying FD immediately.
GSources attached to GMainContext can be processed during
vdagent_virtio_port_flush() call.
Apart from that, the behavior stays th
Add a set of helper functions built around GIO that can be used to
easily write messages to and read from the given FD.
Since VDAgentConnection uses GIO,
it integrates well with GMainLoop.
Read messages must begin with a header of a fixed size.
Message body size can vary.
User of VDAgentConnecti
In console-kit.c, FD polling is handled internally by GDBus.
In systemd-login.c, FD is integrated into the GMainLoop
using GIOChannel.
Users of session-info.h are notified by ActiveSessionChangeCb
when the active session changes.
This renders the session_info_get_fd() obsolete.
---
src/vdagentd
Divide creation of udscs_server into 2 steps:
1) udscs_server_new()
- allocates new udscs_server struct and inits it
2) udscs_server_listen_to_fd(), udscs_server_listen_to_address()
- starts accepting new connections
udscs_server_listen_to_address() creates a new socket and
binds it to the giv
Require UNIX-specific GIO package to build spice-vdagent.
This includes
- GDBus which is going to be used instead of libdbus in
console-kit.c and systemd-login.c
- I/O stream classes and networking APIs that are going to
be used in udscs.c and virtio-port.c instead of low-level
POSIX I/O f
This is necessary for the following GDBus integration,
which drops session_info_get_fd(). The vdagentd therefore
won't be able to detect session changes using select().
---
src/vdagentd/console-kit.c| 2 +-
src/vdagentd/dummy-session-info.c | 2 +-
src/vdagentd/session-info.h | 4 +++
Rewrite systemd-login.c using GDBus instead of libdbus.
Use GDBusProxy. This is less efficient than the previous impl
as it caches all properties and listens to all signals of the
given DBus object. But it is just so much more convenient...
---
src/vdagentd/systemd-login.c | 254 +++--
Rewrite udscs.c using VDAgentConnection to integrate it
into GMainLoop and simplify the code.
udscs_destroy_connection() does NOT close
the underlying FD immediately.
Apart from that, the behavior stays the same.
Drop support for select() in udscs_server, remove
udscs_server_fill_fds(), udscs_ser
If the virtio port is destroyed explicitly
by calling vdagent_virtio_port_destroy(),
by_user is set to TRUE, otherwise to FALSE.
This will be used later with GMainLoop.
---
src/vdagentd/virtio-port.c | 24 +++-
src/vdagentd/virtio-port.h | 10 +-
2 files changed, 20 in
All command line options now have long names
as they are required by GLib.
Change types of some global variables that hold the options:
- const char * --> gchar *
- int --> gboolean
Define DEFAULT_UINPUT_DEVICE as "/dev/uinput",
since there's already DEFAULT_VIRTIO_PORT_PATH, VDAGENTD_SO
This is purely a preparatory patch as it renders
the vdagentd non-functional.
Remove main while loop with FD polling.
udscs, virtio-port and session-info will be integrated
into the GMainLoop in the following commits.
Use g_unix_signal_add() to handle SIGINT, SIGHUP, SIGTERM.
SIGQUIT handling is
Hi all,
as the subject suggests, this series further deepens the integration of GLib
into spice-vdagent(d).
Change summary:
* udscs.c and virtio-port.c are built upon common base, VDAgentConnection, that
handles the lower-level I/O operations
* systemd-login.c and console-kit.c use GDBus inste
On Mon, 2018-08-13 at 12:18 +0200, Christophe Fergeau wrote:
> Signed-off-by: Christophe Fergeau
> ---
> NEWS | 33 +
> configure.ac | 2 +-
> 2 files changed, 34 insertions(+), 1 deletion(-)
>
> diff --git a/NEWS b/NEWS
> index 51bb4a2e6..1577b013f 10064
Acked-by: Jonathon Jongsma
On Mon, 2018-08-13 at 12:18 +0200, Christophe Fergeau wrote:
> Signed-off-by: Christophe Fergeau
> ---
> docs/manual/manual.txt | 51
> ++
> 1 file changed, 51 insertions(+)
>
> diff --git a/docs/manual/manual.txt b/docs/manua
On Tue, Aug 14, 2018 at 09:03:30AM +0200, Jorge Olmos wrote:
> Hi,
> On Mon, 2018-08-13 at 17:25 +0200, Christophe Fergeau wrote:
> > Hey,
> >
> > On Mon, Aug 13, 2018 at 03:45:06PM +0200, Jorge Olmos wrote:
> > > This patch configures libusb in the context used in win-usb-dev,
> > > with
> > > th
On Mon, Aug 13, 2018 at 12:28:32PM -0400, Frediano Ziglio wrote:
> >
> > On Mon, Aug 13, 2018 at 11:41:31AM -0400, Frediano Ziglio wrote:
> > > >
> > > > Hey,
> > > >
> > > > At this point, I don't think we have anything which is going to give us
> > > > an h265 encoded stream, and latest spice-
Ninja parallel build fails in certain environments, so it is required
specify those files as a dependency so that vncdisplaykeymap.c will only
be built after all files it depends on are generated.
The problem is that, when using meson declare_dependency() function,
only header files should be used
Signed-off-by: Eduardo Lima (Etrunko)
---
.gitlab-ci.yml | 22 +-
1 file changed, 21 insertions(+), 1 deletion(-)
diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index a34eb0e..e5e90af 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -3,7 +3,7 @@ image: fedora:latest
before_
in v5:
* Added missing -Wno-cast-function-type build flag
* No need to declare_dependency() for generated vncdisplaykeymap files
now that extension has been renamed to '.h'
Eduardo Lima (Etrunko) (3):
Rename generated vncdisplaykeymap_*.c files to .h
Add support for building with meson/ni
In a comparison with current autotools build system, meson/ninja
provides a huge improvement in build speed, while keeping the same
functionalities currently available and being considered more user
friendly.
The new system coexists within the same repository with the current one,
so we can do mor
On Tue, 2018-08-14 at 10:35 +0100, Frediano Ziglio wrote:
> As documented in PK11_ReadRawAttribute data should be freed with
> PORT_Free, not free or g_free.
Good catch. This might even make the valgrind suppression file slimmer
in the end.
> Signed-off-by: Frediano Ziglio
> ---
> src/cac.c
The simpletlv_* functions return lengths as an int, but the test is
casting it to a size_t before comparing it against -1. This is not
giving the expected result on i686, causing the test to fail. The
variables used in the test do not need to be size_t, and can be changed
to int to match the return
Signed-off-by: Eduardo Lima (Etrunko)
---
src/meson.build | 15 ++-
1 file changed, 6 insertions(+), 9 deletions(-)
diff --git a/src/meson.build b/src/meson.build
index c9429f5..b9ee039 100644
--- a/src/meson.build
+++ b/src/meson.build
@@ -301,21 +301,18 @@ if spice_gtk_has_gtk
Sorry, this one is missing at least one hunk, will send a v3.
Christophe
On Tue, Aug 14, 2018 at 02:40:58PM +0200, Christophe Fergeau wrote:
> The simpletlv_* functions return lengths as an int, but the test is
> casting it to a size_t before comparing it against -1. This is not
> giving the expe
On 14/08/18 04:47, Frediano Ziglio wrote:
>> On 13/08/18 07:23, Frediano Ziglio wrote:
Ninja parallel build fails in certain environments, so it is required
specify those files as a dependency so that vncdisplaykeymap.c will only
be built after all files it depends on are genera
On Tue, 2018-08-14 at 14:41 +0200, Christophe Fergeau wrote:
> Document return value of simpletlv_get_length and simpletlv_encode
> and
> fix a few typos.
Sharp eye or good spellchecker. Looks good to me. Ack.
> ---
> src/simpletlv.h | 7 +--
> 1 file changed, 5 insertions(+), 2 deletions(-)
Document return value of simpletlv_get_length and simpletlv_encode and
fix a few typos.
---
src/simpletlv.h | 7 +--
1 file changed, 5 insertions(+), 2 deletions(-)
diff --git a/src/simpletlv.h b/src/simpletlv.h
index bf785e4..d12f52d 100644
--- a/src/simpletlv.h
+++ b/src/simpletlv.h
@@ -48,
The simpletlv_* functions return lengths as an int, but the test is
casting it to a size_t before comparing it against -1. This is not
giving the expected result on i686, causing the test to fail. The
variables used in the test do not need to be size_t, and can be changed
to int to match the return
Porprietary -> Proprietary
Signed-off-by: Frediano Ziglio
---
src/card_7816.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/card_7816.c b/src/card_7816.c
index dc12138..bf8bdce 100644
--- a/src/card_7816.c
+++ b/src/card_7816.c
@@ -669,7 +669,7 @@ vcard7816_vm_process_a
Using a static writeable buffer and threads could lead to race
conditions (like multiple threads calling the same function).
The fci_template buffer is changed some lines below with a memcpy.
Signed-off-by: Frediano Ziglio
---
src/card_7816.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
As documented in PK11_ReadRawAttribute data should be freed with
PORT_Free, not free or g_free.
Signed-off-by: Frediano Ziglio
---
src/cac.c| 3 ++-
src/vcard_emul_nss.c | 2 +-
2 files changed, 3 insertions(+), 2 deletions(-)
diff --git a/src/cac.c b/src/cac.c
index ba63685..6c49d9
On Mon, Aug 13, 2018 at 07:12:24PM +0200, Marc-André Lureau wrote:
> On Mon, Aug 13, 2018 at 6:11 PM, Christophe Fergeau
> wrote:
> > Without it, make check from tarballs will fail.
>
> Actually make distcheck passes, but the tests are skipped, right?
When running make check from a tarball in r
>
> The simpletlv_* functions return lengths as an int, but the test is
> casting it to a size_t before comparing it against -1. This is not
> giving the expected result on i686, causing the test to fail.
>
> Signed-off-by: Christophe Fergeau
Maybe a bit OT. Do we support arm32 ?
I'm asking bec
On Mon, 2018-08-13 at 19:15 +0200, Marc-André Lureau wrote:
> On Mon, Aug 13, 2018 at 6:11 PM, Christophe Fergeau <
> cferg...@redhat.com> wrote:
> > The simpletlv_* functions return lengths as an int, but the test is
> > casting it to a size_t before comparing it against -1. This is not
> > giving
>
> The handler is meant to ignore the X11 error, name it properly.
>
> Signed-off-by: Lukáš Hrázký
For the series
Acked-by: Frediano Ziglio
Frediano
> ---
> src/vdagent/x11-randr.c | 10 +-
> 1 file changed, 5 insertions(+), 5 deletions(-)
>
> diff --git a/src/vdagent/x11-randr.c
> On 13/08/18 07:23, Frediano Ziglio wrote:
> >>
> >> Ninja parallel build fails in certain environments, so it is required
> >> specify those files as a dependency so that vncdisplaykeymap.c will only
> >> be built after all files it depends on are generated.
> >>
> >> The problem is that, when us
Hi,
On Mon, 2018-08-13 at 17:25 +0200, Christophe Fergeau wrote:
> Hey,
>
> On Mon, Aug 13, 2018 at 03:45:06PM +0200, Jorge Olmos wrote:
> > This patch configures libusb in the context used in win-usb-dev,
> > with
> > the same backend as the context initialized in usb-device-manager.
> >
> > Oth
45 matches
Mail list logo