"buffers" and "marshallers" are always pointing to the static
buffers inside the same structure.
Use single array fields to avoid having initialize and have them.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Julien Rope <jrope@redhat.com>
This macro was suggested to simplify hot path expensive checks
which should be disable in production environments.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Francesco Giudici <fgiudici@redhat.com>
Allows to check QUIC algorithm and code with multiple input/output
formats.
Was limited to just RGB and RGBA.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Remove some system checks.
Specifically:
- functions in the standard C (memset, memmove);
- functions we use but we assume in the specific implementation
present (dup2, fork);
- functions we don't use (error_at_line);
- headers in the standard C (stddef.h, stdint.h, string.h);
- headers present in all systems we support if are not checked
in the source files (malloc.h);
- types we assume being present (XintYY_t, pid_t).
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Include headers required for implementation only in the
implementation file.
This remove dependency and avoid the user component having to
check for header flags (in this case HAVE_CELT051 and HAVE_OPUS).
This make easier integration in other components.
This also make compilation faster.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
quic_config is used only by the implementation so include it
only from the implementation file.
This remove dependency.
This make easier integration in other components.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Allow the user to disable tests through -Dtests=false
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>
Allow the user to disable tests through --disable-tests, this is
especially useful for example to disable gdk-pixbuf dependency
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>
Make sure that guard do no change building out-of-tree or
with Meson.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Uri Lublin <uril@redhat.com>
The agent-interface is an experimental instrumentation library for
capturing and sharing Spice performance indicators with an external
agent.
--enable-instrumentation=[recorder/agent/no]
Enable instrumentation [default=no]
The former configuration option '--enable-recorder' is transformed
into '--enable-instrumentation=recorder'.
Signed-off-by: Kevin Pouget <kpouget@redhat.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>
agent_interface_start: this function allows launching the agent
interface (ie, its listening socket) on a given port.
agent_interface_set_on_connect_cb: this function allows passing a
callback function that will be triggered when a client (a Local Agent)
connects to the Agent Interface socket.
agent_interface_set_forward_quality_cb: this function allows SPICE to
provide a function that will forward Quality messages received by the
Agent Interface towards SPICE server, for a centralized processing of
the messages.
Signed-off-by: Kevin Pouget <kpouget@redhat.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>
When initialized (recorder_initialization), the Agent Interface launch
a GThread (handle_communications) that opens a TCP server socket and
waits for Smart Local Agent connections. When a Local Agent connects
to the sockets, the communication is initialized
(agent_initialize_communication), the communication socket is stored
and the list of Recorders is sent. In return, the local agent
indicates which recorders to enable.
On the SPICE side, the Agent Interface handles the record() calls
(recorder_append*). When a record is received from SPICE, and if the
recorder is enabled, the record entry is sent through the TCP
connection. Otherwise, the record is dropped.
Signed-off-by: Kevin Pouget <kpouget@redhat.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>
Check that combination of fields for an array does not
lead to unsafe code.
check_valid method came from generate_c_declaration with
some more check and it's use in demarshaller to validate
the array if the structure is not generated.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Victor Toso <victortoso@redhat.com>
Output attributes are the attributes that specify how to store
that field in the C structure.
There can be only one output type specified.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Victor Toso <victortoso@redhat.com>
The argument requires Meson 0.50 however it's already "false"
by default if "install_dir" is not provided so remove it
to keep compatibility with Meson 0.48.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Eduardo Lima (Etrunko) <etrunko@redhat.com>
This check make sure that output fields for member with @end (arrays)
are declared as empty arrays in output C structure.
This avoids output fields to be declared as pointer or other
invalid types.
The check is a compile time check so no code in object file
is generated.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Uri Lublin <uril@redhat.com>
This was some left-over during development of C structure
generations (the single structure generation was skipped).
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Uri Lublin <uril@redhat.com>
Just style, they do the same thing, but is more coherent
with the rest of the code.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Uri Lublin <uril@redhat.com>
Do it by adding @end tag.
Without it the allocated memory has no space for 'name'.
Also fix SpiceMsgMainArrayMessage tests/test-marshallers.h,
replacing int8_t* name with int8_t name[0].
This makes name an "in-structure" array with no pre-defined size
instead of a pointer.
The size is defined by the message size.
Signed-off-by: Uri Lublin <uril@redhat.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>
member = None is set before the if/else condition.
In the else code, when member is set it is checked
and if not-None it breaks out of the loop.
If the code is still in the loop for sure member is None.
Found by covscan.
Signed-off-by: Uri Lublin <uril@redhat.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>
We started disabling Celt support making the option required.
After 2 releases start making it disabled unless explicitly
enabled.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Snir Sheriber <ssheribe@redhat.com>
Pull version update and some features.
Version update changed copyright in files to LGPLv2, more
compatible with SPICE.
One of the feature is the support for absolute time.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
This should always be defined and including config.h is a requirement.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Victor Toso <victortoso@redhat.com>
git.mk seems to have issues with Makefile rules generating multiple
files so add manually this file.
This avoid SPICE server to generate distribution files with
"-dirty" suffix.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Victor Toso <victortoso@redhat.com>
Fixes out-of-tree builds.
For example spice-gtk out-of-tree build fails with:
../subprojects/spice-common/common/generated_client_marshallers.h:19:10:
fatal error: client_marshallers.h: No such file or directory
#include "client_marshallers.h"
^~~~~~~~~~~~~~~~~~~~~~
Signed-off-by: Uri Lublin <uril@redhat.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>
Access the items by names instead of indexes.
Signed-off-by: Eduardo Lima (Etrunko) <etrunko@redhat.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>
spice-gtk and spice-server will use spice_common_client_dep
and spice_common_server_dep as dependencies.
However they will depend on both spice-common client/server
libraries and their sources causing the sources to be compiled
multiple times and causes linker errors on spice-gtk.
The issue can be observed doing a "find -name \*generated\*.o"
in Meson build directory.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Signed-off-by: Eduardo Lima (Etrunko) <etrunko@redhat.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>
All generated file depends on generated_messages.h which is
generated too.
So add an explicit dependency from all generated file
(except generated_messages.h generator) to generated_messages.h
generator.
This fixes compiling SPICE server where generated_messages.h
was not generated at all.
Add dependency to the include header to make Meson generate the
proper include flag.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Signed-off-by: Eduardo Lima (Etrunko) <etrunko@redhat.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>
RecoderChannel::data message is the same as PlaybackChannel::data
one.
This fixed compilation of spice-gtk using updated spice-common.
Before generating the messages automatically the definition was:
typedef struct SpiceMsgPlaybackPacket {
uint32_t time;
uint8_t *data;
uint32_t data_size;
} SpiceMsgPlaybackPacket, SpiceMsgcRecordPacket;
so both messages were defined as struct SpiceMsgPlaybackPacket with
SpiceMsgcRecordPacket an alias to it.
But spice.proto for RecordChannel::data was changed from
SpiceMsgcRecordPacket (the alias) to SpiceMsgcPlaybackPacket
defining a new structure and breaking the alias.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Victor Toso <victortoso@redhat.com>
This commit adds a SPICE_UNREACHABLE macro (courtesy of Frediano)
so that gcc does not think that code control can go past
spice_return_{val_,}if_fail(), spice_critical() and spice_error()
This avoids this kind of warnings:
fallthrough.c:
#include "log.h"
int main(int argc, char **argv)
{
switch(argc) {
case 1:
spice_critical("foo");
default:
return 0;
}
}
$ gcc -c $(pkg-config --cflags --libs glib-2.0 spice-protocol)
-I common -Wimplicit-fallthrough=5 ./fallthrough.c
In file included from ./fallthrough.c:1:
./fallthrough.c: In function 'main':
common/log.h:73:5: warning: this statement may fall through [-Wimplicit-fallthrough=]
73 | spice_log(G_LOG_LEVEL_CRITICAL, SPICE_STRLOC, __FUNCTION__, "" format, ## __VA_ARGS__); \
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
./fallthrough.c:8:25: note: in expansion of macro 'spice_critical'
8 | spice_critical("foo");
| ^~~~~~~~~~~~~~
./fallthrough.c:9:17: note: here
9 | default:
| ^~~~~~~
Signed-off-by: Christophe Fergeau <cfergeau@redhat.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>
Pull some fixes and features.
One of the feature is the support for @output setting to redirect
log output.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Eduardo Lima (Etrunko) <etrunko@redhat.com>
QUIC_VERSION_MINOR is never used.. Set QUIC_VERSION_MINOR to the same
version as QUIC_VERSION_MAJOR to avoid breaking backwards compatibility,
and fix the QUIC_VERSION macro.
Signed-off-by: Christophe Fergeau <cfergeau@redhat.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>
test-marshallers.h is missing a #define _H_TEST_MARSHALLERS in order to
prevent multiple #include for the same header.
Signed-off-by: Christophe Fergeau <cfergeau@redhat.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>
They were suggested by gcc when using -Wsuggest-attribute=format
Signed-off-by: Christophe Fergeau <cfergeau@redhat.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>
encoder->type is only going to be set by lz_set_sizes() after the
error() call. We can use 'type' directly which is what encoder->type is
going to be set to.
Signed-off-by: Christophe Fergeau <cfergeau@redhat.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>
This fixes a warning about missing prototype for backtrace()
Signed-off-by: Christophe Fergeau <cfergeau@redhat.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>
canvas_base.c is #included by spice-common users. They currently don't
enable this warning, but if/when they do, we don't want code from
spice-common to trigger it.
Signed-off-by: Christophe Fergeau <cfergeau@redhat.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>
Generate and include C declarations.
Next patch will use this facility.
Since none of the spice.proto types are decorated with @declare,
adding the #include in messages.h won't have any bad consequences.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Christophe Fergeau <cfergeau@redhat.com>
