It is used in canvas_get_lz4() which is guarded by USE_LZ4, and
in canvas_get_lz() which is guarded by SW_CANVAS_CACHE.
Acked-by: Christophe Fergeau <cfergeau@redhat.com>
LZ image decompression was broken for 16 bpp:
- stride was computed not computed correctly (as width*4). This caused
also a buffer underflow;
- stride in pixman is always multiple of 4 bytes (so for 16 bpp is
ALIGN(width*2, 4)) so image decompressed by lz_decode as some missing
bytes to be fixed.
The alignment code is reused from LZ4 function.
This fix also https://bugzilla.redhat.com/show_bug.cgi?id=1285469.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Pavel Grunt <pgrunt@redhat.com>
While testing spice-server on EL6, I was getting random crashes in the
glib logging code because g_logv was called recursively even though this
was not visible in a backtrace.
It turns out on older glib versions (EL6 has 2.28), g_logv is not
thread-safe unless g_thread_init() is called first. If g_thread_init()
is not called, the GMutex/GPrivate calls g_logv makes are turned into
no-ops, which is going to cause the recursion issue I was seeing.
This commit adds a call to g_thread_init() for these older glib
versions.
(gdb) bt
0x7fff9f9fb110 "item.type: 114", unused_data=0x0) at gmessages.c:863
format=0x7ffff50e72ac "item.type: %d", args1=0x7fff9f9fb640) at gmessages.c:517
SPICE_LOG_LEVEL_DEBUG, strloc=0x7ffff50e72ba "dcc.c:1652", function=
0x7ffff50e7320 "release_item_before_push", format=0x7ffff50e72ac "item.type: %d", args=
0x7fff9f9fb640) at log.c:163
SPICE_LOG_LEVEL_DEBUG, strloc=0x7ffff50e72ba "dcc.c:1652", function=
0x7ffff50e7320 "release_item_before_push", format=0x7ffff50e72ac "item.type: %d") at log.c:195
at dcc.c:1652
at dcc.c:1719
at dcc-send.c:2450
at red-channel.c:578
at red-channel.c:1587
at event-loop.c:122
0x7ffff4fb2ef2 <watch_func>, user_data=0x7fff980244e0) at giounix.c:166
0x7ffff86ec770) at gmain.c:3092
This was done through a GOnce called every time spice_log() is called,
now it will always be called at spice-server startup.
This means the unit test needs to be updated as SPICE_DEBUG/ABORT_LEVEL
must now be set before the process starts up rather than before the
first spice_log call, and the deprecation warning these environment
variables trigger cannot be caught using g_test_expect_message() as
they are output before g_test_init() is called.
Avoid having to call function at runtime to inialize static.
Old functions are defined as deprecated for compatibility.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Pavel Grunt <pgrunt@redhat.com>
Allow to define functions executed at program/shared object initialization
or close.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Victor Toso <victortoso@redhat.com>
This reverts commit 7665dcf1bb.
Also revert the related build-sys changes to fix the build.
codegen generated code depends on spice-common code (marshaller,
messages etc), it makes more sense to keep the generator along
this. Otherwise a newer protocol release will fail to build older
projects.
*.proto files are required as well, since it generates code that parent
modules depend on unconditionnaly.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
In some cases, it might be worth to be able to send a message with a -1
fd, has the protocol permits. Change add_fd/get_fd in order to track
if the caller wanted to send -1.
Signed-off-by: Marc-André Lureau <marcandre.lureau@gmail.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>
Code is in the slow path, this reduce space needed for data+code.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Jonathon Jongsma <jjongsma@redhat.com>
avoid to compute the string length twice and use memcpy instead of
strcpy which is faster not having to check for terminator.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Jonathon Jongsma <jjongsma@redhat.com>
Actually not causing problems as when used is always followed by another
terminator but better to fix the definition.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Jonathon Jongsma <jjongsma@redhat.com>
spice-common has been duplicating glib logging methods for a long while.
Now that spice-common is depending on glib, it's more consistent to use
glib logging too. However, the code base is still using spice logging
functions.
This commit aims to make spice logging go through glib logging system,
while keeping the same behaviour for the SPICE_ABORT_LEVEL and
SPICE_DEBUG_LEVEL environment variables. They are deprecated however in
favour of the glib alternatives (G_DEBUG and G_MESSAGES_DEBUG).
Reviewed-by: Jonathon Jongsma <jjongsma@redhat.com>
If header guards are working as expected, there should not be multiple
definitions of these macros. If they are redefined somewhere else, this
is a bug we want to fix.
Acked-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Jonathon Jongsma <jjongsma@redhat.com>
The marshaller can't serialize fd in memory stream. Instead, append the
fd to the marshaller structure. The marshaller user is responsible for
sending the fd when the message is sent. The fd to send can be retrieved
with spice_marshaller_get_fd().
Note: only a single fd is supported with this API, supporting multiple
fd is left for the future if it becomes necessary.
Signed-off-by: Marc-André Lureau <marcandre.lureau@gmail.com>
Acked-by: Frediano Ziglio <fziglio@redhat.com>
This is consistent with the rest of the code making clear
fields are pointers.
Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Acked-by: Uri Lublin <uril@redhat.com>
Acked-by: Victor Toso <victortoso@redhat.com>
On LE machine, color order when creating surface will always be A/XRGB.
On BE machines the color order will sometimes be ARGB and sometimes BGRA/X.
This is because we actually create the surface two times on BE machines.
Once with BE order, and then again with LE order. Copying data inbetween
theese two surfaces will byteswap the colors automatically.
This change introduces cases for BGRA/X color byte orders on BE machines.
After the previous commit, spice_bitmap_try_as_pixman() can be
simplified as its #ifdef WORDS_BIGENDIAN exactly match what the
PIXMAN_LE_ constants do.
When using image compression on PowerPC architecture, colors are in
wrong order ARGB -> BGRA.
This commit introduces macros, that will change the color order
according to machine endianness.
Theese macros are similar to QEMU macros in qemu-pixman.h
libcacard.h requires 2.5.1. Keep compatibility for older versions until
the transition in distros to the standalone version is done.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
[ Christophe: add < 2.5.1 fallback ]
Signed-off-by: Christophe Fergeau <cfergeau@redhat.com>
Change inet_aton function to glib functions.
inet_aton only supported IPv4 addresses, and wasn't available on windows
machines. GInetAddress functions support IPv6 natively, and requires less
boilerplate code then IPv6 gettaddrinfo().
The linearization optimization that avoids copying only one item must
check that there are no further marshallers in the chain.
Just to be clear, we are trying to marshall a message like this:
message {
uint32 data_size;
uint64 *data[data_size] @marshall;
} SomeData;
Where the data field points to an array in dynamic memory. Marshalling
and demarshalling functions look good. The marshalling function creates
a submarshaller for the data field and links it to the root marshaller.
But when it comes to sending the data through the wire, only the
data_size field gets sent. We have observed that, in
spice_marshaller_linearize, execution enters into the optimization that
avoids copying the data when the root marshaller only has one item, but
it ignores the following marshallers in the list. Checking if there are
more marshallers fixes the problem.
Now that spice-protocol ships the needed .proto files as well as the
corresponding python scripts, spice-common can use these in order to
generate the C code for the SPICE (de)marshallers.
Runtime conversion from a string to uint32 is storing the magic with the same
endianness on both LE and BE machines. This requires aditional byte swap
when sending magic between LE/BE machines.
Changing quic magic to a constant will ensure, that it will be always stored in
native endianness, and the second byte swap won't be needed.
Commit d39dfbfe changes lz magic to be always treated as LE when encoded.
Runtime conversion from a string to uint32 is storing the magic with the same
endianness on both LE and BE machines. This requires aditional byte swap
when sending magic between LE/BE machines.
Changing lz magic to a constant will ensure, that it will be always stored in
native endianness, and the second byte swap won't be needed.
This commit reverts d39dfbfe changes in lz.c while keeping the rest.
They will be needed in future commit for quic.c
