sysemu/sysemu.h is a rather unfocused dumping ground for stuff related
to the system-emulator. Evidence:
* It's included widely: in my "build everything" tree, changing
sysemu/sysemu.h still triggers a recompile of some 1100 out of 6600
objects (not counting tests and objects that don't depend on
qemu/osdep.h, down from 5400 due to the previous two commits).
* It pulls in more than a dozen additional headers.
Split stuff related to run state management into its own header
sysemu/runstate.h.
Touching sysemu/sysemu.h now recompiles some 850 objects. qemu/uuid.h
also drops from 1100 to 850, and qapi/qapi-types-run-state.h from 4400
to 4200. Touching new sysemu/runstate.h recompiles some 500 objects.
Since I'm touching MAINTAINERS to add sysemu/runstate.h anyway, also
add qemu/main-loop.h.
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20190812052359.30071-30-armbru@redhat.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
[Unbreak OS-X build]
In my "build everything" tree, changing sysemu/sysemu.h triggers a
recompile of some 5400 out of 6600 objects (not counting tests and
objects that don't depend on qemu/osdep.h).
Almost a third of its inclusions are actually superfluous. Delete
them. Downgrade two more to qapi/qapi-types-run-state.h, and move one
from char/serial.h to char/serial.c.
hw/semihosting/config.c, monitor/monitor.c, qdev-monitor.c, and
stubs/semihost.c define variables declared in sysemu/sysemu.h without
including it. The compiler is cool with that, but include it anyway.
This doesn't reduce actual use much, as it's still included into
widely included headers. The next commit will tackle that.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20190812052359.30071-27-armbru@redhat.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
In my "build everything" tree, changing qemu/main-loop.h triggers a
recompile of some 5600 out of 6600 objects (not counting tests and
objects that don't depend on qemu/osdep.h). It includes block/aio.h,
which in turn includes qemu/event_notifier.h, qemu/notify.h,
qemu/processor.h, qemu/qsp.h, qemu/queue.h, qemu/thread-posix.h,
qemu/thread.h, qemu/timer.h, and a few more.
Include qemu/main-loop.h only where it's needed. Touching it now
recompiles only some 1700 objects. For block/aio.h and
qemu/event_notifier.h, these numbers drop from 5600 to 2800. For the
others, they shrink only slightly.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20190812052359.30071-21-armbru@redhat.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Tested-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Tested-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20190812052359.30071-20-armbru@redhat.com>
In my "build everything" tree, changing hw/hw.h triggers a recompile
of some 2600 out of 6600 objects (not counting tests and objects that
don't depend on qemu/osdep.h).
The previous commits have left only the declaration of hw_error() in
hw/hw.h. This permits dropping most of its inclusions. Touching it
now recompiles less than 200 objects.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20190812052359.30071-19-armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Tested-by: Philippe Mathieu-Daudé <philmd@redhat.com>
docs/devel/tracing.txt explains "since many source files include
trace.h, [the generated trace.h use] a minimum of types and other
header files included to keep the namespace clean and compile times
and dependencies down."
Commit 4815185902 "trace: Add per-vCPU tracing states for events with
the 'vcpu' property" made them all include qom/cpu.h via
control-internal.h. qom/cpu.h in turn includes about thirty headers.
Ouch.
Per-vCPU tracing is currently not supported in sub-directories'
trace-events. In other words, qom/cpu.h can only be used in
trace-root.h, not in any trace.h.
Split trace/control-vcpu.h off trace/control.h and
trace/control-internal.h. Have the generated trace.h include
trace/control.h (which no longer includes qom/cpu.h), and trace-root.h
include trace/control-vcpu.h (which includes it).
The resulting improvement is a bit disappointing: in my "build
everything" tree, some 1100 out of 6600 objects (not counting tests
and objects that don't depend on qemu/osdep.h) depend on a trace.h,
and about 600 of them no longer depend on qom/cpu.h. But more than
1300 others depend on trace-root.h. More work is clearly needed.
Left for another day.
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Tested-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20190812052359.30071-8-armbru@redhat.com>
Some of the generated qapi-types-MODULE.h are included all over the
place. Changing a QAPI type can trigger massive recompiling. Top
scorers recompile more than 1000 out of some 6600 objects (not
counting tests and objects that don't depend on qemu/osdep.h):
6300 qapi/qapi-builtin-types.h
5700 qapi/qapi-types-run-state.h
3900 qapi/qapi-types-common.h
3300 qapi/qapi-types-sockets.h
3000 qapi/qapi-types-misc.h
3000 qapi/qapi-types-crypto.h
3000 qapi/qapi-types-job.h
3000 qapi/qapi-types-block-core.h
2800 qapi/qapi-types-block.h
1300 qapi/qapi-types-net.h
Clean up headers to include generated QAPI headers only where needed.
Impact is negligible except for hw/qdev-properties.h.
This header includes qapi/qapi-types-block.h and
qapi/qapi-types-misc.h. They are used only in expansions of property
definition macros such as DEFINE_PROP_BLOCKDEV_ON_ERROR() and
DEFINE_PROP_OFF_AUTO(). Moving their inclusion from
hw/qdev-properties.h to the users of these macros avoids pointless
recompiles. This is how other property definition macros, such as
DEFINE_PROP_NETDEV(), already work.
Improves things for some of the top scorers:
3600 qapi/qapi-types-common.h
2800 qapi/qapi-types-sockets.h
900 qapi/qapi-types-misc.h
2200 qapi/qapi-types-crypto.h
2100 qapi/qapi-types-job.h
2100 qapi/qapi-types-block-core.h
270 qapi/qapi-types-block.h
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Tested-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20190812052359.30071-3-armbru@redhat.com>
Linux terminal behavior (coming from vt100 I think) is somewhat strange
when it comes to line wraps: When a character is printed to the last
char cell of a line the cursor does NOT jump to the next line but stays
where it is. The line feed happens when the next character is printed.
So the valid range for the cursor position is not 0 .. width-1 but
0 .. width, where x == width represents the state where the line is
full but the cursor didn't jump to the next line yet.
The code for the 'clear from start of line' control sequence (ESC[1K)
fails to handle this corner case correctly and may call
console_clear_xy() with x == width. That will incorrectly clear the
first char cell of the next line, or in case the cursor happens to be on
the last line overflow the cell buffer by one character (three bytes).
Add a check to the loop to fix that.
Didn't spot any other places with the same problem. But it's easy to
miss that corner case, so also allocate one extra cell as precaution, so
in case we have simliar issues lurking elsewhere it at least wouldn't be
a buffer overflow.
v2: squashed in additional checks suggested by Christophe de Dinechin.
Reported-by: Alexander Oleinik <alxndr@bu.edu>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Christophe de Dinechin <dinechin@redhat.com>
Message-id: 20190701075301.14165-1-kraxel@redhat.com
In fullscreen mode, the window property of cocoaView may not be the key
window, and the current implementation would not re-grab cursor by left click
in fullscreen mode after ungrabbed in fullscreen mode with hot-key ctrl-opt-g.
This patch used value of isFullscreen as a short-cirtuit condition for
relative input device grabbing.
Signed-off-by: Chen Zhang <tgfbeta@me.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 2D2F1191-E82F-4B54-A6E7-73FFB953DE93@me.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
On Mojave, absolute input device, i.e. tablet, had trouble re-grabbing
the cursor in re-entry into the virtual screen area. In some cases,
the `window` property of NSEvent object was nil after cursor exiting from
window, hinting that the `-locationInWindow` method would return value in
screen coordinates. The current implementation used raw locations from
NSEvent without considering whether the value was for the window coordinates
or the macOS screen coordinates, nor the zooming factor for Zoom-to-Fit in
fullscreen mode.
In fullscreen mode, the fullscreen cocoa window might not be the key
window, therefore the location of event in virtual coordinates should
suffice.
This patches fixed boundary check methods for cursor in normal
and fullscreen with/without Zoom-to-Fit in Mojave.
Note: CGRect, -convertRectToScreen: and -convertRectFromScreen: were
used in coordinates conversion for compatibility reason.
Signed-off-by: Chen Zhang <tgfbeta@me.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: FA3FBC4F-5379-4118-B997-58FE05CC58F9@me.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
No header includes qemu-common.h after this commit, as prescribed by
qemu-common.h's file comment.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20190523143508.25387-5-armbru@redhat.com>
[Rebased with conflicts resolved automatically, except for
include/hw/arm/xlnx-zynqmp.h hw/arm/nrf51_soc.c hw/arm/msf2-soc.c
block/qcow2-refcount.c block/qcow2-cluster.c block/qcow2-cache.c
target/arm/cpu.h target/lm32/cpu.h target/m68k/cpu.h target/mips/cpu.h
target/moxie/cpu.h target/nios2/cpu.h target/openrisc/cpu.h
target/riscv/cpu.h target/tilegx/cpu.h target/tricore/cpu.h
target/unicore32/cpu.h target/xtensa/cpu.h; bsd-user/main.c and
net/tap-bsd.c fixed up]
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20190523143508.25387-3-armbru@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Check and use QemuDmaBuf->modifier in egl_dmabuf_import_texture()
for dmabuf imports.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Message-id: 20190529072144.26737-5-kraxel@redhat.com
Add modifier parameter to egl_get_fd_for_texture(), to return the used
modifier on dmabuf exports.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Message-id: 20190529072144.26737-4-kraxel@redhat.com
wchar_t may resolve to be an unsigned long on 32-bit architectures.
Using the %x conversion specifier will then give a compiler warning:
ui/curses.c: In function ‘get_ucs’:
ui/curses.c:492:49: error: format ‘%x’ expects argument of type ‘unsigned int’, but argument 3 has type ‘wchar_t’ {aka ‘long int’} [-Werror=format=]
492 | fprintf(stderr, "Could not convert 0x%04x "
| ~~~^
| |
| unsigned int
| %04lx
493 | "from wchar_t to a multibyte character: %s\n",
494 | wch, strerror(errno));
| ~~~
| |
| wchar_t {aka long int}
ui/curses.c:504:49: error: format ‘%x’ expects argument of type ‘unsigned int’, but argument 3 has type ‘wchar_t’ {aka ‘long int’} [-Werror=format=]
504 | fprintf(stderr, "Could not convert 0x%04x "
| ~~~^
| |
| unsigned int
| %04lx
505 | "from a multibyte character to UCS-2 : %s\n",
506 | wch, strerror(errno));
| ~~~
| |
| wchar_t {aka long int}
Fix this by casting the wchar_t value to an unsigned long and using %lx
as the conversion specifier.
Fixes: b7b664a4fe
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Message-id: 20190527142540.23255-1-mreitz@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Use a better interface for random numbers than rand().
Fail gracefully if for some reason we cannot use the crypto system.
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
There were 3 copies of this code, one of which used the wrong
data size for the failure indicator.
Reviewed-by: Laurent Vivier <lvivier@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
When allowing multiple down-events in a row (key autorepeat) we can't
use change_bit() any more to update the state, because autorepeat events
don't change the key state. We have to explicitly use set_bit() and
clear_bit() instead.
Cc: qemu-stable@nongnu.org
Fixes: 3592186015 kbd-state: don't block auto-repeat events
Buglink: https://bugs.launchpad.net/qemu/+bug/1828272
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Message-id: 20190514042443.10735-1-kraxel@redhat.com
In a GVT-g setup with dmabuf and GTK GUI, the current 2D texture at
surface_gl_update_texture is not necessarily
surface->texture. Adding a glBindTexture fixes related crashes and
artifacts, and is generally more secure.
Signed-off-by: HOU Qiming <hqm03ster@gmail.com>
Tested-by: Marcel Apfelbaum<marcel.apfelbaum@gmail.com>
Signed-off-by: Marcel Apfelbaum <marcel.apfelbaum@gmail.com>
Message-id: 20190507080501.26712-1-marcel.apfelbaum@gmail.com
[fixed malformed patch, rebase to master]
Signed-off-by: Marcel Apfelbaum <marcel.apfelbaum@gmail.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
The chars/attr fields are curses internals, setcchar and getcchar have
to be used instead.
Signed-off-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Tested-by: Kamil Rytarowski <n54@gmx.com>
Message-Id: <20190427183307.12796-3-samuel.thibault@ens-lyon.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
E.g. BSD and Solaris even use locale-specific encoding there.
We thus have to go through the native multibyte representation and use
mbrtowc/wcrtomb to make a proper conversion.
Signed-off-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Tested-by: Kamil Rytarowski <n54@gmx.com>
Message-Id: <20190427183307.12796-2-samuel.thibault@ens-lyon.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
On some systems wchar_t is "long int", on others just "int".
So go cast to "long int" and adjust the printf format accordingly.
Reported-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20190402073018.17747-1-kraxel@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Some trace points are attributed to the wrong source file. Happens
when we neglect to update trace-events for code motion, or add events
in the wrong place, or misspell the file name.
Clean up with help of cleanup-trace-events.pl. Same funnies as in the
previous commit, of course. Manually shorten its change to
linux-user/trace-events to */signal.c.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-id: 20190314180929.27722-6-armbru@redhat.com
Message-Id: <20190314180929.27722-6-armbru@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
We spell out sub/dir/ in sub/dir/trace-events' comments pointing to
source files. That's because when trace-events got split up, the
comments were moved verbatim.
Delete the sub/dir/ part from these comments. Gets rid of several
misspellings.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20190314180929.27722-3-armbru@redhat.com
Message-Id: <20190314180929.27722-3-armbru@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
The iconv_t are opened but never closed.
Spotted by Coverity: CID 1399708
Spotted by Coverity: CID 1399709
Spotted by Coverity: CID 1399713
Signed-off-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Message-Id: <20190314172524.9290-1-samuel.thibault@ens-lyon.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
cchar_t can contain not only attr and chars fields, but also ext_color.
Initialize the whole structure to zero instead of enumerating fields.
Spotted by Coverity: CID 1399711
Signed-off-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Message-Id: <20190315130932.26094-1-samuel.thibault@ens-lyon.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
When adding '-fsanitize=undefined' in compiling configuration
and connect VM with vnc, it reports following error:
ui/vnc-enc-tight.c:910:13: runtime error: load of
misaligned address 0x621000466513 for type 'uint32_t',
which requires 4 byte alignment
This patch fix this issue.
Signed-off-by: Li Qiang <liq3ea@163.com>
Message-id: 20190318010442.14897-1-liq3ea@163.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
This uses iconv to convert glyphs from the specified VGA font encoding to
unicode, and makes use of cchar_t instead of chtype when using ncursesw,
which allows to store all wide char as well as the WACS values. The default
charset is made CP437 since that is the charset of the hardware default VGA
font. This also makes the curses backend set the LC_CTYPE locale to "" to
allow curses to emit wide characters.
Signed-off-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Cc: Eddie Kohler <ekohler@gmail.com>
Acked-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20190311135127.2229-3-samuel.thibault@ens-lyon.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
curses will use it for proper wide output support.
Signed-off-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Message-Id: <20190311135127.2229-2-samuel.thibault@ens-lyon.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
I had to include an enum for audio sampling formats into qapi, but that
meant duplicating the audfmt_e enum. This patch replaces audfmt_e and
associated values with the qapi generated AudioFormat enum.
This patch is mostly a search-and-replace, except for switches where the
qapi generated AUDIO_FORMAT_MAX caused problems.
Signed-off-by: Kővágó, Zoltán <DirtY.iCE.hu@gmail.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-id: 01251b2758a1679c66842120b77c0fb46d7d0eaf.1552083282.git.DirtY.iCE.hu@gmail.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
The VNC server has historically had support for ACLs to check both the
SASL username and the TLS x509 distinguished name. The VNC server was
responsible for creating the initial ACL, and the client app was then
responsible for populating it with rules using the HMP 'acl_add' command.
This is not satisfactory for a variety of reasons. There is no way to
populate the ACLs from the command line, users are forced to use the
HMP. With multiple network services all supporting TLS and ACLs now, it
is desirable to be able to define a single ACL that is referenced by all
services.
To address these limitations, two new options are added to the VNC
server CLI. The 'tls-authz' option takes the ID of a QAuthZ object to
use for checking TLS x509 distinguished names, and the 'sasl-authz'
option takes the ID of another object to use for checking SASL usernames.
In this example, we setup two authorization rules. The first allows any
client with a certificate issued by the 'RedHat' organization in the
'London' locality. The second ACL allows clients with either the
'joe@REDHAT.COM' or 'fred@REDHAT.COM' kerberos usernames. Both checks
must pass for the user to be allowed.
$QEMU -object tls-creds-x509,id=tls0,dir=/home/berrange/qemutls,\
endpoint=server,verify-peer=yes \
-object authz-simple,id=authz0,policy=deny,\
rules.0.match=O=RedHat,,L=London,rules.0.policy=allow \
-object authz-simple,id=authz1,policy=deny,\
rules.0.match=fred@REDHAT.COM,rules.0.policy=allow \
rules.0.match=joe@REDHAT.COM,rules.0.policy=allow \
-vnc 0.0.0.0:1,tls-creds=tls0,tls-authz=authz0,
sasl,sasl-authz=authz1 \
...other QEMU args...
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-id: 20190227145755.26556-2-berrange@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
vnc aborts display update jobs on video mode switches and page flips.
That can cause vnc update stalls in case an unfinished vnc job gets
aborted. The vnc client will never receive the requested update then.
Fix that by copying the state from job_update back to update in that
case.
Reports complain about stalls with two or more clients being connected
at the same time, on some but not all connections. I suspect it can
also happen with a single connection, multiple connections only make
this more much likely to happen.
Buglink: https://bugzilla.redhat.com/show_bug.cgi?id=1662260
Reported-by: Ying Fang <fangying1@huawei.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Ying Fang <fangying1@huawei.com>
Message-id: 20190305130930.24516-1-kraxel@redhat.com
This makes use of wide curses functions instead of 8bit functions. This
allows to type e.g. accented letters.
Unfortunately, key codes are then returned with values that could be
confused with wide characters by ncurses, so we need to add a maybe_keycode
variable to know whether the returned value is a key code or a character
(curses with wide support), or possibly both (curses without wide support).
The translation tables thus also need to be separated into key code
translation and character translation. The curses2foo helper makes it easier
to use them.
Signed-off-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Message-id: 20190304210532.7840-1-samuel.thibault@ens-lyon.org
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
By default, curses will only report single ESC key event after 1s delay,
since ESC is also used for keypad escape sequences. This however makes
users believe that ESC is not working. Reducing to 25ms provides good user
experience, while still allowing 25ms for keypad sequences to get in, which
should be enough.
Signed-off-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Message-Id: <20190303172557.17139-1-samuel.thibault@ens-lyon.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
The OSX Mojave release is more picky about enforcing the Cocoa API
restriction that only the main thread may perform UI calls. To
accommodate this we need to restructure the Cocoa code:
* the special OSX main() creates a second thread and uses
that to call the vl.c qemu_main(); the original main
thread goes into the OSX event loop
* the refresh, switch and update callbacks asynchronously
tell the main thread to do the necessary work
* the refresh callback no longer does the "get events from the
UI event queue and handle them" loop, since we now use
the stock OSX event loop. Instead our NSApplication sendEvent
method will either deal with them or pass them on to OSX
All these things have to be changed in one commit, to avoid
breaking bisection.
Note that since we use dispatch_get_main_queue(), this bumps
our minimum version requirement to OSX 10.10 Yosemite (released
in 2014, unsupported by Apple since 2017).
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Roman Bolshakov <r.bolshakov@yadro.com>
Tested-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-id: 20190225102433.22401-8-peter.maydell@linaro.org
Message-id: 20190214102816.3393-8-peter.maydell@linaro.org
When we switch away from our custom event handling, we still want to
be able to have first go at any events our application receives,
because in full-screen mode we want to send key events to the guest,
even if they would be menu item activation events. There are several
ways we could do that, but one simple approach is to subclass
NSApplication so we can implement a custom sendEvent method.
Do that, but for the moment have our sendEvent just invoke the
superclass method.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: Roman Bolshakov <r.bolshakov@yadro.com>
Tested-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-id: 20190225102433.22401-7-peter.maydell@linaro.org
Message-id: 20190214102816.3393-7-peter.maydell@linaro.org
Currently the handleEvent method will directly call the NSApp
sendEvent method for any events that we want to let OSX deal
with. When we rearrange the event handling code, the way that
we say "let OSX have this event" is going to change. Prepare
for that by refactoring so that handleEvent returns a flag
indicating whether it consumed the event.
Suggested-by: BALATON Zoltan <balaton@eik.bme.hu>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: Roman Bolshakov <r.bolshakov@yadro.com>
Tested-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-id: 20190225102433.22401-6-peter.maydell@linaro.org
Message-id: 20190214102816.3393-6-peter.maydell@linaro.org
Move the console/device menu creation code functions
further up in the source file, next to the code which
creates the initial menus. We're going to want to
change the location we call these functions from in
the next patch.
This commit is a pure code move with no other changes.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: Roman Bolshakov <r.bolshakov@yadro.com>
Tested-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-id: 20190225102433.22401-5-peter.maydell@linaro.org
Message-id: 20190214102816.3393-5-peter.maydell@linaro.org
Factor out the long code sequence in main() which creates
the initial set of menus. This will make later patches
which move initialization code around a bit clearer.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: Roman Bolshakov <r.bolshakov@yadro.com>
Tested-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-id: 20190225102433.22401-4-peter.maydell@linaro.org
Message-id: 20190214102816.3393-4-peter.maydell@linaro.org
Currently the switchSurface method takes a DisplaySurface. We want
to change our DisplayChangeListener's dpy_gfx_switch callback
to do this work asynchronously on a different thread. The caller
of the switch callback will free the old DisplaySurface
immediately the callback returns, so to ensure that the
other thread doesn't access freed data we need to switch
to using the underlying pixman image instead. The pixman
image is reference counted, so we will be able to take
a reference to it to avoid it vanishing too early.
In this commit we only change the switchSurface method
to take a pixman image, and keep the flow of control
synchronous for now.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: Roman Bolshakov <r.bolshakov@yadro.com>
Tested-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-id: 20190225102433.22401-3-peter.maydell@linaro.org
Message-id: 20190214102816.3393-3-peter.maydell@linaro.org
The Cocoa UI should run on the main thread; this is enforced
in OSX Mojave. In order to be able to run on the main thread,
we need to make sure we hold the iothread lock whenever we
call into various QEMU UI midlayer functions.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Roman Bolshakov <r.bolshakov@yadro.com>
Tested-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-id: 20190225102433.22401-2-peter.maydell@linaro.org
Message-id: 20190214102816.3393-2-peter.maydell@linaro.org
The 'qemu_acl' type was a previous non-QOM based attempt to provide an
authorization facility in QEMU. Because it is non-QOM based it cannot be
created via the command line and requires special monitor commands to
manipulate it.
The new QAuthZ subclasses provide a superset of the functionality in
qemu_acl, so the latter can now be deleted. The HMP 'acl_*' monitor
commands are converted to use the new QAuthZSimple data type instead
in order to provide temporary backwards compatibility.
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Tested-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Add a new display backend that will configure Spice to allow a remote
client to control QEMU in a similar fashion as other QEMU display
backend/UI like GTK.
For this to work, it will set up Spice server with a unix socket, and
register a VC chardev that will be exposed as Spice ports. A QMP
monitor is also exposed as a Spice port, this allows the remote client
fuller qemu control and state handling.
- doesn't handle VC set_echo() - this doesn't seem a strong
requirement, very few front-end use it
- spice options can be tweaked with other -spice arguments
- Windows support shouldn't be hard to do, but will probably use a TCP
port instead
- we may want to watch the child process to quit automatically if it
crashed
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Tested-by: Victor Toso <victortoso@redhat.com>
Message-id: 20190221110703.5775-12-marcandre.lureau@redhat.com
[ kraxel: squash incremental fix ]
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
If no -name is given, let's use a friendly "QEMU version" server
name. This is sometime exposed on spice client side, for example on
remote-viewer title.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Tested-by: Victor Toso <victortoso@redhat.com>
Message-id: 20190221110703.5775-11-marcandre.lureau@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
spice_server_vm_start/stop() was added to help migration state (commit
f5bb039c6d).
However, a paused VM could keep running the spice server. This will
allow a Spice client to keep sending commands to a spice chardev. This
allows to stop/cont a VM from a Spice monitor port. Character
devices (vdagent/usb/smartcard/..) should not read from Spice when the
VM is paused.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Tested-by: Victor Toso <victortoso@redhat.com>
Message-id: 20190221110703.5775-6-marcandre.lureau@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Passing several -spice options to qemu command line, or calling
several time qemu_opts_set() will ignore all but the first option
list. Since the spice server is a singleton, it makes sense to merge
all the options, the last value being the one taken into account.
This changes the behaviour from, for ex:
$ qemu... -spice port=5900 -spice port=5901 -> port: 5900
to:
$ qemu... -spice port=5900 -spice port=5901 -> port: 5901
(if necessary we could instead produce an error when an option is
given twice, although this makes handling default values and such more
complicated)
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Tested-by: Victor Toso <victortoso@redhat.com>
Message-id: 20190221110703.5775-5-marcandre.lureau@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
