aboutsummaryrefslogtreecommitdiff
AgeCommit message (Collapse)AuthorFilesLines
2023-09-04ui/gtk: fix leaks found wtih fuzzingDmitry Frolov1-1/+2
It is true, that there is no problem during runtime from the first sight, because the memory is lost just before qemu exits. Nevertheless, this change is necessary, because AddressSanitizer is not able to recognize this situation and produces crash-report (which is false-positive in fact). Lots of False-Positive warnings are davaluing problems, found with fuzzing, and thus the whole methodology of dynamic analysis. This patch eliminates such False-Positive reports, and makes every problem, found with fuzzing, more valuable. Fixes: 060ab76356 ("gtk: don't exit early in case gtk init fails") Signed-off-by: Dmitry Frolov <frolov@swemel.ru> Reviewed-by: Michael Tokarev <mjt@tls.msk.ru> Message-Id: <20230825115818.1091936-1-frolov@swemel.ru>
2023-09-04ui/vdagent: Unregister input handler of mouse during finalizationGuoyi Tu1-0/+3
Input handler resource should be released when VDAgentChardev object finalize Signed-off-by: Guoyi Tu <tugy@chinatelecom.cn> Signed-off-by: dengpengcheng <dengpc12@chinatelecom.cn> Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com> Message-Id: <e7f5e172abf797d454e00a4bbe53af83e4aa4497.1692281173.git.tugy@chinatelecom.cn>
2023-09-04ui/vdagent: call vdagent_disconnect() when agent connection is lostGuoyi Tu1-0/+3
when the agent connection is lost, the input handler of the mouse doesn't deactivate, which results in unresponsive mouse events in VNC windows. To fix this issue, call vdagent_disconnect() to reset the state each time the frontend disconncect Signed-off-by: Guoyi Tu <tugy@chinatelecom.cn> Signed-off-by: dengpengcheng <dengpc12@chinatelecom.cn> Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com> Message-Id: <71fd5a58fd09f10cdb35f167b2edb5669300116e.1692281173.git.tugy@chinatelecom.cn>
2023-09-04ui/dbus: implement damage regions for GLBilal Elmoussaoui1-7/+25
Currently, when using `-display dbus,gl=on` all updates to the client become "full scanout" updates, meaning there is no way for the client to limit damage regions to the display server. Instead of using an "update count", this patch tracks the damage region and propagates it to the client. This was less of an issue when clients were using GtkGLArea for rendering, as you'd be doing full-surface redraw. To be efficient, the client needs both a DMA-BUF and the damage region to be updated. Co-authored-by: Christian Hergert <chergert@redhat.com> Signed-off-by: Bilal Elmoussaoui <belmouss@redhat.com> Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com> Message-Id: <20230814125802.102160-1-belmouss@redhat.com>
2023-09-04ui/dbus: Properly dispose touch/mouse dbus objectsBilal Elmoussaoui1-0/+2
Fixes: 142ca628a7 ("ui: add a D-Bus display backend") Fixes: de9f844ce2 ("ui/dbus: Expose a touch device interface") Signed-off-by: Bilal Elmoussaoui <belmouss@redhat.com> Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com> Message-Id: <20230901124507.94087-1-belmouss@redhat.com>
2023-09-04ui/vnc-enc-tight: Avoid dynamic stack allocationPhilippe Mathieu-Daudé1-5/+6
Use autofree heap allocation instead of variable-length array on the stack. The codebase has very few VLAs, and if we can get rid of them all we can make the compiler error on new additions. This is a defensive measure against security bugs where an on-stack dynamic allocation isn't correctly size-checked (e.g. CVE-2021-3527). Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com> [PMM: expanded commit message] Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Francisco Iglesias <frasse.iglesias@gmail.com> Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com> Message-Id: <20230818151057.1541189-4-peter.maydell@linaro.org>
2023-09-04ui/vnc-enc-hextile: Use static rather than dynamic length stack arrayPeter Maydell1-1/+7
In the send_hextile_tile_* function we create a variable length array data[]. In fact we know that the client_pf.bytes_per_pixel is at most 4 (enforced by set_pixel_format()), so we can make the array a compile-time fixed length of 1536 bytes. The codebase has very few VLAs, and if we can get rid of them all we can make the compiler error on new additions. This is a defensive measure against security bugs where an on-stack dynamic allocation isn't correctly size-checked (e.g. CVE-2021-3527). Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> [ Marc-André - rename BPP to MAX_BYTES_PER_PIXEL ] Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Message-Id: <20230818151057.1541189-3-peter.maydell@linaro.org>
2023-09-04ui/spice-display: Avoid dynamic stack allocationPeter Maydell1-1/+2
Use an autofree heap allocation instead of a variable-length array on the stack in qemu_spice_create_update(). The codebase has very few VLAs, and if we can get rid of them all we can make the compiler error on new additions. This is a defensive measure against security bugs where an on-stack dynamic allocation isn't correctly size-checked (e.g. CVE-2021-3527). Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com> Message-Id: <20230818151057.1541189-2-peter.maydell@linaro.org>
2023-09-04ui/vc: change the argument for QemuTextConsoleMarc-André Lureau5-18/+16
Those functions are specifc to text/vc console, make that explicit from the argument type. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-45-marcandre.lureau@redhat.com>
2023-09-04ui/vc: do not parse VC-specific options in Spice and GTKMarc-André Lureau5-7/+12
In commit 6f974c843c ("gtk: overwrite the console.c char driver"), I shared the VC console parse handler with GTK. And later on in commit d8aec9d9 ("display: add -display spice-app launching a Spice client"), I also used it to handle spice-app VC. This is not necessary, the VC console options (width/height/cols/rows) are specific, and unused by tty-level GTK/Spice VC. This is not a breaking change, as those options are still being parsed by QAPI ChardevVC. Adjust the documentation about it. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-44-marcandre.lureau@redhat.com>
2023-09-04ui/vc: move text console invalidate in helperMarc-André Lureau2-4/+10
This will allow to split the VC code in a separate unit more easily. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-43-marcandre.lureau@redhat.com>
2023-09-04ui/console: minor stylistic changesMarc-André Lureau1-6/+4
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-42-marcandre.lureau@redhat.com>
2023-09-04ui/vc: skip text console resize when possibleMarc-André Lureau1-3/+9
This function is called on invalidate, on each cursor blink. Avoid the extra copy when the console size didn't change. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-41-marcandre.lureau@redhat.com>
2023-09-04ui/console: fold text_console_update_cursor_timerMarc-André Lureau1-8/+2
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Message-Id: <20230830093843.3531473-40-marcandre.lureau@redhat.com>
2023-09-04ui/console: assert(surface) where appropriateMarc-André Lureau1-0/+4
The QemuTextConsole code paths assume a surface is being used as scanout, let's make this more explicit. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-39-marcandre.lureau@redhat.com>
2023-09-04ui/console: rename vga_ functions with qemu_console_Marc-André Lureau1-13/+12
They are not specific to VGA. Let's use the object type name as prefix instead, to avoid confusion. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Message-Id: <20230830093843.3531473-38-marcandre.lureau@redhat.com>
2023-09-04ui/console: use QEMU_PIXMAN_COLOR helpersMarc-André Lureau2-21/+24
QEMU_RGB macro is actually defining a pixman color. Make this explicit in the macro name. Move it to qemu-pixman.h so it can be used elsewhere, as done in the following patch. Finally, define QEMU_PIXMAN_COLOR_{BLACK,GRAY}, to avoid need to look up the VGA color table from the QemuConsole placeholder surface rendering. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Message-Id: <20230830093843.3531473-37-marcandre.lureau@redhat.com>
2023-09-04ui/console: declare console types in console.hMarc-André Lureau2-19/+21
We are going to split the console.c unit next, and implement separately. But we need to check the underlying type in various places. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-36-marcandre.lureau@redhat.com>
2023-09-04ui/vc: use common text console surface creationMarc-André Lureau1-7/+2
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-35-marcandre.lureau@redhat.com>
2023-09-04ui/console: remove need for g_width/g_heightMarc-André Lureau1-5/+3
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-34-marcandre.lureau@redhat.com>
2023-09-04ui/console: simplify getting active_console sizeMarc-André Lureau1-4/+2
We can get the active console dimension regardless of its kind, by simply giving NULL as argument. It will fallback with the given value when the dimensions aren't known. This will also allow to move the code in a separate unit more easily. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-33-marcandre.lureau@redhat.com>
2023-09-04ui/vc: move some text console initialization to qom handlersMarc-André Lureau1-11/+11
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-32-marcandre.lureau@redhat.com>
2023-09-04ui/vc: fold text_console_do_init() in vc_chr_open()Marc-André Lureau1-49/+31
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Message-Id: <20230830093843.3531473-31-marcandre.lureau@redhat.com>
2023-09-04ui/console: move graphic fields to QemuGraphicConsoleMarc-André Lureau1-46/+64
Move fields specific to graphic console to the console subclass. qemu_console_get_head() is adapated to accomodate QemuTextConsole, and always returns 0. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-30-marcandre.lureau@redhat.com>
2023-09-04ui/vc: move text fields to QemuTextConsoleMarc-André Lureau1-86/+91
Now we can instantiate the specific console with its own fields. Pass the most appropriate type to the various functions, and cast up to QEMU_CONSOLE as necessary. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-29-marcandre.lureau@redhat.com>
2023-09-04ui/console: free more QemuConsole resourcesMarc-André Lureau1-1/+4
This code path is probably not executed at this point, since console aren't being released. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-28-marcandre.lureau@redhat.com>
2023-09-04ui/vc: move cursor_timer initialization to QemuTextConsole classMarc-André Lureau1-2/+4
The timer is only relevant when a text console exists. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-27-marcandre.lureau@redhat.com>
2023-09-04ui/console: allocate ui_timer in QemuConsoleMarc-André Lureau1-2/+6
Although at this point only QemuGraphicConsole have hw_ops that implements ui_info() callback, it makes sense to keep the code in the base QemuConsole, to simplify conditions for the caller. As of now, the code didn't reach a NULL timer because dpy_set_ui_info() checks if dpy_ui_info_supported() (hw_ops->ui_info != NULL), which is false for text_console_ops. This is a bit fragile, let simply allocate and free the timer in the base class. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-26-marcandre.lureau@redhat.com>
2023-09-04ui/console: update the head from unused QemuConsoleMarc-André Lureau1-1/+1
When recycling unused QemuConsole, we should still set the associated head number for correct information and lookups. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-25-marcandre.lureau@redhat.com>
2023-09-04ui/console: specialize console_lookup_unused()Marc-André Lureau2-4/+4
graphics_console_init() is expected to return a graphic console. The function doesn't need to be exported. We are going to specialize further QemuGraphicConsole & QemuTextConsole. The two will not be interchangeable anymore. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-24-marcandre.lureau@redhat.com>
2023-09-04ui/console: remove new_console()Marc-André Lureau1-14/+5
The constructor helper isn't of much used now. "head" is only specified for graphic console (and default to 0), and we are going to move it to QemuGraphicConsole next. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-23-marcandre.lureau@redhat.com>
2023-09-04ui/console: register the console from qemu_console_init()Marc-André Lureau1-2/+1
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-22-marcandre.lureau@redhat.com>
2023-09-04ui/console: instantiate a specific console typeMarc-André Lureau1-28/+19
This will allow to move code/data to the specific console types. Replace console_type_t with object type check. QemuConsole can be abstract. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-21-marcandre.lureau@redhat.com>
2023-09-04ui/console: introduce different console objectsMarc-André Lureau1-0/+84
Boilerplate code to introduce different object types for the different console types. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-20-marcandre.lureau@redhat.com>
2023-09-04ui/console: change new_console() to use object initializationMarc-André Lureau1-36/+56
Object construction should be done in respective object instance and class handlers. Introduce qemu_console_register() to split out the registration logic. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-19-marcandre.lureau@redhat.com>
2023-09-04ui/console: use OBJECT_DEFINE_TYPE for QemuConsoleMarc-André Lureau1-14/+17
The following patch will move some object initialization to the corresponding handlers. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-18-marcandre.lureau@redhat.com>
2023-09-04ui/vc: move VCCharDev specific fields out of QemuConsoleMarc-André Lureau1-74/+73
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-17-marcandre.lureau@redhat.com>
2023-09-04ui/vc: pass VCCharDev to VC-specific functionsMarc-André Lureau1-31/+39
Even though they actually use more of QemuConsole at this point, it makes it clearer those functions are only used from the chardev implementation. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-16-marcandre.lureau@redhat.com>
2023-09-04ui/vc: fold text_update_xy()Marc-André Lureau1-9/+4
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-15-marcandre.lureau@redhat.com>
2023-09-04ui/vc: replace variable with static text attributes defaultMarc-André Lureau1-17/+13
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-14-marcandre.lureau@redhat.com>
2023-09-04ui/vc: move VCChardev declaration at the topMarc-André Lureau1-6/+6
To allow easier refactoring in following patches. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-13-marcandre.lureau@redhat.com>
2023-09-04ui/vc: VC always has a DisplayState nowMarc-André Lureau1-4/+0
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-12-marcandre.lureau@redhat.com>
2023-09-04ui/console: new_console() cannot failMarc-André Lureau1-5/+0
There is no code path that could allow a NULL return there. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-11-marcandre.lureau@redhat.com>
2023-09-04ui/console: get the DisplayState from new_console()Marc-André Lureau1-19/+8
There is no obvious reason to defer text console initialization. We can simply take the global display state in new_console(). This simplify somewhat the code to allow moving the VC to a separate unit. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-10-marcandre.lureau@redhat.com>
2023-09-01ui/console: drop have_gfxMarc-André Lureau1-7/+0
All usages have been removed. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-9-marcandre.lureau@redhat.com>
2023-09-01ui/console: call dpy_gfx_update() regardless of have_gfxMarc-André Lureau1-1/+1
The function will handle the case when no listeners are gfx, without extra meaningful cost. This allows to get rid of DisplayState dependency in VC implementation. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-8-marcandre.lureau@redhat.com>
2023-09-01ui/console: console_select() regardless of have_gfxMarc-André Lureau1-6/+4
Even if we don't have a gfx listener, we should call displaychangelistener_display_console() which handle that case correctly. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-7-marcandre.lureau@redhat.com>
2023-09-01ui/vc: drop have_textMarc-André Lureau1-29/+13
If there are no "text" listener, the callback will simply be ignored. The rest of text handling can be done cheaply. This allows to remove some dependency on DisplayState from VC implementation. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-6-marcandre.lureau@redhat.com>
2023-09-01ui/vc: replace vc_chr_write() with generic qemu_chr_write()Marc-André Lureau1-3/+3
We shouldn't call the callback directly, but use the chardev API, unless there is a clear reason. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Message-Id: <20230830093843.3531473-5-marcandre.lureau@redhat.com>
2023-09-01ui/qmp: move screendump to ui-qmp-cmds.cMarc-André Lureau3-195/+205
console.c unit is over-crowded. This code is specific to the handling of the QMP screendump command, so move it in ui-qmp-cmds. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Message-Id: <20230830093843.3531473-4-marcandre.lureau@redhat.com>
generated by cgit v1.1 at 2024-07-18 06:20:09 +0000