Found these FDs unintentionally propagated into forked processes (e.g.
updater) during my local debugging. Use O_CLOEXEC to avoid that.
Test: Dump open FDs in updater process.
Test: Sideload a package on taimen (which uses framebuffer).
Test: Run graphics test on taimen.
Change-Id: I1fb65dd4a103c47d28b15973def9e9359c251ca3
GRSurface::Create() doesn't need to rely on caller specifying the buffer
size, as it can compute that info based on the given args.
This CL also uses `size_t` for all the parameters in
GRSurface::Create().
Test: Run recovery_unit_test on marlin.
Test: Build and boot into blueline recovery. `Run graphics test`.
Test: Build and boot into blueline charger mode.
Change-Id: Idec9381079196abf13553a475006fefcfca10950
This is a bug introduced while refactoring init_display_surface(), in
[1]. As a result, user of res_create_multi_display_surface(), which is
effectively `charger` right now, crashes due to buffer overrun.
This CL fixes the wrong arg and adds a sanity test for
res_create_multi_display_surface(). The testdata (battery_scale.png) is
copied from system/core/healthd/images/battery_scale.png.
[1] commit 44820ac1e3.
Bug: 119122296
Test: Run recovery_unit_test on marlin.
Test: Build and boot into charger mode on blueline. Verify that
`charger` no longer crashes.
Change-Id: Ib6d083e1512a9c3c6eb63874d26d22658921d693
Clone() allows duplicating the image that's stored in the GRSurface.
Test: Run recovery_unit_test.
Change-Id: Ia50d507c6200f2de5f17143775de805247a60e1f
As well as all the derived classes. Instances must be created with
Create(). A default copy ctor would mess up the ownership of the
mapped or allocated buffer in these classes, so that has been explicitly
removed.
Test: mmma -j bootable/recovery
Test: Run recovery_unit_test on marlin.
Test: `Run graphics test` on blueline.
Change-Id: I69ce001a9ec9e3ac851edb6ec4d3fa11f4aaea08
- Adds Create() that returns a GRSurfaceFbdev instance.
- Moves away from using the copy ctor (precisely assignment operator) of
GRSurfaceFbdev.
- Moves the GRSurfaceFbdev deallocation code into GRSurfaceFbdev's dtor.
- Manages MinuiBackendFbdev::gr_framebuffer with std::unique_ptr.
Test: mmma -j bootable/recovery
Test: `Run graphics test` on taimen.
Change-Id: I8e67cda7bc3a2feec0790124d035caa36fb58a89
This prepares for the removal of the default and copy ctors, by making
GRSurface::Create() as the only way to get GRSurface instances.
Test: mmma -j bootable/recovery
Test: Run recovery_unit_test on marlin.
Change-Id: I0c34c3f3967e252deb020907c83acbac8a8f36b9
This CL adds a dtor to GRSurfaceDrm that handles the resource
deallocation. It also manages MinuiBackendDrm::GRSurfaceDrms with smart
pointers.
Test: Build and boot into recovery on blueline. `Run graphics test`.
Change-Id: Iff7bbdddbc0b5ab16483d00870794fca9f832bd5
This CL adds GRSurface::Create() and dtor for managing the allocated
memory in GRSurface class. It also adds GRSurface::data() that hides the
underlying implementation, with both of const and non-const overloads.
This allows `const GRSurface&` to be more useful - previously it only
ensured a const member variable of `data`, instead of a read-only buffer
it points to.
It also marks the parameters in gr_texticon() and gr_blit() as const, as
they're incoming source that shouldn't be altered. It corrects the type
of gr_draw, which is the sink to be painted on (an earlier attempt was
made in [1], but didn't get the full picture correctly).
[1] https://android-review.googlesource.com/c/platform/bootable/recovery/+/704757/
Test: mmma -j bootable/recovery
Test: recovery_unit_test on marlin
Test: Run graphics test on marlin (fbdev).
Test: Run graphics test on blueline (drm).
Change-Id: I7904df084cd6c08fa04a9da97d01b4b1a6e3a20c
These properties need to be exposed to both of charger and recovery.
This CL tracks the name change.
Bug: 113567255
Test: Build along with other CLs in the topic (for Makefile and
sepolicy changes). Boot into charger mode.
Test: Boot into recovery. Run graphics test.
Change-Id: I4d90778c1767c9bd7e00830a34ee2d4875ce0b83
Bug: 110380063
Test: `m -j installclean && mmma -j bootable/recovery` with
aosp_taimen-userdebug
Test: Build (`m -j bootimage`) and boot into recovery. Check that
`adb sideload` and `Run graphics test` both work.
Test: Run recovery_unit_test and recovery_component_test on marlin.
Change-Id: Ie6ed0e7cafa352d5faff9d1b6ccef724a0415e65
The DRM_FORMAT settings in DrmCreateSurface should match with the
correct PixelFormat
Bug: 112245291
Test: build and flash boot image to device.
Change-Id: I3ee0de38256d1109cd5174b666e2f47dceb637e6
Note that although libminui says "shared_libs" in Android.bp file,
`recovery` binary will still statically link the libs, until we drop the
`LOCAL_FORCE_STATIC_EXECUTABLE := true` flag from `recovery` module.
Bug: 110380063
Test: `mmma -j bootable/recovery` with aosp_taimen-userdebug
Test: Boot into recovery and `Run graphics test`.
Change-Id: I48785d161dfe52eb01667e938d9c3f0a497b61a2
This CL removes the use of the following build time macros, and uses the
runtime property values instead.
- TARGET_RECOVERY_PIXEL_FORMAT
- TARGET_RECOVERY_OVERSCAN_PERCENT
- TARGET_RECOVERY_DEFAULT_ROTATION
Bug: 110380063
Test: Set up taimen with `TARGET_RECOVERY_DEFAULT_ROTATION := ROTATION_LEFT`.
Build and check recovery UI.
Test: Set up taimen with `TARGET_RECOVERY_PIXEL_FORMAT := ABGR_8888`.
Build and check recovery UI.
Change-Id: I4439556a03fde4805a18011ef72eff1373f31d47
This CL uses scoped enum for GRRotation. It also refactors the previous
comparisions to improve readability -- it was hard to tell the meaning
of `if (rotation % 2)` unless by checking the exact values.
It also temporarily adds STRINGFY to expand the macro of
DEFAULT_ROTATION. This will be removed once minui is changed from
reading build time vars to runtime properties.
Test: Build and boot into recovery image on marlin.
Test: Set `TARGET_RECOVERY_DEFAULT_ROTATION := ROTATION_LEFT` and check
the display under recovery with `Run graphics test`.
Change-Id: Ieb08b1cd13c284f8dcf906cb2a1430c6bae2835b
If two consecutive call are made to drmModePageFlip, the second call
may fail with -EBUSY because the first flip is not completed yet.
This patch adds a wait for completion mechanism based on
drmHandleEvent.
Change-Id: Ied13ebefc7523003431b1b307bae70d1a70cb24b
Signed-off-by: Jeremy Compostella <jeremy.compostella@intel.com>
Signed-off-by: Benoit Fradin <benoit.fradin@intel.com>
Also free gr_font in gr_exit(), as it's owned by minui.
Test: mmma -j bootable/recovery system/core/healthd
Test: recovery_unit_test on marlin
Test: Boot into recovery. Run graphics test.
Change-Id: I00457bb97ed3dc95ef5aad493827fb5a62a258dc
In a charger mode manual test, we encounter failures from the
MinuiBackendDrm when calling DrmEnableCrtc and Flip. To make the minui
more robust, we should fall back to another backend if drm's SetCrtc
fails. And check the value of gr_draw before dereferencing.
Bug: 80249440
Test: boot to recovery
Change-Id: Ibd1ca1fb1115fe1132684586c54eccd8fb4c3ad9
Some users of minui don't need to draw texts; and we should let them
still be able to use the library without providing a font file. Also add
check for null pointers in gr_measure() and gr_font_size().
Bug: 80535212
Test: boot into recovery without a font file, and the buttons still work
Change-Id: I848e4410f2ce09ea0ab433573e6827b7e9b2c575
In order to support that, this CL adds Paths::set_resource_dir() to
override the default resource dir ("/res/images/") that's only available
under recovery. Note that since there're external modules depending on
libminui, it adds a separate function of res_set_resource_dir(), instead
of requiring the dependency on libotautil for everyone.
Test: mmma -j bootable/recovery
Test: Run recovery_unit_test on marlin.
Change-Id: I0a7dcf4476808bea9e634eaffc9676f6cbaf92b7
As a private header for testing purpose. PngHandler additionally loads a
given filename if the one with '/res/images' prefix is not available. It
also provides color_type/bit_depth that are parsed from the PNG file.
This allows reusing the same code for the ResourcesTest (renamed from
ResourceTest).
Test: Run recovery_manual_test on marlin.
Change-Id: I3f939d79a1cb1b83a899847dbe2d51bde15d16d8
The font_10x18.h is used as a fall back for fonts if the
$(recovery_font) under bootable/recovery/fonts failed to reload. This
font looks very small on the screen and we haven't used it in a while.
Plus, fonts in header doesn't have many benefits if we support only
ascii characters. So we decide to remove this header along with the mkfont.c
that generates the header.
For reference, here's the steps to use mkfont:
1. Generate a picture of fonts with apps on host
2. Open the pic with GIMP
3. Export the pic as '.c' file, and rename it into '.h' file
4. Compile the generated '.h' file with mkfont.c
Bug: 76420958
Test: build and check the recovery menu on angler
Change-Id: I6c108aa43a07bf4994115ef764275bd84df725e1
Use libdrm_platform which is a platform variant of libdrm.
Bug: 63741047
Bug: 37342627
Test: BOARD_VNDK_VERSION=current m -j libminui
Test: ryu recovery graphics test shows various graphics
Change-Id: Ifd2c1432781a96538585cdf818aa728d628a2f5a
Merged-In: Ifd2c1432781a96538585cdf818aa728d628a2f5a
The open_png() function used to open the png file but didn't close it;
and this caused the leak of fd. However, we cannot close the file inside
open_png() because the png file needs to remain open until the outer
function finishes parsing the file and destroys the png struct.
This CL addresses this issue by implementing a PngReader class to handle
the creation/destruction of the png struct.
Bug: 67010912
Test: Run graphic tests; also run locale tests and check fd.
Change-Id: I9a803b3cd8c16f16a9ffe8f0acc7fe0f42e95eb0
Add a new option "Run locale test" to check the background text
images (i.e. texts for "erasing", "error", "no_command" and "installing"
with different locales.)
Use volume up/down button to cycle through all the locales embedded in
the png file, and power button to go back to recovery main menu.
Test: Run locale test with bullhead.
Change-Id: Ib16e119f372110cdb5e611ef497b0f9b9b418f51
Turn on -Wall for all modules. Also remove the obsolete file_cmp() in
apply_patch test and now() in wear_ui.
The only exception is lib_edify due to the unused functions in the
intermediate cpp files generated from the lex files. It will be handled
in a seperate CL.
Bug: 64939312
Test: mma, unit tests pass
Change-Id: Ic53f76b60b6401ab20db3d98130d674c08e3702f
- Added detection for EV_ABS events in minui/events.cpp, if it's
allowed;
- Added listening and processing touch inputs in ui.cpp;
- Fixed an issue in recognizing swipe with multi-touch protocol A;
- Changed the logic in RecoveryUI::ProcessKey() to be swipe-aware. It
now allows turning on text mode with <power> + <swipe-up>.
The last change also fixed an issue on devices with protocol A: prior
to this CL, user may accidentally toggle the text mode during an OTA.
Because it was considered as a single-button device, a long tap that
sent BTN_TOUCH event would turn on text mode.
Test: Allow detecting touch inputs. Swiping (up, down, enter) works on
angler, angelfish, dorado respectively.
Bug: 36169090
Change-Id: I4bc882b99114ce4ab414f8bdb8f4f7a525b8a8fd
When graphics_adf was refactored in 557fa1f, a class member was introduced
that was not initialized to zero. This meant that the first time Flip()
was called, current_surface would have a junk value and cause a bad
pointer rereference, crashing recovery.
Make sure current_surface is initialized to 0 for the first Flip().
Test: Ran recovery on a device using the ADF backend.
Change-Id: I9b8fac0a4d48cac990e5e9808a071c232de1ebfb
Switch the locale header in the png files from Locale.toString() to
Locale.toLanguageTag(). For example, en_US --> en-us and sr__#Latn
--> sr-Latn. Also clean up recovery a bit to expect the new locale
format.
Bug: 35215015
Test: sr-Latn shows correctly under graphic tests && recovery tests pass
Change-Id: Ic62bab7756cdc6e5f98f26076f7c2dd046f811db
For the use of std::function and std::bind. They were relying on the
transitive inclusion from <minui/minui.h>.
Test: mmma bootable/recovery
Change-Id: Ia138e1cbdd035b11d6cdca9e16c5591303b6ee13
