[root cause ] miniui has no mechanism to protect
the buffer which is displaying
[changes ] recovery
[side effects] ui show
[self test ] sc9850-2 general operation has not display abnormal
[reviewers ] xinglong.zhu
Signed-off-by: xinglong.zhu <xinglong.zhu@spreadtrum.com>
[change_type ] AOB --> google_original
[tag_product ] common
Change-Id: I989a0b2943ff6070a0e98718cfbe95144510d3a2
* Use const reference parameter type to avoid unnecessary copy.
* Use more efficient overloaded string methods.
Bug: 30407689
Bug: 30411878
Change-Id: Iefab05c077367f272abf545036b853e8a295c8cd
Test: build with WITH_TIDY=1
am: 32c2e78726
* commit '32c2e78726fe9c7cdc927c8435352b2f8e824735':
Keep ADF device alive for the lifetime of the minui backend
Change-Id: If28a4b4c7098f1a5fcf8e46e03d0df575fc68fce
Some ADF drivers do some amount of state cleanup when the ADF device
node is closed, making and attempts to draw using it fail.
This changes the minui ADF backend to keep the adf_device open until it
is exited, fixing issues on such devices.
matches_locale was expecting input locale string to have at most one
underscore; as a result "zh_CN_#Hans" ignores "zh_CN" and matches into
"zh". Fix the match function and add unit tests.
Bug: 27837319
Change-Id: I4e8a66f91cae6ac2a46b6bf21f670d5ea564c7c8
Switch to a Wear-like intro/loop system. We don't have an intro yet,
but hopefully this will let Wear delete more code when they move to N.
Unlike them, we don't hard-code the number of frames: we just look to
see what we have available. We do hard-code the fps though.
Also add a graphics test mode so you can see a demo of the UI components
without having to actually apply an OTA.
Also fix a bug where default locale is null rather than en-US: it's
more useful to show _some_ text if we don't have a locale (which should
only be during development anyway).
Bug: http://b/26548285
Change-Id: I63422e3fef3c41109f924d96fb5ded0b3ae7815d
bootable/recovery/minui/resources.cpp:321:21: warning: Branch condition evaluates to a garbage value
if (surface[i]) free(surface[i]);
^~~~~~~~~~
bootable/recovery/minui/resources.cpp:424:13: warning: Value stored to 'len' during its initialization is never read
int len = row[4];
^~~ ~~~~~~
Bug: http://b/27264652
Change-Id: Icc1a914c59d6a89dab1b752b2cd2b40549566481
This reverts commit e5879c3639.
The swap in page flip code is not needed any more.
New changes take care of ABGR and BGRA formats swapping bytes in
png and drawing routines
See commit fd778e3e40
Bug: 26243152
Change-Id: I313ee41bee2c143b4e5412515285a65ac394ec77
We allow vendor-specific icon installing image but have defined private
animation_fps that can't be overridden. This CL changes the image
generator to optionally embed FPS (otherwise use the default value of
20) into the generated image.
For wear devices, they are using individual images instead of the
interlaced one. Change the animation_fps from private to protected so
that it can be customized.
Bug: 26009230
Change-Id: I9fbf64ec717029d4c54f72316f6cb079e8dbfb5e
Also add missing TEMP_FAILURE_RETRYs on read, write, and lseek.
Bug: http://b/20625546
Change-Id: I03b198e11c1921b35518ee2dd005a7cfcf4fd94b
(cherry picked from commit 7bad7c4646)
This makes it easier for us to deal with arbitrary information at the
top, and means that headers added by specific commands don't overwrite
the default ones.
Add the fingerprint back, but broken up so it fits even on sprout's
display.
Change-Id: Id71da79ab1aa455a611d72756a3100a97ceb4c1c
Currently fugu has a custom subclass to handle this. The default code
supports devices with trackballs but not all shipping Nexus devices?
That's just silly.
Change-Id: Id2779c91284899a26b4bb1af41e7033aa889df10
Previously most devices would lose the character before a line wrap.
The log's text rendering was starting at offset 4 but none of the
arithmetic was taking this into account. It just happened to work
on the Nexus 9's 1536-pixel wide display (1536/18=85.3) but not on
a device such as the Nexus 5 (1080/18=60).
The only active part of this change is the change from 4 to 0 in the
gr_text call. The rest is just a few bits of trivial cleanup while I
was working out what was going on.
Change-Id: I9279ae323c77bc8b6ea87dc0fe009aaaec6bfa0e
[Problem]
Droid and animation color in recovery mode are incorrect
[Modify]
- Add support for flipping (zero copy) with RECOVERY_ABGR.
- Decodes PNG files to BGRA directly, and other fills, text and alpha blending are also done directly in BGRA (i.e. blits can still bypass conversion)
- Remove the BGRA workaround added previous for single buffer mode (f766396)
Bug:19216535
Change-Id: Ie864419fc6da776ff58b2d02e130f203c194500f
Signed-off-by: Tony Kuo <tony.kuo@mediatek.com>
The re-designed recovery graphics code only supports RGB{_,X,A}
framebuffer, and this patch adds support for BGRA framebuffer.
Change-Id: I3780c8288088f497fa248f3492c54f43834a8598
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
Reviewed-by: Jovanovic, Radivoje <radivoje.jovanovic@intel.com>
Reviewed-by: Parkinson, TimothyX L <timothyx.l.parkinson@intel.com>
Tested-by: Parkinson, TimothyX L <timothyx.l.parkinson@intel.com>
Reviewed-by: Boie, Andrew P <andrew.p.boie@intel.com>
Reviewed-by: Gumbel, Matthew K <matthew.k.gumbel@intel.com>
Reduce the number of copies of libpng boilerplate. Rename
res_create_* functions to be more clear. Make explicit the use of the
framebuffer pixel format for images, and handle more combinations of
input and output (eg, loading a grayscale image for display rather
than use as a text alpha channel).
Change-Id: I3d41c800a8f4c22b2f0167967ce6ee4d6b2b8846
Recovery now draws directly to the framebuffer by rolling its own
graphics code, rather than depending on libpixelflinger.
The recovery UI is modified slightly to eliminate operations that are
slow with the software implementation: when the text display / menu is
turned on, it now appears on a black background instead of a dimmed
version of the recovery icon.
There's probably substantial room for optimization of the graphics
operations.
Bug: 12131110
Change-Id: Iab6520e0a7aaec39e2ce39377c10aef82ae0c595
Conflicts:
minui/resources.c
Instead of one 'base' installing image and a number of overlay images
that are drawn on top of it, we represent the installing animation
with one PNG that contains all the animation frames, interlaced by
row. The PNG is expected to have a text chunk with the keyword
'Frames' and a value that's the number of frames (as an ascii
string). This representation provides better compression, removes the
need to subclass ScreenRecoveryUI just to change the position of the
overlay or number of frames, and doesn't require gr_blit() to support
an alpha channel.
We also remove the 'indeterminate' progress bar used when wiping data
and/or cache. The main animation serves the same purpose (showing
that the device is still alive); the spinning progress bar has been
redundant for a while.
This changes the default recovery animation to include the
antenna-wiggling and gear-turning that's used in the Nexus 5 recovery
animation.
Change-Id: I51930a76035ac09969a25472f4e572b289418729
Conflicts:
screen_ui.cpp
screen_ui.h
Recovery now draws directly to the framebuffer by rolling its own
graphics code, rather than depending on libpixelflinger.
The recovery UI is modified slightly to eliminate operations that are
slow with the software implementation: when the text display / menu is
turned on, it now appears on a black background instead of a dimmed
version of the recovery icon.
There's probably substantial room for optimization of the graphics
operations.
Bug: 12131110
Change-Id: Iab6520e0a7aaec39e2ce39377c10aef82ae0c595
Instead of one 'base' installing image and a number of overlay images
that are drawn on top of it, we represent the installing animation
with one PNG that contains all the animation frames, interlaced by
row. The PNG is expected to have a text chunk with the keyword
'Frames' and a value that's the number of frames (as an ascii
string). This representation provides better compression, removes the
need to subclass ScreenRecoveryUI just to change the position of the
overlay or number of frames, and doesn't require gr_blit() to support
an alpha channel.
We also remove the 'indeterminate' progress bar used when wiping data
and/or cache. The main animation serves the same purpose (showing
that the device is still alive); the spinning progress bar has been
redundant for a while.
This changes the default recovery animation to include the
antenna-wiggling and gear-turning that's used in the Nexus 5 recovery
animation.
Change-Id: I51930a76035ac09969a25472f4e572b289418729
Use intptr_t/uintptr_t to cast between pointer and int to allow
building with -Werror=pointer-to-int-cast and
Werror=int-to-pointer-cast turned on.
Cast to char* instead of unsigned int for pointer arithmetic.
Change-Id: Ia862306fdcca53866b330e8cf726f3d62f2248a0
Help enable external main loop combined with ev_*() key event processing.
Specify EPOLLWAKEUP to hold a wakelock on any event, assuming this is needed
(may need to make this optional).
Convert callback events parameter to unsigned int.
Change-Id: Ib5e09abbd7724ffd830e2cf8e25e7eb59d3aa072
Recovery currently has a random mix of messages printed to stdout and
messages printed to stderr, which can make logs hard to read. Move
everything to stdout.
Change-Id: Ie33bd4a9e1272e731302569cdec918e0534c48a6
Get rid of the notion of a font's "ascent"; the reference point for
drawing is the top-left corner of the character box rather than the
baseline. Add some more space between the menu entries and make the
highlight bar around the text.
Replace the default font.png with two images; the build system will
include one or the other based on the resolutions of the device.
Restore the original compiled-in bitmap font, to fall back on when
font.png can't be found (eg, in the charger binary).
Add support for bold text (when a font.png image is used).
Change-Id: I6d211a486a3636f20208502b1cd2aeae8b9f5b02
Instead of representing the font used for menus and log messages in
the recovery binary, load it from a resource PNG image. This allows
different devices to substitute their own font images.
Change-Id: Ib36b86db3d01298aa7ae2b62a26ca29e6ef18014
If your screen is a TV, it may not actually be displaying the edges of
the framebuffer. Allow specifying an overscan percentage, and move
each edge of the framebuffer in by that percent of the width/height.
(The gr_* layer just lies to the caller about the size of the
framebuffer, telling the caller it's smaller than it really is, and
offsets all drawing commands to match.)
Change-Id: I11bb2feb39ae522bd3e957a14ebdecf3609e0fdc
- protect against missing/malformed bitmaps: fail to display them but
don't crash.
- don't draw animation overlays until the overlay offset is computed.
- logging cleanup
Change-Id: Ieb1c155cfbb11e643000bdb5d1a57900c8757739
- change locale filename to "last_locale" so the main system doesn't
delete it
- clean up some chatty logging
- update images with real German (other languages TBD)
Change-Id: I2ebb4ed4e054bd1808a3042d9efbb2c18f3a044d
- recovery takes a --locale argument, which will be passed by the main
system
- the locale is saved in cache, in case the --locale argument is
missing (eg, when recovery is started from fastboot)
- we include images that have prerendered text for many locales
- we split the background states into four (installing update,
erasing, no command, error) so that appropriate text can be shown.
Change-Id: I731b8108e83d5ccc09a4aacfc1dbf7e86b397aaf
After fixing three trivial warnings, libminui builds cleanly with clang.
Change-Id: Id7c6228295427a5ed3c774c5f90e88a28336c1f7
Author: Edwin Vane <edwin.vane@intel.com>
Reviewed-by: Kevin P Schoedel <kevin.p.schoedel@intel.com>
yres_virtual value is set incorrectly, causing serveral images to be skipped.
Change this value according to the number of buffers to fix this issue.
from: codeaurora.org
minui: Display the battery charging image correctly
commit: 581a4dead6b96579a13ff22e2454c1f329731679
Change-Id: I10f5d1c6cc37705f0287c7dd517082de2e11d264
Signed-off-by: Iliyan Malchev <malchev@google.com>
When we don't have enough video memory for double buffering we
fallback to "single buffering".
Change-Id: I8bfab6d8cd6b54f0cc6c67edc41a4c37d8fbd4ba
Signed-off-by: Octavian Purdila <octavian.purdila@intel.com>
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
If a key is down prior to the time of initialization, we would not get the
down event for the key, and thus think that the key is not pressed.
Add an interface that allows one to provide a callback to execute
on all keys that are currently down.
Change-Id: I2a4096c0cb4c7c7a9a80d207835f168a0b418413
Signed-off-by: Dima Zavin <dima@android.com>
Events are now delivered through a callback mechanism during
a call to ev_dispatch(). This will allow us to extend the events
code to handle other devices/fds, not just input. One such example
is the ability to process uevents.
During initialization, we provide an input callback to ev_init
that gets called when a new event is encountered during dispatch.
ev_get has been removed and replaced with ev_get_input() helper
function that can be called from inside the callback to attempt
to get an input event.
The existing client of ev_get in recovery has been split up such
that the input thread just calls ev_wait(); ev_dispatch(); and
the input_callback handles individual events by using the
ev_get_input() helper.
Change-Id: I24d8e71bd1533876b4ab1ae751ba200fea43c049
Signed-off-by: Dima Zavin <dima@android.com>
