The earlier commit in 2dfc1a3898
unintentionally changed the behavior. It gives a different result when
looking up non-existent mount points (e.g. /cache on marlin).
The logic behind volume_for_path("/xyz") is unclear:
- It's fine to return non-null value if it's called by
ensure_path_mounted() before accessing that file "/xyz". (Just based
on the function name, we're not actually having this case.)
- It should return nullptr if the caller is interested in the existence
of that particular mount point "/xyz".
This CL renames the function to volume_for_mount_point(), which does an
exact match by querying the given mount point from libfs_mgr. The former
volume_for_path() has been moved down to function scope for serving
ensure_path_mounted() only.
Test: Build and boot into recovery on bullhead and marlin respectively.
'View recovery logs'.
Test: 'Mount /system'
Test: 'Apply update from ADB'
Change-Id: I1a16390f57540cae08a2b8f3d439d17886975217
And add the missing include of <string.h> (e.g. for strcmp(3)).
Minor update to the arg of fs_mgr_get_entry_for_mount_point(), which now
accepts std::string.
Test: mmma bootable/recovery
Change-Id: I9cb8c31fe71b5a053f4d84bf1aba00e96c02ed03
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
Commit cc323958f99e40fea06c511656c69c0b2e2d47f7 in system/core has
changed fs_mgr_get_entry_for_mount_point() to do an exact match only,
which breaks the behavior in volume_for_path().
This CL changes the volume_for_path() implementation to parse and pass
prefixes locally. For a given path like "/cache/recovery/last_log", it
will in turn attempt the prefixes of "/cache/recovery/last_log",
"/cache/recovery", "/cache", "/" and return the first hit.
Bug: 63912287
Test: Build and boot into recovery image on bullhead. 'View recovery
logs' works.
Change-Id: Ic8635b0939649dd5cc9ca501ebc3a2d1fbf5849d
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
So it falls back (from 32) to the default value in Google style:
PenaltyExcessCharacter: 1000000
This way it no longer allows lines exceeding the 100-char limit.
const char* x[] = { "long enough so that the line has 101 chars", nullptr };
would become
const char* x[] = { "long enough so that the line has 101 chars",
nullptr };
Test: clang-format the line above.
Change-Id: If578dbdae55d6d81f5e47889890e2afa12b4199a
Add a mandatory option in imgdiff to write the split info (i.e.
patch_size, tgt_size, src_ranges) to file when handling large apks.
Therefore, the caller of imgdiff can create split transfers based on
the info.
Bug: 63542719
Test: unit tests pass
Change-Id: I853d55d1f999fd576474faa81077f7307f4d856d
This reverts commit 8be0f39fec to reland
the change that removes EXPAND/STRINGIFY macros.
It's error-prone by putting anything into a string (e.g.
EXPAND(RECOVERY_API_VERSION) would become "RECOVER_API_VERSION" if we
forgot to pass -DRECOVERY_API_VERSION=3).
The initial attempt put RECOVERY_API_VERSION into common.h, which might
be included by device-specific codes but without defining that when
compiling the module. This CL avoids the issue by using a constant
in the header, with a static_assert in recovery.cpp that guards the
consistency.
Test: recovery_component_test
Test: Sideload OTAs on bullhead and sailfish respectively.
Change-Id: I12af3f73392a85554ba703f04970ec9d984ccbaa
While calculating volume size, get_block_device_size() returns
u64 value but the returned value is assigned in ssize_t variable.
This may cause integer overflow if the volume size is beyond
ssize_t limit.
Use int64_t instead of ssize_t in get_file_size() and explicitly check
for overflow to fix the issue.
Bug: 65001754
Test: mmma bootable/recovery
Change-Id: I91eb30bff0bf7dcc48678efc2f414d2b79af6d0d
Otherwise the test may fail after a large number of iterations due to
file open failure.
Bug: 65430057
Test: run recovery_component_test on sailfish for 2000 iterations.
Change-Id: I0d456284d6064467038911d63eade95740cbec2c
Localized texts only make sense on devices with screens.
Test: Run fake OTA on angler; check the on-screen texts.
Change-Id: I3a644294c8b1f2056cfb78b2d61a598b8ddf2acf
By grouping similar kinds together, in an order of types, constants,
ctor/dtor, all other methods and data members.
Also rename ScreenRecoveryUI::density_ to ScreenRecoveryUI::kDensity to
align with others.
Test: mmma bootable/recovery
Change-Id: I1ba2d15c05ba7be8c39762f3d9dadf1fb2130de4
After the cleanup to WearRecoveryUI, text_top_ now always equals to
((text_row_ + 1) % text_rows_).
Test: Check the recovery UI and 'View recovery logs'.
Change-Id: I69a7f377bbd990db2194f9d3efae257c323c06a8
They're mostly identical to the ones in ScreenRecoveryUI, except for the
(legacy) use of 'text_top_'. Because wear_ui.cpp misses the change in
[1] that uses an alternate screen for viewing recovery logs.
Also clean up the included headers.
[1] commit c049163234 ('Add an alternate
screen for viewing recovery logs.').
Test: Build a wearable target recovery; `View recovery logs`.
Change-Id: Ic9208c42a11c037469f5b073ef7d9b721c14d1f3
Prior to this CL, menu_ is allocated with a fixed length of text_rows_.
However, because we support scrollable menu in wear_ui, there might be
more menu entries than text_rows_, which would lead to out-of-bounds
array access. This CL addresses the issue by switching to std::vector.
Bug: 65416558
Test: Run 'View recovery logs' on angler.
Test: Set large margin height that leaves text_rows less than 21. Then
run 'View recovery logs' with 21 menu entries.
Change-Id: I5d4e3a0a097039e1104eda7d494c6269053dc894
Due to the cache size limit for OTA generation, we used to split large
zip files linearly into pieces and do bsdiff on them. As a result, i) we
lose the advantage of imgdiff; ii) if there's an accidental order change
of some huge files inside the zip, we'll create an insanely large patch.
This patch splits the src&tgt more smartly based on the zip entry_name.
If the entry_name is empty or no matching source is found for a target
chunk, we'll skip adding its source and later do a bsdiff against the
whole split source image (this rarely happens in our use cases except
for the metadata inside a ziparchive).
After the split, the target pieces are continuous and block aligned,
while the sources pieces are mutually exclusive. (Some of the source
blocks may not be used if there's no matching entry_name in the target.)
Then we will generate patches accordingly between each split image
pairs.
Afterwards, if we apply imgpatch to each pair of split source/target
images and add up the patched result, we can get back the original
target image.
For example:
Input: [src_image, tgt_image]
Split: [src-0,tgt-0; src-1,tgt-1, src-2,tgt-2]
Diff: [ patch-0; patch-1; patch-2]
Patch: [(src-0,patch-0)=tgt-0; (src-1,patch-1)=tgt-1;
(src-2,patch-2)=tgt-2;]
Append: [tgt-0 + tgt-1 + tgt-2 = tgt_image]
Peformance:
For the small package in b/34220646, we decrease the patch size of
chrome.apk dramatically from 30M to 400K due to the order change of
two big .so files.
On two versions of angler, I also observe decent patch size decrease.
For chrome.apk, we reduced the size from 5.9M to 3.2M; and for
vevlet.apk from 8.0M to 6.5M.
Bug: 34220646
Test: recovery component test && apply imgdiff & imgpatch on two
chrome.apk
Change-Id: I145d802984fa805efbbac9d01a2e64d82ef9728b
