Clean up the recovery image and switch to libbase logging.
Bug: 28191554
Change-Id: Icd999c3cc832f0639f204b5c36cea8afe303ad35
Merged-In: Icd999c3cc832f0639f204b5c36cea8afe303ad35
We might end up in an infinite loop if read(2) reached EOF unexpectedly.
The problematic code in uncrypt mentioned in the bug has been fixed
by switching to libbase ReadFully(). So I grepped through the recovery
code and fixed some other occurences of the issue.
Bug: 31073201
Change-Id: Ib867029158ba23363b8f85d61c25058a635c5a6b
adb_server_main in adb refers to the adb server on the host, not adbd.
Since there doesn't seem to be a good reason to reuse the declaration
from adb's headers, give minadbd a main function of its own.
Change-Id: I748f1a6822dc14c726cb73ef3b533c57a6615608
Keys for package verification is loaded after the update
package is mmaped into memory. This mmaped area needs
to be freed when exiting the function.
Another approach would be to mmap after loading the keys.
Change-Id: Ib77711a8acd5c363b5517da12dc311fb8f9f4605
Signed-off-by: WiZarD <WiZarD.Devel@gmail.com>
For A/B devices, "view recovery logs" doesn't work due to the lack
of cache partition. To help debugging, we'll show /tmp/recovery.log
instead if /cache is not found.
Change-Id: Idb77c3a4c30388148a210b38d732a7b27e757bba
Test: Tested on an A/B device and /tmp/recovery.log showed up.
Bug: 30905700
If mount fails and the partition has the formattable flag set in fstab
it is formatted and another attempt at the mount is made.
Change-Id: I44edafac7d7ff663489072c66b7e210466316b14
The fopen_path will fail here e.g. if /cache is corrupt and could
not be mounted. Not properly handling an error code from
fopen_path() results in a boot loop into recovery as fwrite() on
a NULL FILE object will crash the recovery app.
Change-Id: I1ccff5a8cb620a32f32adfe97a3fae8e5e908f39
[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
A Certificate is a pair of an RSAPublicKey and a particular hash. So v1
and v3 differ in the hash algorithm (SHA-1 vs SHA-256), similarly for
v2 and v4.
In verifier testcases, we used to load v1/v2 keys with an explicit
argument of "sha256" to test the v3/v4 keys. This CL switches to loading
v3/v4 keys directly and lets load_keys() to handle that, which is the
actual flow we use in practice.
Also remove the "fallback to v1 key" in the testcases, which is not the
actual behavior.
Bug: 30415901
Test: Run the verifier_test component test on device.
Change-Id: I3a2baa64826f1b6c4c367a560090df384c4521bb
After process the text images with pngquant, I encountered a
image-loading error when doing graphic tests on angler. The command
"pngcrush -c 0 $imagename" resolves this error and gives almost the
same compression rate. So regenerate all the text images with "pngcrush
-c 0"
Bug: 30415666
Test: Run graphics test on angler and the background text looks good.
Change-Id: I5d989c1230e826407287265fc559349a7bd29303
Update the text images according to the new translations of
"b+sr+Latn", "be-rBY", and "bs-rBA". Also switch to pngquant
which gives a better compression rate.
Bug: 30415666
Change-Id: I6eae5dd7be4c7f1f93c50113822caac9868e0f95
* 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
Tianjie Xu
Treehugger Robot
Josh Gao
Chih-Hung Hsieh
Tao Bao
WiZarD
Johan Harvyl
Elliott Hughes
xinglong.zhu
Bill Yi