When IOServiceGetMatchingServices returns an invalid
iterator, IOIteratorReset doesn't help. Break the loop
to prevent being stuck in the reset loop forever.
This should be fine as the open_device() call in the
upper layer is already doing retries.
Bug: 286495045
Test: fastboot update
Change-Id: Ia58fa9285640be34aa22cd77e4f58cc9092679a4
images are being flashed twice, because of an additional collectImages.
This is resulting in a ~50% increase in flashing time
Test: fastboot flashall -w
Change-Id: I6c271e1e1456cd789f37ebd67cefd221cabc6e7a
This fixes the following compiler errors with gcc
storage.cpp:45:20: error: 'ostream_iterator' is not a member of 'std'
storage.cpp:50:10: error: 'istream_iterator' is not a member of 'std'
Change-Id: I570e332d100bb18e445644a9d57dc14d80f104b5
This fixes the following compiler error with gcc
fastboot_driver.h:51:10: error: ‘function’ in namespace ‘std’ does not name a template type
Change-Id: I3a4d9b48bfdc2c76e443e2efe1761ec9503c6dc5
The ANDROID_SERIAL env is useful for managing multiple
directly connected devices at one time.
Unfortunately for network connected devices, its harder to use.
This is because both adb and fastboot use the same ANDROID_SERIAL
environment value to identify the device to use, and while that
one value works for directly connected devices, when using
network connected devices, the fastboot and adb port numbers may
differ for the same device.
So if I set: `ANDROID_SERIAL="tcp:127.0.0.1:44403"`
fastboot will work, but when I type `adb shell`, I'll get:
adb: device 'tcp:127.0.0.1:44403' not found
As `adb devices` outputs:
List of devices attached
localhost:36697 device
To resolve this, we need separate environment variables,
so introduce a FASTBOOT_DEVICE variable for fastboot.
If FASTBOOT_DEVICE is set, it will use that, and if not
it will fall back to the ANDROID_SERIAL if it is set.
Using an explicit -s argument will still override both.
Change-Id: Icd7db6f29e51ed2decd219e35537f6ed98d698b8
Signed-off-by: John Stultz <jstultz@google.com>
Using for easier debugging. Replacing old getter methods in flash task
with GetInfo(key)
Test: fastboot_test
Change-Id: If42df86c4b0278c3dc5525b1058f1947f0b34794
if we call -w and a reboot task runs beforehand, we don't actually wipe
the given partitions. We want the wipe to occur before reboots.
Test: fastboot flashall -w
Change-Id: If327c5de0858e222cf5fc14531bc20fb741c2960
Falling back on old flashing code until fastboot-info is fixed
Test: m fastboot, fastboot flashall
Change-Id: I10146ec75f75d57d4a4049d35def370533619401
Old should_flash_in_userspace doesn't work if $ANDROID_PRODUCT_OUT is
not set. Also adding in a check to see if resizetasks were added
correctly
Test: fastboot update update.zip without $ANDROID_PRODUCT_OUT and
removing the flashsuperlayout code.
Bug: 283330320
Change-Id: Ib72f6a1cf07745daf70fffae3d1a6b8352e3f79c
This check isn't needed as super_flash_helper.cpp calls a different
should_flash_in_userspace that isn't reliant on $ANDROID_PRODUCT_OUT
being set. Current code will fail when calling fastboot update
Test: fastboot update update.zip without calling lunch
Bug: 283330320
Change-Id: Icefe6092befb747b7f419ea997e0834602808d69
Old changes didn't actually take into account sources. fastboot update
{update.zip} should read fastboot-info from the update package. It was
instead flashing local images + reading local fastboot-info.txt
Test: m fastboot, fastboot update, fastboot flashall
Bug: 283330320
Change-Id: I9587a7735ddfbfb661153eb12ebc3caa7c32f141
If fastboot-info.txt has wrong format we should not fall back on
hardcoded list silently. Instead we should LOG(FATAL) and fail.
Hardcoded list should only be used when fastboot-info.txt is not found
or is empty
Test: fastboot flashall
Bug: 194686221
Change-Id: I1cada102f3ff12c1f3002d0b61d3785fc25543c1
Replacing check with PLATFORM_TOOLS_VERSION with FASTBOOT_INFO_VERSION
to indicate waht version of fastboot_info.txt we currently support. This
makes mor sense since PLATFORM_TOOLS_VERSION won't be updated every time
me make a change to the host tool
Test: fastboot flashall
Bug: 194686221
Change-Id: I621b62c92ba129f402857463dae9112a0797ab07
Updating version to just be a single number. Reason for updating is
to keep format the same as Flashstation's
Test: fastboot_test
Bug: 194686221
Change-Id: I21ab0747e620d3f6d05c5170c3e55707eed0288a
Removing as we're no longer support reboot userspace in
fastboot-info.txt. Use reboot fastboot instead
Test: fastboot reboot fastboot
Change-Id: Ibd8bf597615dacde23ad446bde4f23837686710b
Adding const decorator to class constructor to ensure Flashing Plan
isn't modified during task construction
Bug: 194686221
Test: tested Flashall on raven
Change-Id: I99b71b2b756beaf871a31a1a035545d754d128e8
Updating old clean up slot logic change. Some devices don't know current
slot so we can't always just call get_current_slot()
Test: tested flashall on raven
Change-Id: Id655f1c1ee0b1ff09331f0d91d77412836d0af8d
cleaning up main() code in fastboot.cpp to run off one task vector. The
eventual goal is to just have all functionality run off this vector.
Test: tested commands on raven
Bug: 194686221
Change-Id: I98d0ba33091eff31ff1bc338766c3689ffb54746
Modifying fastboot-info.txt to directly call reboot fastboot instead of
reboot userspace -> so no longer need "userspace" in reboot task. Also
moving skip_reboot check to task since in old code we can have
skip_reboot be true and still reboot to bootloader, fastboot, or
recovery
Test: tested reboot commands on Raven
Change-Id: I98ba51747bf7229a88ba737540c81d20b42e5981
Doesn't look like we actually need current_slot, since we can always
just call function get_current_slot
Test: tested flashall on raven
Change-Id: Icba64bc9fdd08f36d44965e32558a73d0beab0e1
Changed so if we call wipe on a partition that doesn't exist we'll get
some helpful error messages.
Test: tested wipe test
Bug: 194686221
Change-Id: I19dc0396a7a98d0b2be7859edbffeb8acdc9850a
is_logical shouldn't work in bootloader for some devices, so we also need to check
using should_flash_in_userspace to see if partition is dynamic
Test: tested to flash {partition} on raven and checked to see avb
footers are not copied from dynamic partitions
Change-Id: Iabb3ea535fa80b26cf1c08040beb3d4ea5e8c2ae
sparse_file_len is not actually infallible; on Windows it's pretty easy
to make it fail by embedding large files in the stream. fastboot didn't
handle this anywhere, leading to bad sparse images when libsparse
runs out of address space.
Bug: 273933042
Bug: 268872725
Test: fastboot flashall on Windows
Change-Id: Ie68aed2f1970e820350d9f97aa89a6c0242229b8
Fastboot has a lot of flags that are used by many different functions
including the Flashall Class and newly added FlashSuperTask. Passing all
of these flags as paramaters is cumbersome, so adding a Flashing Plan
that contains these flags simplifies the code.
Test: tested Flashall and update img.zip on raven
Change-Id: I9c842f25389a20b852d55f684e1b86040af1d86a
Bug: 194686221
Introduced positive and negative unit tests to cover
ParseNetworkSerial logic. Alongside with that move
result related stuff to the separate header.
Test: atest fastboot_test
Test: manually checked basic functionality works fine
Bug: 271155012
Change-Id: Icac6053c11b5a36daa64555209555826ea28cc61
Signed-off-by: Dmitrii Merkurev <dimorinny@google.com>
Use more universal lseek instead of lseek64
Test: atest fastboot_vendor_boot_img_utils_test
Test: mm on mac os
Bug: 271152365
Change-Id: I4d094dd2c24e4ffec8ea7fe2f3b355122fd8cd19
Signed-off-by: Dmitrii Merkurev <dimorinny@google.com>
During fastboot connect / disconnect introduction, we
completely broke the IPv6 support (it was considering
all IPv6 addresses as a USB serial).
Makeing sure this problem isn't reproducible anymore
alongside with fixing EXCPECT causing process crash
and improve network serial error detection.
Bug: 271152365
Change-Id: Ic52aa5fff1948a64ac3d2672f3cf4d2b022e5cea
Signed-off-by: Dmitrii Merkurev <dimorinny@google.com>
Test: tested on raven device
Change-Id: I4935d720f567e70da95ea8da37f3404b80b313c8
Bug: 194686221
Changed reboot {target} to work off tasks. reboot-{target} commands are
also supported.
Test: tested reboot on raven device
Change-Id: I05aed269d121a5d651c1ab1180a1b4878ae213fd
Modified load_buf to be able to find images in $OUT directory
Test: tested flash {partition} and flash {partition} {img_name} on raven
device
Change-Id: I3879792d11ad15bc910670853d2a7fe200fcc66f
Trusty Benchmarks need to be evaluated in ABL which is much more
controlled than linux. However fastboot prints evry atomic message from
trusty/abl on its own line starting with "(bootloader)"
Bug: 263454481
Test: - "fastboot oem trusty runtest trusty.hwrng.bench"
Change-Id: I99847a8cc54457c8ec809e219736325dce0ac891
This test checks that the sparsing logic matches a canonical sample
output.
Bug: N/A
Test: fastboot_test
Change-Id: Ic7bad08a79e4223f8352db2397b741c2179e76bb
