Programmer param can be NULL and this is a valid case which can
be covered by unit test.
`run_lifecycle` needs an adjustment to handle NULL as programmer
param, which is also included in the patch.
Change-Id: I409f1c9ac832943e54107f7cf8652d1f46ac67df
Signed-off-by: Anastasia Klimchuk <aklm@chromium.org>
Reviewed-on: https://review.coreboot.org/c/flashrom/+/67642
Reviewed-by: Edward O'Callaghan <quasisec@chromium.org>
Reviewed-by: Felix Singer <felixsinger@posteo.net>
Reviewed-by: Thomas Heijligen <src@posteo.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
fileno can be a function and a macro in FreeBSD, depending on whether
the environment in multi-threaded or single-threaded. For single
thread, macro is used. Macro is expanded into a inline code which
accesses private field of file descriptor. This is impossible to
mock in tests.
Pretending that environment is multi-threaded makes fileno function
to be called instead of a macro. Function can be mocked for tests.
BUG=b:237606255
TEST=ninja test
tests pass on two environments:
1) FreeBSD 13.1-RELEASE-p2 GENERIC amd64
2) Debian 5.17.11 x86_64 GNU/Linux
Change-Id: I3789ea9107a4cf8033cf59bb96d3c82aa58de026
TICKET: https://ticket.coreboot.org/issues/411
Signed-off-by: Anastasia Klimchuk <aklm@chromium.org>
Reviewed-on: https://review.coreboot.org/c/flashrom/+/67312
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Thomas Heijligen <src@posteo.de>
Reviewed-by: Peter Marheine <pmarheine@chromium.org>
This implements a positive selection choice of which programmers should
be built.
- Each programmer is represented through an entry in the programmer
dictionary
- The entry contains:
- A list of systems and CPU families where the programmer can run on
- A list of required dependencies
- A list of sources needed to build the programmer
- A list of compiler flags
- A flag to determin if the programmer should be build on 'auto'
- If an entry is not given it is set to the default value
- If a programmer gets selected, an 'active' flag is added to the entry
on runtime
- All programmers with an 'active' flag will be included in the build
- One or more programmers can be selected through '-Dprogrammer=<>'
- 'auto' enables all programmers which are available, deps are found
and have the 'default' flag
- 'all' enables all programmers which are available and deps are found
- 'group_***' enables all programmers which are available, deps are
found and the programmer belongs to the selected group
- '_programmer_name_' forces the programmer to be built or the build
will fail.
Change-Id: Ib44b26e3748fc71f116184082b4aed0bb208b4c1
Signed-off-by: Thomas Heijligen <thomas.heijligen@secunet.com>
Reviewed-on: https://review.coreboot.org/c/flashrom/+/63724
Reviewed-by: Anastasia Klimchuk <aklm@chromium.org>
Reviewed-by: Felix Singer <felixsinger@posteo.net>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This reverts commit 99eca0899b.
The environment provided by our Jenkins builder doesn't have cmocka.
It's not exactly known why it worked as a subproject. However, if it
allows us to enforce tests in the CI, we should revert now and drop
the subproject again once the environment is updated.
Change-Id: Iafdab92eb8ab96d02fb2c29524d785bdc7a54034
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: https://review.coreboot.org/c/flashrom/+/67345
Reviewed-by: Felix Singer <felixsinger@posteo.net>
Reviewed-by: Thomas Heijligen <src@posteo.de>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-by: Anastasia Klimchuk <aklm@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
`run_init_error_path` tests the scenario when init function of
a programmer fails. Init can fail at different phases and depending
on a specific test scenario it could be before or after shutdown
function has been registered.
If shutdown function has already been registered, it needs to run
because it cleans up the resources allocated during init. This patch
prevents memory leaks.
BUG=b:181803212
TEST=ninja test
Change-Id: I604edff18e35b7c044b73d3a8adfa8c800eddfd2
Signed-off-by: Anastasia Klimchuk <aklm@chromium.org>
Reviewed-on: https://review.coreboot.org/c/flashrom/+/67198
Reviewed-by: Edward O'Callaghan <quasisec@chromium.org>
Reviewed-by: Felix Singer <felixsinger@posteo.net>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Previously, parade lspcon unit test had no custom emulation and
was running basic lifecycle with default mocks (default mocks do
nothing and return success). I have discovered that it does not
work in all environments.
Specifically, in functions `parade_lspcon_wait_command_done`
and `parade_lspcon_wait_rom_free` there is a local variable
`uint8_t val` which is declared and not explicitly initialised. It can
get different initial values depending on the environment/toolchain
(for example gcc vs clang).
Later during the code execution, this variable is used as a holder
for writing/reading data from register(s). For unit test, reading and
writing data from registers is emulated. With default mocks, initial
value of variable was propagated further as if it was "read" from
a register. So the unit test could pass or fail depending on the
initial value of local variable, which in turn depends on environment
and toolchain.
If initial value was 0 the test reliably passed (this is the case with
upstream build environment, locally I have gcc 11.3.0 on x86_64).
If it was 1 the test reliably failed (this is the case of chromium
flashrom tree, this is clang 15.0 under chroot, tested on x86_64 and
arm boards).
If it was any other value then it could be anything.
This patch adds custom mocks for ioctl/read/write operations
and emulates a successful scenario of running a lifecycle. Local
variable is initialised by reading from the register, same as it
happens in the real (non-test) run for this programmer.
BUG=b:242816982
TEST=ninja test
Change-Id: I98f52507a0ddbbfbeb390038d14192cacc2be683
Signed-off-by: Anastasia Klimchuk <aklm@chromium.org>
Reviewed-on: https://review.coreboot.org/c/flashrom/+/67161
Reviewed-by: Peter Marheine <pmarheine@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Singer <felixsinger@posteo.net>
Depend exclusive on the cmocka package provided by the build
environment. Cmocka is widely available in Linux distributions and BSD
systems. Besides that, it is suboptimal to fetch own dependencies in the
build process. Most packaging systems even forbid fetching additional
assets at build time.
Change-Id: I751c85d5f72e47356113cf55dfbaec73cbd8028c
Signed-off-by: Thomas Heijligen <thomas.heijligen@secunet.com>
Reviewed-on: https://review.coreboot.org/c/flashrom/+/66696
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-by: Edward O'Callaghan <quasisec@chromium.org>
Reviewed-by: Anastasia Klimchuk <aklm@chromium.org>
Reviewed-by: Felix Singer <felixsinger@posteo.net>
Add tests for i2c programmers that assert that initialisation fails
when allow_brick parameter is not provided.
Example of logs from test run:
[ RUN ] parade_lspcon_no_allow_brick_test_success
Testing init error path for programmer=parade_lspcon with params: bus=254 ...
... init failed with error code -1 as expected
[ OK ] parade_lspcon_no_allow_brick_test_success
BUG=b:181803212
TEST=ninja test
Change-Id: I382f563016502f3342131d5f9c0de41dc665b03a
Signed-off-by: Anastasia Klimchuk <aklm@chromium.org>
Reviewed-on: https://review.coreboot.org/c/flashrom/+/66508
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Singer <felixsinger@posteo.net>
Reviewed-by: Thomas Heijligen <src@posteo.de>
This unit test does not require any additional mocks because init
and shutdown of i2c infra was covered in realtek programmer lifecycle.
Default mocking of i2c is sufficient to run a basic lifecycle.
To run the test, config option for the programmer needs to be
enabled explicitly, since by default this programmer is disabled.
BUG=b:238816884
TEST=meson configure -Dconfig_mediatek_i2c_spi=true
ninja test
Change-Id: I98a12067d165c90013d33ffc45d20dab5c364c83
Signed-off-by: Anastasia Klimchuk <aklm@chromium.org>
Reviewed-on: https://review.coreboot.org/c/flashrom/+/66261
Reviewed-by: Edward O'Callaghan <quasisec@chromium.org>
Reviewed-by: Felix Singer <felixsinger@posteo.net>
Reviewed-by: Thomas Heijligen <src@posteo.de>
Reviewed-by: Alexander Goncharov <chat@joursoir.net>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This unit test does not require any additional mocks because init
and shutdown of i2c infra was covered in realtek programmer lifecycle.
Default mocking of i2c is sufficient to run a basic lifecycle.
To run the test, config option for the programmer needs to be
enabled explicitly, since by default this programmer is disabled.
BUG=b:238816889
TEST=meson configure -Dconfig_parade_lspcon=true
ninja test
Change-Id: I0dcfae4a58c64b2e8d56ec51b4b050534cbb4cd2
Signed-off-by: Anastasia Klimchuk <aklm@chromium.org>
Reviewed-on: https://review.coreboot.org/c/flashrom/+/66003
Reviewed-by: Edward O'Callaghan <quasisec@chromium.org>
Reviewed-by: Felix Singer <felixsinger@posteo.net>
Reviewed-by: Thomas Heijligen <src@posteo.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Tests can be filtered by providing patterns on the command line. The
pattern is the first argument provided to the test binary, if present.
Only tests matching the string provided are run, wildcards * and ?
match any characters, or one character respectively.
`meson test` or `ninja test` will continue to run all tests, as they do
not provide an argument to the test binary.
https://api.cmocka.org/group__cmocka.html
TEST=tests/flashrom_unit_tests 'layout_*'
Change-Id: I45f4ac5ef0cfb74156408022a19769d6598ad2ea
Signed-off-by: Evan Benn <evanbenn@chromium.org>
Reviewed-on: https://review.coreboot.org/c/flashrom/+/65998
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Edward O'Callaghan <quasisec@chromium.org>
Reviewed-by: Anastasia Klimchuk <aklm@chromium.org>
Reviewed-by: Felix Singer <felixsinger@posteo.net>
- Make run_command() calls check for failures to fix warnings about
ignoring errors, see https://github.com/mesonbuild/meson/issues/9300.
- Remove `include_directories('../subprojects')` from tests/meson.build.
It isn't necessary and caused build warnings due to referencing files
outside the tests/ directory.
- Fix indent level and formatting in a few places.
BUG=none
BRANCH=none
TEST=meson; ninja; ninja test
Change-Id: I17ae0c51d68ed004772a237641f08345f4893200
Signed-off-by: Nikolai Artemiev <nartemiev@google.com>
Reviewed-on: https://review.coreboot.org/c/flashrom/+/66060
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Thomas Heijligen <src@posteo.de>
Reviewed-by: Anastasia Klimchuk <aklm@chromium.org>
Currently i2c programmers do not have a safe allow listing
mechanism via board_enable to facilitate fully qualified
chip detection.
Since i2c addresses alone can overlap a user may make the mistake
of using the wrong programmer. Although unlikely, it is within the
realm of possibility that a user could accidently somehow program
another chip on their board.
Change-Id: Ifb303989fdb67f7267002bd0425f3d050450ec93
Signed-off-by: Edward O'Callaghan <quasisec@google.com>
Reviewed-on: https://review.coreboot.org/c/flashrom/+/65545
Reviewed-by: Thomas Heijligen <src@posteo.de>
Reviewed-by: Anastasia Klimchuk <aklm@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Unit tests had an unconditional dependency on libusb and this was
a) strictly speaking not needed, b) blocking one build system effort.
This patch is a temporary solution to unblock one build system effort,
specifically CB:63724. It creates a condition so that libusb is only
included when it is required, not always.
This workaround is based on the fact that at the moment only
2 lifecycle unit tests are using libusb symbols: dediprog and
raiden_debug.
BUG=b:237606255
TEST=the following scenarios run tests successfully
1) dediprog and raiden_debug programmers enabled, libusb.h present
result:
all test run and pass
2) dediprog disabled, libusb.h present
result:
dediprog test skipped, all other tests run and pass
3) dediprog and raiden_debug both disabled,
libusb.h changed to libusbabcd.h
result:
dediprog and raiden_debug tests are skipped,
all other tests run and pass
Change-Id: Iec8a1826951fd6ae586e90fde1a55170e7de41a8
Signed-off-by: Anastasia Klimchuk <aklm@chromium.org>
Reviewed-on: https://review.coreboot.org/c/flashrom/+/65627
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Singer <felixsinger@posteo.net>
Reviewed-by: Thomas Heijligen <src@posteo.de>
This patch creates individual files for each programmer's lifecycle
tests. Common functions that are reusable for all tests are
gathered in lifecycle.c. Each individual file needs to include
lifecycle.h
BUG=b:237606255
TEST=ninja test
Change-Id: If2307699dcbb3a085b91a2dcd41156e6fd07f812
Signed-off-by: Anastasia Klimchuk <aklm@chromium.org>
Reviewed-on: https://review.coreboot.org/c/flashrom/+/65543
Reviewed-by: Peter Marheine <pmarheine@chromium.org>
Reviewed-by: Felix Singer <felixsinger@posteo.net>
Reviewed-by: Edward O'Callaghan <quasisec@chromium.org>
Reviewed-by: Thomas Heijligen <src@posteo.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Anastasia Klimchuk
Thomas Heijligen
Nico Huber
Edward O'Callaghan
Felix Singer
Evan Benn
Nikolai Artemiev