[cryptolib] csrng driver #14372
[cryptolib] csrng driver #14372
Conversation
| #include "sw/device/lib/testing/test_framework/ottf_main.h" | ||
|
|
||
| OTTF_DEFINE_TEST_CONFIG(); | ||
| bool test_main(void) { |
There was a problem hiding this comment.
nit: not related to this PR but I think the return type of test_main() could be status_t, wdyt @timothytrippel ?
There was a problem hiding this comment.
It could, but I'm not sure what that would give us as the return type here gets piped through the OTTF to a test_status_t that gets backdoor read by the testbench (which has aligned SV code) or piped over UART for the FPGA, and all we really care about is whether the test passed or not don't we?
| do { | ||
| reg = abs_mmio_read32(kBaseCsrng + CSRNG_SW_CMD_STS_REG_OFFSET); | ||
| cmd_ready = bitfield_bit32_read(reg, CSRNG_SW_CMD_STS_CMD_RDY_BIT); | ||
| } while (!cmd_ready); |
| * Number of words set in `data`. CSRNG will extend the `data` to zeros if the | ||
| * provided value is less than 12. |
There was a problem hiding this comment.
nit: will CSRNG pad at the start or end? Can we just require this from the clients for the sake of simplicity or do you anticipate a considerable difference in performance?
There was a problem hiding this comment.
This is done by hardware at the moment. The padding is added at the end.
| @@ -0,0 +1,169 @@ | |||
| // Copyright lowRISC contributors. | |||
There was a problem hiding this comment.
nit: I know "kat" stands for "known answer test" but wdyt about renaming this file to entropy_kat_test.c for the sake of clarity?
There was a problem hiding this comment.
I didn't want to use the _test.c to avoid giving the impression that it is a functional test. wdyt?
moidx
force-pushed
the
cryptolib-driver-entropy
branch
from
269ea40
to
c27dca5
Compare
Move the required args to the end and name them the "test commands." Pass the test commands via an environment variable, and place them at the end of the invocation. This makes the test commands unable to be overridden by test_arg arguments, and simultaneously, it allows the user to specify test_arg arguments on the bazel command line to modify arguments to opentitantool and friends. Those test_arg arguments can override or supplement the test harness's arguments provided by the bazel rule. Signed-off-by: Alexander Williams <[email protected]> Signed-off-by: Miguel Osorio <[email protected]>
moidx
force-pushed
the
cryptolib-driver-entropy
branch
from
c27dca5
to
f349042
Compare
This commit introduces the a driver for the software instance of the csrng. The driver interface supports regular DRBG functions, i.e. instantiate, reseed, update and generate functions. The focus of this commit is functionality, so there are a few things that will be addressed in future commits: - Handling of blocking operations. This is currently an area where we need to come up with a strategy for the crypto library as indefinite blocking operations may need an unconditional timeout when called in higher application layers. - alert and state checks. Signed-off-by: Miguel Osorio <[email protected]>
moidx
force-pushed
the
cryptolib-driver-entropy
branch
from
f349042
to
3643f63
Compare
This commit introduces the a driver for the software instance of the
csrng. The driver interface supports regular DRBG functions, i.e.
instantiate, reseed, update and generate functions.
This driver uses the
status.hmodule for error reporting/handling.The focus of this commit is basic functionality. There are a few things that
will be implemented in follow up commits:
need to come up with a strategy for the crypto library as indefinite
blocking operations may need an unconditional timeout when called in
higher application layers.
Signed-off-by: Miguel Osorio [email protected]