On 12/5/23 9:22 AM, Nishanth Menon wrote:
On 08:46-20231205, Andrew Davis wrote:
On 12/4/23 1:29 PM, Nishanth Menon wrote:
On 15:59-20231201, Dhruva Gole wrote:
Delete the flag CFG_WITH_SOFTWARE_PRNG as it's not necessary/ boot requirement for this SoC
Signed-off-by: Dhruva Gole d-gole@ti.com
doc/board/beagle/am62x_beagleplay.rst | 1 - 1 file changed, 1 deletion(-)
diff --git a/doc/board/beagle/am62x_beagleplay.rst b/doc/board/beagle/am62x_beagleplay.rst index 7784e62b0b71..50d7d3c620d7 100644 --- a/doc/board/beagle/am62x_beagleplay.rst +++ b/doc/board/beagle/am62x_beagleplay.rst @@ -63,7 +63,6 @@ Set the variables corresponding to this platform: # we dont use any extra TFA parameters unset TFA_EXTRA_ARGS export OPTEE_PLATFORM=k3-am62x
- export OPTEE_EXTRA_ARGS="CFG_WITH_SOFTWARE_PRNG=y" .. include:: ../ti/am62x_sk.rst :start-after: .. am62x_evm_rst_include_start_build_steps
-- 2.34.1
NAK. RNG is needed to seed standard distros.
You have this backwards, setting WITH_SOFTWARE_PRNG=y forces the SW RNG, disabling the HW RNG. Without this line the HW RNG is the default.
That is not the rationale with which the series was posted. I would prefer we use HW RNG by default. but as I understand there are a bunch of f/w bugs preventing us from doing so. if they are resolved, then the commit message argument should be that the bugs are fixed, so we can easily use then with f/w version x.y.z onwards.
There was a single FW bug that caused suspend/resume to fail when OP-TEE was using the HW RNG. The HW RNG always worked, disabling it was a hack that allowed us to still demo suspend/resume, not sure how that ended up in this documentation.
The fact we disabled a security feature to workaround a non-security bug shows a lack of good judgement on our part IMHO. Product security is our top priority.
Andrew