Hi all,
On 2026-08-08 17:36, Simon Glass wrote:
Good catch. This also means the RNG path has never worked: with the
inverted check, dm_rng_read() was only called when uclass_get_device()
failed, using an uninitialised devp, hence the abort. Boards with a
working RNG silently fell back to the MAC seed. You could mention that
in the commit message. I believe it is:
Fixes: ea707dc0aaaf ("net: Use NDRNG device in srand_mac()")
Apologies for the delay while I was traveling. I see it's moot now (my
initial patch was picked up) - but you're correct, and an amended
message would have been better.
Thanks for the review.
cheers,
Graeme
