On Sat, Jun 28, 2025 at 04:29:18AM +0000, Jay Wang wrote: > This commit fixes a timing issue introduced in the previous commit > "crypto: rng - Override drivers/char/random in FIPS mode" where the crypto RNG > was attempting to override the drivers/char/random interface before the default > RNG became available. The previous implementation would immediately register the > external RNG during module initialization, which could fail if the default RNG > wasn't ready. Why isn't this just part of the previous commit so that there is no regression anywhere? thanks, greg k-h
