Does leaving CONFIG_RANDOM_TRUST_CPU unset when building Linux actually plug a hardware backdoor or am I being trolled into making the kernel slower?
The kernel uses other inputs as well. This just controls if it uses additional hardware RNGs, like RDRAND.Also it's not like RDRAND is fast by any measure. There's a good chance you're making the kernel slower by *enabling* it.
>>108350842AI slop
>>108350889RDRAND can save you from boot hangs if you have a configuration that truly doesn't have any other entropy source, but that's not likely for a desktop user.
>>108350926Yeah, but if you don't you're literally adding thousands of cycles of TP and LAT costs to your kernel on every single RDRAND.
>>108350842
>>108350967All fairness, Intel RDRAND 'only' takes ~500 cycles. It's AMD's putting it in microcode that takes 1k+.
>>108350983>~500 cyclesI need to see a source for this.
>>108351020Reciprocal throughput in cycles is the second to last column.https://www.agner.org/optimize/instruction_tables.pdf
>>108350842Lucky Star is a baby anime
>>108351075Yeah, I'm not buying it.
