IRC channel logs

2024-10-29.log

back to list of logs

<fossy>ooh interesting
<fossy>hmmmmm
<fossy>i'm still very conflicted on my thoughts about the hardware bootstrapping problem
<fossy>i don't take the view including bootstrapping the whole fabrication process - my position is to treat the fabrication process as a black box that introduces some kind of distrust (and then on-die firmware like intel ME as a lot more distrust)
<fossy>& so depending on the level of distrust you're comfortable with you have different starting points
<fossy>but i can't decide whether something like a fpga-emulated riscv core is sufficient (which also introduces the problem of compiling the core)
<fossy>or whether actual hand-built hardware is at all worthwhile
<lanodan>I think in terms of trust I'd rather use cross-compiling so it's a bit like Diverse Double Compiling, hardware edition. Although there should still be the goal of finding a trustworthy computer.
<fossy>yeah, I tend to agree in terms of mitigating distrust of hardware
<fossy>only really feasible for open architectures I think though
<fossy>or otherwise many-impls
<fossy>(some problem with DCC now, clang/gcc duopoly in the (open) compiler land makes it difficult to effectively execute DCC)
<lanodan>And part of the problem is even if say you'd have a trustworthy computer but only capable of disassembling programs to check the work of less trustworthy computers, you'd then need a lot of human time for inspection.
<fossy>right
<fossy>these are difficult problems, that's for sure :D