This mitigates spectre variant #2, branch target injection. We also have a mitigation for meltdown, namely KPTI. Is there a known mitigation for spectre variant

This mitigates spectre variant #2, branch target injection. We also have a mitigation for meltdown, namely KPTI. Is there a known mitigation for spectre variant #1, bounds check bypass? Maybe I'm being naive, but would a simple modulo instruction work? Consider the example code from https://googleprojectzero.blogspot.com/2018/01/reading-privileged-memory-with-side.html https://googleprojectzero.blogspot.com/2018/01/reading-privi...: unsigned long untrusted_offset_from_caller = ...; if (untrusted_offset_from_caller < arr1->length) { unsigned char value = arr1->data[untrusted_offset_from_caller]; ... } If instead we did: unsigned char value = arr1->data[untrusted_offset_from_caller % arr1->length]; Would this produce a data dependency that prevents speculative execution from reading an out-of-bounds memory address? (Ignore for the moment that a sufficiently smart compiler might "optimize" out the modulo here.)