in the PM spec--"Alternatively, an implementation may choose to "waste" TLB capacity by having up to 4 duplicate entries for each page." Why are there duplicate entries for each page?

[chara811]

When vsatp.mode=bare and hgatp.mode= sv48, the ptw_pa looked up by the G-stage is {hgatp.ppn[35:2], GPA[49:39]}, with a size of 4K. When GPA[49:48] is 00, 01, 10, or 11, the page table entries retrieved should be different based on the value of GPA[49:48].

[SiFiveHolland]

@martinmaas Since instruction fetches are not subject to pointer masking, they would need to resolve to a possibly different underlying page when GPA[49:48] is nonzero. Is it really a good idea to recommend duplicate TLB entries? The TLB would need to return different results for loads/stores vs instruction fetches, so it would somehow need to know what instruction caused the TLB lookup. And this would break virtually-indexed caches as well.