- Jun 04, 2022
-
-
Tim Newsome authored
Give it the same interface as load_slow_path(), so there's no need to worry about endianness in either of those two functions.
-
Andrew Waterman authored
Move all misaligned handling into the slow path. This is justifiable on the basis that, on average, they account for a significant minority of accesses even on systems that support them with reasonable performance.
-
Tim Newsome authored
It already throws the exception when it needs to.
-
Tim Newsome authored
Fixes the other half of #971. Compared the start of this set of changes, it now takes 11% longer to run the towers benchmark (with 22 discs).
-
Tim Newsome authored
Easier to read, maintaing, compiler errors, etc.
-
Tim Newsome authored
When the slow path occurs, don't check for alignment in load_fast(). I haven't tested different endianness. With this change, #971 is resolved for loads.
-
- May 27, 2022
-
-
Tim Newsome authored
Easier to edit, better error messages from compiler, etc.
-
Tim Newsome authored
-
Tim Newsome authored
Slightly more readable than sizeof(type##_t) everywhere.
-
Tim Newsome authored
Only in slow path right now.
-
Tim Newsome authored
In preparation for checking accesses before they happen, when we only know the address and not the data.
-
Tim Newsome authored
-
- May 16, 2022
-
-
Andrew Waterman authored
Include recently added headers in riscv/riscv.mk.in
-
Pirmin Vogel authored
-
- May 14, 2022
-
-
Andrew Waterman authored
Simplify decode_insn and insn_desc_t
-
Andrew Waterman authored
Iterate over the instruction chains in reverse order, prioritizing the last call to `disassembler_t::add_insn`. To preserve behavior for the standard instructions, reverse the order in which we add instructions in the `disassembler_t` constructor. Supersedes #995.
-
Andrew Waterman authored
We do not condition them on Zicbop/Zihintpause because, definitionally, all implementations provide them.
-
- May 13, 2022
-
-
Andrew Waterman authored
-
Andrew Waterman authored
-
Andrew Waterman authored
-
Andrew Waterman authored
Fixes bug introduced in 5b7cdbe1
-
Andrew Waterman authored
-
Andrew Waterman authored
The field is rendered unnecessary by 11f5942b. Undoes some changes from 750f008e.
-
Andrew Waterman authored
These add to the length of the instruction list without providing an apparent benefit.
-
- May 12, 2022
-
-
Andrew Waterman authored
Fix for issue #990, implements the PBMTE bit in the henvcfg /menvcfg reisters
-
Ryan Buchner authored
See #990.
-
Ryan Buchner authored
Fix issue #990.
-
Ryan Buchner authored
Also make PBMTE set on reset for backward compatibility. Since before Spike proceeded as if these bits were set if the extension was enabled.
-
Ryan Buchner authored
To do so implemented henvcfg_csr_t. henvcfg.PBMTE will be read only 0 if menvcfg.PBMTE = 0.
-
Andrew Waterman authored
rvv: fix the checking eew and elen for index load
-
- May 11, 2022
-
-
Chih-Min Chao authored
eew of index register can't be larger than elen ex: elen = 32, vloxei64.v is illegal Signed-off-by:Chih-Min Chao <chihmin.chao@sifive.com>
-
- May 10, 2022
-
-
Andrew Waterman authored
Factor out P and V extension macros into their own headers
-
- May 06, 2022
-
-
Andrew Waterman authored
No functional change.
-
Andrew Waterman authored
No functional change.
-
- May 05, 2022
-
-
Scott Johnson authored
Implement Smepmp extension
-
soberl authored
-
soberl@nvidia.com authored
-
soberl@nvidia.com authored
-
soberl@nvidia.com authored
-
soberl@nvidia.com authored
-