[PATCH -next v15 00/19] riscv: Add vector ISA support

Björn Töpel bjorn at kernel.org
Thu Mar 23 03:44:54 PDT 2023


Andy Chiu <andy.chiu at sifive.com> writes:

> This patchset is implemented based on vector 1.0 spec to add vector support
> in riscv Linux kernel. There are some assumptions for this implementations.
>
> 1. We assume all harts has the same ISA in the system.
> 2. We disable vector in both kernel andy user space [1] by default. Only
>    enable an user's vector after an illegal instruction trap where it
>    actually starts executing vector (the first-use trap [2]).
> 3. We detect "riscv,isa" to determine whether vector is support or not.
>
> We defined a new structure __riscv_v_ext_state in struct thread_struct to
> save/restore the vector related registers. It is used for both kernel space
> and user space.
>  - In kernel space, the datap pointer in __riscv_v_ext_state will be
>    allocated to save vector registers.
>  - In user space,
> 	- In signal handler of user space, the structure is placed
> 	  right after __riscv_ctx_hdr, which is embedded in fp reserved
> 	  aera. This is required to avoid ABI break [2]. And datap points
> 	  to the end of __riscv_v_ext_state.
> 	- In ptrace, the data will be put in ubuf in which we use
> 	  riscv_vr_get()/riscv_vr_set() to get or set the
> 	  __riscv_v_ext_state data structure from/to it, datap pointer
> 	  would be zeroed and vector registers will be copied to the
> 	  address right after the __riscv_v_ext_state structure in ubuf.
>
> This patchset is rebased to v6.3-rc1 and it is tested by running several
> vector programs simultaneously. It delivers signals correctly in a test
> where we can see a valid ucontext_t in a signal handler, and a correct V
> context returing back from it. And the ptrace interface is tested by
> PTRACE_{GET,SET}REGSET. Lastly, KVM is tested by running above tests in
> a guest using the same kernel image. All tests are done on an rv64gcv
> virt QEMU.
>
> Note: please apply the patch at [4] due to a regression introduced by
> commit 596ff4a09b89 ("cpumask: re-introduce constant-sized cpumask
> optimizations") before testing the series.
>
> Specail thanks to Conor and Vineet for kindly giving help on- and off-list.
>
> Source tree:
> https://github.com/sifive/riscv-linux/tree/riscv/for-next/vector-v15
>
> Links:
>  - [1] https://lore.kernel.org/all/20220921214439.1491510-17-stillson@rivosinc.com/
>  - [2] https://lore.kernel.org/all/73c0124c-4794-6e40-460c-b26df407f322@rivosinc.com/T/#u
>  - [3] https://lore.kernel.org/all/20230128082847.3055316-1-apatel@ventanamicro.com/
>  - [4] https://lore.kernel.org/all/CAHk-=wiAxtKyxs6BPEzirrXw1kXJ-7ZyGpgOrbzhmC=ud-6jBA@mail.gmail.com/
> ---
> Changelog V15
>  - Rebase to risc-v -next (v6.3-rc1)
>  - Make V depend on FD in Kconfig according to the spec and shut off v
>    properly.
>  - Fix a syntax error for clang build. But mark RISCV_ISA_V GAS only due
>    to https://reviews.llvm.org/D123515
>  - Use scratch reg in inline asm instead of t4.
>  - Refine code.
>  - Cleanup per-patch changelogs.

Andy, I think the series is in a good shape! Thanks for the hard work!

To summarize; AFAIU the outstanding issues are:

 * sparse, patch 13
 * Anup's KVM comments, patch 18
 * Nathan's suggestion, patch 19

Anything else? If not, it would be amazing for a quick v16 turnaround,
addressing the points above. Hopefully the next version can land in the
upcoming release.


Björn



More information about the linux-riscv mailing list