[GIT,PULL] KVM/riscv changes for 6.8 part #1

Pull-request

https://github.com/kvm-riscv/linux.git tags/kvm-riscv-6.8-1

Message

Anup Patel Dec. 31, 2023, 5:33 a.m. UTC

Hi Paolo,

We have the following KVM RISC-V changes for 6.8:
1) KVM_GET_REG_LIST improvement for vector registers
2) Generate ISA extension reg_list using macros in get-reg-list selftest
3) Steal time account support along with selftest

Please pull.

Please note that I will be sending another PR for 6.8 which will
include two more changes:
1) KVM RISC-V report more ISA extensions through ONE_REG
2) RISC-V SBI v2.0 PMU improvements and Perf sampling in KVM guest

Two separate PRs are because #1 (above) depends on a series
merged by Palmer for 6.8 and #2 (above) requires little more testing.
I hope you are okay with two separate PRs for 6.8.

Regards,
Anup

The following changes since commit 861deac3b092f37b2c5e6871732f3e11486f7082:

  Linux 6.7-rc7 (2023-12-23 16:25:56 -0800)

are available in the Git repository at:

  https://github.com/kvm-riscv/linux.git tags/kvm-riscv-6.8-1

for you to fetch changes up to aad86da229bc9d0390dc2c02eb0db9ab1f50d059:

  RISC-V: KVM: selftests: Add get-reg-list test for STA registers
(2023-12-30 11:26:47 +0530)

----------------------------------------------------------------
KVM/riscv changes for 6.8 part #1

- KVM_GET_REG_LIST improvement for vector registers
- Generate ISA extension reg_list using macros in get-reg-list selftest
- Steal time account support along with selftest

----------------------------------------------------------------
Andrew Jones (19):
      RISC-V: KVM: Don't add SBI multi regs in get-reg-list
      KVM: riscv: selftests: Drop SBI multi registers
      RISC-V: KVM: Make SBI uapi consistent with ISA uapi
      KVM: riscv: selftests: Add RISCV_SBI_EXT_REG
      KVM: riscv: selftests: Use register subtypes
      RISC-V: KVM: selftests: Treat SBI ext regs like ISA ext regs
      RISC-V: paravirt: Add skeleton for pv-time support
      RISC-V: Add SBI STA extension definitions
      RISC-V: paravirt: Implement steal-time support
      RISC-V: KVM: Add SBI STA extension skeleton
      RISC-V: KVM: Add steal-update vcpu request
      RISC-V: KVM: Add SBI STA info to vcpu_arch
      RISC-V: KVM: Add support for SBI extension registers
      RISC-V: KVM: Add support for SBI STA registers
      RISC-V: KVM: Implement SBI STA extension
      RISC-V: KVM: selftests: Move sbi_ecall to processor.c
      RISC-V: KVM: selftests: Add guest_sbi_probe_extension
      RISC-V: KVM: selftests: Add steal_time test support
      RISC-V: KVM: selftests: Add get-reg-list test for STA registers

Anup Patel (2):
      KVM: riscv: selftests: Generate ISA extension reg_list using macros
      RISC-V: KVM: Fix indentation in kvm_riscv_vcpu_set_reg_csr()

Chao Du (1):
      RISC-V: KVM: remove a redundant condition in kvm_arch_vcpu_ioctl_run()

Clément Léger (2):
      riscv: kvm: Use SYM_*() assembly macros instead of deprecated ones
      riscv: kvm: use ".L" local labels in assembly when applicable

Daniel Henrique Barboza (3):
      RISC-V: KVM: set 'vlenb' in kvm_riscv_vcpu_alloc_vector_context()
      RISC-V: KVM: add 'vlenb' Vector CSR
      RISC-V: KVM: add vector registers and CSRs in KVM_GET_REG_LIST

 Documentation/admin-guide/kernel-parameters.txt    |   6 +-
 arch/riscv/Kconfig                                 |  19 +
 arch/riscv/include/asm/kvm_host.h                  |  10 +
 arch/riscv/include/asm/kvm_vcpu_sbi.h              |  20 +-
 arch/riscv/include/asm/paravirt.h                  |  28 +
 arch/riscv/include/asm/paravirt_api_clock.h        |   1 +
 arch/riscv/include/asm/sbi.h                       |  17 +
 arch/riscv/include/uapi/asm/kvm.h                  |  13 +
 arch/riscv/kernel/Makefile                         |   1 +
 arch/riscv/kernel/paravirt.c                       | 135 +++++
 arch/riscv/kernel/time.c                           |   3 +
 arch/riscv/kvm/Kconfig                             |   1 +
 arch/riscv/kvm/Makefile                            |   1 +
 arch/riscv/kvm/vcpu.c                              |  10 +-
 arch/riscv/kvm/vcpu_onereg.c                       | 135 +++--
 arch/riscv/kvm/vcpu_sbi.c                          | 142 +++--
 arch/riscv/kvm/vcpu_sbi_replace.c                  |   2 +-
 arch/riscv/kvm/vcpu_sbi_sta.c                      | 208 ++++++++
 arch/riscv/kvm/vcpu_switch.S                       |  32 +-
 arch/riscv/kvm/vcpu_vector.c                       |  16 +
 tools/testing/selftests/kvm/Makefile               |   5 +-
 .../testing/selftests/kvm/include/kvm_util_base.h  |   1 +
 .../selftests/kvm/include/riscv/processor.h        |  62 ++-
 tools/testing/selftests/kvm/lib/riscv/processor.c  |  49 +-
 tools/testing/selftests/kvm/lib/riscv/ucall.c      |  26 -
 tools/testing/selftests/kvm/riscv/get-reg-list.c   | 588 ++++++++++-----------
 tools/testing/selftests/kvm/steal_time.c           |  99 ++++
 27 files changed, 1184 insertions(+), 446 deletions(-)
 create mode 100644 arch/riscv/include/asm/paravirt.h
 create mode 100644 arch/riscv/include/asm/paravirt_api_clock.h
 create mode 100644 arch/riscv/kernel/paravirt.c
 create mode 100644 arch/riscv/kvm/vcpu_sbi_sta.c

Comments

Paolo Bonzini Jan. 2, 2024, 6:24 p.m. UTC | #1

On Sun, Dec 31, 2023 at 6:33 AM Anup Patel <[e-mail]> wrote:
>
> Hi Paolo,
>
> We have the following KVM RISC-V changes for 6.8:
> 1) KVM_GET_REG_LIST improvement for vector registers
> 2) Generate ISA extension reg_list using macros in get-reg-list selftest
> 3) Steal time account support along with selftest

Just one small thing I noticed on (3), do you really need cpu_to_le64
and le64_to_cpu on RISC-V? It seems that it was copied from aarch64.
No need to resend the PR anyway, of course.

> Please pull.
>
> Please note that I will be sending another PR for 6.8 which will
> include two more changes:
> 1) KVM RISC-V report more ISA extensions through ONE_REG
> 2) RISC-V SBI v2.0 PMU improvements and Perf sampling in KVM guest
>
> Two separate PRs are because #1 (above) depends on a series
> merged by Palmer for 6.8 and #2 (above) requires little more testing.
> I hope you are okay with two separate PRs for 6.8.

Yes, sure. The more the merrier. :)  If you want to send only #1, that
may be better?

Paolo

Andrew Jones Jan. 2, 2024, 8:25 p.m. UTC | #2

On Tue, Jan 02, 2024 at 07:24:26PM +0100, Paolo Bonzini wrote:
> On Sun, Dec 31, 2023 at 6:33 AM Anup Patel <[e-mail]> wrote:
> >
> > Hi Paolo,
> >
> > We have the following KVM RISC-V changes for 6.8:
> > 1) KVM_GET_REG_LIST improvement for vector registers
> > 2) Generate ISA extension reg_list using macros in get-reg-list selftest
> > 3) Steal time account support along with selftest
> 
> Just one small thing I noticed on (3), do you really need cpu_to_le64
> and le64_to_cpu on RISC-V? It seems that it was copied from aarch64.
> No need to resend the PR anyway, of course.

While Linux/KVM is only LE, the arch doesn't prohibit S-mode being
configured to use BE memory accesses, so I kept the conversions. They
at least provide some self-documenting of the code. The biggest
problem with them, though, is that I didn't use __le64 types and now
sparse is yelling at me. I patched that this morning, but didn't get
a chance to post yet. I could instead rip out the conversions to
quiet sparse, if that would be preferred.

Thanks,
drew

Anup Patel Jan. 4, 2024, 5:30 a.m. UTC | #3

Hi Paolo,

On Wed, Jan 3, 2024 at 1:55 AM Andrew Jones <[e-mail]> wrote:
>
> On Tue, Jan 02, 2024 at 07:24:26PM +0100, Paolo Bonzini wrote:
> > On Sun, Dec 31, 2023 at 6:33 AM Anup Patel <[e-mail]> wrote:
> > >
> > > Hi Paolo,
> > >
> > > We have the following KVM RISC-V changes for 6.8:
> > > 1) KVM_GET_REG_LIST improvement for vector registers
> > > 2) Generate ISA extension reg_list using macros in get-reg-list selftest
> > > 3) Steal time account support along with selftest
> >
> > Just one small thing I noticed on (3), do you really need cpu_to_le64
> > and le64_to_cpu on RISC-V? It seems that it was copied from aarch64.
> > No need to resend the PR anyway, of course.
>
> While Linux/KVM is only LE, the arch doesn't prohibit S-mode being
> configured to use BE memory accesses, so I kept the conversions. They
> at least provide some self-documenting of the code. The biggest
> problem with them, though, is that I didn't use __le64 types and now
> sparse is yelling at me. I patched that this morning, but didn't get
> a chance to post yet. I could instead rip out the conversions to
> quiet sparse, if that would be preferred.

The SBI spec is quite explicit about endianness of data in shared
memory. Also, the RISC-V priv spec allows BE load/store operations
so eventually we might see BE platforms. I suggest keeping the LE
conversion macros and __le32/__le64 data types.

Regarding the sparse errors, Drew can send fix patches which I
will include in the second PR for 6.8.

I hope this is okay.

Regards,
Anup

Anup Patel Jan. 9, 2024, 6:02 a.m. UTC | #4

Hi Paolo,

On Thu, Jan 4, 2024 at 11:00 AM Anup Patel <[e-mail]> wrote:
>
> Hi Paolo,
>
> On Wed, Jan 3, 2024 at 1:55 AM Andrew Jones <[e-mail]> wrote:
> >
> > On Tue, Jan 02, 2024 at 07:24:26PM +0100, Paolo Bonzini wrote:
> > > On Sun, Dec 31, 2023 at 6:33 AM Anup Patel <[e-mail]> wrote:
> > > >
> > > > Hi Paolo,
> > > >
> > > > We have the following KVM RISC-V changes for 6.8:
> > > > 1) KVM_GET_REG_LIST improvement for vector registers
> > > > 2) Generate ISA extension reg_list using macros in get-reg-list selftest
> > > > 3) Steal time account support along with selftest
> > >
> > > Just one small thing I noticed on (3), do you really need cpu_to_le64
> > > and le64_to_cpu on RISC-V? It seems that it was copied from aarch64.
> > > No need to resend the PR anyway, of course.
> >
> > While Linux/KVM is only LE, the arch doesn't prohibit S-mode being
> > configured to use BE memory accesses, so I kept the conversions. They
> > at least provide some self-documenting of the code. The biggest
> > problem with them, though, is that I didn't use __le64 types and now
> > sparse is yelling at me. I patched that this morning, but didn't get
> > a chance to post yet. I could instead rip out the conversions to
> > quiet sparse, if that would be preferred.
>
> The SBI spec is quite explicit about endianness of data in shared
> memory. Also, the RISC-V priv spec allows BE load/store operations
> so eventually we might see BE platforms. I suggest keeping the LE
> conversion macros and __le32/__le64 data types.
>
> Regarding the sparse errors, Drew can send fix patches which I
> will include in the second PR for 6.8.
>
> I hope this is okay.
>

Friendly ping ?

Regards,
Anup

Paolo Bonzini Jan. 9, 2024, 11:34 a.m. UTC | #5

On Tue, Jan 9, 2024 at 7:02 AM Anup Patel <[e-mail]> wrote:
> > > On Tue, Jan 02, 2024 at 07:24:26PM +0100, Paolo Bonzini wrote:
> > > > On Sun, Dec 31, 2023 at 6:33 AM Anup Patel <[e-mail]> wrote:
> > > > > We have the following KVM RISC-V changes for 6.8:
> > > > > 1) KVM_GET_REG_LIST improvement for vector registers
> > > > > 2) Generate ISA extension reg_list using macros in get-reg-list selftest
> > > > > 3) Steal time account support along with selftest
> > > >
> > > > Just one small thing I noticed on (3), do you really need cpu_to_le64
> > > > and le64_to_cpu on RISC-V? It seems that it was copied from aarch64.
> > > > No need to resend the PR anyway, of course.
>
> Friendly ping ?

Hi, I had already pulled but I was waiting for the last x86 tests to
finish. Everything is now in kvm.git.

I'll send the PR to Linus once a bunch of conflicting trees have been pulled.

Paolo