Re: Re: [RFC PATCH 0/3] RISC-V: KVM: Guest Debug Support

["Chao Du" <duchao@eswincomputing.com>]

On 2024-01-04 19:59, Anup Patel <anup@brainfault.org> wrote:
> 
> On Fri, Dec 22, 2023 at 1:59 PM Chao Du <duchao@eswincomputing.com> wrote:
> >
> > On 2023-12-21 21:01, Anup Patel <apatel@ventanamicro.com> wrote:
> > >
> > > On Thu, Dec 21, 2023 at 3:21 PM Chao Du <duchao@eswincomputing.com> wrote:
> > > >
> > > > This series implements KVM Guest Debug on RISC-V. Currently, we can
> > > > debug RISC-V KVM guest from the host side, with software breakpoints.
> > > >
> > > > A brief test was done on QEMU RISC-V hypervisor emulator.
> > > >
> > > > A TODO list which will be added later:
> > > > 1. HW breakpoints support
> > > > 2. Test cases
> > >
> > > Himanshu has already done the complete HW breakpoint implementation
> > > in OpenSBI, Linux RISC-V, and KVM RISC-V. This is based on the upcoming
> > > SBI debug trigger extension draft proposal.
> > > (Refer, https://lists.riscv.org/g/tech-debug/message/1261)
> > >
> > > There are also RISE projects to track these efforts:
> > > https://wiki.riseproject.dev/pages/viewpage.action?pageId=394541
> > > https://wiki.riseproject.dev/pages/viewpage.action?pageId=394545
> > >
> > > Currently, we are in the process of upstreaming the OpenSBI support
> > > for SBI debug trigger extension. The Linux RISC-V and KVM RISC-V
> > > patches require SBI debug trigger extension and Sdtrig extension to
> > > be frozen which will happen next year 2024.
> > >
> > > Regards,
> > > Anup
> > >
> >
> > Hi Anup,
> >
> > Thank you for the information and your great work on the SBI
> > Debug Trigger Extension proposal.
> >
> > So I think that 'HW breakpoints support' in the above TODO list
> > will be taken care of by Himanshu following the extension proposal.
> >
> > On the other hand, if I understand correctly, the software
> > breakpoint part of KVM Guest Debug has no dependency on the new
> > extension since it does not use the trigger module. Just an
> > ebreak substitution is made.
> >
> > So may I know your suggestion about this RFC? Both in KVM and QEMU.
> 
> Sorry for the delay in response due to holiday season other
> stuff keeping me busy.

No problem. :)

> 
> If this is about ebreak instruction virtualization then this series
> needs following changes:
> 1) Update cover letter to indicate this series focus on ebreak
>      instruction virtualization
> 2) PATCH1 and PATCH2 can be merged into one PATCH1
> 3) Include a new patch which adds KVM selftest for ebreak
>     based guest debug. This selftest will test both:
>     A) Taking "ebreak" trap from guest as KVM_EXIT_DEBUG
>          in host user-space
>     B) Taking "ebreak" trap from guest as BREAKPOINT
>         exception in guest

Sure, I will prepare a V2 series.

Thanks.

> 
> Regards,
> Anup
> 
> >
> > Regards,
> > Chao
> >
> > > >
> > > > This series is based on Linux 6.7-rc6 and is also available at:
> > > > https://github.com/Du-Chao/linux/tree/riscv_gd_sw
> > > >
> > > > The matched QEMU is available at:
> > > > https://github.com/Du-Chao/qemu/tree/riscv_gd_sw
> > > >
> > > > Chao Du (3):
> > > >   RISC-V: KVM: Enable the KVM_CAP_SET_GUEST_DEBUG capability
> > > >   RISC-V: KVM: Implement kvm_arch_vcpu_ioctl_set_guest_debug()
> > > >   RISC-V: KVM: Handle breakpoint exits for VCPU
> > > >
> > > >  arch/riscv/include/uapi/asm/kvm.h |  1 +
> > > >  arch/riscv/kvm/vcpu.c             | 15 +++++++++++++--
> > > >  arch/riscv/kvm/vcpu_exit.c        |  4 ++++
> > > >  arch/riscv/kvm/vm.c               |  1 +
> > > >  4 files changed, 19 insertions(+), 2 deletions(-)
> > > >
> > > > --
> > > > 2.17.1
> > > >
> > > >
> > > > --
> > > > kvm-riscv mailing list
> > > > kvm-riscv@lists.infradead.org
> > > > http://lists.infradead.org/mailman/listinfo/kvm-riscv