Re: [PATCH 1/3] target/riscv/cpu: ignore TCG init for KVM CPUs in reset_hold

On 2/24/25 8:47 AM, Peter Maydell wrote: On Mon, 24 Feb 2025 at 11:29, Daniel Henrique Barboza wrote: On 2/24/25 6:59 AM, Peter Maydell wrote: On Thu, 20 Feb 2025 at 16:14, Daniel Henrique Barboza wrote: riscv_cpu_reset_hold() does a lot of TCG-related initializations that aren't rele

Re: [PATCH 1/3] target/riscv/cpu: ignore TCG init for KVM CPUs in reset_hold

On Mon, 24 Feb 2025 at 11:29, Daniel Henrique Barboza wrote: > > > > On 2/24/25 6:59 AM, Peter Maydell wrote: > > On Thu, 20 Feb 2025 at 16:14, Daniel Henrique Barboza > > wrote: > >> > >> riscv_cpu_reset_hold() does a lot of TCG-related initializations that > >> aren't relevant for KVM, but neve

Re: [PATCH 1/3] target/riscv/cpu: ignore TCG init for KVM CPUs in reset_hold

On 2/24/25 6:59 AM, Peter Maydell wrote: On Thu, 20 Feb 2025 at 16:14, Daniel Henrique Barboza wrote: riscv_cpu_reset_hold() does a lot of TCG-related initializations that aren't relevant for KVM, but nevertheless are impacting the reset state of KVM vcpus. When running a KVM guest, kvm_ri

Re: [PATCH 1/3] target/riscv/cpu: ignore TCG init for KVM CPUs in reset_hold

On Thu, 20 Feb 2025 at 16:14, Daniel Henrique Barboza wrote: > > riscv_cpu_reset_hold() does a lot of TCG-related initializations that > aren't relevant for KVM, but nevertheless are impacting the reset state > of KVM vcpus. > > When running a KVM guest, kvm_riscv_reset_vcpu() is called at the end

Re: [PATCH 1/3] target/riscv/cpu: ignore TCG init for KVM CPUs in reset_hold

On Fri, Feb 21, 2025 at 2:14 AM Daniel Henrique Barboza wrote: > > riscv_cpu_reset_hold() does a lot of TCG-related initializations that > aren't relevant for KVM, but nevertheless are impacting the reset state > of KVM vcpus. > > When running a KVM guest, kvm_riscv_reset_vcpu() is called at the e

Re: [PATCH 1/3] target/riscv/cpu: ignore TCG init for KVM CPUs in reset_hold

On Thu, Feb 20, 2025 at 01:13:11PM -0300, Daniel Henrique Barboza wrote: > riscv_cpu_reset_hold() does a lot of TCG-related initializations that > aren't relevant for KVM, but nevertheless are impacting the reset state > of KVM vcpus. > > When running a KVM guest, kvm_riscv_reset_vcpu() is called

[PATCH 1/3] target/riscv/cpu: ignore TCG init for KVM CPUs in reset_hold

riscv_cpu_reset_hold() does a lot of TCG-related initializations that aren't relevant for KVM, but nevertheless are impacting the reset state of KVM vcpus. When running a KVM guest, kvm_riscv_reset_vcpu() is called at the end of reset_hold(). At that point env->mstatus is initialized to a non-zero