Re: [PATCH] KVM: arm64: nv: Set ISTATUS for emulated timers, If timer expired

[Marc Zyngier <maz@kernel.org>]

On Tue, 18 Feb 2025 07:33:11 +0000,
Ganapatrao Kulkarni <gankulkarni@os.amperecomputing.com> wrote:
> 
> 
> Hi Marc,
> 
> On 15-02-2025 11:20 pm, Marc Zyngier wrote:
> > On Mon, 10 Feb 2025 18:26:48 +0000,
> > Eric Auger <eauger@redhat.com> wrote:
> >> 
> >> Hi Marc,
> >> 
> >> On 2/7/25 7:38 PM, Marc Zyngier wrote:
> >>> On Fri, 07 Feb 2025 18:09:58 +0000,
> >>> Oliver Upton <oliver.upton@linux.dev> wrote:
> >>>> 
> >>>> Hey,
> >>>> 
> >>>> On Fri, Feb 07, 2025 at 05:45:33PM +0000, Marc Zyngier wrote:
> >>>>> I found at least one issue that could fail the migration. Before the
> >>>>> VM starts running, we limit the feature set to the subset we actually
> >>>>> support with NV.
> >>>>> 
> >>>>> By doing this, we also change the value of IDreg fields that are not
> >>>>> writable, because they describe features that we don't support.
> >>>>> Obviously, that fails on restore.
> >>>>> 
> >>>>> I need to have a think...
> >>>> 
> >>>> We spoke about this a while ago (and I forgot til now), but I was
> >>>> wondering if we could use vCPU feature flags to describe NV, including
> >>>> the selection between FEAT_E2H0 and FEAT_VHE.
> >>>> 
> >>>> I think this might match userspace expectations a bit more closely where
> >>>> the state of the ID registers after init gives the actual feature set
> >>>> supported by the VM.
> >>> 
> >>> I'm not sure that's enough. Let me give you an example:
> >>> 
> >>> My host has FEAT_XNX, described in ID_AA64MMFR1_EL1.XNX. For whatever
> >>> reason, we don't allow this field to be written to, even out of NV
> >>> context. This is odd, because for an EL1 VM, this field means nothing
> >>> at all.
> >> So the curprit fields for me look like
> >> 
> >> - ID_AA64MMFR1_EL1.XNX
> >> - ID_AA64DFR0_EL1.DoubleLock
> >> - ID_AA64PFR0_EL1.RAS
> >> 
> >> This is still based on your nv-next branch from Jan 9
> >> https://github.com/eauger/linux/tree/nv_next_jan9_2025
> > 
> > I have now pushed out a new nv-next branch with the new and improved
> > UAPI. I expect migration to work a bit better, or at least not to
> > explode on ID register restore. You will notice that things have
> > changed a bit (extra flag and cap for FEAT_E2H0), but nothing really
> > major.
> > 
> 
> Tried nv-next branch and it is breaking(kernel Oops) for normal VM
> boot itself with qemu. Looks like this is happening since qemu is
> trying to write to ID_UNALLOCATED mapped registers as part of
> save-restore of registers.

My take on this problem ends up being more consolidation, and make
sure that the individual macros only override the default callbacks
for idregs.

Additionally, ID_UNALLOCATED gets a name matching the architectural
encoding.

	M.

diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c
index e6f4599dca48e..2e14562b5841f 100644
--- a/arch/arm64/kvm/sys_regs.c
+++ b/arch/arm64/kvm/sys_regs.c
@@ -2261,24 +2261,26 @@ static bool bad_redir_trap(struct kvm_vcpu *vcpu,
  * from userspace.
  */
 
-#define ID_DESC(name)				\
-	SYS_DESC(SYS_##name),			\
+#define ID_DESC_DEFAULT_CALLBACKS		\
 	.access	= access_id_reg,		\
 	.get_user = get_id_reg,			\
+	.set_user = set_id_reg,			\
+	.visibility = id_visibility,		\
 	.reset = kvm_read_sanitised_id_reg
 
+#define ID_DESC(name)				\
+	SYS_DESC(SYS_##name),			\
+	ID_DESC_DEFAULT_CALLBACKS
+
 /* sys_reg_desc initialiser for known cpufeature ID registers */
 #define ID_SANITISED(name) {			\
 	ID_DESC(name),				\
-	.set_user = set_id_reg,			\
-	.visibility = id_visibility,		\
 	.val = 0,				\
 }
 
 /* sys_reg_desc initialiser for known cpufeature ID registers */
 #define AA32_ID_SANITISED(name) {		\
 	ID_DESC(name),				\
-	.set_user = set_id_reg,			\
 	.visibility = aa32_id_visibility,	\
 	.val = 0,				\
 }
@@ -2286,8 +2288,6 @@ static bool bad_redir_trap(struct kvm_vcpu *vcpu,
 /* sys_reg_desc initialiser for writable ID registers */
 #define ID_WRITABLE(name, mask) {		\
 	ID_DESC(name),				\
-	.set_user = set_id_reg,			\
-	.visibility = id_visibility,		\
 	.val = mask,				\
 }
 
@@ -2295,7 +2295,6 @@ static bool bad_redir_trap(struct kvm_vcpu *vcpu,
 #define ID_FILTERED(sysreg, name, mask) {	\
 	ID_DESC(sysreg),				\
 	.set_user = set_##name,				\
-	.visibility = id_visibility,			\
 	.val = (mask),					\
 }
 
@@ -2305,10 +2304,9 @@ static bool bad_redir_trap(struct kvm_vcpu *vcpu,
  * (1 <= crm < 8, 0 <= Op2 < 8).
  */
 #define ID_UNALLOCATED(crm, op2) {			\
+	.name = "S3_0_0_" #crm "_" #op2,		\
 	Op0(3), Op1(0), CRn(0), CRm(crm), Op2(op2),	\
-	.access = access_id_reg,			\
-	.get_user = get_id_reg,				\
-	.set_user = set_id_reg,				\
+	ID_DESC_DEFAULT_CALLBACKS,			\
 	.visibility = raz_visibility,			\
 	.val = 0,					\
 }
@@ -2320,7 +2318,6 @@ static bool bad_redir_trap(struct kvm_vcpu *vcpu,
  */
 #define ID_HIDDEN(name) {			\
 	ID_DESC(name),				\
-	.set_user = set_id_reg,			\
 	.visibility = raz_visibility,		\
 	.val = 0,				\
 }

-- 
Without deviation from the norm, progress is not possible.