[PATCH v4 0/2] KVM: powerpc: Use generic xfer to guest work function

[PATCH v4 0/2] KVM: powerpc: Use generic xfer to guest work function

[Vishal Chourasia]

This series fixes a KVM scheduling bug on Book3S HV where a guest VM
under a cpu.max bandwidth limit can run arbitrarily past its quota and
then appear frozen for minutes afterwards. 

== Problem ==

Since commit 2cd571245b43 ("sched/fair: Add related data structure for
task based throttle"), merged in v6.18, CFS bandwidth throttling no
longer dequeues a task directly. Instead it queues a task_work item via
task_work_add(..., TWA_RESUME), sets TIF_NOTIFY_RESUME, and relies on
that work running on the return path to actually dequeue the task.

The powerpc KVM run loops only test TIF_SIGPENDING and TIF_NEED_RESCHED
before re-entering the guest; TIF_NOTIFY_RESUME is never checked. For a
CPU-bound guest that generates few KVM exits back to userspace, the vCPU
thread never returns to user mode, so the deferred throttle task_work
never runs. The guest keeps running unchecked while its
runtime_remaining goes increasingly negative, and once it finally does
exit to userspace it is legitimately throttled for minutes while the
accrued debt is repaid at the bandwidth-timer replenishment rate.

The generic xfer-to-guest-mode infrastructure (commit 935ace2fb5cc,
"entry: Provide infrastructure for work before transitioning to guest
mode") exists precisely to handle this kind of work before each guest
entry. A full trace-backed root-cause analysis was posted with v1 [2].

== Fix ==

Opt powerpc KVM into VIRT_XFER_TO_GUEST_WORK and use the generic
xfer_to_guest_mode helpers to check for and handle pending guest-mode
work (reschedule, signals, and TIF_NOTIFY_RESUME task_work such as the
deferred CFS throttle) on every guest re-entry:

- Book3S HV: both run loops — kvmhv_run_single_vcpu() for POWER9+ and
  kvmppc_run_vcpu() for pre-POWER9.
- Book3S PR and BookE: the common kvmppc_prepare_to_enter(), which
  likewise only checked need_resched()/signal_pending().

== Changes from v3 ==
- BookE: add local_irq_disable() before hard_irq_disable() on the emulated
  MTMSR[WE] path so lockdep's irq-tracing state is resynced, fixing a
  lockdep_assert_irqs_disabled() splat in the following
  xfer_to_guest_mode_prepare(). (Reported by Sashiko AI review)

== Changes from v2 ==
- Add xfer_to_guest_mode_prepare() check for pending rcuog wakeup before
  entering guest
- While at it patch all exit points for kvmhv_run_single_vcpu() via
  done label matching the corresponding trace_kvmppc_run_vcpu_enter()
  with trace_kvmppc_run_vcpu_exit() tracepoint. (Reported by Sashiko AI
  review)
- Include posix cpu timer enablement patch by Shrikanth as part of this
  series

== Changes from v1 ==
- Extend the fix beyond Book3S HV to the shared powerpc KVM entry path:
  also convert the common kvmppc_prepare_to_enter() used by Book3S PR
  and BookE. (Shrikanth Hegde)
- Move "select VIRT_XFER_TO_GUEST_WORK" from KVM_BOOK3S_64_HV up to the
  common "config KVM" so every powerpc KVM variant gets the
  infrastructure.
- Drop the redundant signal_pending() recheck and its sigpend label in
  kvmhv_run_single_vcpu(); xfer_to_guest_mode_work_pending() is a
  superset of it.
- Preserve the E500 CONFIG_KVM_EXIT_TIMING histogram on the signal path
  via an explicit kvmppc_set_exit_type(SIGNAL_EXITS).

[1] https://lore.kernel.org/all/20250421102837.78515-2-sshegde@linux.ibm.com/
[2] https://lore.kernel.org/all/20260626105449.2897924-2-vishalc@linux.ibm.com/

Shrikanth Hegde (1):
  powerpc: enable to run posix cpu timers in task context

Vishal Chourasia (1):
  KVM: powerpc: Use generic xfer to guest work function

 arch/powerpc/Kconfig         |  1 +
 arch/powerpc/kvm/Kconfig     |  1 +
 arch/powerpc/kvm/book3s_hv.c | 39 +++++++++++++++++++++++-------------
 arch/powerpc/kvm/booke.c     |  1 +
 arch/powerpc/kvm/powerpc.c   | 37 +++++++++++++++++++++++++---------
 5 files changed, 56 insertions(+), 23 deletions(-)

-- 
2.54.0

[PATCH v4 1/2] KVM: powerpc: Use generic xfer to guest work function

[Vishal Chourasia]

Since commit 2cd571245b43 ("sched/fair: Add related data structure for
task based throttle") in v6.18, CFS bandwidth throttling no longer
dequeues a task directly; it queues task_work via TWA_RESUME and sets
TIF_NOTIFY_RESUME, relying on that work running before the task returns
to guest/user mode. The powerpc KVM run loops only checked for reschedule
and signals, never TIF_NOTIFY_RESUME, so the deferred throttle never ran
while a vCPU stayed in the run loop: a CPU-bound guest that rarely exits
to userspace ran far past its cpu.max quota and then appeared frozen for
minutes while the accrued throttle debt was repaid.

Use the generic infrastructure to check for and handle pending work
before transitioning into guest mode, replacing the open-coded
need_resched() and cond_resched() checks in the Book3S HV run loops and
in the common kvmppc_prepare_to_enter() used by the Book3S PR and BookE
run loops. The redundant signal_pending() recheck (and its sigpend label)
in kvmhv_run_single_vcpu() is also dropped, as
xfer_to_guest_mode_work_pending() is a superset of it.

This picks up handling for TIF_NOTIFY_RESUME, which was previously
ignored, meaning task work will now be correctly handled on every
guest re-entry.

Selecting VIRT_XFER_TO_GUEST_WORK disables RCU's last-resort self-IPI
fallback for vCPU tasks (see rcu_irq_work_resched()), which on
nohz_full CPUs was what forced a reschedule for deferred rcuog wakeups
queued right before guest entry. Take over that obligation the same way
x86 and s390 do: call xfer_to_guest_mode_prepare() with IRQs disabled
immediately before the final xfer_to_guest_mode_work_pending() check at
each guest-entry gate (kvmhv_run_single_vcpu(), kvmppc_run_core() and
kvmppc_prepare_to_enter()).

In kvmppc_prepare_to_enter(), IRQs are now disabled with
local_irq_disable() before hard_irq_disable(): on 32-bit,
hard_irq_disable() is a raw MSR[EE] clear that bypasses the
lockdep/irq-tracing state, and the strict xfer_to_guest_mode helpers
assert that IRQs are seen as disabled. This also allows upgrading the
racy __xfer_to_guest_mode_work_pending() check to the asserting
variant, as this loop is the terminal gate for the PR and BookE paths.

In kvmhv_run_single_vcpu(), the -EINTR exit and the pre-existing
kvmhv_setup_mmu() failure exit now leave via the done label instead of
returning directly, keeping the run_vcpu enter/exit tracepoints
balanced and vcpu->arch.ret consistent with the returned value.

In kvmppc_prepare_to_enter() the generic helper accounts the signal exit
(vcpu->stat.signal_exits and KVM_EXIT_INTR) but does not set the exit
type, so kvmppc_set_exit_type(SIGNAL_EXITS) is retained on the signal
path to preserve the E500 CONFIG_KVM_EXIT_TIMING histogram; it is a no-op
otherwise.

Signed-off-by: Vishal Chourasia <vishalc@linux.ibm.com>
---
 arch/powerpc/kvm/Kconfig     |  1 +
 arch/powerpc/kvm/book3s_hv.c | 39 +++++++++++++++++++++++-------------
 arch/powerpc/kvm/booke.c     |  1 +
 arch/powerpc/kvm/powerpc.c   | 37 +++++++++++++++++++++++++---------
 4 files changed, 55 insertions(+), 23 deletions(-)

diff --git a/arch/powerpc/kvm/Kconfig b/arch/powerpc/kvm/Kconfig
index 9a0d1c1aca6c..b6bc2fc86dca 100644
--- a/arch/powerpc/kvm/Kconfig
+++ b/arch/powerpc/kvm/Kconfig
@@ -22,6 +22,7 @@ config KVM
 	select KVM_COMMON
 	select KVM_VFIO
 	select HAVE_KVM_IRQ_BYPASS
+	select VIRT_XFER_TO_GUEST_WORK
 
 config KVM_BOOK3S_HANDLER
 	bool
diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c
index 61dbeea317f3..3cfe9a7be9c6 100644
--- a/arch/powerpc/kvm/book3s_hv.c
+++ b/arch/powerpc/kvm/book3s_hv.c
@@ -3853,7 +3853,8 @@ static noinline void kvmppc_run_core(struct kvmppc_vcore *vc)
 	 */
 	local_irq_disable();
 	hard_irq_disable();
-	if (lazy_irq_pending() || need_resched() ||
+	xfer_to_guest_mode_prepare();
+	if (lazy_irq_pending() || xfer_to_guest_mode_work_pending() ||
 	    recheck_signals_and_mmu(&core_info)) {
 		local_irq_enable();
 		vc->vcore_state = VCORE_INACTIVE;
@@ -4824,10 +4825,16 @@ static int kvmppc_run_vcpu(struct kvm_vcpu *vcpu)
 		vc->runner = vcpu;
 		if (n_ceded == vc->n_runnable) {
 			kvmppc_vcore_blocked(vc);
-		} else if (need_resched()) {
+		} else if (__xfer_to_guest_mode_work_pending()) {
 			kvmppc_vcore_preempt(vc);
-			/* Let something else run */
-			cond_resched_lock(&vc->lock);
+			/*
+			 * Let something else run. The raw helper is used as
+			 * signal exits are accounted by this path already;
+			 * it may schedule(), so drop the vcore lock.
+			 */
+			spin_unlock(&vc->lock);
+			xfer_to_guest_mode_handle_work();
+			spin_lock(&vc->lock);
 			if (vc->vcore_state == VCORE_PREEMPT)
 				kvmppc_vcore_end_preempt(vc);
 		} else {
@@ -4895,12 +4902,16 @@ int kvmhv_run_single_vcpu(struct kvm_vcpu *vcpu, u64 time_limit,
 			run->exit_reason = KVM_EXIT_FAIL_ENTRY;
 			run->fail_entry.hardware_entry_failure_reason = 0;
 			vcpu->arch.ret = r;
-			return r;
+			goto done;
 		}
 	}
 
-	if (need_resched())
-		cond_resched();
+	r = kvm_xfer_to_guest_mode_handle_work(vcpu);
+	if (r) {
+		/* -EINTR: signal pending, exit to userspace (KVM_EXIT_INTR) */
+		vcpu->arch.ret = r;
+		goto done;
+	}
 
 	kvmppc_update_vpas(vcpu);
 
@@ -4914,9 +4925,13 @@ int kvmhv_run_single_vcpu(struct kvm_vcpu *vcpu, u64 time_limit,
 
 	vcpu->arch.state = KVMPPC_VCPU_RUNNABLE;
 
-	if (signal_pending(current))
-		goto sigpend;
-	if (need_resched() || !kvm->arch.mmu_ready)
+	xfer_to_guest_mode_prepare();
+
+	/*
+	 * IRQs are disabled here, so on pending work bail to the outer loop,
+	 * which handles it via kvm_xfer_to_guest_mode_handle_work() above.
+	 */
+	if (xfer_to_guest_mode_work_pending() || !kvm->arch.mmu_ready)
 		goto out;
 
 	vcpu->cpu = pcpu;
@@ -5068,10 +5083,6 @@ int kvmhv_run_single_vcpu(struct kvm_vcpu *vcpu, u64 time_limit,
 
 	return vcpu->arch.ret;
 
- sigpend:
-	vcpu->stat.signal_exits++;
-	run->exit_reason = KVM_EXIT_INTR;
-	vcpu->arch.ret = -EINTR;
  out:
 	vcpu->cpu = -1;
 	vcpu->arch.thread_cpu = -1;
diff --git a/arch/powerpc/kvm/booke.c b/arch/powerpc/kvm/booke.c
index f3ddb24ece74..5fba199dfdd6 100644
--- a/arch/powerpc/kvm/booke.c
+++ b/arch/powerpc/kvm/booke.c
@@ -722,6 +722,7 @@ int kvmppc_core_prepare_to_enter(struct kvm_vcpu *vcpu)
 	if (vcpu->arch.shared->msr & MSR_WE) {
 		local_irq_enable();
 		kvm_vcpu_halt(vcpu);
+		local_irq_disable();
 		hard_irq_disable();
 
 		kvmppc_set_exit_type(vcpu, EMULATED_MTMSRWE_EXITS);
diff --git a/arch/powerpc/kvm/powerpc.c b/arch/powerpc/kvm/powerpc.c
index 00302399fc37..be5e48ae0c6c 100644
--- a/arch/powerpc/kvm/powerpc.c
+++ b/arch/powerpc/kvm/powerpc.c
@@ -81,23 +81,41 @@ int kvmppc_prepare_to_enter(struct kvm_vcpu *vcpu)
 	int r;
 
 	WARN_ON(irqs_disabled());
+	/*
+	 * local_irq_disable() first: on 32-bit, hard_irq_disable() alone is a
+	 * raw MSR[EE] clear that bypasses the lockdep/irq-tracing state, and
+	 * the xfer_to_guest_mode helpers assert IRQs are seen as disabled.
+	 */
+	local_irq_disable();
 	hard_irq_disable();
 
 	while (true) {
-		if (need_resched()) {
+		xfer_to_guest_mode_prepare();
+
+		if (xfer_to_guest_mode_work_pending()) {
+			/*
+			 * The helper must run with IRQs enabled and may
+			 * schedule(). On a pending signal it returns -EINTR
+			 * with run->exit_reason and vcpu->stat.signal_exits
+			 * already set, so just return to userspace.
+			 */
 			local_irq_enable();
-			cond_resched();
+			r = kvm_xfer_to_guest_mode_handle_work(vcpu);
+			local_irq_disable();
 			hard_irq_disable();
+			if (r) {
+				/*
+				 * The generic helper does not set the exit
+				 * type; record it for the E500
+				 * CONFIG_KVM_EXIT_TIMING histogram (a no-op
+				 * otherwise).
+				 */
+				kvmppc_set_exit_type(vcpu, SIGNAL_EXITS);
+				break;
+			}
 			continue;
 		}
 
-		if (signal_pending(current)) {
-			kvmppc_account_exit(vcpu, SIGNAL_EXITS);
-			vcpu->run->exit_reason = KVM_EXIT_INTR;
-			r = -EINTR;
-			break;
-		}
-
 		vcpu->mode = IN_GUEST_MODE;
 
 		/*
@@ -116,6 +134,7 @@ int kvmppc_prepare_to_enter(struct kvm_vcpu *vcpu)
 			local_irq_enable();
 			trace_kvm_check_requests(vcpu);
 			r = kvmppc_core_check_requests(vcpu);
+			local_irq_disable();
 			hard_irq_disable();
 			if (r > 0)
 				continue;
-- 
2.54.0

[PATCH v4 2/2] powerpc: enable to run posix cpu timers in task context

[Vishal Chourasia]

From: Shrikanth Hegde <sshegde@linux.ibm.com>

Now that all kvm entry to guest paths handle the task work
using the generic framework, enable HAVE_POSIX_CPU_TIMERS_TASK_WORK
which allows running posix cpu timers in task context instead of running
them in hardirq. This would is a necessary step towards enabling
PREEMPT_RT on powerNV systems.

Signed-off-by: Shrikanth Hegde <sshegde@linux.ibm.com>
Signed-off-by: Vishal Chourasia <vishalc@linux.ibm.com>
---
 arch/powerpc/Kconfig | 1 +
 1 file changed, 1 insertion(+)

diff --git a/arch/powerpc/Kconfig b/arch/powerpc/Kconfig
index f7ce5fff81f0..51555a0b1a26 100644
--- a/arch/powerpc/Kconfig
+++ b/arch/powerpc/Kconfig
@@ -281,6 +281,7 @@ config PPC
 	select HAVE_PERF_REGS
 	select HAVE_PERF_USER_STACK_DUMP
 	select HAVE_PREEMPT_DYNAMIC_KEY
+	select HAVE_POSIX_CPU_TIMERS_TASK_WORK
 	select HAVE_RETHOOK			if KPROBES
 	select HAVE_REGS_AND_STACK_ACCESS_API
 	select HAVE_RELIABLE_STACKTRACE
-- 
2.54.0