[PATCH v2 0/4] KVM: selftests: xen_shinfo cleanups and slowpath test

mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed

* [PATCH v2 0/4] KVM: selftests: xen_shinfo cleanups and slowpath test
@ 2023-02-04  2:41 Sean Christopherson
  2023-02-04  2:41 ` [PATCH v2 1/4] KVM: selftests: Move the guts of kvm_hypercall() to a separate macro Sean Christopherson
                   ` (4 more replies)
  0 siblings, 5 replies; 6+ messages in thread
From: Sean Christopherson @ 2023-02-04  2:41 UTC (permalink / raw)
  To: Paolo Bonzini; +Cc: kvm, linux-kernel, Sean Christopherson, David Woodhouse

David's new testcase to validate the slow path for EVTCHNOP_send, on top
of cleanups to reduce the amount of copy+paste in the test.

Applies on top of `kvm-x86 selftests`.

David Woodhouse (2):
  KVM: selftests: Use enum for test numbers in xen_shinfo_test
  KVM: selftests: Add EVTCHNOP_send slow path test to xen_shinfo_test

Sean Christopherson (2):
  KVM: selftests: Move the guts of kvm_hypercall() to a separate macro
  KVM: selftests: Add helpers to make Xen-style VMCALL/VMMCALL
    hypercalls

 .../selftests/kvm/include/x86_64/processor.h  |   2 +
 .../selftests/kvm/lib/x86_64/processor.c      |  37 ++-
 .../selftests/kvm/x86_64/xen_shinfo_test.c    | 221 +++++++++---------
 3 files changed, 145 insertions(+), 115 deletions(-)


base-commit: 531f33c5a6edf259da4960de694293e458262d14
-- 
2.39.1.519.gcb327c4b5f-goog


^ permalink raw reply	[flat|nested] 6+ messages in thread

* [PATCH v2 1/4] KVM: selftests: Move the guts of kvm_hypercall() to a separate macro
  2023-02-04  2:41 [PATCH v2 0/4] KVM: selftests: xen_shinfo cleanups and slowpath test Sean Christopherson
@ 2023-02-04  2:41 ` Sean Christopherson
  2023-02-04  2:41 ` [PATCH v2 2/4] KVM: selftests: Add helpers to make Xen-style VMCALL/VMMCALL hypercalls Sean Christopherson
                   ` (3 subsequent siblings)
  4 siblings, 0 replies; 6+ messages in thread
From: Sean Christopherson @ 2023-02-04  2:41 UTC (permalink / raw)
  To: Paolo Bonzini; +Cc: kvm, linux-kernel, Sean Christopherson, David Woodhouse

Extract the guts of kvm_hypercall() to a macro so that Xen hypercalls,
which have a different register ABI, can reuse the VMCALL vs. VMMCALL
logic.

No functional change intended.

Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 .../selftests/kvm/lib/x86_64/processor.c      | 29 +++++++++++--------
 1 file changed, 17 insertions(+), 12 deletions(-)

diff --git a/tools/testing/selftests/kvm/lib/x86_64/processor.c b/tools/testing/selftests/kvm/lib/x86_64/processor.c
index ae1e573d94ce..ff901cb47ffc 100644
--- a/tools/testing/selftests/kvm/lib/x86_64/processor.c
+++ b/tools/testing/selftests/kvm/lib/x86_64/processor.c
@@ -1139,21 +1139,26 @@ const struct kvm_cpuid_entry2 *get_cpuid_entry(const struct kvm_cpuid2 *cpuid,
 	return NULL;
 }
 
+#define X86_HYPERCALL(inputs...)					\
+({									\
+	uint64_t r;							\
+									\
+	asm volatile("test %[use_vmmcall], %[use_vmmcall]\n\t"		\
+		     "jnz 1f\n\t"					\
+		     "vmcall\n\t"					\
+		     "jmp 2f\n\t"					\
+		     "1: vmmcall\n\t"					\
+		     "2:"						\
+		     : "=a"(r)						\
+		     : [use_vmmcall] "r" (host_cpu_is_amd), inputs);	\
+									\
+	r;								\
+})
+
 uint64_t kvm_hypercall(uint64_t nr, uint64_t a0, uint64_t a1, uint64_t a2,
 		       uint64_t a3)
 {
-	uint64_t r;
-
-	asm volatile("test %[use_vmmcall], %[use_vmmcall]\n\t"
-		     "jnz 1f\n\t"
-		     "vmcall\n\t"
-		     "jmp 2f\n\t"
-		     "1: vmmcall\n\t"
-		     "2:"
-		     : "=a"(r)
-		     : "a"(nr), "b"(a0), "c"(a1), "d"(a2), "S"(a3),
-		       [use_vmmcall] "r" (host_cpu_is_amd));
-	return r;
+	return X86_HYPERCALL("a"(nr), "b"(a0), "c"(a1), "d"(a2), "S"(a3));
 }
 
 const struct kvm_cpuid2 *kvm_get_supported_hv_cpuid(void)
-- 
2.39.1.519.gcb327c4b5f-goog


^ permalink raw reply	[flat|nested] 6+ messages in thread

* [PATCH v2 2/4] KVM: selftests: Add helpers to make Xen-style VMCALL/VMMCALL hypercalls
  2023-02-04  2:41 [PATCH v2 0/4] KVM: selftests: xen_shinfo cleanups and slowpath test Sean Christopherson
  2023-02-04  2:41 ` [PATCH v2 1/4] KVM: selftests: Move the guts of kvm_hypercall() to a separate macro Sean Christopherson
@ 2023-02-04  2:41 ` Sean Christopherson
  2023-02-04  2:41 ` [PATCH v2 3/4] KVM: selftests: Use enum for test numbers in xen_shinfo_test Sean Christopherson
                   ` (2 subsequent siblings)
  4 siblings, 0 replies; 6+ messages in thread
From: Sean Christopherson @ 2023-02-04  2:41 UTC (permalink / raw)
  To: Paolo Bonzini; +Cc: kvm, linux-kernel, Sean Christopherson, David Woodhouse

Add wrappers to do hypercalls using VMCALL/VMMCALL and Xen's register ABI
(as opposed to full Xen-style hypercalls through a hypervisor provided
page).  Using the common helpers dedups a pile of code, and uses the
native hypercall instruction when running on AMD.

Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 .../selftests/kvm/include/x86_64/processor.h  |  2 +
 .../selftests/kvm/lib/x86_64/processor.c      | 10 +++
 .../selftests/kvm/x86_64/xen_shinfo_test.c    | 63 +++----------------
 3 files changed, 21 insertions(+), 54 deletions(-)

diff --git a/tools/testing/selftests/kvm/include/x86_64/processor.h b/tools/testing/selftests/kvm/include/x86_64/processor.h
index 53ffa43c90db..90387ddcb2a9 100644
--- a/tools/testing/selftests/kvm/include/x86_64/processor.h
+++ b/tools/testing/selftests/kvm/include/x86_64/processor.h
@@ -1063,6 +1063,8 @@ uint64_t *vm_get_page_table_entry(struct kvm_vm *vm, uint64_t vaddr);
 
 uint64_t kvm_hypercall(uint64_t nr, uint64_t a0, uint64_t a1, uint64_t a2,
 		       uint64_t a3);
+uint64_t __xen_hypercall(uint64_t nr, uint64_t a0, void *a1);
+void xen_hypercall(uint64_t nr, uint64_t a0, void *a1);
 
 void __vm_xsave_require_permission(int bit, const char *name);
 
diff --git a/tools/testing/selftests/kvm/lib/x86_64/processor.c b/tools/testing/selftests/kvm/lib/x86_64/processor.c
index ff901cb47ffc..c39a4353ba19 100644
--- a/tools/testing/selftests/kvm/lib/x86_64/processor.c
+++ b/tools/testing/selftests/kvm/lib/x86_64/processor.c
@@ -1161,6 +1161,16 @@ uint64_t kvm_hypercall(uint64_t nr, uint64_t a0, uint64_t a1, uint64_t a2,
 	return X86_HYPERCALL("a"(nr), "b"(a0), "c"(a1), "d"(a2), "S"(a3));
 }
 
+uint64_t __xen_hypercall(uint64_t nr, uint64_t a0, void *a1)
+{
+	return X86_HYPERCALL("a"(nr), "D"(a0), "S"(a1));
+}
+
+void xen_hypercall(uint64_t nr, uint64_t a0, void *a1)
+{
+	GUEST_ASSERT(!__xen_hypercall(nr, a0, a1));
+}
+
 const struct kvm_cpuid2 *kvm_get_supported_hv_cpuid(void)
 {
 	static struct kvm_cpuid2 *cpuid;
diff --git a/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c b/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c
index 6a838df69174..1380947af1fe 100644
--- a/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c
+++ b/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c
@@ -228,15 +228,8 @@ static void guest_code(void)
 
 	/* Our turn. Deliver event channel (to ourselves) with
 	 * EVTCHNOP_send hypercall. */
-	unsigned long rax;
 	struct evtchn_send s = { .port = 127 };
-	__asm__ __volatile__ ("vmcall" :
-			      "=a" (rax) :
-			      "a" (__HYPERVISOR_event_channel_op),
-			      "D" (EVTCHNOP_send),
-			      "S" (&s));
-
-	GUEST_ASSERT(rax == 0);
+	xen_hypercall(__HYPERVISOR_event_channel_op, EVTCHNOP_send, &s);
 
 	guest_wait_for_irq();
 
@@ -245,24 +238,15 @@ static void guest_code(void)
 	/* Deliver "outbound" event channel to an eventfd which
 	 * happens to be one of our own irqfds. */
 	s.port = 197;
-	__asm__ __volatile__ ("vmcall" :
-			      "=a" (rax) :
-			      "a" (__HYPERVISOR_event_channel_op),
-			      "D" (EVTCHNOP_send),
-			      "S" (&s));
-
-	GUEST_ASSERT(rax == 0);
+	xen_hypercall(__HYPERVISOR_event_channel_op, EVTCHNOP_send, &s);
 
 	guest_wait_for_irq();
 
 	GUEST_SYNC(13);
 
 	/* Set a timer 100ms in the future. */
-	__asm__ __volatile__ ("vmcall" :
-			      "=a" (rax) :
-			      "a" (__HYPERVISOR_set_timer_op),
-			      "D" (rs->state_entry_time + 100000000));
-	GUEST_ASSERT(rax == 0);
+	xen_hypercall(__HYPERVISOR_set_timer_op,
+		      rs->state_entry_time + 100000000, NULL);
 
 	GUEST_SYNC(14);
 
@@ -284,37 +268,19 @@ static void guest_code(void)
 		.timeout = 0,
 	};
 
-	__asm__ __volatile__ ("vmcall" :
-			      "=a" (rax) :
-			      "a" (__HYPERVISOR_sched_op),
-			      "D" (SCHEDOP_poll),
-			      "S" (&p));
-
-	GUEST_ASSERT(rax == 0);
+	xen_hypercall(__HYPERVISOR_sched_op, SCHEDOP_poll, &p);
 
 	GUEST_SYNC(17);
 
 	/* Poll for an unset port and wait for the timeout. */
 	p.timeout = 100000000;
-	__asm__ __volatile__ ("vmcall" :
-			      "=a" (rax) :
-			      "a" (__HYPERVISOR_sched_op),
-			      "D" (SCHEDOP_poll),
-			      "S" (&p));
-
-	GUEST_ASSERT(rax == 0);
+	xen_hypercall(__HYPERVISOR_sched_op, SCHEDOP_poll, &p);
 
 	GUEST_SYNC(18);
 
 	/* A timer will wake the masked port we're waiting on, while we poll */
 	p.timeout = 0;
-	__asm__ __volatile__ ("vmcall" :
-			      "=a" (rax) :
-			      "a" (__HYPERVISOR_sched_op),
-			      "D" (SCHEDOP_poll),
-			      "S" (&p));
-
-	GUEST_ASSERT(rax == 0);
+	xen_hypercall(__HYPERVISOR_sched_op, SCHEDOP_poll, &p);
 
 	GUEST_SYNC(19);
 
@@ -322,13 +288,7 @@ static void guest_code(void)
 	 * actual interrupt, while we're polling on a different port. */
 	ports[0]++;
 	p.timeout = 0;
-	__asm__ __volatile__ ("vmcall" :
-			      "=a" (rax) :
-			      "a" (__HYPERVISOR_sched_op),
-			      "D" (SCHEDOP_poll),
-			      "S" (&p));
-
-	GUEST_ASSERT(rax == 0);
+	xen_hypercall(__HYPERVISOR_sched_op, SCHEDOP_poll, &p);
 
 	guest_wait_for_irq();
 
@@ -363,12 +323,7 @@ static void guest_code(void)
 	 * timer IRQ is dropped due to an invalid event channel.
 	 */
 	for (i = 0; i < 100 && !guest_saw_irq; i++)
-		asm volatile("vmcall"
-			     : "=a" (rax)
-			     : "a" (__HYPERVISOR_sched_op),
-			       "D" (SCHEDOP_poll),
-			       "S" (&p)
-			     : "memory");
+		__xen_hypercall(__HYPERVISOR_sched_op, SCHEDOP_poll, &p);
 
 	/*
 	 * Re-send the timer IRQ if it was (likely) dropped due to the timer
-- 
2.39.1.519.gcb327c4b5f-goog


^ permalink raw reply	[flat|nested] 6+ messages in thread

* [PATCH v2 3/4] KVM: selftests: Use enum for test numbers in xen_shinfo_test
  2023-02-04  2:41 [PATCH v2 0/4] KVM: selftests: xen_shinfo cleanups and slowpath test Sean Christopherson
  2023-02-04  2:41 ` [PATCH v2 1/4] KVM: selftests: Move the guts of kvm_hypercall() to a separate macro Sean Christopherson
  2023-02-04  2:41 ` [PATCH v2 2/4] KVM: selftests: Add helpers to make Xen-style VMCALL/VMMCALL hypercalls Sean Christopherson
@ 2023-02-04  2:41 ` Sean Christopherson
  2023-02-04  2:41 ` [PATCH v2 4/4] KVM: selftests: Add EVTCHNOP_send slow path test to xen_shinfo_test Sean Christopherson
  2023-03-14 13:43 ` [PATCH v2 0/4] KVM: selftests: xen_shinfo cleanups and slowpath test Paolo Bonzini
  4 siblings, 0 replies; 6+ messages in thread
From: Sean Christopherson @ 2023-02-04  2:41 UTC (permalink / raw)
  To: Paolo Bonzini; +Cc: kvm, linux-kernel, Sean Christopherson, David Woodhouse

From: David Woodhouse <dwmw@amazon.co.uk>

The xen_shinfo_test started off with very few iterations, and the numbers
we used in GUEST_SYNC() were precisely mapped to the RUNSTATE_xxx values
anyway to start with.

It has since grown quite a few more tests, and it's kind of awful to be
handling them all as bare numbers. Especially when I want to add a new
test in the middle. Define an enum for the test stages, and use it both
in the guest code and the host switch statement.

No functional change, if I can count to 24.

Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 .../selftests/kvm/x86_64/xen_shinfo_test.c    | 133 +++++++++++-------
 1 file changed, 82 insertions(+), 51 deletions(-)

diff --git a/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c b/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c
index 1380947af1fe..0459126365c9 100644
--- a/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c
+++ b/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c
@@ -44,6 +44,36 @@
 #define EVTCHN_TEST2 66
 #define EVTCHN_TIMER 13
 
+enum {
+	TEST_INJECT_VECTOR = 0,
+	TEST_RUNSTATE_runnable,
+	TEST_RUNSTATE_blocked,
+	TEST_RUNSTATE_offline,
+	TEST_RUNSTATE_ADJUST,
+	TEST_RUNSTATE_DATA,
+	TEST_STEAL_TIME,
+	TEST_EVTCHN_MASKED,
+	TEST_EVTCHN_UNMASKED,
+	TEST_EVTCHN_SLOWPATH,
+	TEST_EVTCHN_SEND_IOCTL,
+	TEST_EVTCHN_HCALL,
+	TEST_EVTCHN_HCALL_EVENTFD,
+	TEST_TIMER_SETUP,
+	TEST_TIMER_WAIT,
+	TEST_TIMER_RESTORE,
+	TEST_POLL_READY,
+	TEST_POLL_TIMEOUT,
+	TEST_POLL_MASKED,
+	TEST_POLL_WAKE,
+	TEST_TIMER_PAST,
+	TEST_LOCKING_SEND_RACE,
+	TEST_LOCKING_POLL_RACE,
+	TEST_LOCKING_POLL_TIMEOUT,
+	TEST_DONE,
+
+	TEST_GUEST_SAW_IRQ,
+};
+
 #define XEN_HYPERCALL_MSR	0x40000000
 
 #define MIN_STEAL_TIME		50000
@@ -147,7 +177,7 @@ static void evtchn_handler(struct ex_regs *regs)
 	vi->evtchn_pending_sel = 0;
 	guest_saw_irq = true;
 
-	GUEST_SYNC(0x20);
+	GUEST_SYNC(TEST_GUEST_SAW_IRQ);
 }
 
 static void guest_wait_for_irq(void)
@@ -168,41 +198,41 @@ static void guest_code(void)
 	);
 
 	/* Trigger an interrupt injection */
-	GUEST_SYNC(0);
+	GUEST_SYNC(TEST_INJECT_VECTOR);
 
 	guest_wait_for_irq();
 
 	/* Test having the host set runstates manually */
-	GUEST_SYNC(RUNSTATE_runnable);
+	GUEST_SYNC(TEST_RUNSTATE_runnable);
 	GUEST_ASSERT(rs->time[RUNSTATE_runnable] != 0);
 	GUEST_ASSERT(rs->state == 0);
 
-	GUEST_SYNC(RUNSTATE_blocked);
+	GUEST_SYNC(TEST_RUNSTATE_blocked);
 	GUEST_ASSERT(rs->time[RUNSTATE_blocked] != 0);
 	GUEST_ASSERT(rs->state == 0);
 
-	GUEST_SYNC(RUNSTATE_offline);
+	GUEST_SYNC(TEST_RUNSTATE_offline);
 	GUEST_ASSERT(rs->time[RUNSTATE_offline] != 0);
 	GUEST_ASSERT(rs->state == 0);
 
 	/* Test runstate time adjust */
-	GUEST_SYNC(4);
+	GUEST_SYNC(TEST_RUNSTATE_ADJUST);
 	GUEST_ASSERT(rs->time[RUNSTATE_blocked] == 0x5a);
 	GUEST_ASSERT(rs->time[RUNSTATE_offline] == 0x6b6b);
 
 	/* Test runstate time set */
-	GUEST_SYNC(5);
+	GUEST_SYNC(TEST_RUNSTATE_DATA);
 	GUEST_ASSERT(rs->state_entry_time >= 0x8000);
 	GUEST_ASSERT(rs->time[RUNSTATE_runnable] == 0);
 	GUEST_ASSERT(rs->time[RUNSTATE_blocked] == 0x6b6b);
 	GUEST_ASSERT(rs->time[RUNSTATE_offline] == 0x5a);
 
 	/* sched_yield() should result in some 'runnable' time */
-	GUEST_SYNC(6);
+	GUEST_SYNC(TEST_STEAL_TIME);
 	GUEST_ASSERT(rs->time[RUNSTATE_runnable] >= MIN_STEAL_TIME);
 
 	/* Attempt to deliver a *masked* interrupt */
-	GUEST_SYNC(7);
+	GUEST_SYNC(TEST_EVTCHN_MASKED);
 
 	/* Wait until we see the bit set */
 	struct shared_info *si = (void *)SHINFO_VADDR;
@@ -210,21 +240,21 @@ static void guest_code(void)
 		__asm__ __volatile__ ("rep nop" : : : "memory");
 
 	/* Now deliver an *unmasked* interrupt */
-	GUEST_SYNC(8);
+	GUEST_SYNC(TEST_EVTCHN_UNMASKED);
 
 	guest_wait_for_irq();
 
 	/* Change memslots and deliver an interrupt */
-	GUEST_SYNC(9);
+	GUEST_SYNC(TEST_EVTCHN_SLOWPATH);
 
 	guest_wait_for_irq();
 
 	/* Deliver event channel with KVM_XEN_HVM_EVTCHN_SEND */
-	GUEST_SYNC(10);
+	GUEST_SYNC(TEST_EVTCHN_SEND_IOCTL);
 
 	guest_wait_for_irq();
 
-	GUEST_SYNC(11);
+	GUEST_SYNC(TEST_EVTCHN_HCALL);
 
 	/* Our turn. Deliver event channel (to ourselves) with
 	 * EVTCHNOP_send hypercall. */
@@ -233,7 +263,7 @@ static void guest_code(void)
 
 	guest_wait_for_irq();
 
-	GUEST_SYNC(12);
+	GUEST_SYNC(TEST_EVTCHN_HCALL_EVENTFD);
 
 	/* Deliver "outbound" event channel to an eventfd which
 	 * happens to be one of our own irqfds. */
@@ -242,23 +272,23 @@ static void guest_code(void)
 
 	guest_wait_for_irq();
 
-	GUEST_SYNC(13);
+	GUEST_SYNC(TEST_TIMER_SETUP);
 
 	/* Set a timer 100ms in the future. */
 	xen_hypercall(__HYPERVISOR_set_timer_op,
 		      rs->state_entry_time + 100000000, NULL);
 
-	GUEST_SYNC(14);
+	GUEST_SYNC(TEST_TIMER_WAIT);
 
 	/* Now wait for the timer */
 	guest_wait_for_irq();
 
-	GUEST_SYNC(15);
+	GUEST_SYNC(TEST_TIMER_RESTORE);
 
 	/* The host has 'restored' the timer. Just wait for it. */
 	guest_wait_for_irq();
 
-	GUEST_SYNC(16);
+	GUEST_SYNC(TEST_POLL_READY);
 
 	/* Poll for an event channel port which is already set */
 	u32 ports[1] = { EVTCHN_TIMER };
@@ -270,19 +300,19 @@ static void guest_code(void)
 
 	xen_hypercall(__HYPERVISOR_sched_op, SCHEDOP_poll, &p);
 
-	GUEST_SYNC(17);
+	GUEST_SYNC(TEST_POLL_TIMEOUT);
 
 	/* Poll for an unset port and wait for the timeout. */
 	p.timeout = 100000000;
 	xen_hypercall(__HYPERVISOR_sched_op, SCHEDOP_poll, &p);
 
-	GUEST_SYNC(18);
+	GUEST_SYNC(TEST_POLL_MASKED);
 
 	/* A timer will wake the masked port we're waiting on, while we poll */
 	p.timeout = 0;
 	xen_hypercall(__HYPERVISOR_sched_op, SCHEDOP_poll, &p);
 
-	GUEST_SYNC(19);
+	GUEST_SYNC(TEST_POLL_WAKE);
 
 	/* A timer wake an *unmasked* port which should wake us with an
 	 * actual interrupt, while we're polling on a different port. */
@@ -292,17 +322,17 @@ static void guest_code(void)
 
 	guest_wait_for_irq();
 
-	GUEST_SYNC(20);
+	GUEST_SYNC(TEST_TIMER_PAST);
 
 	/* Timer should have fired already */
 	guest_wait_for_irq();
 
-	GUEST_SYNC(21);
+	GUEST_SYNC(TEST_LOCKING_SEND_RACE);
 	/* Racing host ioctls */
 
 	guest_wait_for_irq();
 
-	GUEST_SYNC(22);
+	GUEST_SYNC(TEST_LOCKING_POLL_RACE);
 	/* Racing vmcall against host ioctl */
 
 	ports[0] = 0;
@@ -330,12 +360,12 @@ static void guest_code(void)
 	 * expiring while the event channel was invalid.
 	 */
 	if (!guest_saw_irq) {
-		GUEST_SYNC(23);
+		GUEST_SYNC(TEST_LOCKING_POLL_TIMEOUT);
 		goto wait_for_timer;
 	}
 	guest_saw_irq = false;
 
-	GUEST_SYNC(24);
+	GUEST_SYNC(TEST_DONE);
 }
 
 static int cmp_timespec(struct timespec *a, struct timespec *b)
@@ -604,25 +634,26 @@ int main(int argc, char *argv[])
 					    "runstate times don't add up");
 
 			switch (uc.args[1]) {
-			case 0:
+			case TEST_INJECT_VECTOR:
 				if (verbose)
 					printf("Delivering evtchn upcall\n");
 				evtchn_irq_expected = true;
 				vinfo->evtchn_upcall_pending = 1;
 				break;
 
-			case RUNSTATE_runnable...RUNSTATE_offline:
+			case TEST_RUNSTATE_runnable...TEST_RUNSTATE_offline:
 				TEST_ASSERT(!evtchn_irq_expected, "Event channel IRQ not seen");
 				if (!do_runstate_tests)
 					goto done;
 				if (verbose)
 					printf("Testing runstate %s\n", runstate_names[uc.args[1]]);
 				rst.type = KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT;
-				rst.u.runstate.state = uc.args[1];
+				rst.u.runstate.state = uc.args[1] + RUNSTATE_runnable -
+					TEST_RUNSTATE_runnable;
 				vcpu_ioctl(vcpu, KVM_XEN_VCPU_SET_ATTR, &rst);
 				break;
 
-			case 4:
+			case TEST_RUNSTATE_ADJUST:
 				if (verbose)
 					printf("Testing RUNSTATE_ADJUST\n");
 				rst.type = KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST;
@@ -637,7 +668,7 @@ int main(int argc, char *argv[])
 				vcpu_ioctl(vcpu, KVM_XEN_VCPU_SET_ATTR, &rst);
 				break;
 
-			case 5:
+			case TEST_RUNSTATE_DATA:
 				if (verbose)
 					printf("Testing RUNSTATE_DATA\n");
 				rst.type = KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA;
@@ -649,7 +680,7 @@ int main(int argc, char *argv[])
 				vcpu_ioctl(vcpu, KVM_XEN_VCPU_SET_ATTR, &rst);
 				break;
 
-			case 6:
+			case TEST_STEAL_TIME:
 				if (verbose)
 					printf("Testing steal time\n");
 				/* Yield until scheduler delay exceeds target */
@@ -659,7 +690,7 @@ int main(int argc, char *argv[])
 				} while (get_run_delay() < rundelay);
 				break;
 
-			case 7:
+			case TEST_EVTCHN_MASKED:
 				if (!do_eventfd_tests)
 					goto done;
 				if (verbose)
@@ -669,7 +700,7 @@ int main(int argc, char *argv[])
 				alarm(1);
 				break;
 
-			case 8:
+			case TEST_EVTCHN_UNMASKED:
 				if (verbose)
 					printf("Testing unmasked event channel\n");
 				/* Unmask that, but deliver the other one */
@@ -680,7 +711,7 @@ int main(int argc, char *argv[])
 				alarm(1);
 				break;
 
-			case 9:
+			case TEST_EVTCHN_SLOWPATH:
 				TEST_ASSERT(!evtchn_irq_expected,
 					    "Expected event channel IRQ but it didn't happen");
 				shinfo->evtchn_pending[1] = 0;
@@ -693,7 +724,7 @@ int main(int argc, char *argv[])
 				alarm(1);
 				break;
 
-			case 10:
+			case TEST_EVTCHN_SEND_IOCTL:
 				TEST_ASSERT(!evtchn_irq_expected,
 					    "Expected event channel IRQ but it didn't happen");
 				if (!do_evtchn_tests)
@@ -713,7 +744,7 @@ int main(int argc, char *argv[])
 				alarm(1);
 				break;
 
-			case 11:
+			case TEST_EVTCHN_HCALL:
 				TEST_ASSERT(!evtchn_irq_expected,
 					    "Expected event channel IRQ but it didn't happen");
 				shinfo->evtchn_pending[1] = 0;
@@ -724,7 +755,7 @@ int main(int argc, char *argv[])
 				alarm(1);
 				break;
 
-			case 12:
+			case TEST_EVTCHN_HCALL_EVENTFD:
 				TEST_ASSERT(!evtchn_irq_expected,
 					    "Expected event channel IRQ but it didn't happen");
 				shinfo->evtchn_pending[0] = 0;
@@ -735,7 +766,7 @@ int main(int argc, char *argv[])
 				alarm(1);
 				break;
 
-			case 13:
+			case TEST_TIMER_SETUP:
 				TEST_ASSERT(!evtchn_irq_expected,
 					    "Expected event channel IRQ but it didn't happen");
 				shinfo->evtchn_pending[1] = 0;
@@ -744,7 +775,7 @@ int main(int argc, char *argv[])
 					printf("Testing guest oneshot timer\n");
 				break;
 
-			case 14:
+			case TEST_TIMER_WAIT:
 				memset(&tmr, 0, sizeof(tmr));
 				tmr.type = KVM_XEN_VCPU_ATTR_TYPE_TIMER;
 				vcpu_ioctl(vcpu, KVM_XEN_VCPU_GET_ATTR, &tmr);
@@ -758,7 +789,7 @@ int main(int argc, char *argv[])
 				alarm(1);
 				break;
 
-			case 15:
+			case TEST_TIMER_RESTORE:
 				TEST_ASSERT(!evtchn_irq_expected,
 					    "Expected event channel IRQ but it didn't happen");
 				shinfo->evtchn_pending[0] = 0;
@@ -772,7 +803,7 @@ int main(int argc, char *argv[])
 				alarm(1);
 				break;
 
-			case 16:
+			case TEST_POLL_READY:
 				TEST_ASSERT(!evtchn_irq_expected,
 					    "Expected event channel IRQ but it didn't happen");
 
@@ -782,14 +813,14 @@ int main(int argc, char *argv[])
 				alarm(1);
 				break;
 
-			case 17:
+			case TEST_POLL_TIMEOUT:
 				if (verbose)
 					printf("Testing SCHEDOP_poll timeout\n");
 				shinfo->evtchn_pending[0] = 0;
 				alarm(1);
 				break;
 
-			case 18:
+			case TEST_POLL_MASKED:
 				if (verbose)
 					printf("Testing SCHEDOP_poll wake on masked event\n");
 
@@ -798,7 +829,7 @@ int main(int argc, char *argv[])
 				alarm(1);
 				break;
 
-			case 19:
+			case TEST_POLL_WAKE:
 				shinfo->evtchn_pending[0] = shinfo->evtchn_mask[0] = 0;
 				if (verbose)
 					printf("Testing SCHEDOP_poll wake on unmasked event\n");
@@ -815,7 +846,7 @@ int main(int argc, char *argv[])
 				alarm(1);
 				break;
 
-			case 20:
+			case TEST_TIMER_PAST:
 				TEST_ASSERT(!evtchn_irq_expected,
 					    "Expected event channel IRQ but it didn't happen");
 				/* Read timer and check it is no longer pending */
@@ -832,7 +863,7 @@ int main(int argc, char *argv[])
 				alarm(1);
 				break;
 
-			case 21:
+			case TEST_LOCKING_SEND_RACE:
 				TEST_ASSERT(!evtchn_irq_expected,
 					    "Expected event channel IRQ but it didn't happen");
 				alarm(0);
@@ -854,7 +885,7 @@ int main(int argc, char *argv[])
 					__vm_ioctl(vm, KVM_XEN_HVM_EVTCHN_SEND, &uxe);
 				break;
 
-			case 22:
+			case TEST_LOCKING_POLL_RACE:
 				TEST_ASSERT(!evtchn_irq_expected,
 					    "Expected event channel IRQ but it didn't happen");
 
@@ -869,7 +900,7 @@ int main(int argc, char *argv[])
 				vcpu_ioctl(vcpu, KVM_XEN_VCPU_SET_ATTR, &tmr);
 				break;
 
-			case 23:
+			case TEST_LOCKING_POLL_TIMEOUT:
 				/*
 				 * Optional and possibly repeated sync point.
 				 * Injecting the timer IRQ may fail if the
@@ -891,7 +922,7 @@ int main(int argc, char *argv[])
 							 SHINFO_RACE_TIMEOUT * 1000000000ULL;
 				vcpu_ioctl(vcpu, KVM_XEN_VCPU_SET_ATTR, &tmr);
 				break;
-			case 24:
+			case TEST_DONE:
 				TEST_ASSERT(!evtchn_irq_expected,
 					    "Expected event channel IRQ but it didn't happen");
 
@@ -902,7 +933,7 @@ int main(int argc, char *argv[])
 				TEST_ASSERT(ret == 0, "pthread_join() failed: %s", strerror(ret));
 				goto done;
 
-			case 0x20:
+			case TEST_GUEST_SAW_IRQ:
 				TEST_ASSERT(evtchn_irq_expected, "Unexpected event channel IRQ");
 				evtchn_irq_expected = false;
 				break;
-- 
2.39.1.519.gcb327c4b5f-goog


^ permalink raw reply	[flat|nested] 6+ messages in thread

* [PATCH v2 4/4] KVM: selftests: Add EVTCHNOP_send slow path test to xen_shinfo_test
  2023-02-04  2:41 [PATCH v2 0/4] KVM: selftests: xen_shinfo cleanups and slowpath test Sean Christopherson
                   ` (2 preceding siblings ...)
  2023-02-04  2:41 ` [PATCH v2 3/4] KVM: selftests: Use enum for test numbers in xen_shinfo_test Sean Christopherson
@ 2023-02-04  2:41 ` Sean Christopherson
  2023-03-14 13:43 ` [PATCH v2 0/4] KVM: selftests: xen_shinfo cleanups and slowpath test Paolo Bonzini
  4 siblings, 0 replies; 6+ messages in thread
From: Sean Christopherson @ 2023-02-04  2:41 UTC (permalink / raw)
  To: Paolo Bonzini; +Cc: kvm, linux-kernel, Sean Christopherson, David Woodhouse

From: David Woodhouse <dwmw@amazon.co.uk>

When kvm_xen_evtchn_send() takes the slow path because the shinfo GPC
needs to be revalidated, it used to violate the SRCU vs. kvm->lock
locking rules and potentially cause a deadlock.

Now that lockdep is learning to catch such things, make sure that code
path is exercised by the selftest.

Link: https://lore.kernel.org/all/20230113124606.10221-2-dwmw2@infradead.org
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 .../selftests/kvm/x86_64/xen_shinfo_test.c    | 27 +++++++++++++++++++
 1 file changed, 27 insertions(+)

diff --git a/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c b/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c
index 0459126365c9..f8baef107a1b 100644
--- a/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c
+++ b/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c
@@ -29,6 +29,9 @@
 #define DUMMY_REGION_GPA	(SHINFO_REGION_GPA + (3 * PAGE_SIZE))
 #define DUMMY_REGION_SLOT	11
 
+#define DUMMY_REGION_GPA_2	(SHINFO_REGION_GPA + (4 * PAGE_SIZE))
+#define DUMMY_REGION_SLOT_2	12
+
 #define SHINFO_ADDR	(SHINFO_REGION_GPA)
 #define VCPU_INFO_ADDR	(SHINFO_REGION_GPA + 0x40)
 #define PVTIME_ADDR	(SHINFO_REGION_GPA + PAGE_SIZE)
@@ -57,6 +60,7 @@ enum {
 	TEST_EVTCHN_SLOWPATH,
 	TEST_EVTCHN_SEND_IOCTL,
 	TEST_EVTCHN_HCALL,
+	TEST_EVTCHN_HCALL_SLOWPATH,
 	TEST_EVTCHN_HCALL_EVENTFD,
 	TEST_TIMER_SETUP,
 	TEST_TIMER_WAIT,
@@ -263,6 +267,16 @@ static void guest_code(void)
 
 	guest_wait_for_irq();
 
+	GUEST_SYNC(TEST_EVTCHN_HCALL_SLOWPATH);
+
+	/*
+	 * Same again, but this time the host has messed with memslots so it
+	 * should take the slow path in kvm_xen_set_evtchn().
+	 */
+	xen_hypercall(__HYPERVISOR_event_channel_op, EVTCHNOP_send, &s);
+
+	guest_wait_for_irq();
+
 	GUEST_SYNC(TEST_EVTCHN_HCALL_EVENTFD);
 
 	/* Deliver "outbound" event channel to an eventfd which
@@ -755,6 +769,19 @@ int main(int argc, char *argv[])
 				alarm(1);
 				break;
 
+			case TEST_EVTCHN_HCALL_SLOWPATH:
+				TEST_ASSERT(!evtchn_irq_expected,
+					    "Expected event channel IRQ but it didn't happen");
+				shinfo->evtchn_pending[0] = 0;
+
+				if (verbose)
+					printf("Testing guest EVTCHNOP_send direct to evtchn after memslot change\n");
+				vm_userspace_mem_region_add(vm, VM_MEM_SRC_ANONYMOUS,
+							    DUMMY_REGION_GPA_2, DUMMY_REGION_SLOT_2, 1, 0);
+				evtchn_irq_expected = true;
+				alarm(1);
+				break;
+
 			case TEST_EVTCHN_HCALL_EVENTFD:
 				TEST_ASSERT(!evtchn_irq_expected,
 					    "Expected event channel IRQ but it didn't happen");
-- 
2.39.1.519.gcb327c4b5f-goog


^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH v2 0/4] KVM: selftests: xen_shinfo cleanups and slowpath test
  2023-02-04  2:41 [PATCH v2 0/4] KVM: selftests: xen_shinfo cleanups and slowpath test Sean Christopherson
                   ` (3 preceding siblings ...)
  2023-02-04  2:41 ` [PATCH v2 4/4] KVM: selftests: Add EVTCHNOP_send slow path test to xen_shinfo_test Sean Christopherson
@ 2023-03-14 13:43 ` Paolo Bonzini
  4 siblings, 0 replies; 6+ messages in thread
From: Paolo Bonzini @ 2023-03-14 13:43 UTC (permalink / raw)
  To: Sean Christopherson; +Cc: kvm, linux-kernel, David Woodhouse

Queued, thanks.

Paolo



^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2023-03-14 13:47 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-02-04  2:41 [PATCH v2 0/4] KVM: selftests: xen_shinfo cleanups and slowpath test Sean Christopherson
2023-02-04  2:41 ` [PATCH v2 1/4] KVM: selftests: Move the guts of kvm_hypercall() to a separate macro Sean Christopherson
2023-02-04  2:41 ` [PATCH v2 2/4] KVM: selftests: Add helpers to make Xen-style VMCALL/VMMCALL hypercalls Sean Christopherson
2023-02-04  2:41 ` [PATCH v2 3/4] KVM: selftests: Use enum for test numbers in xen_shinfo_test Sean Christopherson
2023-02-04  2:41 ` [PATCH v2 4/4] KVM: selftests: Add EVTCHNOP_send slow path test to xen_shinfo_test Sean Christopherson
2023-03-14 13:43 ` [PATCH v2 0/4] KVM: selftests: xen_shinfo cleanups and slowpath test Paolo Bonzini

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®