[PATCH v2] x86/paravirt: Don't make vcpu_is_preempted() a callee-save function

Subsystems: kvm paravirt (kvm/paravirt), paravirt_ops interface, the rest, x86 architecture (32-bit and 64-bit), xen hypervisor x86

STALE3470d

13 messages, 4 authors, 2017-02-13 · open the first message on its own page

[PATCH v2] x86/paravirt: Don't make vcpu_is_preempted() a callee-save function

From: Waiman Long <longman@redhat.com>
Date: 2017-02-10 15:56:09

It was found when running fio sequential write test with a XFS ramdisk
on a VM running on a 2-socket x86-64 system, the %CPU times as reported
by perf were as follows:

 69.75%  0.59%  fio  [k] down_write
 69.15%  0.01%  fio  [k] call_rwsem_down_write_failed
 67.12%  1.12%  fio  [k] rwsem_down_write_failed
 63.48% 52.77%  fio  [k] osq_lock
  9.46%  7.88%  fio  [k] __raw_callee_save___kvm_vcpu_is_preempt
  3.93%  3.93%  fio  [k] __kvm_vcpu_is_preempted

Making vcpu_is_preempted() a callee-save function has a relatively
high cost on x86-64 primarily due to at least one more cacheline of
data access from the saving and restoring of registers (8 of them)
to and from stack as well as one more level of function call. As
vcpu_is_preempted() is called within the spinlock, mutex and rwsem
slowpaths, there isn't much to gain by making it callee-save. So it
is now changed to a normal function call instead.

With this patch applied on both bare-metal & KVM guest on a 2-socekt
16-core 32-thread system with 16 parallel jobs (8 on each socket), the
aggregrate bandwidth of the fio test on an XFS ramdisk were as follows:

                       Bare Metal                KVM Guest
   I/O Type      w/o patch    with patch   w/o patch    with patch
   --------      ---------    ----------   ---------    ----------
   random read   8650.5 MB/s  8560.9 MB/s  7602.9 MB/s  8196.1 MB/s  
   seq read      9104.8 MB/s  9397.2 MB/s  8293.7 MB/s  8566.9 MB/s
   random write  1623.8 MB/s  1626.7 MB/s  1590.6 MB/s  1700.7 MB/s
   seq write     1626.4 MB/s  1624.9 MB/s  1604.8 MB/s  1726.3 MB/s

The perf data (on KVM guest) now became:

 70.78%  0.58%  fio  [k] down_write
 70.20%  0.01%  fio  [k] call_rwsem_down_write_failed
 69.70%  1.17%  fio  [k] rwsem_down_write_failed
 59.91% 55.42%  fio  [k] osq_lock
 10.14% 10.14%  fio  [k] __kvm_vcpu_is_preempted

On bare metal, the patch doesn't introduce any performance
regression. On KVM guest, it produces noticeable performance
improvement (up to 7%).

Signed-off-by: Waiman Long <longman@redhat.com>
---
 v1->v2:
  - Rerun the fio test on a different system on both bare-metal and a
    KVM guest. Both sockets were utilized in this test.
  - The commit log was updated with new performance numbers, but the
    patch wasn't changed.
  - Drop patch 2.

 arch/x86/include/asm/paravirt.h       | 2 +-
 arch/x86/include/asm/paravirt_types.h | 2 +-
 arch/x86/kernel/kvm.c                 | 7 ++-----
 arch/x86/kernel/paravirt-spinlocks.c  | 6 ++----
 arch/x86/xen/spinlock.c               | 4 +---
 5 files changed, 7 insertions(+), 14 deletions(-)
diff --git a/arch/x86/include/asm/paravirt.h b/arch/x86/include/asm/paravirt.h
index 864f57b..2515885 100644
--- a/arch/x86/include/asm/paravirt.h
+++ b/arch/x86/include/asm/paravirt.h
@@ -676,7 +676,7 @@ static __always_inline void pv_kick(int cpu)
 
 static __always_inline bool pv_vcpu_is_preempted(int cpu)
 {
-	return PVOP_CALLEE1(bool, pv_lock_ops.vcpu_is_preempted, cpu);
+	return PVOP_CALL1(bool, pv_lock_ops.vcpu_is_preempted, cpu);
 }
 
 #endif /* SMP && PARAVIRT_SPINLOCKS */
diff --git a/arch/x86/include/asm/paravirt_types.h b/arch/x86/include/asm/paravirt_types.h
index bb2de45..88dc852 100644
--- a/arch/x86/include/asm/paravirt_types.h
+++ b/arch/x86/include/asm/paravirt_types.h
@@ -309,7 +309,7 @@ struct pv_lock_ops {
 	void (*wait)(u8 *ptr, u8 val);
 	void (*kick)(int cpu);
 
-	struct paravirt_callee_save vcpu_is_preempted;
+	bool (*vcpu_is_preempted)(int cpu);
 };
 
 /* This contains all the paravirt structures: we get a convenient
diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
index 099fcba..eb3753d 100644
--- a/arch/x86/kernel/kvm.c
+++ b/arch/x86/kernel/kvm.c
@@ -595,7 +595,6 @@ __visible bool __kvm_vcpu_is_preempted(int cpu)
 
 	return !!src->preempted;
 }
-PV_CALLEE_SAVE_REGS_THUNK(__kvm_vcpu_is_preempted);
 
 /*
  * Setup pv_lock_ops to exploit KVM_FEATURE_PV_UNHALT if present.
@@ -614,10 +613,8 @@ void __init kvm_spinlock_init(void)
 	pv_lock_ops.wait = kvm_wait;
 	pv_lock_ops.kick = kvm_kick_cpu;
 
-	if (kvm_para_has_feature(KVM_FEATURE_STEAL_TIME)) {
-		pv_lock_ops.vcpu_is_preempted =
-			PV_CALLEE_SAVE(__kvm_vcpu_is_preempted);
-	}
+	if (kvm_para_has_feature(KVM_FEATURE_STEAL_TIME))
+		pv_lock_ops.vcpu_is_preempted = __kvm_vcpu_is_preempted;
 }
 
 #endif	/* CONFIG_PARAVIRT_SPINLOCKS */
diff --git a/arch/x86/kernel/paravirt-spinlocks.c b/arch/x86/kernel/paravirt-spinlocks.c
index 6259327..da050bc 100644
--- a/arch/x86/kernel/paravirt-spinlocks.c
+++ b/arch/x86/kernel/paravirt-spinlocks.c
@@ -24,12 +24,10 @@ __visible bool __native_vcpu_is_preempted(int cpu)
 {
 	return false;
 }
-PV_CALLEE_SAVE_REGS_THUNK(__native_vcpu_is_preempted);
 
 bool pv_is_native_vcpu_is_preempted(void)
 {
-	return pv_lock_ops.vcpu_is_preempted.func ==
-		__raw_callee_save___native_vcpu_is_preempted;
+	return pv_lock_ops.vcpu_is_preempted == __native_vcpu_is_preempted;
 }
 
 struct pv_lock_ops pv_lock_ops = {
@@ -38,7 +36,7 @@ struct pv_lock_ops pv_lock_ops = {
 	.queued_spin_unlock = PV_CALLEE_SAVE(__native_queued_spin_unlock),
 	.wait = paravirt_nop,
 	.kick = paravirt_nop,
-	.vcpu_is_preempted = PV_CALLEE_SAVE(__native_vcpu_is_preempted),
+	.vcpu_is_preempted = __native_vcpu_is_preempted,
 #endif /* SMP */
 };
 EXPORT_SYMBOL(pv_lock_ops);
diff --git a/arch/x86/xen/spinlock.c b/arch/x86/xen/spinlock.c
index 25a7c43..c85bb8f 100644
--- a/arch/x86/xen/spinlock.c
+++ b/arch/x86/xen/spinlock.c
@@ -114,8 +114,6 @@ void xen_uninit_lock_cpu(int cpu)
 	per_cpu(irq_name, cpu) = NULL;
 }
 
-PV_CALLEE_SAVE_REGS_THUNK(xen_vcpu_stolen);
-
 /*
  * Our init of PV spinlocks is split in two init functions due to us
  * using paravirt patching and jump labels patching and having to do
@@ -138,7 +136,7 @@ void __init xen_init_spinlocks(void)
 	pv_lock_ops.queued_spin_unlock = PV_CALLEE_SAVE(__pv_queued_spin_unlock);
 	pv_lock_ops.wait = xen_qlock_wait;
 	pv_lock_ops.kick = xen_qlock_kick;
-	pv_lock_ops.vcpu_is_preempted = PV_CALLEE_SAVE(xen_vcpu_stolen);
+	pv_lock_ops.vcpu_is_preempted = xen_vcpu_stolen;
 }
 
 static __init int xen_parse_nopvspin(char *arg)
-- 
1.8.3.1

Re: [PATCH v2] x86/paravirt: Don't make vcpu_is_preempted() a callee-save function

From: Peter Zijlstra <peterz@infradead.org>
Date: 2017-02-10 16:20:40

On Fri, Feb 10, 2017 at 10:43:09AM -0500, Waiman Long wrote:
It was found when running fio sequential write test with a XFS ramdisk
on a VM running on a 2-socket x86-64 system, the %CPU times as reported
by perf were as follows:

 69.75%  0.59%  fio  [k] down_write
 69.15%  0.01%  fio  [k] call_rwsem_down_write_failed
 67.12%  1.12%  fio  [k] rwsem_down_write_failed
 63.48% 52.77%  fio  [k] osq_lock
  9.46%  7.88%  fio  [k] __raw_callee_save___kvm_vcpu_is_preempt
  3.93%  3.93%  fio  [k] __kvm_vcpu_is_preempted
Thinking about this again, wouldn't something like the below also work?

diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
index 099fcba4981d..6aa33702c15c 100644
--- a/arch/x86/kernel/kvm.c
+++ b/arch/x86/kernel/kvm.c
@@ -589,6 +589,7 @@ static void kvm_wait(u8 *ptr, u8 val)
 	local_irq_restore(flags);
 }
 
+#ifdef CONFIG_X86_32
 __visible bool __kvm_vcpu_is_preempted(int cpu)
 {
 	struct kvm_steal_time *src = &per_cpu(steal_time, cpu);
@@ -597,6 +598,31 @@ __visible bool __kvm_vcpu_is_preempted(int cpu)
 }
 PV_CALLEE_SAVE_REGS_THUNK(__kvm_vcpu_is_preempted);
 
+#else
+
+extern bool __raw_callee_save___kvm_vcpu_is_preempted(int);
+
+asm(
+".pushsection .text;"
+".global __raw_callee_save___kvm_vcpu_is_preempted;"
+".type __raw_callee_save___kvm_vcpu_is_preempted, @function;"
+"__raw_callee_save___kvm_vcpu_is_preempted:"
+FRAME_BEGIN
+"push %rdi;"
+"push %rdx;"
+"movslq  %edi, %rdi;"
+"movq    $steal_time+16, %rax;"
+"movq    __per_cpu_offset(,%rdi,8), %rdx;"
+"cmpb    $0, (%rdx,%rax);"
+"setne   %al;"
+"pop %rdx;"
+"pop %rdi;"
+FRAME_END
+"ret;"
+".popsection");
+
+#endif
+
 /*
  * Setup pv_lock_ops to exploit KVM_FEATURE_PV_UNHALT if present.
  */

Re: [PATCH v2] x86/paravirt: Don't make vcpu_is_preempted() a callee-save function

From: Paolo Bonzini <pbonzini@redhat.com>
Date: 2017-02-10 16:23:05


On 10/02/2017 16:43, Waiman Long wrote:
quoted hunk
It was found when running fio sequential write test with a XFS ramdisk
on a VM running on a 2-socket x86-64 system, the %CPU times as reported
by perf were as follows:

 69.75%  0.59%  fio  [k] down_write
 69.15%  0.01%  fio  [k] call_rwsem_down_write_failed
 67.12%  1.12%  fio  [k] rwsem_down_write_failed
 63.48% 52.77%  fio  [k] osq_lock
  9.46%  7.88%  fio  [k] __raw_callee_save___kvm_vcpu_is_preempt
  3.93%  3.93%  fio  [k] __kvm_vcpu_is_preempted

Making vcpu_is_preempted() a callee-save function has a relatively
high cost on x86-64 primarily due to at least one more cacheline of
data access from the saving and restoring of registers (8 of them)
to and from stack as well as one more level of function call. As
vcpu_is_preempted() is called within the spinlock, mutex and rwsem
slowpaths, there isn't much to gain by making it callee-save. So it
is now changed to a normal function call instead.

With this patch applied on both bare-metal & KVM guest on a 2-socekt
16-core 32-thread system with 16 parallel jobs (8 on each socket), the
aggregrate bandwidth of the fio test on an XFS ramdisk were as follows:

                       Bare Metal                KVM Guest
   I/O Type      w/o patch    with patch   w/o patch    with patch
   --------      ---------    ----------   ---------    ----------
   random read   8650.5 MB/s  8560.9 MB/s  7602.9 MB/s  8196.1 MB/s  
   seq read      9104.8 MB/s  9397.2 MB/s  8293.7 MB/s  8566.9 MB/s
   random write  1623.8 MB/s  1626.7 MB/s  1590.6 MB/s  1700.7 MB/s
   seq write     1626.4 MB/s  1624.9 MB/s  1604.8 MB/s  1726.3 MB/s

The perf data (on KVM guest) now became:

 70.78%  0.58%  fio  [k] down_write
 70.20%  0.01%  fio  [k] call_rwsem_down_write_failed
 69.70%  1.17%  fio  [k] rwsem_down_write_failed
 59.91% 55.42%  fio  [k] osq_lock
 10.14% 10.14%  fio  [k] __kvm_vcpu_is_preempted

On bare metal, the patch doesn't introduce any performance
regression. On KVM guest, it produces noticeable performance
improvement (up to 7%).

Signed-off-by: Waiman Long <longman@redhat.com>
---
 v1->v2:
  - Rerun the fio test on a different system on both bare-metal and a
    KVM guest. Both sockets were utilized in this test.
  - The commit log was updated with new performance numbers, but the
    patch wasn't changed.
  - Drop patch 2.

 arch/x86/include/asm/paravirt.h       | 2 +-
 arch/x86/include/asm/paravirt_types.h | 2 +-
 arch/x86/kernel/kvm.c                 | 7 ++-----
 arch/x86/kernel/paravirt-spinlocks.c  | 6 ++----
 arch/x86/xen/spinlock.c               | 4 +---
 5 files changed, 7 insertions(+), 14 deletions(-)
diff --git a/arch/x86/include/asm/paravirt.h b/arch/x86/include/asm/paravirt.h
index 864f57b..2515885 100644
--- a/arch/x86/include/asm/paravirt.h
+++ b/arch/x86/include/asm/paravirt.h
@@ -676,7 +676,7 @@ static __always_inline void pv_kick(int cpu)
 
 static __always_inline bool pv_vcpu_is_preempted(int cpu)
 {
-	return PVOP_CALLEE1(bool, pv_lock_ops.vcpu_is_preempted, cpu);
+	return PVOP_CALL1(bool, pv_lock_ops.vcpu_is_preempted, cpu);
 }
 
 #endif /* SMP && PARAVIRT_SPINLOCKS */
diff --git a/arch/x86/include/asm/paravirt_types.h b/arch/x86/include/asm/paravirt_types.h
index bb2de45..88dc852 100644
--- a/arch/x86/include/asm/paravirt_types.h
+++ b/arch/x86/include/asm/paravirt_types.h
@@ -309,7 +309,7 @@ struct pv_lock_ops {
 	void (*wait)(u8 *ptr, u8 val);
 	void (*kick)(int cpu);
 
-	struct paravirt_callee_save vcpu_is_preempted;
+	bool (*vcpu_is_preempted)(int cpu);
 };
 
 /* This contains all the paravirt structures: we get a convenient
diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
index 099fcba..eb3753d 100644
--- a/arch/x86/kernel/kvm.c
+++ b/arch/x86/kernel/kvm.c
@@ -595,7 +595,6 @@ __visible bool __kvm_vcpu_is_preempted(int cpu)
 
 	return !!src->preempted;
 }
-PV_CALLEE_SAVE_REGS_THUNK(__kvm_vcpu_is_preempted);
 
 /*
  * Setup pv_lock_ops to exploit KVM_FEATURE_PV_UNHALT if present.
@@ -614,10 +613,8 @@ void __init kvm_spinlock_init(void)
 	pv_lock_ops.wait = kvm_wait;
 	pv_lock_ops.kick = kvm_kick_cpu;
 
-	if (kvm_para_has_feature(KVM_FEATURE_STEAL_TIME)) {
-		pv_lock_ops.vcpu_is_preempted =
-			PV_CALLEE_SAVE(__kvm_vcpu_is_preempted);
-	}
+	if (kvm_para_has_feature(KVM_FEATURE_STEAL_TIME))
+		pv_lock_ops.vcpu_is_preempted = __kvm_vcpu_is_preempted;
 }
 
 #endif	/* CONFIG_PARAVIRT_SPINLOCKS */
diff --git a/arch/x86/kernel/paravirt-spinlocks.c b/arch/x86/kernel/paravirt-spinlocks.c
index 6259327..da050bc 100644
--- a/arch/x86/kernel/paravirt-spinlocks.c
+++ b/arch/x86/kernel/paravirt-spinlocks.c
@@ -24,12 +24,10 @@ __visible bool __native_vcpu_is_preempted(int cpu)
 {
 	return false;
 }
-PV_CALLEE_SAVE_REGS_THUNK(__native_vcpu_is_preempted);
 
 bool pv_is_native_vcpu_is_preempted(void)
 {
-	return pv_lock_ops.vcpu_is_preempted.func ==
-		__raw_callee_save___native_vcpu_is_preempted;
+	return pv_lock_ops.vcpu_is_preempted == __native_vcpu_is_preempted;
 }
 
 struct pv_lock_ops pv_lock_ops = {
@@ -38,7 +36,7 @@ struct pv_lock_ops pv_lock_ops = {
 	.queued_spin_unlock = PV_CALLEE_SAVE(__native_queued_spin_unlock),
 	.wait = paravirt_nop,
 	.kick = paravirt_nop,
-	.vcpu_is_preempted = PV_CALLEE_SAVE(__native_vcpu_is_preempted),
+	.vcpu_is_preempted = __native_vcpu_is_preempted,
 #endif /* SMP */
 };
 EXPORT_SYMBOL(pv_lock_ops);
diff --git a/arch/x86/xen/spinlock.c b/arch/x86/xen/spinlock.c
index 25a7c43..c85bb8f 100644
--- a/arch/x86/xen/spinlock.c
+++ b/arch/x86/xen/spinlock.c
@@ -114,8 +114,6 @@ void xen_uninit_lock_cpu(int cpu)
 	per_cpu(irq_name, cpu) = NULL;
 }
 
-PV_CALLEE_SAVE_REGS_THUNK(xen_vcpu_stolen);
-
 /*
  * Our init of PV spinlocks is split in two init functions due to us
  * using paravirt patching and jump labels patching and having to do
@@ -138,7 +136,7 @@ void __init xen_init_spinlocks(void)
 	pv_lock_ops.queued_spin_unlock = PV_CALLEE_SAVE(__pv_queued_spin_unlock);
 	pv_lock_ops.wait = xen_qlock_wait;
 	pv_lock_ops.kick = xen_qlock_kick;
-	pv_lock_ops.vcpu_is_preempted = PV_CALLEE_SAVE(xen_vcpu_stolen);
+	pv_lock_ops.vcpu_is_preempted = xen_vcpu_stolen;
 }
 
 static __init int xen_parse_nopvspin(char *arg)
Acked-by: Paolo Bonzini <pbonzini@redhat.com>

Thank you very much!

Paolo

Re: [PATCH v2] x86/paravirt: Don't make vcpu_is_preempted() a callee-save function

From: Waiman Long <longman@redhat.com>
Date: 2017-02-10 17:00:49

On 02/10/2017 11:35 AM, Waiman Long wrote:
On 02/10/2017 11:19 AM, Peter Zijlstra wrote:
quoted
On Fri, Feb 10, 2017 at 10:43:09AM -0500, Waiman Long wrote:
quoted
It was found when running fio sequential write test with a XFS ramdisk
on a VM running on a 2-socket x86-64 system, the %CPU times as reported
by perf were as follows:

 69.75%  0.59%  fio  [k] down_write
 69.15%  0.01%  fio  [k] call_rwsem_down_write_failed
 67.12%  1.12%  fio  [k] rwsem_down_write_failed
 63.48% 52.77%  fio  [k] osq_lock
  9.46%  7.88%  fio  [k] __raw_callee_save___kvm_vcpu_is_preempt
  3.93%  3.93%  fio  [k] __kvm_vcpu_is_preempted
Thinking about this again, wouldn't something like the below also work?

diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
index 099fcba4981d..6aa33702c15c 100644
--- a/arch/x86/kernel/kvm.c
+++ b/arch/x86/kernel/kvm.c
@@ -589,6 +589,7 @@ static void kvm_wait(u8 *ptr, u8 val)
 	local_irq_restore(flags);
 }
 
+#ifdef CONFIG_X86_32
 __visible bool __kvm_vcpu_is_preempted(int cpu)
 {
 	struct kvm_steal_time *src = &per_cpu(steal_time, cpu);
@@ -597,6 +598,31 @@ __visible bool __kvm_vcpu_is_preempted(int cpu)
 }
 PV_CALLEE_SAVE_REGS_THUNK(__kvm_vcpu_is_preempted);
 
+#else
+
+extern bool __raw_callee_save___kvm_vcpu_is_preempted(int);
+
+asm(
+".pushsection .text;"
+".global __raw_callee_save___kvm_vcpu_is_preempted;"
+".type __raw_callee_save___kvm_vcpu_is_preempted, @function;"
+"__raw_callee_save___kvm_vcpu_is_preempted:"
+FRAME_BEGIN
+"push %rdi;"
+"push %rdx;"
+"movslq  %edi, %rdi;"
+"movq    $steal_time+16, %rax;"
+"movq    __per_cpu_offset(,%rdi,8), %rdx;"
+"cmpb    $0, (%rdx,%rax);"
+"setne   %al;"
+"pop %rdx;"
+"pop %rdi;"
+FRAME_END
+"ret;"
+".popsection");
+
+#endif
+
 /*
  * Setup pv_lock_ops to exploit KVM_FEATURE_PV_UNHALT if present.
  */
That should work for now. I have done something similar for
__pv_queued_spin_unlock. However, this has the problem of creating a
dependency on the exact layout of the steal_time structure. Maybe the
constant 16 can be passed in as a parameter offsetof(struct
kvm_steal_time, preempted) to the asm call.

Cheers,
Longman
One more thing, that will improve KVM performance, but it won't help Xen.

I looked into the assembly code for rwsem_spin_on_owner, It need to save
and restore 2 additional registers with my patch. Doing it your way,
will transfer the save and restore overhead to the assembly code.
However, __kvm_vcpu_is_preempted() is called multiple times per
invocation of rwsem_spin_on_owner. That function is simple enough that
making __kvm_vcpu_is_preempted() callee-save won't produce much compiler
optimization opportunity. The outer function rwsem_down_write_failed()
does appear to be a bit bigger (from 866 bytes to 884 bytes) though.

Cheers,
Longman

Re: [PATCH v2] x86/paravirt: Don't make vcpu_is_preempted() a callee-save function

From: Waiman Long <longman@redhat.com>
Date: 2017-02-10 18:06:51

On 02/10/2017 11:19 AM, Peter Zijlstra wrote:
quoted hunk
On Fri, Feb 10, 2017 at 10:43:09AM -0500, Waiman Long wrote:
quoted
It was found when running fio sequential write test with a XFS ramdisk
on a VM running on a 2-socket x86-64 system, the %CPU times as reported
by perf were as follows:

 69.75%  0.59%  fio  [k] down_write
 69.15%  0.01%  fio  [k] call_rwsem_down_write_failed
 67.12%  1.12%  fio  [k] rwsem_down_write_failed
 63.48% 52.77%  fio  [k] osq_lock
  9.46%  7.88%  fio  [k] __raw_callee_save___kvm_vcpu_is_preempt
  3.93%  3.93%  fio  [k] __kvm_vcpu_is_preempted
Thinking about this again, wouldn't something like the below also work?

diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
index 099fcba4981d..6aa33702c15c 100644
--- a/arch/x86/kernel/kvm.c
+++ b/arch/x86/kernel/kvm.c
@@ -589,6 +589,7 @@ static void kvm_wait(u8 *ptr, u8 val)
 	local_irq_restore(flags);
 }
 
+#ifdef CONFIG_X86_32
 __visible bool __kvm_vcpu_is_preempted(int cpu)
 {
 	struct kvm_steal_time *src = &per_cpu(steal_time, cpu);
@@ -597,6 +598,31 @@ __visible bool __kvm_vcpu_is_preempted(int cpu)
 }
 PV_CALLEE_SAVE_REGS_THUNK(__kvm_vcpu_is_preempted);
 
+#else
+
+extern bool __raw_callee_save___kvm_vcpu_is_preempted(int);
+
+asm(
+".pushsection .text;"
+".global __raw_callee_save___kvm_vcpu_is_preempted;"
+".type __raw_callee_save___kvm_vcpu_is_preempted, @function;"
+"__raw_callee_save___kvm_vcpu_is_preempted:"
+FRAME_BEGIN
+"push %rdi;"
+"push %rdx;"
+"movslq  %edi, %rdi;"
+"movq    $steal_time+16, %rax;"
+"movq    __per_cpu_offset(,%rdi,8), %rdx;"
+"cmpb    $0, (%rdx,%rax);"
+"setne   %al;"
+"pop %rdx;"
+"pop %rdi;"
+FRAME_END
+"ret;"
+".popsection");
+
+#endif
+
 /*
  * Setup pv_lock_ops to exploit KVM_FEATURE_PV_UNHALT if present.
  */
That should work for now. I have done something similar for
__pv_queued_spin_unlock. However, this has the problem of creating a
dependency on the exact layout of the steal_time structure. Maybe the
constant 16 can be passed in as a parameter offsetof(struct
kvm_steal_time, preempted) to the asm call.

Cheers,
Longman

Re: [PATCH v2] x86/paravirt: Don't make vcpu_is_preempted() a callee-save function

From: Peter Zijlstra <peterz@infradead.org>
Date: 2017-02-13 10:48:40

On Fri, Feb 10, 2017 at 12:00:43PM -0500, Waiman Long wrote:
quoted
quoted
+asm(
+".pushsection .text;"
+".global __raw_callee_save___kvm_vcpu_is_preempted;"
+".type __raw_callee_save___kvm_vcpu_is_preempted, @function;"
+"__raw_callee_save___kvm_vcpu_is_preempted:"
+FRAME_BEGIN
+"push %rdi;"
+"push %rdx;"
+"movslq  %edi, %rdi;"
+"movq    $steal_time+16, %rax;"
+"movq    __per_cpu_offset(,%rdi,8), %rdx;"
+"cmpb    $0, (%rdx,%rax);"
Could we not put the $steal_time+16 displacement as an immediate in the
cmpb and save a whole register here?

That way we'd end up with something like:

asm("
push %rdi;
movslq %edi, %rdi;
movq __per_cpu_offset(,%rdi,8), %rax;
cmpb $0, %[offset](%rax);
setne %al;
pop %rdi;
" : : [offset] "i" (((unsigned long)&steal_time) + offsetof(struct steal_time, preempted)));

And if we could get rid of the sign extend on edi we could avoid all the
push-pop nonsense, but I'm not sure I see how to do that (then again,
this asm foo isn't my strongest point).
quoted
quoted
+"setne   %al;"
+"pop %rdx;"
+"pop %rdi;"
+FRAME_END
+"ret;"
+".popsection");
+
+#endif
+
 /*
  * Setup pv_lock_ops to exploit KVM_FEATURE_PV_UNHALT if present.
  */
That should work for now. I have done something similar for
__pv_queued_spin_unlock. However, this has the problem of creating a
dependency on the exact layout of the steal_time structure. Maybe the
constant 16 can be passed in as a parameter offsetof(struct
kvm_steal_time, preempted) to the asm call.
Yeah it should be well possible to pass that in. But ideally we'd have
GCC grow something like __attribute__((callee_saved)) or somesuch and it
would do all this for us.
One more thing, that will improve KVM performance, but it won't help Xen.
People still use Xen? ;-) In any case, their implementation looks very
similar and could easily crib this.
I looked into the assembly code for rwsem_spin_on_owner, It need to save
and restore 2 additional registers with my patch. Doing it your way,
will transfer the save and restore overhead to the assembly code.
However, __kvm_vcpu_is_preempted() is called multiple times per
invocation of rwsem_spin_on_owner. That function is simple enough that
making __kvm_vcpu_is_preempted() callee-save won't produce much compiler
optimization opportunity.
This is because of that noinline, right? Otherwise it would've been
folded and register pressure would be much higher.
The outer function rwsem_down_write_failed()
does appear to be a bit bigger (from 866 bytes to 884 bytes) though.
I suspect GCC is being clever and since all this is static it plays
games with the calling convention and pushes these clobbers out.

Re: [PATCH v2] x86/paravirt: Don't make vcpu_is_preempted() a callee-save function

From: Peter Zijlstra <peterz@infradead.org>
Date: 2017-02-13 10:54:09

On Mon, Feb 13, 2017 at 11:47:16AM +0100, Peter Zijlstra wrote:
That way we'd end up with something like:

asm("
push %rdi;
movslq %edi, %rdi;
movq __per_cpu_offset(,%rdi,8), %rax;
cmpb $0, %[offset](%rax);
setne %al;
pop %rdi;
" : : [offset] "i" (((unsigned long)&steal_time) + offsetof(struct steal_time, preempted)));

And if we could get rid of the sign extend on edi we could avoid all the
push-pop nonsense, but I'm not sure I see how to do that (then again,
this asm foo isn't my strongest point).
Maybe:

movsql %edi, %rax;
movq __per_cpu_offset(,%rax,8), %rax;
cmpb $0, %[offset](%rax);
setne %al;

?

Re: [PATCH v2] x86/paravirt: Don't make vcpu_is_preempted() a callee-save function

From: Waiman Long <longman@redhat.com>
Date: 2017-02-13 19:41:06

On 02/13/2017 05:47 AM, Peter Zijlstra wrote:
On Fri, Feb 10, 2017 at 12:00:43PM -0500, Waiman Long wrote:
quoted
quoted
quoted
+asm(
+".pushsection .text;"
+".global __raw_callee_save___kvm_vcpu_is_preempted;"
+".type __raw_callee_save___kvm_vcpu_is_preempted, @function;"
+"__raw_callee_save___kvm_vcpu_is_preempted:"
+FRAME_BEGIN
+"push %rdi;"
+"push %rdx;"
+"movslq  %edi, %rdi;"
+"movq    $steal_time+16, %rax;"
+"movq    __per_cpu_offset(,%rdi,8), %rdx;"
+"cmpb    $0, (%rdx,%rax);"
Could we not put the $steal_time+16 displacement as an immediate in the
cmpb and save a whole register here?

That way we'd end up with something like:

asm("
push %rdi;
movslq %edi, %rdi;
movq __per_cpu_offset(,%rdi,8), %rax;
cmpb $0, %[offset](%rax);
setne %al;
pop %rdi;
" : : [offset] "i" (((unsigned long)&steal_time) + offsetof(struct steal_time, preempted)));

And if we could get rid of the sign extend on edi we could avoid all the
push-pop nonsense, but I'm not sure I see how to do that (then again,
this asm foo isn't my strongest point).
Yes, I think that can work. I will try to ran this patch to see how
thing goes.
quoted
quoted
quoted
+"setne   %al;"
+"pop %rdx;"
+"pop %rdi;"
+FRAME_END
+"ret;"
+".popsection");
+
+#endif
+
 /*
  * Setup pv_lock_ops to exploit KVM_FEATURE_PV_UNHALT if present.
  */
That should work for now. I have done something similar for
__pv_queued_spin_unlock. However, this has the problem of creating a
dependency on the exact layout of the steal_time structure. Maybe the
constant 16 can be passed in as a parameter offsetof(struct
kvm_steal_time, preempted) to the asm call.
Yeah it should be well possible to pass that in. But ideally we'd have
GCC grow something like __attribute__((callee_saved)) or somesuch and it
would do all this for us.
That will be really nice too. I am not too fond of working in assembly.
quoted
One more thing, that will improve KVM performance, but it won't help Xen.
People still use Xen? ;-) In any case, their implementation looks very
similar and could easily crib this.
In Red Hat, my focus will be on KVM performance. I do believe that there
are still Xen users out there. So we still need to keep their interest
into consideration. Given that, I am OK to make it work better in KVM
first and then think about Xen later.
quoted
I looked into the assembly code for rwsem_spin_on_owner, It need to save
and restore 2 additional registers with my patch. Doing it your way,
will transfer the save and restore overhead to the assembly code.
However, __kvm_vcpu_is_preempted() is called multiple times per
invocation of rwsem_spin_on_owner. That function is simple enough that
making __kvm_vcpu_is_preempted() callee-save won't produce much compiler
optimization opportunity.
This is because of that noinline, right? Otherwise it would've been
folded and register pressure would be much higher.
Yes, I guess so. The noinline is there so that we know what the CPU time
is for spinning rather than other activities within the slowpath.
quoted
The outer function rwsem_down_write_failed()
does appear to be a bit bigger (from 866 bytes to 884 bytes) though.
I suspect GCC is being clever and since all this is static it plays
games with the calling convention and pushes these clobbers out.
Cheers,
Longman

Re: [PATCH v2] x86/paravirt: Don't make vcpu_is_preempted() a callee-save function

From: Waiman Long <longman@redhat.com>
Date: 2017-02-13 19:42:18

On 02/13/2017 05:53 AM, Peter Zijlstra wrote:
On Mon, Feb 13, 2017 at 11:47:16AM +0100, Peter Zijlstra wrote:
quoted
That way we'd end up with something like:

asm("
push %rdi;
movslq %edi, %rdi;
movq __per_cpu_offset(,%rdi,8), %rax;
cmpb $0, %[offset](%rax);
setne %al;
pop %rdi;
" : : [offset] "i" (((unsigned long)&steal_time) + offsetof(struct steal_time, preempted)));

And if we could get rid of the sign extend on edi we could avoid all the
push-pop nonsense, but I'm not sure I see how to do that (then again,
this asm foo isn't my strongest point).
Maybe:

movsql %edi, %rax;
movq __per_cpu_offset(,%rax,8), %rax;
cmpb $0, %[offset](%rax);
setne %al;

?
Yes, that looks good to me.

Cheers,
Longman

Re: [PATCH v2] x86/paravirt: Don't make vcpu_is_preempted() a callee-save function

From: hpa@zytor.com
Date: 2017-02-13 20:13:09

On February 13, 2017 2:53:43 AM PST, Peter Zijlstra [off-list ref] wrote:
On Mon, Feb 13, 2017 at 11:47:16AM +0100, Peter Zijlstra wrote:
quoted
That way we'd end up with something like:

asm("
push %rdi;
movslq %edi, %rdi;
movq __per_cpu_offset(,%rdi,8), %rax;
cmpb $0, %[offset](%rax);
setne %al;
pop %rdi;
" : : [offset] "i" (((unsigned long)&steal_time) + offsetof(struct
steal_time, preempted)));
quoted
And if we could get rid of the sign extend on edi we could avoid all
the
quoted
push-pop nonsense, but I'm not sure I see how to do that (then again,
this asm foo isn't my strongest point).
Maybe:

movsql %edi, %rax;
movq __per_cpu_offset(,%rax,8), %rax;
cmpb $0, %[offset](%rax);
setne %al;

?
We could kill the zero or sign extend by changing the calling interface to pass an unsigned long instead of an int.  It is much more likely that a zero extend is free for the caller than a sign extend.
-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.

Re: [PATCH v2] x86/paravirt: Don't make vcpu_is_preempted() a callee-save function

From: Peter Zijlstra <peterz@infradead.org>
Date: 2017-02-13 21:59:44

On Mon, Feb 13, 2017 at 12:06:44PM -0800, hpa@zytor.com wrote:
quoted
Maybe:

movsql %edi, %rax;
movq __per_cpu_offset(,%rax,8), %rax;
cmpb $0, %[offset](%rax);
setne %al;

?
We could kill the zero or sign extend by changing the calling
interface to pass an unsigned long instead of an int.  It is much more
likely that a zero extend is free for the caller than a sign extend.
Right, Boris and me talked about that on IRC. I was wondering if the
argument was u32 if we could assume the top 32 bits are 0 and then use
rdi without prior movzx.

That would allow reducing the thing one more instruction.

Also, PVOP_CALL_ARG#() have an (unsigned long) cast in them that doesn't
make sense. That cast ends up resulting in the calling code doing
explicit sign or zero extends into the full 64bit register for no good
reason.

If one removes that cast things still compile, but I worry something
somehow relies on this weird behaviour and will come apart.

Re: [PATCH v2] x86/paravirt: Don't make vcpu_is_preempted() a callee-save function

From: Waiman Long <longman@redhat.com>
Date: 2017-02-13 22:24:41

On 02/13/2017 03:06 PM, hpa@zytor.com wrote:
On February 13, 2017 2:53:43 AM PST, Peter Zijlstra [off-list ref] wrote:
quoted
On Mon, Feb 13, 2017 at 11:47:16AM +0100, Peter Zijlstra wrote:
quoted
That way we'd end up with something like:

asm("
push %rdi;
movslq %edi, %rdi;
movq __per_cpu_offset(,%rdi,8), %rax;
cmpb $0, %[offset](%rax);
setne %al;
pop %rdi;
" : : [offset] "i" (((unsigned long)&steal_time) + offsetof(struct
steal_time, preempted)));
quoted
And if we could get rid of the sign extend on edi we could avoid all
the
quoted
push-pop nonsense, but I'm not sure I see how to do that (then again,
this asm foo isn't my strongest point).
Maybe:

movsql %edi, %rax;
movq __per_cpu_offset(,%rax,8), %rax;
cmpb $0, %[offset](%rax);
setne %al;

?
We could kill the zero or sign extend by changing the calling interface to pass an unsigned long instead of an int.  It is much more likely that a zero extend is free for the caller than a sign extend.
I have thought of that too. However, the goal is to eliminate memory
read/write from/to stack. Eliminating a register sign-extend instruction
won't help much in term of performance.

Cheers,
Longman

Re: [PATCH v2] x86/paravirt: Don't make vcpu_is_preempted() a callee-save function

From: Peter Zijlstra <peterz@infradead.org>
Date: 2017-02-13 22:31:50

On Mon, Feb 13, 2017 at 05:24:36PM -0500, Waiman Long wrote:
quoted
quoted
movsql %edi, %rax;
movq __per_cpu_offset(,%rax,8), %rax;
cmpb $0, %[offset](%rax);
setne %al;
I have thought of that too. However, the goal is to eliminate memory
read/write from/to stack. Eliminating a register sign-extend instruction
won't help much in term of performance.
Problem here is that all instructions have dependencies, so if you can
get rid of the sign extend mov you kill a bunch of stall cycles (I would
expect).

But yes, peanuts vs the stack load/stores.
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help