Skip to contentSkip to conversation

Xen on Hyper-V: Implement L0 assisted TLB flush

PATCH Superseded Wei Liu 4 participants 10 messages Feb 19, 2020 – Mar 10, 2020

v4 of 5 Superseded Latest is v5 →

lore
Share & tools
lore
author
#1 lore raw

PATCH v4 1/3 x86/hypervisor: pass flags to hypervisor_flush_tlb

In reply to Wei Liu

Hyper-V's L0 assisted flush has fine-grained control over what gets flushed. We need all the flags available to make the best decisions possible.

No functional change because Xen's implementation doesn't care about what is passed to it.

---
v2:
1. Introduce FLUSH_TLB_FLAGS_MASK

---
 xen/arch/x86/guest/hypervisor.c        |  7 +++++--
 xen/arch/x86/guest/xen/xen.c           |  2 +-
 xen/arch/x86/smp.c                     |  5 ++---
 xen/include/asm-x86/flushtlb.h         |  3 +++
 xen/include/asm-x86/guest/hypervisor.h | 10 +++++-----
 5 files changed, 16 insertions(+), 11 deletions(-)
xen/arch/x86/guest/hypervisor.c +5 -2
index 47e938e287..6ee28c9df1 100644
@@ -75,10 +75,13 @@ void __init hypervisor_e820_fixup(struct e820map *e820)
 }
 
 int hypervisor_flush_tlb(const cpumask_t *mask, const void *va,
-                         unsigned int order)
+                         unsigned int flags)
 {
+    if ( flags & ~FLUSH_TLB_FLAGS_MASK )
+        return -EINVAL;
+
     if ( ops.flush_tlb )
-        return alternative_call(ops.flush_tlb, mask, va, order);
+        return alternative_call(ops.flush_tlb, mask, va, flags);
 
     return -ENOSYS;
 }
xen/arch/x86/guest/xen/xen.c +1 -1
index 5d3427a713..0eb1115c4d 100644
@@ -324,7 +324,7 @@ static void __init e820_fixup(struct e820map *e820)
         pv_shim_fixup_e820(e820);
 }
 
-static int flush_tlb(const cpumask_t *mask, const void *va, unsigned int order)
+static int flush_tlb(const cpumask_t *mask, const void *va, unsigned int flags)
 {
     return xen_hypercall_hvm_op(HVMOP_flush_tlbs, NULL);
 }
xen/arch/x86/smp.c +2 -3
index c7caf5bc26..4dab74c0d5 100644
@@ -258,9 +258,8 @@ void flush_area_mask(const cpumask_t *mask, const void *va, unsigned int flags)
          !cpumask_subset(mask, cpumask_of(cpu)) )
     {
         if ( cpu_has_hypervisor &&
-             !(flags & ~(FLUSH_TLB | FLUSH_TLB_GLOBAL | FLUSH_VA_VALID |
-                         FLUSH_ORDER_MASK)) &&
-             !hypervisor_flush_tlb(mask, va, flags & FLUSH_ORDER_MASK) )
+             !(flags & ~FLUSH_TLB_FLAGS_MASK) &&
+             !hypervisor_flush_tlb(mask, va, flags) )
         {
             if ( tlb_clk_enabled )
                 tlb_clk_enabled = false;
xen/include/asm-x86/flushtlb.h +3 -0
index 9773014320..a4de317452 100644
@@ -123,6 +123,9 @@ void switch_cr3_cr4(unsigned long cr3, unsigned long cr4);
  /* Flush all HVM guests linear TLB (using ASID/VPID) */
 #define FLUSH_GUESTS_TLB 0x4000
 
+#define FLUSH_TLB_FLAGS_MASK (FLUSH_TLB | FLUSH_TLB_GLOBAL | FLUSH_VA_VALID | \
+                              FLUSH_ORDER_MASK)
+
 /* Flush local TLBs/caches. */
 unsigned int flush_area_local(const void *va, unsigned int flags);
 #define flush_local(flags) flush_area_local(NULL, flags)
xen/include/asm-x86/guest/hypervisor.h +5 -5
index 432e57c2a0..48d54735d2 100644
@@ -35,7 +35,7 @@ struct hypervisor_ops {
     /* Fix up e820 map */
     void (*e820_fixup)(struct e820map *e820);
     /* L0 assisted TLB flush */
-    int (*flush_tlb)(const cpumask_t *mask, const void *va, unsigned int order);
+    int (*flush_tlb)(const cpumask_t *mask, const void *va, unsigned int flags);
 };
 
 #ifdef CONFIG_GUEST
@@ -48,11 +48,11 @@ void hypervisor_e820_fixup(struct e820map *e820);
 /*
  * L0 assisted TLB flush.
  * mask: cpumask of the dirty vCPUs that should be flushed.
- * va: linear address to flush, or NULL for global flushes.
- * order: order of the linear address pointed by va.
+ * va: linear address to flush, or NULL for entire address space.
+ * flags: flags for flushing, including the order of va.
  */
 int hypervisor_flush_tlb(const cpumask_t *mask, const void *va,
-                         unsigned int order);
+                         unsigned int flags);
 
 #else
 
@@ -65,7 +65,7 @@ static inline int hypervisor_ap_setup(void) { return 0; }
 static inline void hypervisor_resume(void) { ASSERT_UNREACHABLE(); }
 static inline void hypervisor_e820_fixup(struct e820map *e820) {}
 static inline int hypervisor_flush_tlb(const cpumask_t *mask, const void *va,
-                                       unsigned int order)
+                                       unsigned int flags)
 {
     return -ENOSYS;
 }
Signature · 7 lines
2.20.1


_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

Wei Liu

author
#0 lore raw

v4 cover Xen on Hyper-V: Implement L0 assisted TLB flush

Hi all

This seris is based on Roger's L0 assisted flush series.

I have done some testing against a Linux on Hyper-V in a 32-vcpu VM. All builds were done with -j32.

Building Xen on Linux:

real    0m45.376s
user    2m28.156s
sys     0m51.672s

Building Xen on Linux on Xen on Hyper-V, no assisted flush:

real    3m8.762s
user    10m46.787s
sys     30m14.492s

Building Xen on Linux on Xen on Hyper-V, with assisted flush:

real    0m44.369s
user    3m16.231s
sys     3m3.330s

Building Linux x86_64_defconfig on Linux:

real    0m59.698s
user    21m14.014s
sys     2m58.742s

Building Linux x86_64_defconfig on Linux on Xen on Hyper-V, no assisted flush:

real    2m6.284s
user    31m18.706s
sys     20m31.106s

Building Linux x86_64_defconfig on Linux on Xen on Hyper-V, with assisted flush:

real    1m38.968s
user    28m40.398s
sys     11m20.151s

There are various degrees of improvement depending on the workload. Xen can perhaps be optmised a bit more because it currently doesn't pass the address space id (cr3) to Hyper-V, but that requires reworking TLB flush APIs within Xen.

Wei.

Cc Jan Beulich <[email protected]>
Cc Andrew Cooper <[email protected]>
Cc Wei Liu <[email protected]>
Cc Roger Pau Monné <[email protected]>
Cc Michael Kelley <[email protected]>
Cc Paul Durrant <[email protected]>

Wei Liu (3):
x86/hypervisor: pass flags to hypervisor_flush_tlb x86/hyperv: skeleton for L0 assisted TLB flush x86/hyperv: L0 assisted TLB flush

 xen/arch/x86/guest/hyperv/Makefile     |   2 +
 xen/arch/x86/guest/hyperv/hyperv.c     |  17 ++
 xen/arch/x86/guest/hyperv/private.h    |  12 ++
 xen/arch/x86/guest/hyperv/tlb.c        | 214 +++++++++++++++++++++++++
 xen/arch/x86/guest/hyperv/util.c       |  75 +++++++++
 xen/arch/x86/guest/hypervisor.c        |   7 +-
 xen/arch/x86/guest/xen/xen.c           |   2 +-
 xen/arch/x86/smp.c                     |   5 +-
 xen/include/asm-x86/flushtlb.h         |   3 +
 xen/include/asm-x86/guest/hypervisor.h |  10 +-
 10 files changed, 336 insertions(+), 11 deletions(-)
 create mode 100644 xen/arch/x86/guest/hyperv/tlb.c
 create mode 100644 xen/arch/x86/guest/hyperv/util.c
Signature · 7 lines
2.20.1


_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
author
#6 lore raw

PATCH v4 2/3 x86/hyperv: skeleton for L0 assisted TLB flush

Implement a basic hook for L0 assisted TLB flush. The hook needs to check if prerequisites are met. If they are not met, it returns an error number to fall back to native flushes.

Introduce a new variable to indicate if hypercall page is ready.

---
v3:
1. Change hv_hcall_page_ready to hcall_page_ready

---
 xen/arch/x86/guest/hyperv/Makefile  |  1 +
 xen/arch/x86/guest/hyperv/hyperv.c  | 17 ++++++++++++
 xen/arch/x86/guest/hyperv/private.h |  4 +++
 xen/arch/x86/guest/hyperv/tlb.c     | 41 +++++++++++++++++++++++++++++
 4 files changed, 63 insertions(+)
 create mode 100644 xen/arch/x86/guest/hyperv/tlb.c
xen/arch/x86/guest/hyperv/Makefile +1 -0
index 68170109a9..18902c33e9 100644
@@ -1 +1,2 @@
 obj-y += hyperv.o
+obj-y += tlb.o
xen/arch/x86/guest/hyperv/hyperv.c +17 -0
index 70f4cd5ae0..f1b3073712 100644
@@ -33,6 +33,8 @@ DEFINE_PER_CPU_READ_MOSTLY(void *, hv_input_page);
 DEFINE_PER_CPU_READ_MOSTLY(void *, hv_vp_assist);
 DEFINE_PER_CPU_READ_MOSTLY(unsigned int, hv_vp_index);
 
+static bool __read_mostly hcall_page_ready;
+
 static uint64_t generate_guest_id(void)
 {
     union hv_guest_os_id id = {};
@@ -119,6 +121,8 @@ static void __init setup_hypercall_page(void)
     BUG_ON(!hypercall_msr.enable);
 
     set_fixmap_x(FIX_X_HYPERV_HCALL, mfn << PAGE_SHIFT);
+
+    hcall_page_ready = true;
 }
 
 static int setup_hypercall_pcpu_arg(void)
@@ -199,11 +203,24 @@ static void __init e820_fixup(struct e820map *e820)
         panic("Unable to reserve Hyper-V hypercall range\n");
 }
 
+static int flush_tlb(const cpumask_t *mask, const void *va,
+                     unsigned int flags)
+{
+    if ( !(ms_hyperv.hints & HV_X64_REMOTE_TLB_FLUSH_RECOMMENDED) )
+        return -EOPNOTSUPP;
+
+    if ( !hcall_page_ready || !this_cpu(hv_input_page) )
+        return -ENXIO;
+
+    return hyperv_flush_tlb(mask, va, flags);
+}
+
 static const struct hypervisor_ops __initdata ops = {
     .name = "Hyper-V",
     .setup = setup,
     .ap_setup = ap_setup,
     .e820_fixup = e820_fixup,
+    .flush_tlb = flush_tlb,
 };
 
 /*
xen/arch/x86/guest/hyperv/private.h +4 -0
index 956eff831f..509bedaafa 100644
@@ -22,10 +22,14 @@
 #ifndef __XEN_HYPERV_PRIVIATE_H__
 #define __XEN_HYPERV_PRIVIATE_H__
 
+#include <xen/cpumask.h>
 #include <xen/percpu.h>
 
 DECLARE_PER_CPU(void *, hv_input_page);
 DECLARE_PER_CPU(void *, hv_vp_assist);
 DECLARE_PER_CPU(unsigned int, hv_vp_index);
 
+int hyperv_flush_tlb(const cpumask_t *mask, const void *va,
+                     unsigned int flags);
+
 #endif /* __XEN_HYPERV_PRIVIATE_H__  */
xen/arch/x86/guest/hyperv/tlb.c new file +41 -0

41 lines not on this page · raw message

Signature · 7 lines
2.20.1


_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
author
#7 lore raw

PATCH v4 3/3 x86/hyperv: L0 assisted TLB flush

In reply to Wei Liu

Implement L0 assisted TLB flush for Xen on Hyper-V. It takes advantage of several hypercalls:

* HVCALL_FLUSH_VIRTUAL_ADDRESS_LIST
* HVCALL_FLUSH_VIRTUAL_ADDRESS_LIST_EX
* HVCALL_FLUSH_VIRTUAL_ADDRESS_SPACE
* HVCALL_FLUSH_VIRTUAL_ADDRESS_SPACE_EX

Pick the most efficient hypercalls available.

---
v4:
1. Fix bank mask generation.
2. Fix page order calculation.
3. Remove types.h from private.h.
4. Add a note about nmi and mc handling.

v3:
1. Address more comments.
2. Fix usage of max_vp_index.
3. Use the fill_gva_list algorithm from Linux.

v2:
1. Address Roger and Jan's comments re types etc.
2. Fix pointer arithmetic.
3. Misc improvement to code.

---
 xen/arch/x86/guest/hyperv/Makefile  |   1 +
 xen/arch/x86/guest/hyperv/private.h |   8 ++
 xen/arch/x86/guest/hyperv/tlb.c     | 175 +++++++++++++++++++++++++++-
 xen/arch/x86/guest/hyperv/util.c    |  75 ++++++++++++
 4 files changed, 258 insertions(+), 1 deletion(-)
 create mode 100644 xen/arch/x86/guest/hyperv/util.c
xen/arch/x86/guest/hyperv/Makefile +1 -0
index 18902c33e9..0e39410968 100644
@@ -1,2 +1,3 @@
 obj-y += hyperv.o
 obj-y += tlb.o
+obj-y += util.o
xen/arch/x86/guest/hyperv/private.h +8 -0
index 509bedaafa..354fc7f685 100644
@@ -29,7 +29,15 @@ DECLARE_PER_CPU(void *, hv_input_page);
 DECLARE_PER_CPU(void *, hv_vp_assist);
 DECLARE_PER_CPU(unsigned int, hv_vp_index);
 
+static inline unsigned int hv_vp_index(unsigned int cpu)
+{
+    return per_cpu(hv_vp_index, cpu);
+}
+
 int hyperv_flush_tlb(const cpumask_t *mask, const void *va,
                      unsigned int flags);
 
+/* Returns number of banks, -ev if error */
+int cpumask_to_vpset(struct hv_vpset *vpset, const cpumask_t *mask);
+
 #endif /* __XEN_HYPERV_PRIVIATE_H__  */
xen/arch/x86/guest/hyperv/tlb.c +174 -1

191 lines not on this page · raw message

xen/arch/x86/guest/hyperv/util.c new file +75 -0

75 lines not on this page · raw message

Signature · 7 lines
2.20.1


_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Reviewed-by on v4 3/3
#8 lore raw

Re: [Xen-devel] [PATCH v4 3/3] x86/hyperv: L0 assisted TLB flush

Quoted 11 lines · Wei Liu

Implement L0 assisted TLB flush for Xen on Hyper-V. It takes advantage of several hypercalls:

* HVCALL_FLUSH_VIRTUAL_ADDRESS_LIST
* HVCALL_FLUSH_VIRTUAL_ADDRESS_LIST_EX
* HVCALL_FLUSH_VIRTUAL_ADDRESS_SPACE
* HVCALL_FLUSH_VIRTUAL_ADDRESS_SPACE_EX

Pick the most efficient hypercalls available.

LGTM:

Quoted 22 lines
xen/arch/x86/guest/hyperv/util.c new file +22 -0
+int cpumask_to_vpset(struct hv_vpset *vpset,
+                     const cpumask_t *mask)
+{
+    int nr = 1;
+    unsigned int cpu, vcpu_bank, vcpu_offset;
+    unsigned int max_banks = ms_hyperv.max_vp_index / 64;
+
+    /* Up to 64 banks can be represented by valid_bank_mask */
+    if ( max_banks > 64 )
+        return -E2BIG;
+
+    /* Clear all banks to avoid flushing unwanted CPUs */
+    for ( vcpu_bank = 0; vcpu_bank < max_banks; vcpu_bank++ )
+        vpset->bank_contents[vcpu_bank] = 0;
+
+    vpset->format = HV_GENERIC_SET_SPARSE_4K;
+
+    for_each_cpu ( cpu, mask )
+    {
+        unsigned int vcpu = hv_vp_index(cpu);
+
+        vcpu_bank = vcpu / 64;

xen/arch/x86/guest/hyperv/util.c line 62

Quoted 10 lines
xen/arch/x86/guest/hyperv/util.c new file +10 -0
+        vcpu_offset = vcpu % 64;
+
+        __set_bit(vcpu_offset, &vpset->bank_contents[vcpu_bank]);
+
+        if ( vcpu_bank >= nr )
+            nr = vcpu_bank + 1;
+    }
+
+    /* Some banks may be empty but that's ok */
+    vpset->valid_bank_mask = ~0ULL >> (64 - nr);

You could also use the recently introduced bitmap_set here.

Thanks, Roger.

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

Reviewed-by Paul Durrant on v4 3/3
#9 lore raw

Re: [Xen-devel] [PATCH v4 3/3] x86/hyperv: L0 assisted TLB flush

In reply to Wei Liu
Quoted 13 lines

-----Original Message-----
From: Wei Liu <[email protected]> On Behalf Of Wei Liu
Sent: 19 February 2020 11:44
To: Xen Development List <[email protected]>
Cc: Michael Kelley <[email protected]>; Durrant, Paul <[email protected]>; Wei Liu <[email protected]>; Wei Liu <[email protected]>; Jan Beulich <[email protected]>; Andrew Cooper <[email protected]>; Roger Pau Monné <[email protected]>
Subject: [PATCH v4 3/3] x86/hyperv: L0 assisted TLB flush

Implement L0 assisted TLB flush for Xen on Hyper-V. It takes advantage of several hypercalls:

Quoted 8 lines

* HVCALL_FLUSH_VIRTUAL_ADDRESS_LIST
* HVCALL_FLUSH_VIRTUAL_ADDRESS_LIST_EX
* HVCALL_FLUSH_VIRTUAL_ADDRESS_SPACE
* HVCALL_FLUSH_VIRTUAL_ADDRESS_SPACE_EX

Pick the most efficient hypercalls available.

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#2 lore raw

Re: [Xen-devel] [PATCH v4 1/3] x86/hypervisor: pass flags to hypervisor_flush_tlb

In reply to Wei Liu

xen/include/asm-x86/flushtlb.h void switch_cr3_cr4(unsigned long cr3, unsigned long cr4); · line 127

Quoted 8 lines · Wei Liu
xen/include/asm-x86/flushtlb.h +2 -0
@@ -123,6 +123,9 @@ void switch_cr3_cr4(unsigned long cr3, unsigned long cr4);
  /* Flush all HVM guests linear TLB (using ASID/VPID) */
 #define FLUSH_GUESTS_TLB 0x4000
 
+#define FLUSH_TLB_FLAGS_MASK (FLUSH_TLB | FLUSH_TLB_GLOBAL | FLUSH_VA_VALID | \
+                              FLUSH_ORDER_MASK)

I don't think FLUSH_ORDER_MASK should be part of this, as it's not exclusively TLB-flush related.

Jan

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

author
#3 lore raw

Re: [Xen-devel] [PATCH v4 1/3] x86/hypervisor: pass flags to hypervisor_flush_tlb

Quoted 7 lines · Jan Beulich
Quoted 6 lines · Wei Liu
xen/include/asm-x86/flushtlb.h +0 -0
@@ -123,6 +123,9 @@ void switch_cr3_cr4(unsigned long cr3, unsigned long cr4);
  /* Flush all HVM guests linear TLB (using ASID/VPID) */
 #define FLUSH_GUESTS_TLB 0x4000
 
Quoted 5 lines

xen/include/asm-x86/flushtlb.h void switch_cr3_cr4(unsigned long cr3, unsigned long cr4); · line 127

Quoted 2 lines
xen/include/asm-x86/flushtlb.h +2 -0
+#define FLUSH_TLB_FLAGS_MASK (FLUSH_TLB | FLUSH_TLB_GLOBAL | FLUSH_VA_VALID | \
+                              FLUSH_ORDER_MASK)

I don't think FLUSH_ORDER_MASK should be part of this, as it's not exclusively TLB-flush related.

My intention was to capture all flags pertinent to TLB flushes. I didn't mean they were exclusively TLB flush related.

I can remove the order mask from the list of flags. Not a big deal.

Wei.

Quoted 2 lines

Jan

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#4 lore raw

Re: [Xen-devel] [PATCH v4 1/3] x86/hypervisor: pass flags to hypervisor_flush_tlb

Quoted 8 lines · Wei Liu
Quoted 7 lines · Jan Beulich
Quoted 6 lines · Wei Liu
xen/include/asm-x86/flushtlb.h +0 -0
@@ -123,6 +123,9 @@ void switch_cr3_cr4(unsigned long cr3, unsigned long cr4);
  /* Flush all HVM guests linear TLB (using ASID/VPID) */
 #define FLUSH_GUESTS_TLB 0x4000
 
Quoted 8 lines
Quoted 5 lines

xen/include/asm-x86/flushtlb.h void switch_cr3_cr4(unsigned long cr3, unsigned long cr4); · line 127

Quoted 2 lines
xen/include/asm-x86/flushtlb.h +2 -0
+#define FLUSH_TLB_FLAGS_MASK (FLUSH_TLB | FLUSH_TLB_GLOBAL | FLUSH_VA_VALID | \
+                              FLUSH_ORDER_MASK)

I don't think FLUSH_ORDER_MASK should be part of this, as it's not exclusively TLB-flush related.

My intention was to capture all flags pertinent to TLB flushes. I didn't mean they were exclusively TLB flush related.

That's what I understood, but the result is leading to possible ambiguity. Thinking about it again, ...

Quoted 1 line

I can remove the order mask from the list of flags. Not a big deal.

... I'm afraid I also have to ask for FLUSH_VA_VALID to be dropped, as that one's not TLB flush related at all. Instead it identifies whether the linear address provided _still has_ a valid mapping (rather than there being an address provided at all), such that CLFLUSH bases cache flushing would be usable. At that point I wonder whether the separate constant is still of much use at all.

Jan

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

author
#5 lore raw

Re: [Xen-devel] [PATCH v4 1/3] x86/hypervisor: pass flags to hypervisor_flush_tlb

Quoted 21 lines · Jan Beulich
Quoted 8 lines · Wei Liu
Quoted 7 lines · Jan Beulich
Quoted 6 lines · Wei Liu
xen/include/asm-x86/flushtlb.h +0 -0
@@ -123,6 +123,9 @@ void switch_cr3_cr4(unsigned long cr3, unsigned long cr4);
  /* Flush all HVM guests linear TLB (using ASID/VPID) */
 #define FLUSH_GUESTS_TLB 0x4000
 
Quoted 8 lines
Quoted 5 lines

xen/include/asm-x86/flushtlb.h void switch_cr3_cr4(unsigned long cr3, unsigned long cr4); · line 127

Quoted 2 lines
xen/include/asm-x86/flushtlb.h +2 -0
+#define FLUSH_TLB_FLAGS_MASK (FLUSH_TLB | FLUSH_TLB_GLOBAL | FLUSH_VA_VALID | \
+                              FLUSH_ORDER_MASK)

I don't think FLUSH_ORDER_MASK should be part of this, as it's not exclusively TLB-flush related.

My intention was to capture all flags pertinent to TLB flushes. I didn't mean they were exclusively TLB flush related.

That's what I understood, but the result is leading to possible ambiguity. Thinking about it again, ...

Quoted 8 lines
Quoted 1 line

I can remove the order mask from the list of flags. Not a big deal.

... I'm afraid I also have to ask for FLUSH_VA_VALID to be dropped, as that one's not TLB flush related at all. Instead it identifies whether the linear address provided _still has_ a valid mapping (rather than there being an address provided at all), such that CLFLUSH bases cache flushing would be usable. At that point I wonder whether the separate constant is still of much use at all.

Sure, I don't mind dropping it altogether.

Wei.

Quoted 2 lines

Jan

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel