[Qemu-devel] [PULL 02/10] s390x: Vector Register IOCTLs

[Cornelia Huck <cornelia.huck@de.ibm.com>]

From: Eric Farman <farman@linux.vnet.ibm.com>

Handle the actual syncing of the vector registers with kernel space,
via the get/put register IOCTLs.

The vector registers that were introduced with the z13 overlay
the existing floating point registers.  FP registers 0-15 are
the high-halves of vector registers 0-15.  Thus, remove the
freg fields and replace them with the equivalent vector field
to avoid errors in duplication.  Moreover, synchronize either the
vector registers via kvm_sync_regs, or floating point registers
via the GET/SET FPU IOCTLs.

Signed-off-by: Eric Farman <farman@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
---
 target-s390x/cpu.h       |  8 ++++++--
 target-s390x/kvm.c       | 49 ++++++++++++++++++++++++++++++++----------------
 target-s390x/machine.c   | 32 +++++++++++++++----------------
 target-s390x/translate.c |  2 +-
 4 files changed, 56 insertions(+), 35 deletions(-)

diff --git a/target-s390x/cpu.h b/target-s390x/cpu.h
index e968769..708349c 100644
--- a/target-s390x/cpu.h
+++ b/target-s390x/cpu.h
@@ -81,7 +81,11 @@ typedef struct MchkQueue {
 
 typedef struct CPUS390XState {
     uint64_t regs[16];     /* GP registers */
-    CPU_DoubleU fregs[16]; /* FP registers */
+    /*
+     * The floating point registers are part of the vector registers.
+     * vregs[0][0] -> vregs[15][0] are 16 floating point registers
+     */
+    CPU_DoubleU vregs[32][2];  /* vector registers */
     uint32_t aregs[16];    /* access registers */
 
     uint32_t fpc;          /* floating-point control register */
@@ -164,7 +168,7 @@ typedef struct CPUS390XState {
 
 static inline CPU_DoubleU *get_freg(CPUS390XState *cs, int nr)
 {
-    return &cs->fregs[nr];
+    return &cs->vregs[nr][0];
 }
 
 #include "cpu-qom.h"
diff --git a/target-s390x/kvm.c b/target-s390x/kvm.c
index 1a40a94..ca8ffcc 100644
--- a/target-s390x/kvm.c
+++ b/target-s390x/kvm.c
@@ -337,15 +337,24 @@ int kvm_arch_put_registers(CPUState *cs, int level)
         }
     }
 
-    /* Floating point */
-    for (i = 0; i < 16; i++) {
-        fpu.fprs[i] = get_freg(env, i)->ll;
-    }
-    fpu.fpc = env->fpc;
+    if (can_sync_regs(cs, KVM_SYNC_VRS)) {
+        for (i = 0; i < 32; i++) {
+            cs->kvm_run->s.regs.vrs[i][0] = env->vregs[i][0].ll;
+            cs->kvm_run->s.regs.vrs[i][1] = env->vregs[i][1].ll;
+        }
+        cs->kvm_run->s.regs.fpc = env->fpc;
+        cs->kvm_run->kvm_dirty_regs |= KVM_SYNC_VRS;
+    } else {
+        /* Floating point */
+        for (i = 0; i < 16; i++) {
+            fpu.fprs[i] = get_freg(env, i)->ll;
+        }
+        fpu.fpc = env->fpc;
 
-    r = kvm_vcpu_ioctl(cs, KVM_SET_FPU, &fpu);
-    if (r < 0) {
-        return r;
+        r = kvm_vcpu_ioctl(cs, KVM_SET_FPU, &fpu);
+        if (r < 0) {
+            return r;
+        }
     }
 
     /* Do we need to save more than that? */
@@ -468,15 +477,23 @@ int kvm_arch_get_registers(CPUState *cs)
         }
     }
 
-    /* Floating point */
-    r = kvm_vcpu_ioctl(cs, KVM_GET_FPU, &fpu);
-    if (r < 0) {
-        return r;
-    }
-    for (i = 0; i < 16; i++) {
-        get_freg(env, i)->ll = fpu.fprs[i];
+    /* Floating point and vector registers */
+    if (can_sync_regs(cs, KVM_SYNC_VRS)) {
+        for (i = 0; i < 32; i++) {
+            env->vregs[i][0].ll = cs->kvm_run->s.regs.vrs[i][0];
+            env->vregs[i][1].ll = cs->kvm_run->s.regs.vrs[i][1];
+        }
+        env->fpc = cs->kvm_run->s.regs.fpc;
+    } else {
+        r = kvm_vcpu_ioctl(cs, KVM_GET_FPU, &fpu);
+        if (r < 0) {
+            return r;
+        }
+        for (i = 0; i < 16; i++) {
+            get_freg(env, i)->ll = fpu.fprs[i];
+        }
+        env->fpc = fpu.fpc;
     }
-    env->fpc = fpu.fpc;
 
     /* The prefix */
     if (can_sync_regs(cs, KVM_SYNC_PREFIX)) {
diff --git a/target-s390x/machine.c b/target-s390x/machine.c
index 7853e3c..1a9390b 100644
--- a/target-s390x/machine.c
+++ b/target-s390x/machine.c
@@ -47,22 +47,22 @@ const VMStateDescription vmstate_fpu = {
     .version_id = 1,
     .minimum_version_id = 1,
     .fields = (VMStateField[]) {
-        VMSTATE_UINT64(env.fregs[0].ll, S390CPU),
-        VMSTATE_UINT64(env.fregs[1].ll, S390CPU),
-        VMSTATE_UINT64(env.fregs[2].ll, S390CPU),
-        VMSTATE_UINT64(env.fregs[3].ll, S390CPU),
-        VMSTATE_UINT64(env.fregs[4].ll, S390CPU),
-        VMSTATE_UINT64(env.fregs[5].ll, S390CPU),
-        VMSTATE_UINT64(env.fregs[6].ll, S390CPU),
-        VMSTATE_UINT64(env.fregs[7].ll, S390CPU),
-        VMSTATE_UINT64(env.fregs[8].ll, S390CPU),
-        VMSTATE_UINT64(env.fregs[9].ll, S390CPU),
-        VMSTATE_UINT64(env.fregs[10].ll, S390CPU),
-        VMSTATE_UINT64(env.fregs[11].ll, S390CPU),
-        VMSTATE_UINT64(env.fregs[12].ll, S390CPU),
-        VMSTATE_UINT64(env.fregs[13].ll, S390CPU),
-        VMSTATE_UINT64(env.fregs[14].ll, S390CPU),
-        VMSTATE_UINT64(env.fregs[15].ll, S390CPU),
+        VMSTATE_UINT64(env.vregs[0][0].ll, S390CPU),
+        VMSTATE_UINT64(env.vregs[1][0].ll, S390CPU),
+        VMSTATE_UINT64(env.vregs[2][0].ll, S390CPU),
+        VMSTATE_UINT64(env.vregs[3][0].ll, S390CPU),
+        VMSTATE_UINT64(env.vregs[4][0].ll, S390CPU),
+        VMSTATE_UINT64(env.vregs[5][0].ll, S390CPU),
+        VMSTATE_UINT64(env.vregs[6][0].ll, S390CPU),
+        VMSTATE_UINT64(env.vregs[7][0].ll, S390CPU),
+        VMSTATE_UINT64(env.vregs[8][0].ll, S390CPU),
+        VMSTATE_UINT64(env.vregs[9][0].ll, S390CPU),
+        VMSTATE_UINT64(env.vregs[10][0].ll, S390CPU),
+        VMSTATE_UINT64(env.vregs[11][0].ll, S390CPU),
+        VMSTATE_UINT64(env.vregs[12][0].ll, S390CPU),
+        VMSTATE_UINT64(env.vregs[13][0].ll, S390CPU),
+        VMSTATE_UINT64(env.vregs[14][0].ll, S390CPU),
+        VMSTATE_UINT64(env.vregs[15][0].ll, S390CPU),
         VMSTATE_UINT32(env.fpc, S390CPU),
         VMSTATE_END_OF_LIST()
     }
diff --git a/target-s390x/translate.c b/target-s390x/translate.c
index 96b43b3..06fc192 100644
--- a/target-s390x/translate.c
+++ b/target-s390x/translate.c
@@ -187,7 +187,7 @@ void s390x_translate_init(void)
     for (i = 0; i < 16; i++) {
         snprintf(cpu_reg_names[i + 16], sizeof(cpu_reg_names[0]), "f%d", i);
         fregs[i] = tcg_global_mem_new(TCG_AREG0,
-                                      offsetof(CPUS390XState, fregs[i].d),
+                                      offsetof(CPUS390XState, vregs[i][0].d),
                                       cpu_reg_names[i + 16]);
     }
 }
-- 
2.4.2