target-ppc: Enable KVM for ppcemb.

Implement hooks called by generic KVM code.

Also add code that will copy the host's CPU and timebase frequencies to the
guest, which is necessary on KVM because the guest can directly access the
timebase.

Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>
Acked-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@6065 c046a42c-6fe2-441c-8c8c-71466251a162

6 files changed, 343 insertions, 0 deletions

diff --git a/Makefile.target b/Makefile.target
index d80ce2cd37..cbff562047 100644
--- a/Makefile.target
+++ b/Makefile.target
@@ -659,6 +659,9 @@ ifdef FDT_LIBS
 OBJS+= device_tree.o
 LIBS+= $(FDT_LIBS)
 endif
+ifdef CONFIG_KVM
+OBJS+= kvm_ppc.o
+endif
 # virtio support
 OBJS+= virtio.o virtio-blk.o virtio-balloon.o
 endif
diff --git a/configure b/configure
index c4aa8810a3..4f8bc5e5bc 100755
--- a/configure
+++ b/configure
@@ -1491,6 +1491,7 @@ gdb_xml_files=""
 
 # Make sure the target and host cpus are compatible
 if test "$kvm" = "yes" -a ! \( "$target_cpu" = "$cpu" -o \
+  \( "$target_cpu" = "ppcemb" -a "$cpu" = "powerpc" \) -o \
   \( "$target_cpu" = "x86_64" -a "$cpu" = "i386"   \) -o \
   \( "$target_cpu" = "i386"   -a "$cpu" = "x86_64" \) \) ; then
   kvm="no"
@@ -1585,6 +1586,11 @@ case "$target_cpu" in
     echo "#define TARGET_ARCH \"ppcemb\"" >> $config_h
     echo "#define TARGET_PPC 1" >> $config_h
     echo "#define TARGET_PPCEMB 1" >> $config_h
+    if test "$kvm" = "yes" ; then
+      echo "CONFIG_KVM=yes" >> $config_mak
+      echo "KVM_CFLAGS=$kvm_cflags" >> $config_mak
+      echo "#define CONFIG_KVM 1" >> $config_h
+    fi
   ;;
   ppc64)
     echo "TARGET_ARCH=ppc64" >> $config_mak
diff --git a/target-ppc/helper.c b/target-ppc/helper.c
index 538d9f34c4..4676d50d87 100644
--- a/target-ppc/helper.c
+++ b/target-ppc/helper.c
@@ -29,6 +29,7 @@
 #include "exec-all.h"
 #include "helper_regs.h"
 #include "qemu-common.h"
+#include "kvm.h"
 
 //#define DEBUG_MMU
 //#define DEBUG_BATS
@@ -2920,6 +2921,10 @@ CPUPPCState *cpu_ppc_init (const char *cpu_model)
     env->cpu_model_str = cpu_model;
     cpu_ppc_register_internal(env, def);
     cpu_ppc_reset(env);
+
+    if (kvm_enabled())
+        kvm_init_vcpu(env);
+
     return env;
 }
 
diff --git a/target-ppc/kvm.c b/target-ppc/kvm.c
new file mode 100644
index 0000000000..acbb1abd2b
--- /dev/null
+++ b/target-ppc/kvm.c
@@ -0,0 +1,204 @@
+/*
+ * PowerPC implementation of KVM hooks
+ *
+ * Copyright IBM Corp. 2007
+ *
+ * Authors:
+ *  Jerone Young <jyoung5@us.ibm.com>
+ *  Christian Ehrhardt <ehrhardt@linux.vnet.ibm.com>
+ *  Hollis Blanchard <hollisb@us.ibm.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#include <sys/types.h>
+#include <sys/ioctl.h>
+#include <sys/mman.h>
+
+#include <linux/kvm.h>
+
+#include "qemu-common.h"
+#include "qemu-timer.h"
+#include "sysemu.h"
+#include "kvm.h"
+#include "kvm_ppc.h"
+#include "cpu.h"
+#include "device_tree.h"
+
+//#define DEBUG_KVM
+
+#ifdef DEBUG_KVM
+#define dprintf(fmt, ...) \
+    do { fprintf(stderr, fmt, ## __VA_ARGS__); } while (0)
+#else
+#define dprintf(fmt, ...) \
+    do { } while (0)
+#endif
+
+int kvm_arch_init(KVMState *s, int smp_cpus)
+{
+    return 0;
+}
+
+int kvm_arch_init_vcpu(CPUState *cenv)
+{
+    return 0;
+}
+
+int kvm_arch_put_registers(CPUState *env)
+{
+    struct kvm_regs regs;
+    int ret;
+    int i;
+
+    ret = kvm_vcpu_ioctl(env, KVM_GET_REGS, &regs);
+    if (ret < 0)
+        return ret;
+
+    regs.ctr = env->ctr;
+    regs.lr  = env->lr;
+    regs.xer = env->xer;
+    regs.msr = env->msr;
+    regs.pc = env->nip;
+
+    regs.srr0 = env->spr[SPR_SRR0];
+    regs.srr1 = env->spr[SPR_SRR1];
+
+    regs.sprg0 = env->spr[SPR_SPRG0];
+    regs.sprg1 = env->spr[SPR_SPRG1];
+    regs.sprg2 = env->spr[SPR_SPRG2];
+    regs.sprg3 = env->spr[SPR_SPRG3];
+    regs.sprg4 = env->spr[SPR_SPRG4];
+    regs.sprg5 = env->spr[SPR_SPRG5];
+    regs.sprg6 = env->spr[SPR_SPRG6];
+    regs.sprg7 = env->spr[SPR_SPRG7];
+
+    for (i = 0;i < 32; i++)
+        regs.gpr[i] = env->gpr[i];
+
+    ret = kvm_vcpu_ioctl(env, KVM_SET_REGS, &regs);
+    if (ret < 0)
+        return ret;
+
+    return ret;
+}
+
+int kvm_arch_get_registers(CPUState *env)
+{
+    struct kvm_regs regs;
+    uint32_t i, ret;
+
+    ret = kvm_vcpu_ioctl(env, KVM_GET_REGS, &regs);
+    if (ret < 0)
+        return ret;
+
+    env->ctr = regs.ctr;
+    env->lr = regs.lr;
+    env->xer = regs.xer;
+    env->msr = regs.msr;
+    env->nip = regs.pc;
+
+    env->spr[SPR_SRR0] = regs.srr0;
+    env->spr[SPR_SRR1] = regs.srr1;
+
+    env->spr[SPR_SPRG0] = regs.sprg0;
+    env->spr[SPR_SPRG1] = regs.sprg1;
+    env->spr[SPR_SPRG2] = regs.sprg2;
+    env->spr[SPR_SPRG3] = regs.sprg3;
+    env->spr[SPR_SPRG4] = regs.sprg4;
+    env->spr[SPR_SPRG5] = regs.sprg5;
+    env->spr[SPR_SPRG6] = regs.sprg6;
+    env->spr[SPR_SPRG7] = regs.sprg7;
+
+    for (i = 0;i < 32; i++)
+        env->gpr[i] = regs.gpr[i];
+
+    return 0;
+}
+
+int kvm_arch_pre_run(CPUState *env, struct kvm_run *run)
+{
+    int r;
+    unsigned irq;
+
+    /* PowerPC Qemu tracks the various core input pins (interrupt, critical
+     * interrupt, reset, etc) in PPC-specific env->irq_input_state. */
+    if (run->ready_for_interrupt_injection &&
+        (env->interrupt_request & CPU_INTERRUPT_HARD) &&
+        (env->irq_input_state & (1<<PPC40x_INPUT_INT)))
+    {
+        /* For now KVM disregards the 'irq' argument. However, in the
+         * future KVM could cache it in-kernel to avoid a heavyweight exit
+         * when reading the UIC.
+         */
+        irq = -1U;
+
+        dprintf("injected interrupt %d\n", irq);
+        r = kvm_vcpu_ioctl(env, KVM_INTERRUPT, &irq);
+        if (r < 0)
+            printf("cpu %d fail inject %x\n", env->cpu_index, irq);
+    }
+
+    /* We don't know if there are more interrupts pending after this. However,
+     * the guest will return to userspace in the course of handling this one
+     * anyways, so we will get a chance to deliver the rest. */
+    return 0;
+}
+
+int kvm_arch_post_run(CPUState *env, struct kvm_run *run)
+{
+    return 0;
+}
+
+static int kvmppc_handle_halt(CPUState *env)
+{
+    if (!(env->interrupt_request & CPU_INTERRUPT_HARD) && (msr_ee)) {
+        env->halted = 1;
+        env->exception_index = EXCP_HLT;
+    }
+
+    return 1;
+}
+
+/* map dcr access to existing qemu dcr emulation */
+static int kvmppc_handle_dcr_read(CPUState *env, uint32_t dcrn, uint32_t *data)
+{
+    if (ppc_dcr_read(env->dcr_env, dcrn, data) < 0)
+        fprintf(stderr, "Read to unhandled DCR (0x%x)\n", dcrn);
+
+    return 1;
+}
+
+static int kvmppc_handle_dcr_write(CPUState *env, uint32_t dcrn, uint32_t data)
+{
+    if (ppc_dcr_write(env->dcr_env, dcrn, data) < 0)
+        fprintf(stderr, "Write to unhandled DCR (0x%x)\n", dcrn);
+
+    return 1;
+}
+
+int kvm_arch_handle_exit(CPUState *env, struct kvm_run *run)
+{
+    int ret = 0;
+
+    switch (run->exit_reason) {
+    case KVM_EXIT_DCR:
+        if (run->dcr.is_write) {
+            dprintf("handle dcr write\n");
+            ret = kvmppc_handle_dcr_write(env, run->dcr.dcrn, run->dcr.data);
+        } else {
+            dprintf("handle dcr read\n");
+            ret = kvmppc_handle_dcr_read(env, run->dcr.dcrn, &run->dcr.data);
+        }
+        break;
+    case KVM_EXIT_HLT:
+        dprintf("handle halt\n");
+        ret = kvmppc_handle_halt(env);
+        break;
+    }
+
+    return ret;
+}
+
diff --git a/target-ppc/kvm_ppc.c b/target-ppc/kvm_ppc.c
new file mode 100644
index 0000000000..0caa5b98c7
--- /dev/null
+++ b/target-ppc/kvm_ppc.c
@@ -0,0 +1,110 @@
+/*
+ * PowerPC KVM support
+ *
+ * Copyright IBM Corp. 2008
+ *
+ * Authors:
+ *  Hollis Blanchard <hollisb@us.ibm.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#include "qemu-common.h"
+#include "qemu-timer.h"
+#include "kvm_ppc.h"
+#include "device_tree.h"
+
+#define PROC_DEVTREE_PATH "/proc/device-tree"
+
+static QEMUTimer *kvmppc_timer;
+static unsigned int kvmppc_timer_rate;
+
+#ifdef HAVE_FDT
+static int kvmppc_read_host_property(const char *node_path, const char *prop,
+                                     void *val, size_t len)
+{
+    char *path;
+    FILE *f;
+    int ret;
+    int pathlen;
+
+    pathlen = snprintf(NULL, 0, "%s/%s/%s", PROC_DEVTREE_PATH, node_path, prop)
+              + 1;
+    path = qemu_malloc(pathlen);
+    if (path == NULL) {
+        ret = -ENOMEM;
+        goto out;
+    }
+
+    snprintf(path, pathlen, "%s/%s/%s", PROC_DEVTREE_PATH, node_path, prop);
+
+    f = fopen(path, "rb");
+    if (f == NULL) {
+        ret = errno;
+        goto free;
+    }
+
+    len = fread(val, len, 1, f);
+    if (len != 1) {
+        ret = ferror(f);
+        goto close;
+    }
+
+close:
+    fclose(f);
+free:
+    free(path);
+out:
+    return ret;
+}
+
+static int kvmppc_copy_host_cell(void *fdt, const char *node, const char *prop)
+{
+    uint32_t cell;
+    int ret;
+
+    ret = kvmppc_read_host_property(node, prop, &cell, sizeof(cell));
+    if (ret < 0) {
+        fprintf(stderr, "couldn't read host %s/%s\n", node, prop);
+        goto out;
+    }
+
+    ret = qemu_devtree_setprop_cell(fdt, node, prop, cell);
+    if (ret < 0) {
+        fprintf(stderr, "couldn't set guest %s/%s\n", node, prop);
+        goto out;
+    }
+
+out:
+    return ret;
+}
+
+void kvmppc_fdt_update(void *fdt)
+{
+    /* Copy data from the host device tree into the guest. Since the guest can
+     * directly access the timebase without host involvement, we must expose
+     * the correct frequencies. */
+    kvmppc_copy_host_cell(fdt, "/cpus/cpu@0", "clock-frequency");
+    kvmppc_copy_host_cell(fdt, "/cpus/cpu@0", "timebase-frequency");
+}
+#endif
+
+static void kvmppc_timer_hack(void *opaque)
+{
+    qemu_service_io();
+    qemu_mod_timer(kvmppc_timer, qemu_get_clock(vm_clock) + kvmppc_timer_rate);
+}
+
+void kvmppc_init(void)
+{
+    /* XXX The only reason KVM yields control back to qemu is device IO. Since
+     * an idle guest does no IO, qemu's device model will never get a chance to
+     * run. So, until Qemu gains IO threads, we create this timer to ensure
+     * that the device model gets a chance to run. */
+    kvmppc_timer_rate = ticks_per_sec / 10;
+    kvmppc_timer = qemu_new_timer(vm_clock, &kvmppc_timer_hack, NULL);
+    qemu_mod_timer(kvmppc_timer, qemu_get_clock(vm_clock) + kvmppc_timer_rate);
+}
+
diff --git a/target-ppc/kvm_ppc.h b/target-ppc/kvm_ppc.h
new file mode 100644
index 0000000000..e536a881e9
--- /dev/null
+++ b/target-ppc/kvm_ppc.h
@@ -0,0 +1,15 @@
+/*
+ * Copyright 2008 IBM Corporation.
+ * Authors: Hollis Blanchard <hollisb@us.ibm.com>
+ *
+ * This work is licensed under the GNU GPL license version 2 or later.
+ *
+ */
+
+#ifndef __KVM_PPC_H__
+#define __KVM_PPC_H__
+
+void kvmppc_init(void);
+void kvmppc_fdt_update(void *fdt);
+
+#endif /* __KVM_PPC_H__ */