target/arm/arch_dump: Add SVE notes

When dumping a guest with dump-guest-memory also dump the SVE
registers if they are in use.

Signed-off-by: Andrew Jones <drjones@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20200120101832.18781-1-drjones@redhat.com
[PMM: fixed checkpatch nits]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

3 files changed, 147 insertions, 26 deletions

diff --git a/target/arm/arch_dump.c b/target/arm/arch_dump.c
index 26a2c09868..2345dec3c2 100644
--- a/target/arm/arch_dump.c
+++ b/target/arm/arch_dump.c
@@ -62,12 +62,23 @@ struct aarch64_user_vfp_state {
 
 QEMU_BUILD_BUG_ON(sizeof(struct aarch64_user_vfp_state) != 528);
 
+/* struct user_sve_header from arch/arm64/include/uapi/asm/ptrace.h */
+struct aarch64_user_sve_header {
+    uint32_t size;
+    uint32_t max_size;
+    uint16_t vl;
+    uint16_t max_vl;
+    uint16_t flags;
+    uint16_t reserved;
+} QEMU_PACKED;
+
 struct aarch64_note {
     Elf64_Nhdr hdr;
     char name[8]; /* align_up(sizeof("CORE"), 4) */
     union {
         struct aarch64_elf_prstatus prstatus;
         struct aarch64_user_vfp_state vfp;
+        struct aarch64_user_sve_header sve;
     };
 } QEMU_PACKED;
 
@@ -76,6 +87,8 @@ struct aarch64_note {
             (AARCH64_NOTE_HEADER_SIZE + sizeof(struct aarch64_elf_prstatus))
 #define AARCH64_PRFPREG_NOTE_SIZE \
             (AARCH64_NOTE_HEADER_SIZE + sizeof(struct aarch64_user_vfp_state))
+#define AARCH64_SVE_NOTE_SIZE(env) \
+            (AARCH64_NOTE_HEADER_SIZE + sve_size(env))
 
 static void aarch64_note_init(struct aarch64_note *note, DumpState *s,
                               const char *name, Elf64_Word namesz,
@@ -128,11 +141,102 @@ static int aarch64_write_elf64_prfpreg(WriteCoreDumpFunction f,
     return 0;
 }
 
+#ifdef TARGET_AARCH64
+static off_t sve_zreg_offset(uint32_t vq, int n)
+{
+    off_t off = sizeof(struct aarch64_user_sve_header);
+    return ROUND_UP(off, 16) + vq * 16 * n;
+}
+
+static off_t sve_preg_offset(uint32_t vq, int n)
+{
+    return sve_zreg_offset(vq, 32) + vq * 16 / 8 * n;
+}
+
+static off_t sve_fpsr_offset(uint32_t vq)
+{
+    off_t off = sve_preg_offset(vq, 17);
+    return ROUND_UP(off, 16);
+}
+
+static off_t sve_fpcr_offset(uint32_t vq)
+{
+    return sve_fpsr_offset(vq) + sizeof(uint32_t);
+}
+
+static uint32_t sve_current_vq(CPUARMState *env)
+{
+    return sve_zcr_len_for_el(env, arm_current_el(env)) + 1;
+}
+
+static size_t sve_size_vq(uint32_t vq)
+{
+    off_t off = sve_fpcr_offset(vq) + sizeof(uint32_t);
+    return ROUND_UP(off, 16);
+}
+
+static size_t sve_size(CPUARMState *env)
+{
+    return sve_size_vq(sve_current_vq(env));
+}
+
+static int aarch64_write_elf64_sve(WriteCoreDumpFunction f,
+                                   CPUARMState *env, int cpuid,
+                                   DumpState *s)
+{
+    struct aarch64_note *note;
+    ARMCPU *cpu = env_archcpu(env);
+    uint32_t vq = sve_current_vq(env);
+    uint64_t tmp[ARM_MAX_VQ * 2], *r;
+    uint32_t fpr;
+    uint8_t *buf;
+    int ret, i;
+
+    note = g_malloc0(AARCH64_SVE_NOTE_SIZE(env));
+    buf = (uint8_t *)&note->sve;
+
+    aarch64_note_init(note, s, "LINUX", 6, NT_ARM_SVE, sve_size_vq(vq));
+
+    note->sve.size = cpu_to_dump32(s, sve_size_vq(vq));
+    note->sve.max_size = cpu_to_dump32(s, sve_size_vq(cpu->sve_max_vq));
+    note->sve.vl = cpu_to_dump16(s, vq * 16);
+    note->sve.max_vl = cpu_to_dump16(s, cpu->sve_max_vq * 16);
+    note->sve.flags = cpu_to_dump16(s, 1);
+
+    for (i = 0; i < 32; ++i) {
+        r = sve_bswap64(tmp, &env->vfp.zregs[i].d[0], vq * 2);
+        memcpy(&buf[sve_zreg_offset(vq, i)], r, vq * 16);
+    }
+
+    for (i = 0; i < 17; ++i) {
+        r = sve_bswap64(tmp, r = &env->vfp.pregs[i].p[0],
+                        DIV_ROUND_UP(vq * 2, 8));
+        memcpy(&buf[sve_preg_offset(vq, i)], r, vq * 16 / 8);
+    }
+
+    fpr = cpu_to_dump32(s, vfp_get_fpsr(env));
+    memcpy(&buf[sve_fpsr_offset(vq)], &fpr, sizeof(uint32_t));
+
+    fpr = cpu_to_dump32(s, vfp_get_fpcr(env));
+    memcpy(&buf[sve_fpcr_offset(vq)], &fpr, sizeof(uint32_t));
+
+    ret = f(note, AARCH64_SVE_NOTE_SIZE(env), s);
+    g_free(note);
+
+    if (ret < 0) {
+        return -1;
+    }
+
+    return 0;
+}
+#endif
+
 int arm_cpu_write_elf64_note(WriteCoreDumpFunction f, CPUState *cs,
                              int cpuid, void *opaque)
 {
     struct aarch64_note note;
-    CPUARMState *env = &ARM_CPU(cs)->env;
+    ARMCPU *cpu = ARM_CPU(cs);
+    CPUARMState *env = &cpu->env;
     DumpState *s = opaque;
     uint64_t pstate, sp;
     int ret, i;
@@ -163,7 +267,18 @@ int arm_cpu_write_elf64_note(WriteCoreDumpFunction f, CPUState *cs,
         return -1;
     }
 
-    return aarch64_write_elf64_prfpreg(f, env, cpuid, s);
+    ret = aarch64_write_elf64_prfpreg(f, env, cpuid, s);
+    if (ret) {
+        return ret;
+    }
+
+#ifdef TARGET_AARCH64
+    if (cpu_isar_feature(aa64_sve, cpu)) {
+        ret = aarch64_write_elf64_sve(f, env, cpuid, s);
+    }
+#endif
+
+    return ret;
 }
 
 /* struct pt_regs from arch/arm/include/asm/ptrace.h */
@@ -335,6 +450,11 @@ ssize_t cpu_get_note_size(int class, int machine, int nr_cpus)
     if (class == ELFCLASS64) {
         note_size = AARCH64_PRSTATUS_NOTE_SIZE;
         note_size += AARCH64_PRFPREG_NOTE_SIZE;
+#ifdef TARGET_AARCH64
+        if (cpu_isar_feature(aa64_sve, cpu)) {
+            note_size += AARCH64_SVE_NOTE_SIZE(env);
+        }
+#endif
     } else {
         note_size = ARM_PRSTATUS_NOTE_SIZE;
         if (arm_feature(env, ARM_FEATURE_VFP)) {
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index 40f2c45e17..c1aedbeac0 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -980,6 +980,31 @@ void aarch64_sve_narrow_vq(CPUARMState *env, unsigned vq);
 void aarch64_sve_change_el(CPUARMState *env, int old_el,
                            int new_el, bool el0_a64);
 void aarch64_add_sve_properties(Object *obj);
+
+/*
+ * SVE registers are encoded in KVM's memory in an endianness-invariant format.
+ * The byte at offset i from the start of the in-memory representation contains
+ * the bits [(7 + 8 * i) : (8 * i)] of the register value. As this means the
+ * lowest offsets are stored in the lowest memory addresses, then that nearly
+ * matches QEMU's representation, which is to use an array of host-endian
+ * uint64_t's, where the lower offsets are at the lower indices. To complete
+ * the translation we just need to byte swap the uint64_t's on big-endian hosts.
+ */
+static inline uint64_t *sve_bswap64(uint64_t *dst, uint64_t *src, int nr)
+{
+#ifdef HOST_WORDS_BIGENDIAN
+    int i;
+
+    for (i = 0; i < nr; ++i) {
+        dst[i] = bswap64(src[i]);
+    }
+
+    return dst;
+#else
+    return src;
+#endif
+}
+
 #else
 static inline void aarch64_sve_narrow_vq(CPUARMState *env, unsigned vq) { }
 static inline void aarch64_sve_change_el(CPUARMState *env, int o,
diff --git a/target/arm/kvm64.c b/target/arm/kvm64.c
index 876184b8fe..e2da756e65 100644
--- a/target/arm/kvm64.c
+++ b/target/arm/kvm64.c
@@ -877,30 +877,6 @@ static int kvm_arch_put_fpsimd(CPUState *cs)
 }
 
 /*
- * SVE registers are encoded in KVM's memory in an endianness-invariant format.
- * The byte at offset i from the start of the in-memory representation contains
- * the bits [(7 + 8 * i) : (8 * i)] of the register value. As this means the
- * lowest offsets are stored in the lowest memory addresses, then that nearly
- * matches QEMU's representation, which is to use an array of host-endian
- * uint64_t's, where the lower offsets are at the lower indices. To complete
- * the translation we just need to byte swap the uint64_t's on big-endian hosts.
- */
-static uint64_t *sve_bswap64(uint64_t *dst, uint64_t *src, int nr)
-{
-#ifdef HOST_WORDS_BIGENDIAN
-    int i;
-
-    for (i = 0; i < nr; ++i) {
-        dst[i] = bswap64(src[i]);
-    }
-
-    return dst;
-#else
-    return src;
-#endif
-}
-
-/*
  * KVM SVE registers come in slices where ZREGs have a slice size of 2048 bits
  * and PREGS and the FFR have a slice size of 256 bits. However we simply hard
  * code the slice index to zero for now as it's unlikely we'll need more than