Merge branch 'x86-fpu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull x86 fpu updates from Ingo Molnar:
 "The main x86 FPU changes in this cycle were:

   - a large series of cleanups, fixes and enhancements to re-enable the
     XSAVES instruction on Intel CPUs - which is the most advanced
     instruction to do FPU context switches (Yu-cheng Yu, Fenghua Yu)

   - Add FPU tracepoints for the FPU state machine (Dave Hansen)"

* 'x86-fpu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/fpu: Do not BUG_ON() in early FPU code
  x86/fpu/xstate: Re-enable XSAVES
  x86/fpu/xstate: Fix fpstate_init() for XRSTORS
  x86/fpu/xstate: Return NULL for disabled xstate component address
  x86/fpu/xstate: Fix __fpu_restore_sig() for XSAVES
  x86/fpu/xstate: Fix xstate_offsets, xstate_sizes for non-extended xstates
  x86/fpu/xstate: Fix XSTATE component offset print out
  x86/fpu/xstate: Fix PTRACE frames for XSAVES
  x86/fpu/xstate: Fix supervisor xstate component offset
  x86/fpu/xstate: Align xstate components according to CPUID
  x86/fpu/xstate: Copy xstate registers directly to the signal frame when compacted format is in use
  x86/fpu/xstate: Keep init_fpstate.xsave.header.xfeatures as zero for init optimization
  x86/fpu/xstate: Rename 'xstate_size' to 'fpu_kernel_xstate_size', to distinguish it from 'fpu_user_xstate_size'
  x86/fpu/xstate: Define and use 'fpu_user_xstate_size'
  x86/fpu: Add tracepoints to dump FPU state at key points

5 files changed, 464 insertions, 152 deletions

diff --git a/arch/x86/kernel/fpu/core.c b/arch/x86/kernel/fpu/core.c
index 97027545a72d..3fc03a09a93b 100644
--- a/arch/x86/kernel/fpu/core.c
+++ b/arch/x86/kernel/fpu/core.c
@@ -8,10 +8,14 @@
 #include <asm/fpu/internal.h>
 #include <asm/fpu/regset.h>
 #include <asm/fpu/signal.h>
+#include <asm/fpu/types.h>
 #include <asm/traps.h>
 
 #include <linux/hardirq.h>
 
+#define CREATE_TRACE_POINTS
+#include <asm/trace/fpu.h>
+
 /*
  * Represents the initial FPU state. It's mostly (but not completely) zeroes,
  * depending on the FPU hardware format:
@@ -192,6 +196,7 @@ void fpu__save(struct fpu *fpu)
         WARN_ON_FPU(fpu != &current->thread.fpu);
 
         preempt_disable();
+        trace_x86_fpu_before_save(fpu);
         if (fpu->fpregs_active) {
                 if (!copy_fpregs_to_fpstate(fpu)) {
                         if (use_eager_fpu())
@@ -200,6 +205,7 @@ void fpu__save(struct fpu *fpu)
                                 fpregs_deactivate(fpu);
                 }
         }
+        trace_x86_fpu_after_save(fpu);
         preempt_enable();
 }
 EXPORT_SYMBOL_GPL(fpu__save);
@@ -222,7 +228,14 @@ void fpstate_init(union fpregs_state *state)
                 return;
         }
 
-        memset(state, 0, xstate_size);
+        memset(state, 0, fpu_kernel_xstate_size);
+
+        /*
+         * XRSTORS requires that this bit is set in xcomp_bv, or
+         * it will #GP. Make sure it is replaced after the memset().
+         */
+        if (static_cpu_has(X86_FEATURE_XSAVES))
+                state->xsave.header.xcomp_bv = XCOMP_BV_COMPACTED_FORMAT;
 
         if (static_cpu_has(X86_FEATURE_FXSR))
                 fpstate_init_fxstate(&state->fxsave);
@@ -247,7 +260,7 @@ int fpu__copy(struct fpu *dst_fpu, struct fpu *src_fpu)
          * leak into the child task:
          */
         if (use_eager_fpu())
-                memset(&dst_fpu->state.xsave, 0, xstate_size);
+                memset(&dst_fpu->state.xsave, 0, fpu_kernel_xstate_size);
 
         /*
          * Save current FPU registers directly into the child
@@ -266,7 +279,8 @@ int fpu__copy(struct fpu *dst_fpu, struct fpu *src_fpu)
          */
         preempt_disable();
         if (!copy_fpregs_to_fpstate(dst_fpu)) {
-                memcpy(&src_fpu->state, &dst_fpu->state, xstate_size);
+                memcpy(&src_fpu->state, &dst_fpu->state,
+                       fpu_kernel_xstate_size);
 
                 if (use_eager_fpu())
                         copy_kernel_to_fpregs(&src_fpu->state);
@@ -275,6 +289,9 @@ int fpu__copy(struct fpu *dst_fpu, struct fpu *src_fpu)
         }
         preempt_enable();
 
+        trace_x86_fpu_copy_src(src_fpu);
+        trace_x86_fpu_copy_dst(dst_fpu);
+
         return 0;
 }
 
@@ -288,7 +305,9 @@ void fpu__activate_curr(struct fpu *fpu)
 
         if (!fpu->fpstate_active) {
                 fpstate_init(&fpu->state);
+                trace_x86_fpu_init_state(fpu);
 
+                trace_x86_fpu_activate_state(fpu);
                 /* Safe to do for the current task: */
                 fpu->fpstate_active = 1;
         }
@@ -314,7 +333,9 @@ void fpu__activate_fpstate_read(struct fpu *fpu)
         } else {
                 if (!fpu->fpstate_active) {
                         fpstate_init(&fpu->state);
+                        trace_x86_fpu_init_state(fpu);
 
+                        trace_x86_fpu_activate_state(fpu);
                         /* Safe to do for current and for stopped child tasks: */
                         fpu->fpstate_active = 1;
                 }
@@ -347,7 +368,9 @@ void fpu__activate_fpstate_write(struct fpu *fpu)
                 fpu->last_cpu = -1;
         } else {
                 fpstate_init(&fpu->state);
+                trace_x86_fpu_init_state(fpu);
 
+                trace_x86_fpu_activate_state(fpu);
                 /* Safe to do for stopped child tasks: */
                 fpu->fpstate_active = 1;
         }
@@ -432,9 +455,11 @@ void fpu__restore(struct fpu *fpu)
 
         /* Avoid __kernel_fpu_begin() right after fpregs_activate() */
         kernel_fpu_disable();
+        trace_x86_fpu_before_restore(fpu);
         fpregs_activate(fpu);
         copy_kernel_to_fpregs(&fpu->state);
         fpu->counter++;
+        trace_x86_fpu_after_restore(fpu);
         kernel_fpu_enable();
 }
 EXPORT_SYMBOL_GPL(fpu__restore);
@@ -463,6 +488,8 @@ void fpu__drop(struct fpu *fpu)
 
         fpu->fpstate_active = 0;
 
+        trace_x86_fpu_dropped(fpu);
+
         preempt_enable();
 }
 
diff --git a/arch/x86/kernel/fpu/init.c b/arch/x86/kernel/fpu/init.c
index aacfd7a82cec..93982aebb398 100644
--- a/arch/x86/kernel/fpu/init.c
+++ b/arch/x86/kernel/fpu/init.c
@@ -145,8 +145,8 @@ static void __init fpu__init_system_generic(void)
  * This is inherent to the XSAVE architecture which puts all state
  * components into a single, continuous memory block:
  */
-unsigned int xstate_size;
-EXPORT_SYMBOL_GPL(xstate_size);
+unsigned int fpu_kernel_xstate_size;
+EXPORT_SYMBOL_GPL(fpu_kernel_xstate_size);
 
 /* Get alignment of the TYPE. */
 #define TYPE_ALIGN(TYPE) offsetof(struct { char x; TYPE test; }, test)
@@ -178,7 +178,7 @@ static void __init fpu__init_task_struct_size(void)
          * Add back the dynamically-calculated register state
          * size.
          */
-        task_size += xstate_size;
+        task_size += fpu_kernel_xstate_size;
 
         /*
          * We dynamically size 'struct fpu', so we require that
@@ -195,7 +195,7 @@ static void __init fpu__init_task_struct_size(void)
 }
 
 /*
- * Set up the xstate_size based on the legacy FPU context size.
+ * Set up the user and kernel xstate sizes based on the legacy FPU context size.
  *
  * We set this up first, and later it will be overwritten by
  * fpu__init_system_xstate() if the CPU knows about xstates.
@@ -208,7 +208,7 @@ static void __init fpu__init_system_xstate_size_legacy(void)
         on_boot_cpu = 0;
 
         /*
-         * Note that xstate_size might be overwriten later during
+         * Note that xstate sizes might be overwritten later during
          * fpu__init_system_xstate().
          */
 
@@ -219,27 +219,17 @@ static void __init fpu__init_system_xstate_size_legacy(void)
                  */
                 setup_clear_cpu_cap(X86_FEATURE_XSAVE);
                 setup_clear_cpu_cap(X86_FEATURE_XSAVEOPT);
-                xstate_size = sizeof(struct swregs_state);
+                fpu_kernel_xstate_size = sizeof(struct swregs_state);
         } else {
                 if (boot_cpu_has(X86_FEATURE_FXSR))
-                        xstate_size = sizeof(struct fxregs_state);
+                        fpu_kernel_xstate_size =
+                                sizeof(struct fxregs_state);
                 else
-                        xstate_size = sizeof(struct fregs_state);
+                        fpu_kernel_xstate_size =
+                                sizeof(struct fregs_state);
         }
-        /*
-         * Quirk: we don't yet handle the XSAVES* instructions
-         * correctly, as we don't correctly convert between
-         * standard and compacted format when interfacing
-         * with user-space - so disable it for now.
-         *
-         * The difference is small: with recent CPUs the
-         * compacted format is only marginally smaller than
-         * the standard FPU state format.
-         *
-         * ( This is easy to backport while we are fixing
-         *   XSAVES* support. )
-         */
-        setup_clear_cpu_cap(X86_FEATURE_XSAVES);
+
+        fpu_user_xstate_size = fpu_kernel_xstate_size;
 }
 
 /*
diff --git a/arch/x86/kernel/fpu/regset.c b/arch/x86/kernel/fpu/regset.c
index 81422dfb152b..c114b132d121 100644
--- a/arch/x86/kernel/fpu/regset.c
+++ b/arch/x86/kernel/fpu/regset.c
@@ -4,6 +4,7 @@
 #include <asm/fpu/internal.h>
 #include <asm/fpu/signal.h>
 #include <asm/fpu/regset.h>
+#include <asm/fpu/xstate.h>
 
 /*
  * The xstateregs_active() routine is the same as the regset_fpregs_active() routine,
@@ -85,21 +86,26 @@ int xstateregs_get(struct task_struct *target, const struct user_regset *regset,
         if (!boot_cpu_has(X86_FEATURE_XSAVE))
                 return -ENODEV;
 
-        fpu__activate_fpstate_read(fpu);
-
         xsave = &fpu->state.xsave;
 
-        /*
-         * Copy the 48bytes defined by the software first into the xstate
-         * memory layout in the thread struct, so that we can copy the entire
-         * xstateregs to the user using one user_regset_copyout().
-         */
-        memcpy(&xsave->i387.sw_reserved,
-                xstate_fx_sw_bytes, sizeof(xstate_fx_sw_bytes));
-        /*
-         * Copy the xstate memory layout.
-         */
-        ret = user_regset_copyout(&pos, &count, &kbuf, &ubuf, xsave, 0, -1);
+        fpu__activate_fpstate_read(fpu);
+
+        if (using_compacted_format()) {
+                ret = copyout_from_xsaves(pos, count, kbuf, ubuf, xsave);
+        } else {
+                fpstate_sanitize_xstate(fpu);
+                /*
+                 * Copy the 48 bytes defined by the software into the xsave
+                 * area in the thread struct, so that we can copy the whole
+                 * area to user using one user_regset_copyout().
+                 */
+                memcpy(&xsave->i387.sw_reserved, xstate_fx_sw_bytes, sizeof(xstate_fx_sw_bytes));
+
+                /*
+                 * Copy the xstate memory layout.
+                 */
+                ret = user_regset_copyout(&pos, &count, &kbuf, &ubuf, xsave, 0, -1);
+        }
         return ret;
 }
 
@@ -114,11 +120,27 @@ int xstateregs_set(struct task_struct *target, const struct user_regset *regset,
         if (!boot_cpu_has(X86_FEATURE_XSAVE))
                 return -ENODEV;
 
-        fpu__activate_fpstate_write(fpu);
+        /*
+         * A whole standard-format XSAVE buffer is needed:
+         */
+        if ((pos != 0) || (count < fpu_user_xstate_size))
+                return -EFAULT;
 
         xsave = &fpu->state.xsave;
 
-        ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf, xsave, 0, -1);
+        fpu__activate_fpstate_write(fpu);
+
+        if (boot_cpu_has(X86_FEATURE_XSAVES))
+                ret = copyin_to_xsaves(kbuf, ubuf, xsave);
+        else
+                ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf, xsave, 0, -1);
+
+        /*
+         * In case of failure, mark all states as init:
+         */
+        if (ret)
+                fpstate_init(&fpu->state);
+
         /*
          * mxcsr reserved bits must be masked to zero for security reasons.
          */
diff --git a/arch/x86/kernel/fpu/signal.c b/arch/x86/kernel/fpu/signal.c
index 31c6a60505e6..9e231d88bb33 100644
--- a/arch/x86/kernel/fpu/signal.c
+++ b/arch/x86/kernel/fpu/signal.c
@@ -8,8 +8,10 @@
 #include <asm/fpu/internal.h>
 #include <asm/fpu/signal.h>
 #include <asm/fpu/regset.h>
+#include <asm/fpu/xstate.h>
 
 #include <asm/sigframe.h>
+#include <asm/trace/fpu.h>
 
 static struct _fpx_sw_bytes fx_sw_reserved, fx_sw_reserved_ia32;
 
@@ -31,7 +33,7 @@ static inline int check_for_xstate(struct fxregs_state __user *buf,
         /* Check for the first magic field and other error scenarios. */
         if (fx_sw->magic1 != FP_XSTATE_MAGIC1 ||
             fx_sw->xstate_size < min_xstate_size ||
-            fx_sw->xstate_size > xstate_size ||
+            fx_sw->xstate_size > fpu_user_xstate_size ||
             fx_sw->xstate_size > fx_sw->extended_size)
                 return -1;
 
@@ -88,7 +90,8 @@ static inline int save_xstate_epilog(void __user *buf, int ia32_frame)
         if (!use_xsave())
                 return err;
 
-        err |= __put_user(FP_XSTATE_MAGIC2, (__u32 *)(buf + xstate_size));
+        err |= __put_user(FP_XSTATE_MAGIC2,
+                          (__u32 *)(buf + fpu_user_xstate_size));
 
         /*
          * Read the xfeatures which we copied (directly from the cpu or
@@ -125,7 +128,7 @@ static inline int copy_fpregs_to_sigframe(struct xregs_state __user *buf)
         else
                 err = copy_fregs_to_user((struct fregs_state __user *) buf);
 
-        if (unlikely(err) && __clear_user(buf, xstate_size))
+        if (unlikely(err) && __clear_user(buf, fpu_user_xstate_size))
                 err = -EFAULT;
         return err;
 }
@@ -167,7 +170,7 @@ int copy_fpstate_to_sigframe(void __user *buf, void __user *buf_fx, int size)
                         sizeof(struct user_i387_ia32_struct), NULL,
                         (struct _fpstate_32 __user *) buf) ? -1 : 1;
 
-        if (fpregs_active()) {
+        if (fpregs_active() || using_compacted_format()) {
                 /* Save the live register state to the user directly. */
                 if (copy_fpregs_to_sigframe(buf_fx))
                         return -1;
@@ -175,8 +178,19 @@ int copy_fpstate_to_sigframe(void __user *buf, void __user *buf_fx, int size)
                 if (ia32_fxstate)
                         copy_fxregs_to_kernel(&tsk->thread.fpu);
         } else {
+                /*
+                 * It is a *bug* if kernel uses compacted-format for xsave
+                 * area and we copy it out directly to a signal frame. It
+                 * should have been handled above by saving the registers
+                 * directly.
+                 */
+                if (boot_cpu_has(X86_FEATURE_XSAVES)) {
+                        WARN_ONCE(1, "x86/fpu: saving compacted-format xsave area to a signal frame!\n");
+                        return -1;
+                }
+
                 fpstate_sanitize_xstate(&tsk->thread.fpu);
-                if (__copy_to_user(buf_fx, xsave, xstate_size))
+                if (__copy_to_user(buf_fx, xsave, fpu_user_xstate_size))
                         return -1;
         }
 
@@ -250,7 +264,7 @@ static int __fpu__restore_sig(void __user *buf, void __user *buf_fx, int size)
         int ia32_fxstate = (buf != buf_fx);
         struct task_struct *tsk = current;
         struct fpu *fpu = &tsk->thread.fpu;
-        int state_size = xstate_size;
+        int state_size = fpu_kernel_xstate_size;
         u64 xfeatures = 0;
         int fx_only = 0;
 
@@ -282,6 +296,7 @@ static int __fpu__restore_sig(void __user *buf, void __user *buf_fx, int size)
                          */
                         state_size = sizeof(struct fxregs_state);
                         fx_only = 1;
+                        trace_x86_fpu_xstate_check_failed(fpu);
                 } else {
                         state_size = fx_sw_user.xstate_size;
                         xfeatures = fx_sw_user.xfeatures;
@@ -308,9 +323,17 @@ static int __fpu__restore_sig(void __user *buf, void __user *buf_fx, int size)
                  */
                 fpu__drop(fpu);
 
-                if (__copy_from_user(&fpu->state.xsave, buf_fx, state_size) ||
-                    __copy_from_user(&env, buf, sizeof(env))) {
+                if (using_compacted_format()) {
+                        err = copyin_to_xsaves(NULL, buf_fx,
+                                               &fpu->state.xsave);
+                } else {
+                        err = __copy_from_user(&fpu->state.xsave,
+                                               buf_fx, state_size);
+                }
+
+                if (err || __copy_from_user(&env, buf, sizeof(env))) {
                         fpstate_init(&fpu->state);
+                        trace_x86_fpu_init_state(fpu);
                         err = -1;
                 } else {
                         sanitize_restored_xstate(tsk, &env, xfeatures, fx_only);
@@ -341,7 +364,8 @@ static int __fpu__restore_sig(void __user *buf, void __user *buf_fx, int size)
 
 static inline int xstate_sigframe_size(void)
 {
-        return use_xsave() ? xstate_size + FP_XSTATE_MAGIC2_SIZE : xstate_size;
+        return use_xsave() ? fpu_user_xstate_size + FP_XSTATE_MAGIC2_SIZE :
+                        fpu_user_xstate_size;
 }
 
 /*
@@ -385,12 +409,12 @@ fpu__alloc_mathframe(unsigned long sp, int ia32_frame,
  */
 void fpu__init_prepare_fx_sw_frame(void)
 {
-        int size = xstate_size + FP_XSTATE_MAGIC2_SIZE;
+        int size = fpu_user_xstate_size + FP_XSTATE_MAGIC2_SIZE;
 
         fx_sw_reserved.magic1 = FP_XSTATE_MAGIC1;
         fx_sw_reserved.extended_size = size;
         fx_sw_reserved.xfeatures = xfeatures_mask;
-        fx_sw_reserved.xstate_size = xstate_size;
+        fx_sw_reserved.xstate_size = fpu_user_xstate_size;
 
         if (config_enabled(CONFIG_IA32_EMULATION) ||
             config_enabled(CONFIG_X86_32)) {
diff --git a/arch/x86/kernel/fpu/xstate.c b/arch/x86/kernel/fpu/xstate.c
index 4ea2a59483c7..680049aa4593 100644
--- a/arch/x86/kernel/fpu/xstate.c
+++ b/arch/x86/kernel/fpu/xstate.c
@@ -11,6 +11,7 @@
 #include <asm/fpu/internal.h>
 #include <asm/fpu/signal.h>
 #include <asm/fpu/regset.h>
+#include <asm/fpu/xstate.h>
 
 #include <asm/tlbflush.h>
 
@@ -44,6 +45,13 @@ static unsigned int xstate_sizes[XFEATURE_MAX]   = { [ 0 ... XFEATURE_MAX - 1] =
 static unsigned int xstate_comp_offsets[sizeof(xfeatures_mask)*8];
 
 /*
+ * The XSAVE area of kernel can be in standard or compacted format;
+ * it is always in standard format for user mode. This is the user
+ * mode standard format size used for signal and ptrace frames.
+ */
+unsigned int fpu_user_xstate_size;
+
+/*
  * Clear all of the X86_FEATURE_* bits that are unavailable
  * when the CPU has no XSAVE support.
  */
@@ -105,6 +113,27 @@ int cpu_has_xfeatures(u64 xfeatures_needed, const char **feature_name)
 }
 EXPORT_SYMBOL_GPL(cpu_has_xfeatures);
 
+static int xfeature_is_supervisor(int xfeature_nr)
+{
+        /*
+         * We currently do not support supervisor states, but if
+         * we did, we could find out like this.
+         *
+         * SDM says: If state component 'i' is a user state component,
+         * ECX[0] return 0; if state component i is a supervisor
+         * state component, ECX[0] returns 1.
+         */
+        u32 eax, ebx, ecx, edx;
+
+        cpuid_count(XSTATE_CPUID, xfeature_nr, &eax, &ebx, &ecx, &edx);
+        return !!(ecx & 1);
+}
+
+static int xfeature_is_user(int xfeature_nr)
+{
+        return !xfeature_is_supervisor(xfeature_nr);
+}
+
 /*
  * When executing XSAVEOPT (or other optimized XSAVE instructions), if
  * a processor implementation detects that an FPU state component is still
@@ -171,7 +200,7 @@ void fpstate_sanitize_xstate(struct fpu *fpu)
          */
         while (xfeatures) {
                 if (xfeatures & 0x1) {
-                        int offset = xstate_offsets[feature_bit];
+                        int offset = xstate_comp_offsets[feature_bit];
                         int size = xstate_sizes[feature_bit];
 
                         memcpy((void *)fx + offset,
@@ -192,6 +221,15 @@ void fpu__init_cpu_xstate(void)
 {
         if (!boot_cpu_has(X86_FEATURE_XSAVE) || !xfeatures_mask)
                 return;
+        /*
+         * Make it clear that XSAVES supervisor states are not yet
+         * implemented should anyone expect it to work by changing
+         * bits in XFEATURE_MASK_* macros and XCR0.
+         */
+        WARN_ONCE((xfeatures_mask & XFEATURE_MASK_SUPERVISOR),
+                "x86/fpu: XSAVES supervisor states are not yet implemented.\n");
+
+        xfeatures_mask &= ~XFEATURE_MASK_SUPERVISOR;
 
         cr4_set_bits(X86_CR4_OSXSAVE);
         xsetbv(XCR_XFEATURE_ENABLED_MASK, xfeatures_mask);
@@ -217,13 +255,29 @@ static void __init setup_xstate_features(void)
         /* start at the beginnning of the "extended state" */
         unsigned int last_good_offset = offsetof(struct xregs_state,
                                                  extended_state_area);
+        /*
+         * The FP xstates and SSE xstates are legacy states. They are always
+         * in the fixed offsets in the xsave area in either compacted form
+         * or standard form.
+         */
+        xstate_offsets[0] = 0;
+        xstate_sizes[0] = offsetof(struct fxregs_state, xmm_space);
+        xstate_offsets[1] = xstate_sizes[0];
+        xstate_sizes[1] = FIELD_SIZEOF(struct fxregs_state, xmm_space);
 
         for (i = FIRST_EXTENDED_XFEATURE; i < XFEATURE_MAX; i++) {
                 if (!xfeature_enabled(i))
                         continue;
 
                 cpuid_count(XSTATE_CPUID, i, &eax, &ebx, &ecx, &edx);
-                xstate_offsets[i] = ebx;
+
+                /*
+                 * If an xfeature is supervisor state, the offset
+                 * in EBX is invalid. We leave it to -1.
+                 */
+                if (xfeature_is_user(i))
+                        xstate_offsets[i] = ebx;
+
                 xstate_sizes[i] = eax;
                 /*
                  * In our xstate size checks, we assume that the
@@ -233,8 +287,6 @@ static void __init setup_xstate_features(void)
                 WARN_ONCE(last_good_offset > xstate_offsets[i],
                         "x86/fpu: misordered xstate at %d\n", last_good_offset);
                 last_good_offset = xstate_offsets[i];
-
-                printk(KERN_INFO "x86/fpu: xstate_offset[%d]: %4d, xstate_sizes[%d]: %4d\n", i, ebx, i, eax);
         }
 }
 
@@ -263,6 +315,33 @@ static void __init print_xstate_features(void)
 }
 
 /*
+ * This check is important because it is easy to get XSTATE_*
+ * confused with XSTATE_BIT_*.
+ */
+#define CHECK_XFEATURE(nr) do {         \
+        WARN_ON(nr < FIRST_EXTENDED_XFEATURE);  \
+        WARN_ON(nr >= XFEATURE_MAX);    \
+} while (0)
+
+/*
+ * We could cache this like xstate_size[], but we only use
+ * it here, so it would be a waste of space.
+ */
+static int xfeature_is_aligned(int xfeature_nr)
+{
+        u32 eax, ebx, ecx, edx;
+
+        CHECK_XFEATURE(xfeature_nr);
+        cpuid_count(XSTATE_CPUID, xfeature_nr, &eax, &ebx, &ecx, &edx);
+        /*
+         * The value returned by ECX[1] indicates the alignment
+         * of state component 'i' when the compacted format
+         * of the extended region of an XSAVE area is used:
+         */
+        return !!(ecx & 2);
+}
+
+/*
  * This function sets up offsets and sizes of all extended states in
  * xsave area. This supports both standard format and compacted format
  * of the xsave aread.
@@ -299,10 +378,29 @@ static void __init setup_xstate_comp(void)
                 else
                         xstate_comp_sizes[i] = 0;
 
-                if (i > FIRST_EXTENDED_XFEATURE)
+                if (i > FIRST_EXTENDED_XFEATURE) {
                         xstate_comp_offsets[i] = xstate_comp_offsets[i-1]
                                         + xstate_comp_sizes[i-1];
 
+                        if (xfeature_is_aligned(i))
+                                xstate_comp_offsets[i] =
+                                        ALIGN(xstate_comp_offsets[i], 64);
+                }
+        }
+}
+
+/*
+ * Print out xstate component offsets and sizes
+ */
+static void __init print_xstate_offset_size(void)
+{
+        int i;
+
+        for (i = FIRST_EXTENDED_XFEATURE; i < XFEATURE_MAX; i++) {
+                if (!xfeature_enabled(i))
+                        continue;
+                pr_info("x86/fpu: xstate_offset[%d]: %4d, xstate_sizes[%d]: %4d\n",
+                         i, xstate_comp_offsets[i], i, xstate_sizes[i]);
         }
 }
 
@@ -322,13 +420,11 @@ static void __init setup_init_fpu_buf(void)
         setup_xstate_features();
         print_xstate_features();
 
-        if (boot_cpu_has(X86_FEATURE_XSAVES)) {
+        if (boot_cpu_has(X86_FEATURE_XSAVES))
                 init_fpstate.xsave.header.xcomp_bv = (u64)1 << 63 | xfeatures_mask;
-                init_fpstate.xsave.header.xfeatures = xfeatures_mask;
-        }
 
         /*
-         * Init all the features state with header_bv being 0x0
+         * Init all the features state with header.xfeatures being 0x0
          */
         copy_kernel_to_xregs_booting(&init_fpstate.xsave);
 
@@ -339,58 +435,19 @@ static void __init setup_init_fpu_buf(void)
         copy_xregs_to_kernel_booting(&init_fpstate.xsave);
 }
 
-static int xfeature_is_supervisor(int xfeature_nr)
-{
-        /*
-         * We currently do not support supervisor states, but if
-         * we did, we could find out like this.
-         *
-         * SDM says: If state component i is a user state component,
-         * ECX[0] return 0; if state component i is a supervisor
-         * state component, ECX[0] returns 1.
-        u32 eax, ebx, ecx, edx;
-        cpuid_count(XSTATE_CPUID, xfeature_nr, &eax, &ebx, &ecx, &edx;
-        return !!(ecx & 1);
-        */
-        return 0;
-}
-/*
-static int xfeature_is_user(int xfeature_nr)
-{
-        return !xfeature_is_supervisor(xfeature_nr);
-}
-*/
-
-/*
- * This check is important because it is easy to get XSTATE_*
- * confused with XSTATE_BIT_*.
- */
-#define CHECK_XFEATURE(nr) do {         \
-        WARN_ON(nr < FIRST_EXTENDED_XFEATURE);  \
-        WARN_ON(nr >= XFEATURE_MAX);    \
-} while (0)
-
-/*
- * We could cache this like xstate_size[], but we only use
- * it here, so it would be a waste of space.
- */
-static int xfeature_is_aligned(int xfeature_nr)
+static int xfeature_uncompacted_offset(int xfeature_nr)
 {
         u32 eax, ebx, ecx, edx;
 
-        CHECK_XFEATURE(xfeature_nr);
-        cpuid_count(XSTATE_CPUID, xfeature_nr, &eax, &ebx, &ecx, &edx);
         /*
-         * The value returned by ECX[1] indicates the alignment
-         * of state component i when the compacted format
-         * of the extended region of an XSAVE area is used
+         * Only XSAVES supports supervisor states and it uses compacted
+         * format. Checking a supervisor state's uncompacted offset is
+         * an error.
          */
-        return !!(ecx & 2);
-}
-
-static int xfeature_uncompacted_offset(int xfeature_nr)
-{
-        u32 eax, ebx, ecx, edx;
+        if (XFEATURE_MASK_SUPERVISOR & (1 << xfeature_nr)) {
+                WARN_ONCE(1, "No fixed offset for xstate %d\n", xfeature_nr);
+                return -1;
+        }
 
         CHECK_XFEATURE(xfeature_nr);
         cpuid_count(XSTATE_CPUID, xfeature_nr, &eax, &ebx, &ecx, &edx);
@@ -415,7 +472,7 @@ static int xfeature_size(int xfeature_nr)
  * that it is obvious which aspect of 'XSAVES' is being handled
  * by the calling code.
  */
-static int using_compacted_format(void)
+int using_compacted_format(void)
 {
         return boot_cpu_has(X86_FEATURE_XSAVES);
 }
@@ -530,11 +587,12 @@ static void do_extra_xstate_size_checks(void)
                  */
                 paranoid_xstate_size += xfeature_size(i);
         }
-        XSTATE_WARN_ON(paranoid_xstate_size != xstate_size);
+        XSTATE_WARN_ON(paranoid_xstate_size != fpu_kernel_xstate_size);
 }
 
+
 /*
- * Calculate total size of enabled xstates in XCR0/xfeatures_mask.
+ * Get total size of enabled xstates in XCR0/xfeatures_mask.
  *
  * Note the SDM's wording here.  "sub-function 0" only enumerates
  * the size of the *user* states.  If we use it to size a buffer
@@ -544,34 +602,33 @@ static void do_extra_xstate_size_checks(void)
  * Note that we do not currently set any bits on IA32_XSS so
  * 'XCR0 | IA32_XSS == XCR0' for now.
  */
-static unsigned int __init calculate_xstate_size(void)
+static unsigned int __init get_xsaves_size(void)
 {
         unsigned int eax, ebx, ecx, edx;
-        unsigned int calculated_xstate_size;
+        /*
+         * - CPUID function 0DH, sub-function 1:
+         *    EBX enumerates the size (in bytes) required by
+         *    the XSAVES instruction for an XSAVE area
+         *    containing all the state components
+         *    corresponding to bits currently set in
+         *    XCR0 | IA32_XSS.
+         */
+        cpuid_count(XSTATE_CPUID, 1, &eax, &ebx, &ecx, &edx);
+        return ebx;
+}
 
-        if (!boot_cpu_has(X86_FEATURE_XSAVES)) {
-                /*
-                 * - CPUID function 0DH, sub-function 0:
-                 *    EBX enumerates the size (in bytes) required by
-                 *    the XSAVE instruction for an XSAVE area
-                 *    containing all the *user* state components
-                 *    corresponding to bits currently set in XCR0.
-                 */
-                cpuid_count(XSTATE_CPUID, 0, &eax, &ebx, &ecx, &edx);
-                calculated_xstate_size = ebx;
-        } else {
-                /*
-                 * - CPUID function 0DH, sub-function 1:
-                 *    EBX enumerates the size (in bytes) required by
-                 *    the XSAVES instruction for an XSAVE area
-                 *    containing all the state components
-                 *    corresponding to bits currently set in
-                 *    XCR0 | IA32_XSS.
-                 */
-                cpuid_count(XSTATE_CPUID, 1, &eax, &ebx, &ecx, &edx);
-                calculated_xstate_size = ebx;
-        }
-        return calculated_xstate_size;
+static unsigned int __init get_xsave_size(void)
+{
+        unsigned int eax, ebx, ecx, edx;
+        /*
+         * - CPUID function 0DH, sub-function 0:
+         *    EBX enumerates the size (in bytes) required by
+         *    the XSAVE instruction for an XSAVE area
+         *    containing all the *user* state components
+         *    corresponding to bits currently set in XCR0.
+         */
+        cpuid_count(XSTATE_CPUID, 0, &eax, &ebx, &ecx, &edx);
+        return ebx;
 }
 
 /*
@@ -591,7 +648,15 @@ static bool is_supported_xstate_size(unsigned int test_xstate_size)
 static int init_xstate_size(void)
 {
         /* Recompute the context size for enabled features: */
-        unsigned int possible_xstate_size = calculate_xstate_size();
+        unsigned int possible_xstate_size;
+        unsigned int xsave_size;
+
+        xsave_size = get_xsave_size();
+
+        if (boot_cpu_has(X86_FEATURE_XSAVES))
+                possible_xstate_size = get_xsaves_size();
+        else
+                possible_xstate_size = xsave_size;
 
         /* Ensure we have the space to store all enabled: */
         if (!is_supported_xstate_size(possible_xstate_size))
@@ -601,8 +666,13 @@ static int init_xstate_size(void)
          * The size is OK, we are definitely going to use xsave,
          * make it known to the world that we need more space.
          */
-        xstate_size = possible_xstate_size;
+        fpu_kernel_xstate_size = possible_xstate_size;
         do_extra_xstate_size_checks();
+
+        /*
+         * User space is always in standard format.
+         */
+        fpu_user_xstate_size = xsave_size;
         return 0;
 }
 
@@ -644,8 +714,13 @@ void __init fpu__init_system_xstate(void)
         xfeatures_mask = eax + ((u64)edx << 32);
 
         if ((xfeatures_mask & XFEATURE_MASK_FPSSE) != XFEATURE_MASK_FPSSE) {
+                /*
+                 * This indicates that something really unexpected happened
+                 * with the enumeration.  Disable XSAVE and try to continue
+                 * booting without it.  This is too early to BUG().
+                 */
                 pr_err("x86/fpu: FP/SSE not present amongst the CPU's xstate features: 0x%llx.\n", xfeatures_mask);
-                BUG();
+                goto out_disable;
         }
 
         xfeatures_mask &= fpu__get_supported_xfeatures_mask();
@@ -653,21 +728,29 @@ void __init fpu__init_system_xstate(void)
         /* Enable xstate instructions to be able to continue with initialization: */
         fpu__init_cpu_xstate();
         err = init_xstate_size();
-        if (err) {
-                /* something went wrong, boot without any XSAVE support */
-                fpu__init_disable_system_xstate();
-                return;
-        }
+        if (err)
+                goto out_disable;
+
+        /*
+         * Update info used for ptrace frames; use standard-format size and no
+         * supervisor xstates:
+         */
+        update_regset_xstate_info(fpu_user_xstate_size, xfeatures_mask & ~XFEATURE_MASK_SUPERVISOR);
 
-        update_regset_xstate_info(xstate_size, xfeatures_mask);
         fpu__init_prepare_fx_sw_frame();
         setup_init_fpu_buf();
         setup_xstate_comp();
+        print_xstate_offset_size();
 
         pr_info("x86/fpu: Enabled xstate features 0x%llx, context size is %d bytes, using '%s' format.\n",
                 xfeatures_mask,
-                xstate_size,
+                fpu_kernel_xstate_size,
                 boot_cpu_has(X86_FEATURE_XSAVES) ? "compacted" : "standard");
+        return;
+
+out_disable:
+        /* something went wrong, try to boot without any XSAVE support */
+        fpu__init_disable_system_xstate();
 }
 
 /*
@@ -693,6 +776,11 @@ void *__raw_xsave_addr(struct xregs_state *xsave, int xstate_feature_mask)
 {
         int feature_nr = fls64(xstate_feature_mask) - 1;
 
+        if (!xfeature_enabled(feature_nr)) {
+                WARN_ON_FPU(1);
+                return NULL;
+        }
+
         return (void *)xsave + xstate_comp_offsets[feature_nr];
 }
 /*
@@ -887,16 +975,16 @@ int arch_set_user_pkey_access(struct task_struct *tsk, int pkey,
         if (!boot_cpu_has(X86_FEATURE_OSPKE))
                 return -EINVAL;
 
-        /* Set the bits we need in PKRU  */
+        /* Set the bits we need in PKRU:  */
         if (init_val & PKEY_DISABLE_ACCESS)
                 new_pkru_bits |= PKRU_AD_BIT;
         if (init_val & PKEY_DISABLE_WRITE)
                 new_pkru_bits |= PKRU_WD_BIT;
 
-        /* Shift the bits in to the correct place in PKRU for pkey. */
+        /* Shift the bits in to the correct place in PKRU for pkey: */
         new_pkru_bits <<= pkey_shift;
 
-        /* Locate old copy of the state in the xsave buffer */
+        /* Locate old copy of the state in the xsave buffer: */
         old_pkru_state = get_xsave_addr(xsave, XFEATURE_MASK_PKRU);
 
         /*
@@ -909,9 +997,10 @@ int arch_set_user_pkey_access(struct task_struct *tsk, int pkey,
         else
                 new_pkru_state.pkru = old_pkru_state->pkru;
 
-        /* mask off any old bits in place */
+        /* Mask off any old bits in place: */
         new_pkru_state.pkru &= ~((PKRU_AD_BIT|PKRU_WD_BIT) << pkey_shift);
-        /* Set the newly-requested bits */
+
+        /* Set the newly-requested bits: */
         new_pkru_state.pkru |= new_pkru_bits;
 
         /*
@@ -925,8 +1014,168 @@ int arch_set_user_pkey_access(struct task_struct *tsk, int pkey,
          */
         new_pkru_state.pad = 0;
 
-        fpu__xfeature_set_state(XFEATURE_MASK_PKRU, &new_pkru_state,
-                        sizeof(new_pkru_state));
+        fpu__xfeature_set_state(XFEATURE_MASK_PKRU, &new_pkru_state, sizeof(new_pkru_state));
+
+        return 0;
+}
+
+/*
+ * This is similar to user_regset_copyout(), but will not add offset to
+ * the source data pointer or increment pos, count, kbuf, and ubuf.
+ */
+static inline int xstate_copyout(unsigned int pos, unsigned int count,
+                                 void *kbuf, void __user *ubuf,
+                                 const void *data, const int start_pos,
+                                 const int end_pos)
+{
+        if ((count == 0) || (pos < start_pos))
+                return 0;
+
+        if (end_pos < 0 || pos < end_pos) {
+                unsigned int copy = (end_pos < 0 ? count : min(count, end_pos - pos));
+
+                if (kbuf) {
+                        memcpy(kbuf + pos, data, copy);
+                } else {
+                        if (__copy_to_user(ubuf + pos, data, copy))
+                                return -EFAULT;
+                }
+        }
+        return 0;
+}
+
+/*
+ * Convert from kernel XSAVES compacted format to standard format and copy
+ * to a ptrace buffer. It supports partial copy but pos always starts from
+ * zero. This is called from xstateregs_get() and there we check the CPU
+ * has XSAVES.
+ */
+int copyout_from_xsaves(unsigned int pos, unsigned int count, void *kbuf,
+                        void __user *ubuf, struct xregs_state *xsave)
+{
+        unsigned int offset, size;
+        int ret, i;
+        struct xstate_header header;
+
+        /*
+         * Currently copy_regset_to_user() starts from pos 0:
+         */
+        if (unlikely(pos != 0))
+                return -EFAULT;
+
+        /*
+         * The destination is a ptrace buffer; we put in only user xstates:
+         */
+        memset(&header, 0, sizeof(header));
+        header.xfeatures = xsave->header.xfeatures;
+        header.xfeatures &= ~XFEATURE_MASK_SUPERVISOR;
+
+        /*
+         * Copy xregs_state->header:
+         */
+        offset = offsetof(struct xregs_state, header);
+        size = sizeof(header);
+
+        ret = xstate_copyout(offset, size, kbuf, ubuf, &header, 0, count);
+
+        if (ret)
+                return ret;
+
+        for (i = 0; i < XFEATURE_MAX; i++) {
+                /*
+                 * Copy only in-use xstates:
+                 */
+                if ((header.xfeatures >> i) & 1) {
+                        void *src = __raw_xsave_addr(xsave, 1 << i);
+
+                        offset = xstate_offsets[i];
+                        size = xstate_sizes[i];
+
+                        ret = xstate_copyout(offset, size, kbuf, ubuf, src, 0, count);
+
+                        if (ret)
+                                return ret;
+
+                        if (offset + size >= count)
+                                break;
+                }
+
+        }
+
+        /*
+         * Fill xsave->i387.sw_reserved value for ptrace frame:
+         */
+        offset = offsetof(struct fxregs_state, sw_reserved);
+        size = sizeof(xstate_fx_sw_bytes);
+
+        ret = xstate_copyout(offset, size, kbuf, ubuf, xstate_fx_sw_bytes, 0, count);
+
+        if (ret)
+                return ret;
+
+        return 0;
+}
+
+/*
+ * Convert from a ptrace standard-format buffer to kernel XSAVES format
+ * and copy to the target thread. This is called from xstateregs_set() and
+ * there we check the CPU has XSAVES and a whole standard-sized buffer
+ * exists.
+ */
+int copyin_to_xsaves(const void *kbuf, const void __user *ubuf,
+                     struct xregs_state *xsave)
+{
+        unsigned int offset, size;
+        int i;
+        u64 xfeatures;
+        u64 allowed_features;
+
+        offset = offsetof(struct xregs_state, header);
+        size = sizeof(xfeatures);
+
+        if (kbuf) {
+                memcpy(&xfeatures, kbuf + offset, size);
+        } else {
+                if (__copy_from_user(&xfeatures, ubuf + offset, size))
+                        return -EFAULT;
+        }
+
+        /*
+         * Reject if the user sets any disabled or supervisor features:
+         */
+        allowed_features = xfeatures_mask & ~XFEATURE_MASK_SUPERVISOR;
+
+        if (xfeatures & ~allowed_features)
+                return -EINVAL;
+
+        for (i = 0; i < XFEATURE_MAX; i++) {
+                u64 mask = ((u64)1 << i);
+
+                if (xfeatures & mask) {
+                        void *dst = __raw_xsave_addr(xsave, 1 << i);
+
+                        offset = xstate_offsets[i];
+                        size = xstate_sizes[i];
+
+                        if (kbuf) {
+                                memcpy(dst, kbuf + offset, size);
+                        } else {
+                                if (__copy_from_user(dst, ubuf + offset, size))
+                                        return -EFAULT;
+                        }
+                }
+        }
+
+        /*
+         * The state that came in from userspace was user-state only.
+         * Mask all the user states out of 'xfeatures':
+         */
+        xsave->header.xfeatures &= XFEATURE_MASK_SUPERVISOR;
+
+        /*
+         * Add back in the features that came in from userspace:
+         */
+        xsave->header.xfeatures |= xfeatures;
 
         return 0;
 }