3 files changed, 77 insertions, 29 deletions

diff --git a/arch/arm/vfp/entry.S b/arch/arm/vfp/entry.S
index 4fa9903b83c..c1a97840258 100644
--- a/arch/arm/vfp/entry.S
+++ b/arch/arm/vfp/entry.S
@@ -10,7 +10,7 @@
  *
  * Basic entry code, called from the kernel's undefined instruction trap.
  *  r0  = faulted instruction
- *  r5  = faulted PC+4
+ *  r2  = faulted PC+4
  *  r9  = successful return
  *  r10 = thread_info structure
  *  lr  = failure return
@@ -26,6 +26,7 @@ ENTRY(do_vfp)
         str     r11, [r10, #TI_PREEMPT]
 #endif
         enable_irq
+        str     r2, [sp, #S_PC]         @ update regs->ARM_pc for Thumb 2 case
         ldr     r4, .LCvfp
         ldr     r11, [r10, #TI_CPU]     @ CPU number
         add     r10, r10, #TI_VFPSTATE  @ r10 = workspace
diff --git a/arch/arm/vfp/vfphw.S b/arch/arm/vfp/vfphw.S
index 9897dcfc16d..404538ae591 100644
--- a/arch/arm/vfp/vfphw.S
+++ b/arch/arm/vfp/vfphw.S
@@ -77,15 +77,12 @@ ENTRY(vfp_support_entry)
         bne     look_for_VFP_exceptions @ VFP is already enabled
 
         DBGSTR1 "enable %x", r10
-        ldr     r3, last_VFP_context_address
+        ldr     r3, vfp_current_hw_state_address
         orr     r1, r1, #FPEXC_EN       @ user FPEXC has the enable bit set
-        ldr     r4, [r3, r11, lsl #2]   @ last_VFP_context pointer
+        ldr     r4, [r3, r11, lsl #2]   @ vfp_current_hw_state pointer
         bic     r5, r1, #FPEXC_EX       @ make sure exceptions are disabled
-        cmp     r4, r10
-        beq     check_for_exception     @ we are returning to the same
-                                        @ process, so the registers are
-                                        @ still there.  In this case, we do
-                                        @ not want to drop a pending exception.
+        cmp     r4, r10                 @ this thread owns the hw context?
+        beq     vfp_hw_state_valid
 
         VFPFMXR FPEXC, r5               @ enable VFP, disable any pending
                                         @ exceptions, so we can get at the
@@ -116,7 +113,7 @@ ENTRY(vfp_support_entry)
 
 no_old_VFP_process:
         DBGSTR1 "load state %p", r10
-        str     r10, [r3, r11, lsl #2]  @ update the last_VFP_context pointer
+        str     r10, [r3, r11, lsl #2]  @ update the vfp_current_hw_state pointer
                                         @ Load the saved state back into the VFP
         VFPFLDMIA r10, r5               @ reload the working registers while
                                         @ FPEXC is in a safe state
@@ -132,7 +129,8 @@ no_old_VFP_process:
 #endif
         VFPFMXR FPSCR, r5               @ restore status
 
-check_for_exception:
+@ The context stored in the VFP hardware is up to date with this thread
+vfp_hw_state_valid:
         tst     r1, #FPEXC_EX
         bne     process_exception       @ might as well handle the pending
                                         @ exception before retrying branch
@@ -207,8 +205,8 @@ ENTRY(vfp_save_state)
 ENDPROC(vfp_save_state)
 
         .align
-last_VFP_context_address:
-        .word   last_VFP_context
+vfp_current_hw_state_address:
+        .word   vfp_current_hw_state
 
         .macro  tbl_branch, base, tmp, shift
 #ifdef CONFIG_THUMB2_KERNEL
diff --git a/arch/arm/vfp/vfpmodule.c b/arch/arm/vfp/vfpmodule.c
index f25e7ec8941..e381dc68505 100644
--- a/arch/arm/vfp/vfpmodule.c
+++ b/arch/arm/vfp/vfpmodule.c
@@ -21,6 +21,7 @@
 #include <asm/cputype.h>
 #include <asm/thread_notify.h>
 #include <asm/vfp.h>
+#include <asm/cpu_pm.h>
 
 #include "vfpinstr.h"
 #include "vfp.h"
@@ -33,7 +34,13 @@ void vfp_support_entry(void);
 void vfp_null_entry(void);
 
 void (*vfp_vector)(void) = vfp_null_entry;
-union vfp_state *last_VFP_context[NR_CPUS];
+
+/*
+ * The pointer to the vfpstate structure of the thread which currently
+ * owns the context held in the VFP hardware, or NULL if the hardware
+ * context is invalid.
+ */
+union vfp_state *vfp_current_hw_state[NR_CPUS];
 
 /*
  * Dual-use variable.
@@ -57,12 +64,12 @@ static void vfp_thread_flush(struct thread_info *thread)
 
         /*
          * Disable VFP to ensure we initialize it first.  We must ensure
-         * that the modification of last_VFP_context[] and hardware disable
+         * that the modification of vfp_current_hw_state[] and hardware disable
          * are done for the same CPU and without preemption.
          */
         cpu = get_cpu();
-        if (last_VFP_context[cpu] == vfp)
-                last_VFP_context[cpu] = NULL;
+        if (vfp_current_hw_state[cpu] == vfp)
+                vfp_current_hw_state[cpu] = NULL;
         fmxr(FPEXC, fmrx(FPEXC) & ~FPEXC_EN);
         put_cpu();
 }
@@ -73,8 +80,8 @@ static void vfp_thread_exit(struct thread_info *thread)
         union vfp_state *vfp = &thread->vfpstate;
         unsigned int cpu = get_cpu();
 
-        if (last_VFP_context[cpu] == vfp)
-                last_VFP_context[cpu] = NULL;
+        if (vfp_current_hw_state[cpu] == vfp)
+                vfp_current_hw_state[cpu] = NULL;
         put_cpu();
 }
 
@@ -129,9 +136,9 @@ static int vfp_notifier(struct notifier_block *self, unsigned long cmd, void *v)
                  * case the thread migrates to a different CPU. The
                  * restoring is done lazily.
                  */
-                if ((fpexc & FPEXC_EN) && last_VFP_context[cpu]) {
-                        vfp_save_state(last_VFP_context[cpu], fpexc);
-                        last_VFP_context[cpu]->hard.cpu = cpu;
+                if ((fpexc & FPEXC_EN) && vfp_current_hw_state[cpu]) {
+                        vfp_save_state(vfp_current_hw_state[cpu], fpexc);
+                        vfp_current_hw_state[cpu]->hard.cpu = cpu;
                 }
                 /*
                  * Thread migration, just force the reloading of the
@@ -139,7 +146,7 @@ static int vfp_notifier(struct notifier_block *self, unsigned long cmd, void *v)
                  * contain stale data.
                  */
                 if (thread->vfpstate.hard.cpu != cpu)
-                        last_VFP_context[cpu] = NULL;
+                        vfp_current_hw_state[cpu] = NULL;
 #endif
 
                 /*
@@ -169,6 +176,35 @@ static struct notifier_block vfp_notifier_block = {
         .notifier_call  = vfp_notifier,
 };
 
+static int vfp_cpu_pm_notifier(struct notifier_block *self, unsigned long cmd,
+        void *v)
+{
+        u32 fpexc = fmrx(FPEXC);
+        unsigned int cpu = smp_processor_id();
+
+        switch (cmd) {
+        case CPU_PM_ENTER:
+                if (vfp_current_hw_state[cpu]) {
+                        fmxr(FPEXC, fpexc | FPEXC_EN);
+                        vfp_save_state(vfp_current_hw_state[cpu], fpexc);
+                        /* force a reload when coming back from idle */
+                        vfp_current_hw_state[cpu] = NULL;
+                        fmxr(FPEXC, fpexc & ~FPEXC_EN);
+                }
+                break;
+        case CPU_PM_ENTER_FAILED:
+        case CPU_PM_EXIT:
+                /* make sure VFP is disabled when leaving idle */
+                fmxr(FPEXC, fpexc & ~FPEXC_EN);
+                break;
+        }
+        return NOTIFY_OK;
+}
+
+static struct notifier_block vfp_cpu_pm_notifier_block = {
+        .notifier_call = vfp_cpu_pm_notifier,
+};
+
 /*
  * Raise a SIGFPE for the current process.
  * sicode describes the signal being raised.
@@ -405,6 +441,12 @@ static int vfp_pm_suspend(void)
         struct thread_info *ti = current_thread_info();
         u32 fpexc = fmrx(FPEXC);
 
+        /* If lazy disable, re-enable the VFP ready for it to be saved */
+        if (vfp_current_hw_state[ti->cpu] != &ti->vfpstate) {
+                fpexc |= FPEXC_EN;
+                fmxr(FPEXC, fpexc);
+        }
+
         /* if vfp is on, then save state for resumption */
         if (fpexc & FPEXC_EN) {
                 printk(KERN_DEBUG "%s: saving vfp state\n", __func__);
@@ -412,10 +454,14 @@ static int vfp_pm_suspend(void)
 
                 /* disable, just in case */
                 fmxr(FPEXC, fmrx(FPEXC) & ~FPEXC_EN);
+        } else if (vfp_current_hw_state[ti->cpu]) {
+                fmxr(FPEXC, fpexc | FPEXC_EN);
+                vfp_save_state(vfp_current_hw_state[ti->cpu], fpexc);
+                fmxr(FPEXC, fpexc);
         }
 
         /* clear any information we had about last context state */
-        memset(last_VFP_context, 0, sizeof(last_VFP_context));
+        memset(vfp_current_hw_state, 0, sizeof(vfp_current_hw_state));
 
         return 0;
 }
@@ -451,7 +497,7 @@ void vfp_sync_hwstate(struct thread_info *thread)
          * If the thread we're interested in is the current owner of the
          * hardware VFP state, then we need to save its state.
          */
-        if (last_VFP_context[cpu] == &thread->vfpstate) {
+        if (vfp_current_hw_state[cpu] == &thread->vfpstate) {
                 u32 fpexc = fmrx(FPEXC);
 
                 /*
@@ -473,7 +519,7 @@ void vfp_flush_hwstate(struct thread_info *thread)
          * If the thread we're interested in is the current owner of the
          * hardware VFP state, then we need to save its state.
          */
-        if (last_VFP_context[cpu] == &thread->vfpstate) {
+        if (vfp_current_hw_state[cpu] == &thread->vfpstate) {
                 u32 fpexc = fmrx(FPEXC);
 
                 fmxr(FPEXC, fpexc & ~FPEXC_EN);
@@ -482,7 +528,7 @@ void vfp_flush_hwstate(struct thread_info *thread)
                  * Set the context to NULL to force a reload the next time
                  * the thread uses the VFP.
                  */
-                last_VFP_context[cpu] = NULL;
+                vfp_current_hw_state[cpu] = NULL;
         }
 
 #ifdef CONFIG_SMP
@@ -514,7 +560,7 @@ static int vfp_hotplug(struct notifier_block *b, unsigned long action,
 {
         if (action == CPU_DYING || action == CPU_DYING_FROZEN) {
                 unsigned int cpu = (long)hcpu;
-                last_VFP_context[cpu] = NULL;
+                vfp_current_hw_state[cpu] = NULL;
         } else if (action == CPU_STARTING || action == CPU_STARTING_FROZEN)
                 vfp_enable(NULL);
         return NOTIFY_OK;
@@ -563,6 +609,7 @@ static int __init vfp_init(void)
                 vfp_vector = vfp_support_entry;
 
                 thread_register_notifier(&vfp_notifier_block);
+                cpu_pm_register_notifier(&vfp_cpu_pm_notifier_block);
                 vfp_pm_init();
 
                 /*
@@ -582,7 +629,6 @@ static int __init vfp_init(void)
                                 elf_hwcap |= HWCAP_VFPv3D16;
                 }
 #endif
-#ifdef CONFIG_NEON
                 /*
                  * Check for the presence of the Advanced SIMD
                  * load/store instructions, integer and single
@@ -590,10 +636,13 @@ static int __init vfp_init(void)
                  * for NEON if the hardware has the MVFR registers.
                  */
                 if ((read_cpuid_id() & 0x000f0000) == 0x000f0000) {
+#ifdef CONFIG_NEON
                         if ((fmrx(MVFR1) & 0x000fff00) == 0x00011100)
                                 elf_hwcap |= HWCAP_NEON;
-                }
 #endif
+                        if ((fmrx(MVFR1) & 0xf0000000) == 0x10000000)
+                                elf_hwcap |= HWCAP_VFPv4;
+                }
         }
         return 0;
 }