Merge branch 'for-linus' of git://one.firstfloor.org/home/andi/git/linux-2.6

* 'for-linus' of git://one.firstfloor.org/home/andi/git/linux-2.6: (156 commits)
  [PATCH] x86-64: Export smp_call_function_single
  [PATCH] i386: Clean up smp_tune_scheduling()
  [PATCH] unwinder: move .eh_frame to RODATA
  [PATCH] unwinder: fully support linker generated .eh_frame_hdr section
  [PATCH] x86-64: don't use set_irq_regs()
  [PATCH] x86-64: check vector in setup_ioapic_dest to verify if need setup_IO_APIC_irq
  [PATCH] x86-64: Make ix86 default to HIGHMEM4G instead of NOHIGHMEM
  [PATCH] i386: replace kmalloc+memset with kzalloc
  [PATCH] x86-64: remove remaining pc98 code
  [PATCH] x86-64: remove unused variable
  [PATCH] x86-64: Fix constraints in atomic_add_return()
  [PATCH] x86-64: fix asm constraints in i386 atomic_add_return
  [PATCH] x86-64: Correct documentation for bzImage protocol v2.05
  [PATCH] x86-64: replace kmalloc+memset with kzalloc in MTRR code
  [PATCH] x86-64: Fix numaq build error
  [PATCH] x86-64: include/asm-x86_64/cpufeature.h isn't a userspace header
  [PATCH] unwinder: Add debugging output to the Dwarf2 unwinder
  [PATCH] x86-64: Clarify error message in GART code
  [PATCH] x86-64: Fix interrupt race in idle callback (3rd try)
  [PATCH] x86-64: Remove unwind stack pointer alignment forcing again
  ...

Fixed conflict in include/linux/uaccess.h manually

Signed-off-by: Linus Torvalds <torvalds@osdl.org>

1 files changed, 37 insertions, 44 deletions

diff --git a/arch/i386/kernel/process.c b/arch/i386/kernel/process.c
index dd53c58f64f1..99308510a17c 100644
--- a/arch/i386/kernel/process.c
+++ b/arch/i386/kernel/process.c
@@ -56,6 +56,7 @@
 
 #include <asm/tlbflush.h>
 #include <asm/cpu.h>
+#include <asm/pda.h>
 
 asmlinkage void ret_from_fork(void) __asm__("ret_from_fork");
 
@@ -99,22 +100,18 @@ EXPORT_SYMBOL(enable_hlt);
  */
 void default_idle(void)
 {
-        local_irq_enable();
-
         if (!hlt_counter && boot_cpu_data.hlt_works_ok) {
                 current_thread_info()->status &= ~TS_POLLING;
                 smp_mb__after_clear_bit();
-                while (!need_resched()) {
-                        local_irq_disable();
-                        if (!need_resched())
-                                safe_halt();
-                        else
-                                local_irq_enable();
-                }
+                local_irq_disable();
+                if (!need_resched())
+                        safe_halt();    /* enables interrupts racelessly */
+                else
+                        local_irq_enable();
                 current_thread_info()->status |= TS_POLLING;
         } else {
-                while (!need_resched())
-                        cpu_relax();
+                /* loop is done by the caller */
+                cpu_relax();
         }
 }
 #ifdef CONFIG_APM_MODULE
@@ -128,14 +125,7 @@ EXPORT_SYMBOL(default_idle);
  */
 static void poll_idle (void)
 {
-        local_irq_enable();
-
-        asm volatile(
-                "2:"
-                "testl %0, %1;"
-                "rep; nop;"
-                "je 2b;"
-                : : "i"(_TIF_NEED_RESCHED), "m" (current_thread_info()->flags));
+        cpu_relax();
 }
 
 #ifdef CONFIG_HOTPLUG_CPU
@@ -256,8 +246,7 @@ void mwait_idle_with_hints(unsigned long eax, unsigned long ecx)
 static void mwait_idle(void)
 {
         local_irq_enable();
-        while (!need_resched())
-                mwait_idle_with_hints(0, 0);
+        mwait_idle_with_hints(0, 0);
 }
 
 void __devinit select_idle_routine(const struct cpuinfo_x86 *c)
@@ -314,8 +303,8 @@ void show_regs(struct pt_regs * regs)
                 regs->eax,regs->ebx,regs->ecx,regs->edx);
         printk("ESI: %08lx EDI: %08lx EBP: %08lx",
                 regs->esi, regs->edi, regs->ebp);
-        printk(" DS: %04x ES: %04x\n",
-                0xffff & regs->xds,0xffff & regs->xes);
+        printk(" DS: %04x ES: %04x GS: %04x\n",
+               0xffff & regs->xds,0xffff & regs->xes, 0xffff & regs->xgs);
 
         cr0 = read_cr0();
         cr2 = read_cr2();
@@ -346,6 +335,7 @@ int kernel_thread(int (*fn)(void *), void * arg, unsigned long flags)
 
         regs.xds = __USER_DS;
         regs.xes = __USER_DS;
+        regs.xgs = __KERNEL_PDA;
         regs.orig_eax = -1;
         regs.eip = (unsigned long) kernel_thread_helper;
         regs.xcs = __KERNEL_CS | get_kernel_rpl();
@@ -431,7 +421,6 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long esp,
         p->thread.eip = (unsigned long) ret_from_fork;
 
         savesegment(fs,p->thread.fs);
-        savesegment(gs,p->thread.gs);
 
         tsk = current;
         if (unlikely(test_tsk_thread_flag(tsk, TIF_IO_BITMAP))) {
@@ -508,7 +497,7 @@ void dump_thread(struct pt_regs * regs, struct user * dump)
         dump->regs.ds = regs->xds;
         dump->regs.es = regs->xes;
         savesegment(fs,dump->regs.fs);
-        savesegment(gs,dump->regs.gs);
+        dump->regs.gs = regs->xgs;
         dump->regs.orig_eax = regs->orig_eax;
         dump->regs.eip = regs->eip;
         dump->regs.cs = regs->xcs;
@@ -648,22 +637,27 @@ struct task_struct fastcall * __switch_to(struct task_struct *prev_p, struct tas
 
         __unlazy_fpu(prev_p);
 
+
+        /* we're going to use this soon, after a few expensive things */
+        if (next_p->fpu_counter > 5)
+                prefetch(&next->i387.fxsave);
+
         /*
          * Reload esp0.
          */
         load_esp0(tss, next);
 
         /*
-         * Save away %fs and %gs. No need to save %es and %ds, as
-         * those are always kernel segments while inside the kernel.
-         * Doing this before setting the new TLS descriptors avoids
-         * the situation where we temporarily have non-reloadable
-         * segments in %fs and %gs.  This could be an issue if the
-         * NMI handler ever used %fs or %gs (it does not today), or
-         * if the kernel is running inside of a hypervisor layer.
+         * Save away %fs. No need to save %gs, as it was saved on the
+         * stack on entry.  No need to save %es and %ds, as those are
+         * always kernel segments while inside the kernel.  Doing this
+         * before setting the new TLS descriptors avoids the situation
+         * where we temporarily have non-reloadable segments in %fs
+         * and %gs.  This could be an issue if the NMI handler ever
+         * used %fs or %gs (it does not today), or if the kernel is
+         * running inside of a hypervisor layer.
          */
         savesegment(fs, prev->fs);
-        savesegment(gs, prev->gs);
 
         /*
          * Load the per-thread Thread-Local Storage descriptor.
@@ -671,22 +665,14 @@ struct task_struct fastcall * __switch_to(struct task_struct *prev_p, struct tas
         load_TLS(next, cpu);
 
         /*
-         * Restore %fs and %gs if needed.
+         * Restore %fs if needed.
          *
-         * Glibc normally makes %fs be zero, and %gs is one of
-         * the TLS segments.
+         * Glibc normally makes %fs be zero.
          */
         if (unlikely(prev->fs | next->fs))
                 loadsegment(fs, next->fs);
 
-        if (prev->gs | next->gs)
-                loadsegment(gs, next->gs);
-
-        /*
-         * Restore IOPL if needed.
-         */
-        if (unlikely(prev->iopl != next->iopl))
-                set_iopl_mask(next->iopl);
+        write_pda(pcurrent, next_p);
 
         /*
          * Now maybe handle debug registers and/or IO bitmaps
@@ -697,6 +683,13 @@ struct task_struct fastcall * __switch_to(struct task_struct *prev_p, struct tas
 
         disable_tsc(prev_p, next_p);
 
+        /* If the task has used fpu the last 5 timeslices, just do a full
+         * restore of the math state immediately to avoid the trap; the
+         * chances of needing FPU soon are obviously high now
+         */
+        if (next_p->fpu_counter > 5)
+                math_state_restore();
+
         return prev_p;
 }