Merge branch 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/l…

…inux/kernel/git/tip/linux-2.6-tip

* 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (57 commits)
  x86, perf events: Check if we have APIC enabled
  perf_event: Fix variable initialization in other codepaths
  perf kmem: Fix unused argument build warning
  perf symbols: perf_header__read_build_ids() offset'n'size should be u64
  perf symbols: dsos__read_build_ids() should read both user and kernel buildids
  perf tools: Align long options which have no short forms
  perf kmem: Show usage if no option is specified
  sched: Mark sched_clock() as notrace
  perf sched: Add max delay time snapshot
  perf tools: Correct size given to memset
  perf_event: Fix perf_swevent_hrtimer() variable initialization
  perf sched: Fix for getting task's execution time
  tracing/kprobes: Fix field creation's bad error handling
  perf_event: Cleanup for cpu_clock_perf_event_update()
  perf_event: Allocate children's perf_event_ctxp at the right time
  perf_event: Clean up __perf_event_init_context()
  hw-breakpoints: Modify breakpoints without unregistering them
  perf probe: Update perf-probe document
  perf probe: Support --del option
  trace-kprobe: Support delete probe syntax
  ...

arch/x86/Kconfig.debug

@@ -187,8 +187,8 @@ config HAVE_MMIOTRACE_SUPPORT
 	def_bool y
 
 config X86_DECODER_SELFTEST
-     bool "x86 instruction decoder selftest"
-     depends on DEBUG_KERNEL
+	bool "x86 instruction decoder selftest"
+	depends on DEBUG_KERNEL && KPROBES
 	---help---
 	 Perform x86 instruction decoder selftests at build time.
 	 This option is useful for checking the sanity of x86 instruction

arch/x86/kernel/cpu/perf_event.c

@@ -1632,6 +1632,7 @@ static void intel_pmu_drain_bts_buffer(struct cpu_hw_events *cpuc)
 
 	data.period	= event->hw.last_period;
 	data.addr	= 0;
+	data.raw	= NULL;
 	regs.ip		= 0;
 
 	/*
@@ -1749,6 +1750,7 @@ static int p6_pmu_handle_irq(struct pt_regs *regs)
 	u64 val;
 
 	data.addr = 0;
+	data.raw = NULL;
 
 	cpuc = &__get_cpu_var(cpu_hw_events);
 
@@ -1794,6 +1796,7 @@ static int intel_pmu_handle_irq(struct pt_regs *regs)
 	u64 ack, status;
 
 	data.addr = 0;
+	data.raw = NULL;
 
 	cpuc = &__get_cpu_var(cpu_hw_events);
 
@@ -1857,6 +1860,7 @@ static int amd_pmu_handle_irq(struct pt_regs *regs)
 	u64 val;
 
 	data.addr = 0;
+	data.raw = NULL;
 
 	cpuc = &__get_cpu_var(cpu_hw_events);
 
@@ -2062,12 +2066,6 @@ static __init int p6_pmu_init(void)
 
 	x86_pmu = p6_pmu;
 
-	if (!cpu_has_apic) {
-		pr_info("no APIC, boot with the \"lapic\" boot parameter to force-enable it.\n");
-		pr_info("no hardware sampling interrupt available.\n");
-		x86_pmu.apic = 0;
-	}
-
 	return 0;
 }
 
@@ -2159,6 +2157,16 @@ static __init int amd_pmu_init(void)
 	return 0;
 }
 
+static void __init pmu_check_apic(void)
+{
+	if (cpu_has_apic)
+		return;
+
+	x86_pmu.apic = 0;
+	pr_info("no APIC, boot with the \"lapic\" boot parameter to force-enable it.\n");
+	pr_info("no hardware sampling interrupt available.\n");
+}
+
 void __init init_hw_perf_events(void)
 {
 	int err;
@@ -2180,6 +2188,8 @@ void __init init_hw_perf_events(void)
 		return;
 	}
 
+	pmu_check_apic();
+
 	pr_cont("%s PMU driver.\n", x86_pmu.name);
 
 	if (x86_pmu.num_events > X86_PMC_MAX_GENERIC) {
@@ -2287,7 +2297,7 @@ void callchain_store(struct perf_callchain_entry *entry, u64 ip)
 
 static DEFINE_PER_CPU(struct perf_callchain_entry, pmc_irq_entry);
 static DEFINE_PER_CPU(struct perf_callchain_entry, pmc_nmi_entry);
-static DEFINE_PER_CPU(int, in_nmi_frame);
+static DEFINE_PER_CPU(int, in_ignored_frame);
 
 
 static void
@@ -2303,8 +2313,9 @@ static void backtrace_warning(void *data, char *msg)
 
 static int backtrace_stack(void *data, char *name)
 {
-	per_cpu(in_nmi_frame, smp_processor_id()) =
-			x86_is_stack_id(NMI_STACK, name);
+	per_cpu(in_ignored_frame, smp_processor_id()) =
+			x86_is_stack_id(NMI_STACK, name) ||
+			x86_is_stack_id(DEBUG_STACK, name);
 
 	return 0;
 }
@@ -2313,7 +2324,7 @@ static void backtrace_address(void *data, unsigned long addr, int reliable)
 {
 	struct perf_callchain_entry *entry = data;
 
-	if (per_cpu(in_nmi_frame, smp_processor_id()))
+	if (per_cpu(in_ignored_frame, smp_processor_id()))
 		return;
 
 	if (reliable)

arch/x86/kernel/dumpstack_64.c

@@ -103,6 +103,35 @@ static unsigned long *in_exception_stack(unsigned cpu, unsigned long stack,
 	return NULL;
 }
 
+static inline int
+in_irq_stack(unsigned long *stack, unsigned long *irq_stack,
+	     unsigned long *irq_stack_end)
+{
+	return (stack >= irq_stack && stack < irq_stack_end);
+}
+
+/*
+ * We are returning from the irq stack and go to the previous one.
+ * If the previous stack is also in the irq stack, then bp in the first
+ * frame of the irq stack points to the previous, interrupted one.
+ * Otherwise we have another level of indirection: We first save
+ * the bp of the previous stack, then we switch the stack to the irq one
+ * and save a new bp that links to the previous one.
+ * (See save_args())
+ */
+static inline unsigned long
+fixup_bp_irq_link(unsigned long bp, unsigned long *stack,
+		  unsigned long *irq_stack, unsigned long *irq_stack_end)
+{
+#ifdef CONFIG_FRAME_POINTER
+	struct stack_frame *frame = (struct stack_frame *)bp;
+
+	if (!in_irq_stack(stack, irq_stack, irq_stack_end))
+		return (unsigned long)frame->next_frame;
+#endif
+	return bp;
+}
+
 /*
  * x86-64 can have up to three kernel stacks:
  * process stack
@@ -175,7 +204,7 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs,
 			irq_stack = irq_stack_end -
 				(IRQ_STACK_SIZE - 64) / sizeof(*irq_stack);
 
-			if (stack >= irq_stack && stack < irq_stack_end) {
+			if (in_irq_stack(stack, irq_stack, irq_stack_end)) {
 				if (ops->stack(data, "IRQ") < 0)
 					break;
 				bp = print_context_stack(tinfo, stack, bp,
@@ -186,6 +215,8 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs,
 				 * pointer (index -1 to end) in the IRQ stack:
 				 */
 				stack = (unsigned long *) (irq_stack_end[-1]);
+				bp = fixup_bp_irq_link(bp, stack, irq_stack,
+						       irq_stack_end);
 				irq_stack_end = NULL;
 				ops->stack(data, "EOI");
 				continue;

arch/x86/kernel/entry_64.S

@@ -1076,10 +1076,10 @@ ENTRY(\sym)
 	TRACE_IRQS_OFF
 	movq %rsp,%rdi		/* pt_regs pointer */
 	xorl %esi,%esi		/* no error code */
-	PER_CPU(init_tss, %rbp)
-	subq $EXCEPTION_STKSZ, TSS_ist + (\ist - 1) * 8(%rbp)
+	PER_CPU(init_tss, %r12)
+	subq $EXCEPTION_STKSZ, TSS_ist + (\ist - 1) * 8(%r12)
 	call \do_sym
-	addq $EXCEPTION_STKSZ, TSS_ist + (\ist - 1) * 8(%rbp)
+	addq $EXCEPTION_STKSZ, TSS_ist + (\ist - 1) * 8(%r12)
 	jmp paranoid_exit	/* %ebx: no swapgs flag */
 	CFI_ENDPROC
 END(\sym)

arch/x86/kernel/hw_breakpoint.c

@@ -362,8 +362,7 @@ int arch_validate_hwbkpt_settings(struct perf_event *bp,
 		return ret;
 	}
 
-	if (bp->callback)
-		ret = arch_store_info(bp);
+	ret = arch_store_info(bp);
 
 	if (ret < 0)
 		return ret;
@@ -519,7 +518,7 @@ static int __kprobes hw_breakpoint_handler(struct die_args *args)
 			break;
 		}
 
-		(bp->callback)(bp, args->regs);
+		perf_bp_event(bp, args->regs);
 
 		rcu_read_unlock();
 	}

arch/x86/kernel/ptrace.c

@@ -555,7 +555,9 @@ static int genregs_set(struct task_struct *target,
 	return ret;
 }
 
-static void ptrace_triggered(struct perf_event *bp, void *data)
+static void ptrace_triggered(struct perf_event *bp, int nmi,
+			     struct perf_sample_data *data,
+			     struct pt_regs *regs)
 {
 	int i;
 	struct thread_struct *thread = &(current->thread);
@@ -593,32 +595,32 @@ static unsigned long ptrace_get_dr7(struct perf_event *bp[])
 	return dr7;
 }
 
-static struct perf_event *
+static int
 ptrace_modify_breakpoint(struct perf_event *bp, int len, int type,
 			 struct task_struct *tsk, int disabled)
 {
 	int err;
 	int gen_len, gen_type;
-	DEFINE_BREAKPOINT_ATTR(attr);
+	struct perf_event_attr attr;
 
 	/*
 	 * We shoud have at least an inactive breakpoint at this
 	 * slot. It means the user is writing dr7 without having
 	 * written the address register first
 	 */
 	if (!bp)
-		return ERR_PTR(-EINVAL);
+		return -EINVAL;
 
 	err = arch_bp_generic_fields(len, type, &gen_len, &gen_type);
 	if (err)
-		return ERR_PTR(err);
+		return err;
 
 	attr = bp->attr;
 	attr.bp_len = gen_len;
 	attr.bp_type = gen_type;
 	attr.disabled = disabled;
 
-	return modify_user_hw_breakpoint(bp, &attr, bp->callback, tsk);
+	return modify_user_hw_breakpoint(bp, &attr);
 }
 
 /*
@@ -656,28 +658,17 @@ static int ptrace_write_dr7(struct task_struct *tsk, unsigned long data)
 				if (!second_pass)
 					continue;
 
-				thread->ptrace_bps[i] = NULL;
-				bp = ptrace_modify_breakpoint(bp, len, type,
+				rc = ptrace_modify_breakpoint(bp, len, type,
 							      tsk, 1);
-				if (IS_ERR(bp)) {
-					rc = PTR_ERR(bp);
-					thread->ptrace_bps[i] = NULL;
+				if (rc)
 					break;
-				}
-				thread->ptrace_bps[i] = bp;
 			}
 			continue;
 		}
 
-		bp = ptrace_modify_breakpoint(bp, len, type, tsk, 0);
-
-		/* Incorrect bp, or we have a bug in bp API */
-		if (IS_ERR(bp)) {
-			rc = PTR_ERR(bp);
-			thread->ptrace_bps[i] = NULL;
+		rc = ptrace_modify_breakpoint(bp, len, type, tsk, 0);
+		if (rc)
 			break;
-		}
-		thread->ptrace_bps[i] = bp;
 	}
 	/*
 	 * Make a second pass to free the remaining unused breakpoints
@@ -721,9 +712,10 @@ static int ptrace_set_breakpoint_addr(struct task_struct *tsk, int nr,
 {
 	struct perf_event *bp;
 	struct thread_struct *t = &tsk->thread;
-	DEFINE_BREAKPOINT_ATTR(attr);
+	struct perf_event_attr attr;
 
 	if (!t->ptrace_bps[nr]) {
+		hw_breakpoint_init(&attr);
 		/*
 		 * Put stub len and type to register (reserve) an inactive but
 		 * correct bp
@@ -734,26 +726,32 @@ static int ptrace_set_breakpoint_addr(struct task_struct *tsk, int nr,
 		attr.disabled = 1;
 
 		bp = register_user_hw_breakpoint(&attr, ptrace_triggered, tsk);
+
+		/*
+		 * CHECKME: the previous code returned -EIO if the addr wasn't
+		 * a valid task virtual addr. The new one will return -EINVAL in
+		 *  this case.
+		 * -EINVAL may be what we want for in-kernel breakpoints users,
+		 * but -EIO looks better for ptrace, since we refuse a register
+		 * writing for the user. And anyway this is the previous
+		 * behaviour.
+		 */
+		if (IS_ERR(bp))
+			return PTR_ERR(bp);
+
+		t->ptrace_bps[nr] = bp;
 	} else {
+		int err;
+
 		bp = t->ptrace_bps[nr];
-		t->ptrace_bps[nr] = NULL;
 
 		attr = bp->attr;
 		attr.bp_addr = addr;
-		bp = modify_user_hw_breakpoint(bp, &attr, bp->callback, tsk);
+		err = modify_user_hw_breakpoint(bp, &attr);
+		if (err)
+			return err;
 	}
-	/*
-	 * CHECKME: the previous code returned -EIO if the addr wasn't a
-	 * valid task virtual addr. The new one will return -EINVAL in this
-	 * case.
-	 * -EINVAL may be what we want for in-kernel breakpoints users, but
-	 * -EIO looks better for ptrace, since we refuse a register writing
-	 * for the user. And anyway this is the previous behaviour.
-	 */
-	if (IS_ERR(bp))
-		return PTR_ERR(bp);
 
-	t->ptrace_bps[nr] = bp;
 
 	return 0;
 }

arch/x86/lib/Makefile

@@ -5,7 +5,7 @@
 inat_tables_script = $(srctree)/arch/x86/tools/gen-insn-attr-x86.awk
 inat_tables_maps = $(srctree)/arch/x86/lib/x86-opcode-map.txt
 quiet_cmd_inat_tables = GEN     $@
-      cmd_inat_tables = $(AWK) -f $(inat_tables_script) $(inat_tables_maps) > $@
+      cmd_inat_tables = $(AWK) -f $(inat_tables_script) $(inat_tables_maps) > $@ || rm -f $@
 
 $(obj)/inat-tables.c: $(inat_tables_script) $(inat_tables_maps)
 	$(call cmd,inat_tables)
@@ -20,7 +20,7 @@ lib-y := delay.o
 lib-y += thunk_$(BITS).o
 lib-y += usercopy_$(BITS).o getuser.o putuser.o
 lib-y += memcpy_$(BITS).o
-lib-y += insn.o inat.o
+lib-$(CONFIG_KPROBES) += insn.o inat.o
 
 obj-y += msr-reg.o msr-reg-export.o
 

arch/x86/tools/test_get_len.c

@@ -113,7 +113,7 @@ int main(int argc, char **argv)
 	char line[BUFSIZE], sym[BUFSIZE] = "<unknown>";
 	unsigned char insn_buf[16];
 	struct insn insn;
-	int insns = 0, c;
+	int insns = 0;
 	int warnings = 0;
 
 	parse_args(argc, argv);