RE: [RFC][PATCH] Introduce page fault tracepoint

[Seiji Aguchi <seiji.aguchi@hds.com>]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset="utf-8", Size: 16644 bytes --]

Any comment?

> -----Original Message-----
> From: Seiji Aguchi [mailto:seiji.aguchi@hds.com]
> Sent: Tuesday, July 30, 2013 6:53 PM
> To: linux-kernel@vger.kernel.org; x86@kernel.org
> Cc: hpa@zytor.com; rostedt@goodmis.org; mingo@elte.hu; bp@alien8.de; tglx@linutronix.de; fdeslaur@gmail.com;
> raphael.beamonte@gmail.com; dle-develop@lists.sourceforge.net; Tomoki Sekiyama
> Subject: [RFC][PATCH] Introduce page fault tracepoint
> 
> This patch introduces page fault tracepoints to x86 architecture
> by switching IDT.
> 
> [Use case of page fault events]
> 
>   Two events, for user and kernel spaces, are introduced at the beginning of
>   page fault handler.
> 
>   - User space event
>     There is a request of page fault event for user space as below.
> 
>     http://marc.info/?l=linux-mm&m=136807959830182&w=2
>     http://marc.info/?l=linux-mm&m=136807959130175&w=2
> 
>   - Kernel space event:
>     Overhead in kernel space is measurable by enabling it.
> 
> [Creating IDT]
> 
>  A way to create IDT is as below.
> 
>  - Introduce set_intr_gate_raw() to register just non-trace handler to IDT.
>    This is used at boot time which tracing is disabled.
>  - Make set_intr_gate() macro so that it can register trace handler to
>    trace IDT and non-trace handler to normal IDT.
> 
> Signed-off-by: Seiji Aguchi <seiji.aguchi@hds.com>
> ---
>  arch/x86/include/asm/desc.h             |   33 +++++++++++++++++---
>  arch/x86/include/asm/hw_irq.h           |   14 ++++++++-
>  arch/x86/include/asm/trace/exceptions.h |   51 +++++++++++++++++++++++++++++++
>  arch/x86/include/asm/traps.h            |   22 +++++++++++++
>  arch/x86/kernel/entry_32.S              |   10 ++++++
>  arch/x86/kernel/entry_64.S              |   13 +++++++-
>  arch/x86/kernel/head64.c                |    2 +-
>  arch/x86/kernel/irqinit.c               |    2 +-
>  arch/x86/kernel/kvm.c                   |    2 +-
>  arch/x86/kernel/traps.c                 |   28 ++++++++--------
>  arch/x86/mm/Makefile                    |    2 +
>  arch/x86/mm/fault.c                     |   22 +++++++++++++
>  12 files changed, 177 insertions(+), 24 deletions(-)
>  create mode 100644 arch/x86/include/asm/trace/exceptions.h
> 
> diff --git a/arch/x86/include/asm/desc.h b/arch/x86/include/asm/desc.h
> index b90e5df..c04302b 100644
> --- a/arch/x86/include/asm/desc.h
> +++ b/arch/x86/include/asm/desc.h
> @@ -327,10 +327,28 @@ static inline void write_trace_idt_entry(int entry, const gate_desc *gate)
>  {
>  	write_idt_entry(trace_idt_table, entry, gate);
>  }
> +
> +static inline void _trace_set_gate(int gate, unsigned type, void *addr,
> +				   unsigned dpl, unsigned ist, unsigned seg)
> +{
> +	gate_desc s;
> +
> +	pack_gate(&s, type, (unsigned long)addr, dpl, ist, seg);
> +	/*
> +	 * does not need to be atomic because it is only done once at
> +	 * setup time
> +	 */
> +	write_trace_idt_entry(gate, &s);
> +}
>  #else
>  static inline void write_trace_idt_entry(int entry, const gate_desc *gate)
>  {
>  }
> +
> +static inline void _trace_set_gate(int gate, unsigned type, void *addr,
> +				   unsigned dpl, unsigned ist, unsigned seg)
> +{
> +}
>  #endif
> 
>  static inline void _set_gate(int gate, unsigned type, void *addr,
> @@ -353,12 +371,20 @@ static inline void _set_gate(int gate, unsigned type, void *addr,
>   * Pentium F0 0F bugfix can have resulted in the mapped
>   * IDT being write-protected.
>   */
> -static inline void set_intr_gate(unsigned int n, void *addr)
> +static inline void set_intr_gate_raw(unsigned int n, void *addr)
>  {
>  	BUG_ON((unsigned)n > 0xFF);
>  	_set_gate(n, GATE_INTERRUPT, addr, 0, 0, __KERNEL_CS);
>  }
> 
> +#define set_intr_gate(n, addr)						\
> +	do {								\
> +		BUG_ON((unsigned)n > 0xFF);				\
> +		_set_gate(n, GATE_INTERRUPT, addr, 0, 0, __KERNEL_CS);	\
> +		_trace_set_gate(n, GATE_INTERRUPT, trace_##addr, 0, 0,	\
> +				__KERNEL_CS);				\
> +	} while (0)
> +
>  extern int first_system_vector;
>  /* used_vectors is BITMAP for irq is not managed by percpu vector_irq */
>  extern unsigned long used_vectors[];
> @@ -395,10 +421,7 @@ static inline void trace_set_intr_gate(unsigned int gate, void *addr)
>  #define __trace_alloc_intr_gate(n, addr)
>  #endif
> 
> -static inline void __alloc_intr_gate(unsigned int n, void *addr)
> -{
> -	set_intr_gate(n, addr);
> -}
> +#define __alloc_intr_gate(n, addr) set_intr_gate(n, addr)
> 
>  #define alloc_intr_gate(n, addr)				\
>  	do {							\
> diff --git a/arch/x86/include/asm/hw_irq.h b/arch/x86/include/asm/hw_irq.h
> index e4ac559..fbd73b7 100644
> --- a/arch/x86/include/asm/hw_irq.h
> +++ b/arch/x86/include/asm/hw_irq.h
> @@ -89,10 +89,22 @@ extern void trace_reschedule_interrupt(void);
>  extern void trace_threshold_interrupt(void);
>  extern void trace_call_function_interrupt(void);
>  extern void trace_call_function_single_interrupt(void);
> +#else /* CONFIG_TRACING */
> +#define trace_apic_timer_interrupt apic_timer_interrupt
> +#define trace_x86_platform_ipi x86_platform_ipi
> +#define trace_error_interrupt error_interrupt
> +#define trace_irq_work_interrupt irq_work_interrupt
> +#define trace_spurious_interrupt spurious_interrupt
> +#define trace_thermal_interrupt thermal_interrupt
> +#define trace_reschedule_interrupt reschedule_interrupt
> +#define trace_threshold_interrupt threshold_interrupt
> +#define trace_call_function_interrupt call_function_interrupt
> +#define trace_call_function_single_interrupt call_function_single_interrupt
> +#endif
> +
>  #define trace_irq_move_cleanup_interrupt  irq_move_cleanup_interrupt
>  #define trace_reboot_interrupt  reboot_interrupt
>  #define trace_kvm_posted_intr_ipi kvm_posted_intr_ipi
> -#endif /* CONFIG_TRACING */
> 
>  /* IOAPIC */
>  #define IO_APIC_IRQ(x) (((x) >= NR_IRQS_LEGACY) || ((1<<(x)) & io_apic_irqs))
> diff --git a/arch/x86/include/asm/trace/exceptions.h b/arch/x86/include/asm/trace/exceptions.h
> new file mode 100644
> index 0000000..660fcf1
> --- /dev/null
> +++ b/arch/x86/include/asm/trace/exceptions.h
> @@ -0,0 +1,51 @@
> +#undef TRACE_SYSTEM
> +#define TRACE_SYSTEM exceptions
> +
> +#if !defined(_TRACE_PAGE_FAULT_H) || defined(TRACE_HEADER_MULTI_READ)
> +#define _TRACE_PAGE_FAULT_H
> +
> +#include <linux/tracepoint.h>
> +
> +extern void trace_irq_vector_regfunc(void);
> +extern void trace_irq_vector_unregfunc(void);
> +
> +DECLARE_EVENT_CLASS(x86_exceptions,
> +
> +	TP_PROTO(unsigned long address, struct pt_regs *regs,
> +		 unsigned long error_code),
> +
> +	TP_ARGS(address, regs, error_code),
> +
> +	TP_STRUCT__entry(
> +		__field(		unsigned long, address	)
> +		__field(		struct pt_regs *, regs	)
> +		__field(		unsigned long, error_code )
> +	),
> +
> +	TP_fast_assign(
> +		__entry->address = address;
> +		__entry->regs = regs;
> +		__entry->error_code = error_code;
> +	),
> +
> +	TP_printk("address=0x%lx regs=0x%p error_code=0x%lx",
> +		  __entry->address, __entry->regs, __entry->error_code) );
> +
> +#define DEFINE_PAGE_FAULT_EVENT(name)				\
> +DEFINE_EVENT_FN(x86_exceptions, name,				\
> +	TP_PROTO(unsigned long address,	struct pt_regs *regs,	\
> +		 unsigned long error_code),			\
> +	TP_ARGS(address, regs, error_code),			\
> +	trace_irq_vector_regfunc,				\
> +	trace_irq_vector_unregfunc);
> +
> +DEFINE_PAGE_FAULT_EVENT(user_page_fault);
> +DEFINE_PAGE_FAULT_EVENT(kernel_page_fault);
> +
> +#undef TRACE_INCLUDE_PATH
> +#define TRACE_INCLUDE_PATH .
> +#define TRACE_INCLUDE_FILE exceptions
> +#endif /*  _TRACE_PAGE_FAULT_H */
> +
> +/* This part must be outside protection */
> +#include <trace/define_trace.h>
> diff --git a/arch/x86/include/asm/traps.h b/arch/x86/include/asm/traps.h
> index 88eae2a..adf9258 100644
> --- a/arch/x86/include/asm/traps.h
> +++ b/arch/x86/include/asm/traps.h
> @@ -41,6 +41,25 @@ asmlinkage void machine_check(void);
>  #endif /* CONFIG_X86_MCE */
>  asmlinkage void simd_coprocessor_error(void);
> 
> +#ifdef CONFIG_TRACING
> +asmlinkage void trace_page_fault(void);
> +#else
> +#define trace_page_fault page_fault
> +#endif
> +#define trace_divide_error divide_error
> +#define trace_bounds bounds
> +#define trace_invalid_op invalid_op
> +#define trace_device_not_available device_not_available
> +#define trace_coprocessor_segment_overrun coprocessor_segment_overrun
> +#define trace_invalid_TSS invalid_TSS
> +#define trace_segment_not_present segment_not_present
> +#define trace_general_protection general_protection
> +#define trace_spurious_interrupt_bug spurious_interrupt_bug
> +#define trace_coprocessor_error coprocessor_error
> +#define trace_alignment_check alignment_check
> +#define trace_simd_coprocessor_error simd_coprocessor_error
> +#define trace_async_page_fault async_page_fault
> +
>  dotraplinkage void do_divide_error(struct pt_regs *, long);
>  dotraplinkage void do_debug(struct pt_regs *, long);
>  dotraplinkage void do_nmi(struct pt_regs *, long);
> @@ -59,6 +78,9 @@ asmlinkage __kprobes struct pt_regs *sync_regs(struct pt_regs *);
>  #endif
>  dotraplinkage void do_general_protection(struct pt_regs *, long);
>  dotraplinkage void do_page_fault(struct pt_regs *, unsigned long);
> +#ifdef CONFIG_TRACING
> +dotraplinkage void trace_do_page_fault(struct pt_regs *, unsigned long);
> +#endif
>  dotraplinkage void do_spurious_interrupt_bug(struct pt_regs *, long);
>  dotraplinkage void do_coprocessor_error(struct pt_regs *, long);
>  dotraplinkage void do_alignment_check(struct pt_regs *, long);
> diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S
> index 2cfbc3a..c9eb4e2 100644
> --- a/arch/x86/kernel/entry_32.S
> +++ b/arch/x86/kernel/entry_32.S
> @@ -1244,6 +1244,16 @@ return_to_handler:
>   */
>  	.pushsection .kprobes.text, "ax"
> 
> +#ifdef CONFIG_TRACING
> +ENTRY(trace_page_fault)
> +	RING0_EC_FRAME
> +	ASM_CLAC
> +	pushl_cfi $trace_do_page_fault
> +	jmp error_code
> +	CFI_ENDPROC
> +END(trace_page_fault)
> +#endif
> +
>  ENTRY(page_fault)
>  	RING0_EC_FRAME
>  	ASM_CLAC
> diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S
> index 1b69951..5136404 100644
> --- a/arch/x86/kernel/entry_64.S
> +++ b/arch/x86/kernel/entry_64.S
> @@ -1295,6 +1295,17 @@ ENTRY(\sym)
>  END(\sym)
>  .endm
> 
> +#ifdef CONFIG_TRACING
> +.macro trace_errorentry sym do_sym
> +errorentry trace(\sym) trace(\do_sym)
> +errorentry \sym \do_sym
> +.endm
> +#else
> +.macro trace_errorentry sym do_sym
> +errorentry \sym \do_sym
> +.endm
> +#endif
> +
>  	/* error code is on the stack already */
>  .macro paranoiderrorentry sym do_sym
>  ENTRY(\sym)
> @@ -1497,7 +1508,7 @@ zeroentry xen_int3 do_int3
>  errorentry xen_stack_segment do_stack_segment
>  #endif
>  errorentry general_protection do_general_protection
> -errorentry page_fault do_page_fault
> +trace_errorentry page_fault do_page_fault
>  #ifdef CONFIG_KVM_GUEST
>  errorentry async_page_fault do_async_page_fault
>  #endif
> diff --git a/arch/x86/kernel/head64.c b/arch/x86/kernel/head64.c
> index 55b6761..67a0649 100644
> --- a/arch/x86/kernel/head64.c
> +++ b/arch/x86/kernel/head64.c
> @@ -162,7 +162,7 @@ void __init x86_64_start_kernel(char * real_mode_data)
>  	clear_bss();
> 
>  	for (i = 0; i < NUM_EXCEPTION_VECTORS; i++)
> -		set_intr_gate(i, &early_idt_handlers[i]);
> +		set_intr_gate_raw(i, &early_idt_handlers[i]);
>  	load_idt((const struct desc_ptr *)&idt_descr);
> 
>  	copy_bootdata(__va(real_mode_data));
> diff --git a/arch/x86/kernel/irqinit.c b/arch/x86/kernel/irqinit.c
> index a2a1fbc..2ca2354 100644
> --- a/arch/x86/kernel/irqinit.c
> +++ b/arch/x86/kernel/irqinit.c
> @@ -206,7 +206,7 @@ void __init native_init_IRQ(void)
>  	i = FIRST_EXTERNAL_VECTOR;
>  	for_each_clear_bit_from(i, used_vectors, NR_VECTORS) {
>  		/* IA32_SYSCALL_VECTOR could be used in trap_init already. */
> -		set_intr_gate(i, interrupt[i - FIRST_EXTERNAL_VECTOR]);
> +		set_intr_gate_raw(i, interrupt[i - FIRST_EXTERNAL_VECTOR]);
>  	}
> 
>  	if (!acpi_ioapic && !of_ioapic)
> diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
> index a96d32c..12b384e 100644
> --- a/arch/x86/kernel/kvm.c
> +++ b/arch/x86/kernel/kvm.c
> @@ -462,7 +462,7 @@ static struct notifier_block kvm_cpu_notifier = {
> 
>  static void __init kvm_apf_trap_init(void)
>  {
> -	set_intr_gate(14, &async_page_fault);
> +	set_intr_gate(14, async_page_fault);
>  }
> 
>  void __init kvm_guest_init(void)
> diff --git a/arch/x86/kernel/traps.c b/arch/x86/kernel/traps.c
> index 1b23a1c..eadd251 100644
> --- a/arch/x86/kernel/traps.c
> +++ b/arch/x86/kernel/traps.c
> @@ -709,7 +709,7 @@ void __init early_trap_init(void)
>  	/* int3 can be called from all */
>  	set_system_intr_gate_ist(X86_TRAP_BP, &int3, DEBUG_STACK);
>  #ifdef CONFIG_X86_32
> -	set_intr_gate(X86_TRAP_PF, &page_fault);
> +	set_intr_gate(X86_TRAP_PF, page_fault);
>  #endif
>  	load_idt(&idt_descr);
>  }
> @@ -717,7 +717,7 @@ void __init early_trap_init(void)
>  void __init early_trap_pf_init(void)
>  {
>  #ifdef CONFIG_X86_64
> -	set_intr_gate(X86_TRAP_PF, &page_fault);
> +	set_intr_gate(X86_TRAP_PF, page_fault);
>  #endif
>  }
> 
> @@ -733,30 +733,30 @@ void __init trap_init(void)
>  	early_iounmap(p, 4);
>  #endif
> 
> -	set_intr_gate(X86_TRAP_DE, &divide_error);
> +	set_intr_gate(X86_TRAP_DE, divide_error);
>  	set_intr_gate_ist(X86_TRAP_NMI, &nmi, NMI_STACK);
>  	/* int4 can be called from all */
>  	set_system_intr_gate(X86_TRAP_OF, &overflow);
> -	set_intr_gate(X86_TRAP_BR, &bounds);
> -	set_intr_gate(X86_TRAP_UD, &invalid_op);
> -	set_intr_gate(X86_TRAP_NM, &device_not_available);
> +	set_intr_gate(X86_TRAP_BR, bounds);
> +	set_intr_gate(X86_TRAP_UD, invalid_op);
> +	set_intr_gate(X86_TRAP_NM, device_not_available);
>  #ifdef CONFIG_X86_32
>  	set_task_gate(X86_TRAP_DF, GDT_ENTRY_DOUBLEFAULT_TSS);
>  #else
>  	set_intr_gate_ist(X86_TRAP_DF, &double_fault, DOUBLEFAULT_STACK);
>  #endif
> -	set_intr_gate(X86_TRAP_OLD_MF, &coprocessor_segment_overrun);
> -	set_intr_gate(X86_TRAP_TS, &invalid_TSS);
> -	set_intr_gate(X86_TRAP_NP, &segment_not_present);
> +	set_intr_gate(X86_TRAP_OLD_MF, coprocessor_segment_overrun);
> +	set_intr_gate(X86_TRAP_TS, invalid_TSS);
> +	set_intr_gate(X86_TRAP_NP, segment_not_present);
>  	set_intr_gate_ist(X86_TRAP_SS, &stack_segment, STACKFAULT_STACK);
> -	set_intr_gate(X86_TRAP_GP, &general_protection);
> -	set_intr_gate(X86_TRAP_SPURIOUS, &spurious_interrupt_bug);
> -	set_intr_gate(X86_TRAP_MF, &coprocessor_error);
> -	set_intr_gate(X86_TRAP_AC, &alignment_check);
> +	set_intr_gate(X86_TRAP_GP, general_protection);
> +	set_intr_gate(X86_TRAP_SPURIOUS, spurious_interrupt_bug);
> +	set_intr_gate(X86_TRAP_MF, coprocessor_error);
> +	set_intr_gate(X86_TRAP_AC, alignment_check);
>  #ifdef CONFIG_X86_MCE
>  	set_intr_gate_ist(X86_TRAP_MC, &machine_check, MCE_STACK);
>  #endif
> -	set_intr_gate(X86_TRAP_XF, &simd_coprocessor_error);
> +	set_intr_gate(X86_TRAP_XF, simd_coprocessor_error);
> 
>  	/* Reserve all the builtin and the syscall vector: */
>  	for (i = 0; i < FIRST_EXTERNAL_VECTOR; i++)
> diff --git a/arch/x86/mm/Makefile b/arch/x86/mm/Makefile
> index 23d8e5f..6a19ad9 100644
> --- a/arch/x86/mm/Makefile
> +++ b/arch/x86/mm/Makefile
> @@ -6,6 +6,8 @@ nostackp := $(call cc-option, -fno-stack-protector)
>  CFLAGS_physaddr.o		:= $(nostackp)
>  CFLAGS_setup_nx.o		:= $(nostackp)
> 
> +CFLAGS_fault.o := -I$(src)/../include/asm/trace
> +
>  obj-$(CONFIG_X86_PAT)		+= pat_rbtree.o
>  obj-$(CONFIG_SMP)		+= tlb.o
> 
> diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c
> index 654be4a..f515154 100644
> --- a/arch/x86/mm/fault.c
> +++ b/arch/x86/mm/fault.c
> @@ -20,6 +20,9 @@
>  #include <asm/kmemcheck.h>		/* kmemcheck_*(), ...		*/
>  #include <asm/fixmap.h>			/* VSYSCALL_START		*/
> 
> +#define CREATE_TRACE_POINTS
> +#include <asm/trace/exceptions.h>
> +
>  /*
>   * Page fault error code bits:
>   *
> @@ -1230,3 +1233,22 @@ do_page_fault(struct pt_regs *regs, unsigned long error_code)
>  	__do_page_fault(regs, error_code);
>  	exception_exit(prev_state);
>  }
> +
> +static void trace_page_fault_entries(struct pt_regs *regs,
> +				     unsigned long error_code)
> +{
> +	if (user_mode(regs))
> +		trace_user_page_fault(read_cr2(), regs, error_code);
> +	else
> +		trace_kernel_page_fault(read_cr2(), regs, error_code);
> +}
> +
> +dotraplinkage void __kprobes
> +trace_do_page_fault(struct pt_regs *regs, unsigned long error_code)
> +{
> +	enum ctx_state prev_state;
> +	prev_state = exception_enter();
> +	trace_page_fault_entries(regs, error_code);
> +	__do_page_fault(regs, error_code);
> +	exception_exit(prev_state);
> +}
> --
> 1.7.1

ÿôèº{.nÇ+‰·Ÿ®‰­†+%ŠËÿ±éݶ\x17¥Šwÿº{.nÇ+‰·¥Š{±þG«éÿŠ{ayº\x1dʇڙë,j\a­¢f£¢·hšïêÿ‘êçz_è®\x03(­éšŽŠÝ¢j"ú\x1a¶^[m§ÿÿ¾\a«þG«éÿ¢¸?™¨è­Ú&£ø§~á¶iO•æ¬z·švØ^\x14\x04\x1a¶^[m§ÿÿÃ\fÿ¶ìÿ¢¸?–I¥