From: wangnan0@huawei.com (Wang Nan)
To: linux-arm-kernel@lists.infradead.org
Subject: [PATCH v8 2/2] ARM: kprobes: enable OPTPROBES for ARM 32
Date: Wed, 19 Nov 2014 15:08:05 +0800 [thread overview]
Message-ID: <546C41D5.6020108@huawei.com> (raw)
In-Reply-To: <1416292359-75893-3-git-send-email-wangnan0@huawei.com>
On 2014/11/18 14:32, Wang Nan wrote:
> This patch introduce kprobeopt for ARM 32.
>
> Limitations:
> - Currently only kernel compiled with ARM ISA is supported.
>
> - Offset between probe point and optinsn slot must not larger than
> 32MiB. Masami Hiramatsu suggests replacing 2 words, it will make
> things complex. Futher patch can make such optimization.
>
> Kprobe opt on ARM is relatively simpler than kprobe opt on x86 because
> ARM instruction is always 4 bytes aligned and 4 bytes long. This patch
> replace probed instruction by a 'b', branch to trampoline code and then
> calls optimized_callback(). optimized_callback() calls opt_pre_handler()
> to execute kprobe handler. It also emulate/simulate replaced instruction.
>
> When unregistering kprobe, the deferred manner of unoptimizer may leave
> branch instruction before optimizer is called. Different from x86_64,
> which only copy the probed insn after optprobe_template_end and
> reexecute them, this patch call singlestep to emulate/simulate the insn
> directly. Futher patch can optimize this behavior.
>
> v1 -> v2:
>
> - Improvement: if replaced instruction is conditional, generate a
> conditional branch instruction for it;
>
> - Introduces RELATIVEJUMP_OPCODES due to ARM kprobe_opcode_t is 4
> bytes;
>
> - Removes size field in struct arch_optimized_insn;
>
> - Use arm_gen_branch() to generate branch instruction;
>
> - Remove all recover logic: ARM doesn't use tail buffer, no need
> recover replaced instructions like x86;
>
> - Remove incorrect CONFIG_THUMB checking;
>
> - can_optimize() always returns true if address is well aligned;
>
> - Improve optimized_callback: using opt_pre_handler();
>
> - Bugfix: correct range checking code and improve comments;
>
> - Fix commit message.
>
> v2 -> v3:
>
> - Rename RELATIVEJUMP_OPCODES to MAX_COPIED_INSNS;
>
> - Remove unneeded checking:
> arch_check_optimized_kprobe(), can_optimize();
>
> - Add missing flush_icache_range() in arch_prepare_optimized_kprobe();
>
> - Remove unneeded 'return;'.
>
> v3 -> v4:
>
> - Use __mem_to_opcode_arm() to translate copied_insn to ensure it
> works in big endian kernel;
>
> - Replace 'nop' placeholder in trampoline code template with
> '.long 0' to avoid confusion: reader may regard 'nop' as an
> instruction, but it is value in fact.
>
> v4 -> v5:
>
> - Don't optimize stack store operations.
>
> - Introduce prepared field to arch_optimized_insn to indicate whether
> it is prepared. Similar to size field with x86. See v1 -> v2.
>
> v5 -> v6:
>
> - Dynamically reserve stack according to instruction.
>
> - Rename: kprobes-opt.c -> kprobes-opt-arm.c.
>
> - Set op->optinsn.insn after all works are done.
>
> v6 -> v7:
>
> - Using checker to check stack consumption.
>
> v7 -> v8:
>
> - Small code adjustments.
>
> Signed-off-by: Wang Nan <wangnan0@huawei.com>
> Acked-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
> Cc: Jon Medhurst (Tixy) <tixy@linaro.org>
> Cc: Russell King - ARM Linux <linux@arm.linux.org.uk>
> Cc: Will Deacon <will.deacon@arm.com>
> ---
> arch/arm/Kconfig | 1 +
> arch/arm/include/asm/kprobes.h | 26 ++++
> arch/arm/kernel/Makefile | 3 +-
> arch/arm/kernel/kprobes-opt-arm.c | 285 ++++++++++++++++++++++++++++++++++++++
> 4 files changed, 314 insertions(+), 1 deletion(-)
> create mode 100644 arch/arm/kernel/kprobes-opt-arm.c
>
> diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig
> index 89c4b5c..8281cea 100644
> --- a/arch/arm/Kconfig
> +++ b/arch/arm/Kconfig
> @@ -59,6 +59,7 @@ config ARM
> select HAVE_MEMBLOCK
> select HAVE_MOD_ARCH_SPECIFIC if ARM_UNWIND
> select HAVE_OPROFILE if (HAVE_PERF_EVENTS)
> + select HAVE_OPTPROBES if (!THUMB2_KERNEL)
> select HAVE_PERF_EVENTS
> select HAVE_PERF_REGS
> select HAVE_PERF_USER_STACK_DUMP
> diff --git a/arch/arm/include/asm/kprobes.h b/arch/arm/include/asm/kprobes.h
> index 56f9ac6..c1016cb 100644
> --- a/arch/arm/include/asm/kprobes.h
> +++ b/arch/arm/include/asm/kprobes.h
> @@ -50,5 +50,31 @@ int kprobe_fault_handler(struct pt_regs *regs, unsigned int fsr);
> int kprobe_exceptions_notify(struct notifier_block *self,
> unsigned long val, void *data);
>
> +/* optinsn template addresses */
> +extern __visible kprobe_opcode_t optprobe_template_entry;
> +extern __visible kprobe_opcode_t optprobe_template_val;
> +extern __visible kprobe_opcode_t optprobe_template_call;
> +extern __visible kprobe_opcode_t optprobe_template_end;
> +
> +#define MAX_OPTIMIZED_LENGTH (4)
> +#define MAX_OPTINSN_SIZE \
> + (((unsigned long)&optprobe_template_end - \
> + (unsigned long)&optprobe_template_entry))
> +#define RELATIVEJUMP_SIZE (4)
> +
> +struct arch_optimized_insn {
> + /*
> + * copy of the original instructions.
> + * Different from x86, ARM kprobe_opcode_t is u32.
> + */
> +#define MAX_COPIED_INSN ((RELATIVEJUMP_SIZE) / sizeof(kprobe_opcode_t))
> + kprobe_opcode_t copied_insn[MAX_COPIED_INSN];
> + /* detour code buffer */
> + kprobe_opcode_t *insn;
> + /*
> + * we always copies one instruction on arm32,
> + * size always be 4, so no size field.
> + */
> +};
>
> #endif /* _ARM_KPROBES_H */
> diff --git a/arch/arm/kernel/Makefile b/arch/arm/kernel/Makefile
> index 45aed4b..8a16fcf 100644
> --- a/arch/arm/kernel/Makefile
> +++ b/arch/arm/kernel/Makefile
> @@ -52,11 +52,12 @@ obj-$(CONFIG_FUNCTION_GRAPH_TRACER) += ftrace.o insn.o
> obj-$(CONFIG_JUMP_LABEL) += jump_label.o insn.o patch.o
> obj-$(CONFIG_KEXEC) += machine_kexec.o relocate_kernel.o
> obj-$(CONFIG_UPROBES) += probes.o probes-arm.o uprobes.o uprobes-arm.o
> -obj-$(CONFIG_KPROBES) += probes.o kprobes.o kprobes-common.o patch.o probes-checkers-common.o
> +obj-$(CONFIG_KPROBES) += probes.o kprobes.o kprobes-common.o patch.o probes-checkers-common.o insn.o
> ifdef CONFIG_THUMB2_KERNEL
> obj-$(CONFIG_KPROBES) += kprobes-thumb.o probes-thumb.o probes-checkers-thumb.o
> else
> obj-$(CONFIG_KPROBES) += kprobes-arm.o probes-arm.o probes-checkers-arm.o
> +obj-$(CONFIG_OPTPROBES) += kprobes-opt-arm.o
> endif
> obj-$(CONFIG_ARM_KPROBES_TEST) += test-kprobes.o
> test-kprobes-objs := kprobes-test.o
> diff --git a/arch/arm/kernel/kprobes-opt-arm.c b/arch/arm/kernel/kprobes-opt-arm.c
> new file mode 100644
> index 0000000..55eb3ea
> --- /dev/null
> +++ b/arch/arm/kernel/kprobes-opt-arm.c
> @@ -0,0 +1,285 @@
> +/*
> + * Kernel Probes Jump Optimization (Optprobes)
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License as published by
> + * the Free Software Foundation; either version 2 of the License, or
> + * (at your option) any later version.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> + * GNU General Public License for more details.
> + *
> + * You should have received a copy of the GNU General Public License
> + * along with this program; if not, write to the Free Software
> + * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
> + *
> + * Copyright (C) IBM Corporation, 2002, 2004
> + * Copyright (C) Hitachi Ltd., 2012
> + * Copyright (C) Huawei Inc., 2014
> + */
> +
> +#include <linux/kprobes.h>
> +#include <linux/jump_label.h>
> +#include <asm/kprobes.h>
> +#include <asm/cacheflush.h>
> +/* for arm_gen_branch */
> +#include "insn.h"
> +/* for patch_text */
> +#include "patch.h"
> +
> +/*
> + * NOTE: the first sub and add instruction will be modified according
> + * to the stack cost of the instruction.
> + */
> +asm (
> + ".global optprobe_template_entry\n"
> + "optprobe_template_entry:\n"
> + " sub sp, sp, #0xff\n"
> + " stmia sp, {r0 - r14} \n"
> + " add r3, sp, #0xff\n"
> + " str r3, [sp, #52]\n"
> + " mrs r4, cpsr\n"
> + " str r4, [sp, #64]\n"
> + " mov r1, sp\n"
> + " ldr r0, 1f\n"
> + " ldr r2, 2f\n"
> + " blx r2\n"
> + " ldr r1, [sp, #64]\n"
> + " msr cpsr_fs, r1\n"
> + " ldmia sp, {r0 - r15}\n"
> + ".global optprobe_template_val\n"
> + "optprobe_template_val:\n"
> + "1: .long 0\n"
> + ".global optprobe_template_call\n"
> + "optprobe_template_call:\n"
> + "2: .long 0\n"
> + ".global optprobe_template_end\n"
> + "optprobe_template_end:\n");
> +
> +#define TMPL_VAL_IDX \
> + ((long)&optprobe_template_val - (long)&optprobe_template_entry)
> +#define TMPL_CALL_IDX \
> + ((long)&optprobe_template_call - (long)&optprobe_template_entry)
> +#define TMPL_END_IDX \
> + ((long)&optprobe_template_end - (long)&optprobe_template_entry)
> +
> +/*
> + * ARM can always optimize an instruction when using ARM ISA, except
> + * instructions like 'str r0, [sp, r1]' which store to stack and unable
> + * to determine stack space consumption statically.
> + */
> +int arch_prepared_optinsn(struct arch_optimized_insn *optinsn)
> +{
> + return optinsn->insn != NULL;
> +}
> +
> +/*
> + * In ARM ISA, kprobe opt always replace one instruction (4 bytes
> + * aligned and 4 bytes long). It is impossiable to encounter another
> + * kprobe in the address range. So always return 0.
> + */
> +int arch_check_optimized_kprobe(struct optimized_kprobe *op)
> +{
> + return 0;
> +}
> +
> +/* Caller must ensure addr & 3 == 0 */
> +static int can_optimize(struct optimized_kprobe *op)
> +{
> + if (op->kp.ainsn.stack_space < 0)
> + return 0;
> + /*
> + * 255 is the biggest imm can be used in 'sub r0, r0, #<imm>'.
> + * Number larger than 255 needs special encoding.
> + */
> + if (op->kp.ainsn.stack_space > 255 - sizeof(struct pt_regs))
> + return 0;
> + return 1;
> +}
> +
> +/* Free optimized instruction slot */
> +static void
> +__arch_remove_optimized_kprobe(struct optimized_kprobe *op, int dirty)
> +{
> + if (op->optinsn.insn) {
> + free_optinsn_slot(op->optinsn.insn, dirty);
> + op->optinsn.insn = NULL;
> + }
> +}
> +
> +extern void kprobe_handler(struct pt_regs *regs);
> +
> +static void
> +optimized_callback(struct optimized_kprobe *op, struct pt_regs *regs)
> +{
> + unsigned long flags;
> + struct kprobe *p = &op->kp;
> + struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
> +
> + /* Save skipped registers */
> + regs->ARM_pc = (unsigned long)op->kp.addr;
> + regs->ARM_ORIG_r0 = ~0UL;
> +
> + local_irq_save(flags);
> +
> + if (kprobe_running()) {
> + kprobes_inc_nmissed_count(&op->kp);
> + } else {
> + __this_cpu_write(current_kprobe, &op->kp);
> + kcb->kprobe_status = KPROBE_HIT_ACTIVE;
> + opt_pre_handler(&op->kp, regs);
> + __this_cpu_write(current_kprobe, NULL);
> + }
> +
> + /* In each case, we must singlestep the replaced instruction. */
> + op->kp.ainsn.insn_singlestep(p->opcode, &p->ainsn, regs);
> +
> + local_irq_restore(flags);
> +}
> +
> +int arch_prepare_optimized_kprobe(struct optimized_kprobe *op)
> +{
> + u8 *buf;
> + unsigned long *code;
> + unsigned long rel_chk;
> + unsigned long val;
> + unsigned long stack_protect = sizeof(struct pt_regs);
> +
> + if (!can_optimize(op))
> + return -EILSEQ;
> +
> + buf = (u8 *)get_optinsn_slot();
> + if (!buf)
> + return -ENOMEM;
> +
> + /*
> + * Verify if the address gap is in 32MiB range, because this uses
> + * a relative jump.
> + *
> + * kprobe opt use a 'b' instruction to branch to optinsn.insn.
> + * According to ARM manual, branch instruction is:
> + *
> + * 31 28 27 24 23 0
> + * +------+---+---+---+---+----------------+
> + * | cond | 1 | 0 | 1 | 0 | imm24 |
> + * +------+---+---+---+---+----------------+
> + *
> + * imm24 is a signed 24 bits integer. The real branch offset is computed
> + * by: imm32 = SignExtend(imm24:'00', 32);
> + *
> + * So the maximum forward branch should be:
> + * (0x007fffff << 2) = 0x01fffffc = 0x1fffffc
> + * The maximum backword branch should be:
> + * (0xff800000 << 2) = 0xfe000000 = -0x2000000
> + *
> + * We can simply check (rel & 0xfe000003):
> + * if rel is positive, (rel & 0xfe000000) shoule be 0
> + * if rel is negitive, (rel & 0xfe000000) should be 0xfe000000
> + * the last '3' is used for alignment checking.
> + */
> + rel_chk = (unsigned long)((long)buf -
> + (long)op->kp.addr + 8) & 0xfe000003;
> +
> + if ((rel_chk != 0) && (rel_chk != 0xfe000000)) {
> + __arch_remove_optimized_kprobe(op, 0);
Here is a bug. Different from x86_64, ARM code doesn't set op->optinsn.insn
here, so __arch_remove_optimized_kprobe won't do anything. In version 9 patch
series I fix this problem by calling free_optinsn_slot(buf, 0) directly.
However, I forgot to mention it in its commit message.
> + return -ERANGE;
> + }
> +
> + /* Copy arch-dep-instance from template. */
> + memcpy(buf, &optprobe_template_entry, TMPL_END_IDX);
> +
> + /* Adjust buffer according to instruction. */
> + BUG_ON(op->kp.ainsn.stack_space < 0);
> + stack_protect += op->kp.ainsn.stack_space;
> +
> + /* Should have been filtered by can_optimize(). */
> + BUG_ON(stack_protect > 255);
> +
> + /* Create a 'sub sp, sp, #<stack_protect>' */
> + code = (unsigned long *)(buf);
> + code[0] = __opcode_to_mem_arm(0xe24dd000 | stack_protect);
> + /* Create a 'add r3, sp, #<stack_protect>' */
> + code[2] = __opcode_to_mem_arm(0xe28d3000 | stack_protect);
> +
> + /* Set probe information */
> + val = (unsigned long)op;
> + memcpy(buf + TMPL_VAL_IDX, &val, sizeof(val));
> +
> + /* Set probe function call */
> + val = (unsigned long)optimized_callback;
> + memcpy(buf + TMPL_CALL_IDX, &val, sizeof(val));
> +
> + flush_icache_range((unsigned long)buf,
> + (unsigned long)buf + TMPL_END_IDX);
> +
> + /* Set op->optinsn.insn means prepared */
> + op->optinsn.insn = (kprobe_opcode_t *)buf;
> + return 0;
> +}
> +
> +void arch_optimize_kprobes(struct list_head *oplist)
> +{
> + struct optimized_kprobe *op, *tmp;
> +
> + list_for_each_entry_safe(op, tmp, oplist, list) {
> + unsigned long insn;
> + WARN_ON(kprobe_disabled(&op->kp));
> +
> + /*
> + * Backup instructions which will be replaced
> + * by jump address
> + */
> + memcpy(op->optinsn.copied_insn, op->kp.addr,
> + RELATIVEJUMP_SIZE);
> +
> + insn = arm_gen_branch((unsigned long)op->kp.addr,
> + (unsigned long)op->optinsn.insn);
> + BUG_ON(insn == 0);
> +
> + /*
> + * Make it a conditional branch if replaced insn
> + * is consitional
> + */
> + insn = (__mem_to_opcode_arm(
> + op->optinsn.copied_insn[0]) & 0xf0000000) |
> + (insn & 0x0fffffff);
> +
> + patch_text(op->kp.addr, insn);
> +
> + list_del_init(&op->list);
> + }
> +}
> +
> +void arch_unoptimize_kprobe(struct optimized_kprobe *op)
> +{
> + arch_arm_kprobe(&op->kp);
> +}
> +
> +/*
> + * Recover original instructions and breakpoints from relative jumps.
> + * Caller must call with locking kprobe_mutex.
> + */
> +void arch_unoptimize_kprobes(struct list_head *oplist,
> + struct list_head *done_list)
> +{
> + struct optimized_kprobe *op, *tmp;
> +
> + list_for_each_entry_safe(op, tmp, oplist, list) {
> + arch_unoptimize_kprobe(op);
> + list_move(&op->list, done_list);
> + }
> +}
> +
> +int arch_within_optimized_kprobe(struct optimized_kprobe *op,
> + unsigned long addr)
> +{
> + return ((unsigned long)op->kp.addr <= addr &&
> + (unsigned long)op->kp.addr + RELATIVEJUMP_SIZE > addr);
> +}
> +
> +void arch_remove_optimized_kprobe(struct optimized_kprobe *op)
> +{
> + __arch_remove_optimized_kprobe(op, 1);
> +}
>
WARNING: multiple messages have this Message-ID (diff)
From: Wang Nan <wangnan0@huawei.com>
To: <tixy@linaro.org>, <masami.hiramatsu.pt@hitachi.com>,
<linux@arm.linux.org.uk>, <will.deacon@arm.com>,
<dave.long@linaro.org>, <taras.kondratiuk@linaro.org>,
<ben.dooks@codethink.co.uk>, <cl@linux.com>, <rabin@rab.in>,
<davem@davemloft.net>
Cc: <lizefan@huawei.com>, <linux-kernel@vger.kernel.org>,
<linux-arm-kernel@lists.infradead.org>
Subject: Re: [PATCH v8 2/2] ARM: kprobes: enable OPTPROBES for ARM 32
Date: Wed, 19 Nov 2014 15:08:05 +0800 [thread overview]
Message-ID: <546C41D5.6020108@huawei.com> (raw)
In-Reply-To: <1416292359-75893-3-git-send-email-wangnan0@huawei.com>
On 2014/11/18 14:32, Wang Nan wrote:
> This patch introduce kprobeopt for ARM 32.
>
> Limitations:
> - Currently only kernel compiled with ARM ISA is supported.
>
> - Offset between probe point and optinsn slot must not larger than
> 32MiB. Masami Hiramatsu suggests replacing 2 words, it will make
> things complex. Futher patch can make such optimization.
>
> Kprobe opt on ARM is relatively simpler than kprobe opt on x86 because
> ARM instruction is always 4 bytes aligned and 4 bytes long. This patch
> replace probed instruction by a 'b', branch to trampoline code and then
> calls optimized_callback(). optimized_callback() calls opt_pre_handler()
> to execute kprobe handler. It also emulate/simulate replaced instruction.
>
> When unregistering kprobe, the deferred manner of unoptimizer may leave
> branch instruction before optimizer is called. Different from x86_64,
> which only copy the probed insn after optprobe_template_end and
> reexecute them, this patch call singlestep to emulate/simulate the insn
> directly. Futher patch can optimize this behavior.
>
> v1 -> v2:
>
> - Improvement: if replaced instruction is conditional, generate a
> conditional branch instruction for it;
>
> - Introduces RELATIVEJUMP_OPCODES due to ARM kprobe_opcode_t is 4
> bytes;
>
> - Removes size field in struct arch_optimized_insn;
>
> - Use arm_gen_branch() to generate branch instruction;
>
> - Remove all recover logic: ARM doesn't use tail buffer, no need
> recover replaced instructions like x86;
>
> - Remove incorrect CONFIG_THUMB checking;
>
> - can_optimize() always returns true if address is well aligned;
>
> - Improve optimized_callback: using opt_pre_handler();
>
> - Bugfix: correct range checking code and improve comments;
>
> - Fix commit message.
>
> v2 -> v3:
>
> - Rename RELATIVEJUMP_OPCODES to MAX_COPIED_INSNS;
>
> - Remove unneeded checking:
> arch_check_optimized_kprobe(), can_optimize();
>
> - Add missing flush_icache_range() in arch_prepare_optimized_kprobe();
>
> - Remove unneeded 'return;'.
>
> v3 -> v4:
>
> - Use __mem_to_opcode_arm() to translate copied_insn to ensure it
> works in big endian kernel;
>
> - Replace 'nop' placeholder in trampoline code template with
> '.long 0' to avoid confusion: reader may regard 'nop' as an
> instruction, but it is value in fact.
>
> v4 -> v5:
>
> - Don't optimize stack store operations.
>
> - Introduce prepared field to arch_optimized_insn to indicate whether
> it is prepared. Similar to size field with x86. See v1 -> v2.
>
> v5 -> v6:
>
> - Dynamically reserve stack according to instruction.
>
> - Rename: kprobes-opt.c -> kprobes-opt-arm.c.
>
> - Set op->optinsn.insn after all works are done.
>
> v6 -> v7:
>
> - Using checker to check stack consumption.
>
> v7 -> v8:
>
> - Small code adjustments.
>
> Signed-off-by: Wang Nan <wangnan0@huawei.com>
> Acked-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
> Cc: Jon Medhurst (Tixy) <tixy@linaro.org>
> Cc: Russell King - ARM Linux <linux@arm.linux.org.uk>
> Cc: Will Deacon <will.deacon@arm.com>
> ---
> arch/arm/Kconfig | 1 +
> arch/arm/include/asm/kprobes.h | 26 ++++
> arch/arm/kernel/Makefile | 3 +-
> arch/arm/kernel/kprobes-opt-arm.c | 285 ++++++++++++++++++++++++++++++++++++++
> 4 files changed, 314 insertions(+), 1 deletion(-)
> create mode 100644 arch/arm/kernel/kprobes-opt-arm.c
>
> diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig
> index 89c4b5c..8281cea 100644
> --- a/arch/arm/Kconfig
> +++ b/arch/arm/Kconfig
> @@ -59,6 +59,7 @@ config ARM
> select HAVE_MEMBLOCK
> select HAVE_MOD_ARCH_SPECIFIC if ARM_UNWIND
> select HAVE_OPROFILE if (HAVE_PERF_EVENTS)
> + select HAVE_OPTPROBES if (!THUMB2_KERNEL)
> select HAVE_PERF_EVENTS
> select HAVE_PERF_REGS
> select HAVE_PERF_USER_STACK_DUMP
> diff --git a/arch/arm/include/asm/kprobes.h b/arch/arm/include/asm/kprobes.h
> index 56f9ac6..c1016cb 100644
> --- a/arch/arm/include/asm/kprobes.h
> +++ b/arch/arm/include/asm/kprobes.h
> @@ -50,5 +50,31 @@ int kprobe_fault_handler(struct pt_regs *regs, unsigned int fsr);
> int kprobe_exceptions_notify(struct notifier_block *self,
> unsigned long val, void *data);
>
> +/* optinsn template addresses */
> +extern __visible kprobe_opcode_t optprobe_template_entry;
> +extern __visible kprobe_opcode_t optprobe_template_val;
> +extern __visible kprobe_opcode_t optprobe_template_call;
> +extern __visible kprobe_opcode_t optprobe_template_end;
> +
> +#define MAX_OPTIMIZED_LENGTH (4)
> +#define MAX_OPTINSN_SIZE \
> + (((unsigned long)&optprobe_template_end - \
> + (unsigned long)&optprobe_template_entry))
> +#define RELATIVEJUMP_SIZE (4)
> +
> +struct arch_optimized_insn {
> + /*
> + * copy of the original instructions.
> + * Different from x86, ARM kprobe_opcode_t is u32.
> + */
> +#define MAX_COPIED_INSN ((RELATIVEJUMP_SIZE) / sizeof(kprobe_opcode_t))
> + kprobe_opcode_t copied_insn[MAX_COPIED_INSN];
> + /* detour code buffer */
> + kprobe_opcode_t *insn;
> + /*
> + * we always copies one instruction on arm32,
> + * size always be 4, so no size field.
> + */
> +};
>
> #endif /* _ARM_KPROBES_H */
> diff --git a/arch/arm/kernel/Makefile b/arch/arm/kernel/Makefile
> index 45aed4b..8a16fcf 100644
> --- a/arch/arm/kernel/Makefile
> +++ b/arch/arm/kernel/Makefile
> @@ -52,11 +52,12 @@ obj-$(CONFIG_FUNCTION_GRAPH_TRACER) += ftrace.o insn.o
> obj-$(CONFIG_JUMP_LABEL) += jump_label.o insn.o patch.o
> obj-$(CONFIG_KEXEC) += machine_kexec.o relocate_kernel.o
> obj-$(CONFIG_UPROBES) += probes.o probes-arm.o uprobes.o uprobes-arm.o
> -obj-$(CONFIG_KPROBES) += probes.o kprobes.o kprobes-common.o patch.o probes-checkers-common.o
> +obj-$(CONFIG_KPROBES) += probes.o kprobes.o kprobes-common.o patch.o probes-checkers-common.o insn.o
> ifdef CONFIG_THUMB2_KERNEL
> obj-$(CONFIG_KPROBES) += kprobes-thumb.o probes-thumb.o probes-checkers-thumb.o
> else
> obj-$(CONFIG_KPROBES) += kprobes-arm.o probes-arm.o probes-checkers-arm.o
> +obj-$(CONFIG_OPTPROBES) += kprobes-opt-arm.o
> endif
> obj-$(CONFIG_ARM_KPROBES_TEST) += test-kprobes.o
> test-kprobes-objs := kprobes-test.o
> diff --git a/arch/arm/kernel/kprobes-opt-arm.c b/arch/arm/kernel/kprobes-opt-arm.c
> new file mode 100644
> index 0000000..55eb3ea
> --- /dev/null
> +++ b/arch/arm/kernel/kprobes-opt-arm.c
> @@ -0,0 +1,285 @@
> +/*
> + * Kernel Probes Jump Optimization (Optprobes)
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License as published by
> + * the Free Software Foundation; either version 2 of the License, or
> + * (at your option) any later version.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> + * GNU General Public License for more details.
> + *
> + * You should have received a copy of the GNU General Public License
> + * along with this program; if not, write to the Free Software
> + * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
> + *
> + * Copyright (C) IBM Corporation, 2002, 2004
> + * Copyright (C) Hitachi Ltd., 2012
> + * Copyright (C) Huawei Inc., 2014
> + */
> +
> +#include <linux/kprobes.h>
> +#include <linux/jump_label.h>
> +#include <asm/kprobes.h>
> +#include <asm/cacheflush.h>
> +/* for arm_gen_branch */
> +#include "insn.h"
> +/* for patch_text */
> +#include "patch.h"
> +
> +/*
> + * NOTE: the first sub and add instruction will be modified according
> + * to the stack cost of the instruction.
> + */
> +asm (
> + ".global optprobe_template_entry\n"
> + "optprobe_template_entry:\n"
> + " sub sp, sp, #0xff\n"
> + " stmia sp, {r0 - r14} \n"
> + " add r3, sp, #0xff\n"
> + " str r3, [sp, #52]\n"
> + " mrs r4, cpsr\n"
> + " str r4, [sp, #64]\n"
> + " mov r1, sp\n"
> + " ldr r0, 1f\n"
> + " ldr r2, 2f\n"
> + " blx r2\n"
> + " ldr r1, [sp, #64]\n"
> + " msr cpsr_fs, r1\n"
> + " ldmia sp, {r0 - r15}\n"
> + ".global optprobe_template_val\n"
> + "optprobe_template_val:\n"
> + "1: .long 0\n"
> + ".global optprobe_template_call\n"
> + "optprobe_template_call:\n"
> + "2: .long 0\n"
> + ".global optprobe_template_end\n"
> + "optprobe_template_end:\n");
> +
> +#define TMPL_VAL_IDX \
> + ((long)&optprobe_template_val - (long)&optprobe_template_entry)
> +#define TMPL_CALL_IDX \
> + ((long)&optprobe_template_call - (long)&optprobe_template_entry)
> +#define TMPL_END_IDX \
> + ((long)&optprobe_template_end - (long)&optprobe_template_entry)
> +
> +/*
> + * ARM can always optimize an instruction when using ARM ISA, except
> + * instructions like 'str r0, [sp, r1]' which store to stack and unable
> + * to determine stack space consumption statically.
> + */
> +int arch_prepared_optinsn(struct arch_optimized_insn *optinsn)
> +{
> + return optinsn->insn != NULL;
> +}
> +
> +/*
> + * In ARM ISA, kprobe opt always replace one instruction (4 bytes
> + * aligned and 4 bytes long). It is impossiable to encounter another
> + * kprobe in the address range. So always return 0.
> + */
> +int arch_check_optimized_kprobe(struct optimized_kprobe *op)
> +{
> + return 0;
> +}
> +
> +/* Caller must ensure addr & 3 == 0 */
> +static int can_optimize(struct optimized_kprobe *op)
> +{
> + if (op->kp.ainsn.stack_space < 0)
> + return 0;
> + /*
> + * 255 is the biggest imm can be used in 'sub r0, r0, #<imm>'.
> + * Number larger than 255 needs special encoding.
> + */
> + if (op->kp.ainsn.stack_space > 255 - sizeof(struct pt_regs))
> + return 0;
> + return 1;
> +}
> +
> +/* Free optimized instruction slot */
> +static void
> +__arch_remove_optimized_kprobe(struct optimized_kprobe *op, int dirty)
> +{
> + if (op->optinsn.insn) {
> + free_optinsn_slot(op->optinsn.insn, dirty);
> + op->optinsn.insn = NULL;
> + }
> +}
> +
> +extern void kprobe_handler(struct pt_regs *regs);
> +
> +static void
> +optimized_callback(struct optimized_kprobe *op, struct pt_regs *regs)
> +{
> + unsigned long flags;
> + struct kprobe *p = &op->kp;
> + struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
> +
> + /* Save skipped registers */
> + regs->ARM_pc = (unsigned long)op->kp.addr;
> + regs->ARM_ORIG_r0 = ~0UL;
> +
> + local_irq_save(flags);
> +
> + if (kprobe_running()) {
> + kprobes_inc_nmissed_count(&op->kp);
> + } else {
> + __this_cpu_write(current_kprobe, &op->kp);
> + kcb->kprobe_status = KPROBE_HIT_ACTIVE;
> + opt_pre_handler(&op->kp, regs);
> + __this_cpu_write(current_kprobe, NULL);
> + }
> +
> + /* In each case, we must singlestep the replaced instruction. */
> + op->kp.ainsn.insn_singlestep(p->opcode, &p->ainsn, regs);
> +
> + local_irq_restore(flags);
> +}
> +
> +int arch_prepare_optimized_kprobe(struct optimized_kprobe *op)
> +{
> + u8 *buf;
> + unsigned long *code;
> + unsigned long rel_chk;
> + unsigned long val;
> + unsigned long stack_protect = sizeof(struct pt_regs);
> +
> + if (!can_optimize(op))
> + return -EILSEQ;
> +
> + buf = (u8 *)get_optinsn_slot();
> + if (!buf)
> + return -ENOMEM;
> +
> + /*
> + * Verify if the address gap is in 32MiB range, because this uses
> + * a relative jump.
> + *
> + * kprobe opt use a 'b' instruction to branch to optinsn.insn.
> + * According to ARM manual, branch instruction is:
> + *
> + * 31 28 27 24 23 0
> + * +------+---+---+---+---+----------------+
> + * | cond | 1 | 0 | 1 | 0 | imm24 |
> + * +------+---+---+---+---+----------------+
> + *
> + * imm24 is a signed 24 bits integer. The real branch offset is computed
> + * by: imm32 = SignExtend(imm24:'00', 32);
> + *
> + * So the maximum forward branch should be:
> + * (0x007fffff << 2) = 0x01fffffc = 0x1fffffc
> + * The maximum backword branch should be:
> + * (0xff800000 << 2) = 0xfe000000 = -0x2000000
> + *
> + * We can simply check (rel & 0xfe000003):
> + * if rel is positive, (rel & 0xfe000000) shoule be 0
> + * if rel is negitive, (rel & 0xfe000000) should be 0xfe000000
> + * the last '3' is used for alignment checking.
> + */
> + rel_chk = (unsigned long)((long)buf -
> + (long)op->kp.addr + 8) & 0xfe000003;
> +
> + if ((rel_chk != 0) && (rel_chk != 0xfe000000)) {
> + __arch_remove_optimized_kprobe(op, 0);
Here is a bug. Different from x86_64, ARM code doesn't set op->optinsn.insn
here, so __arch_remove_optimized_kprobe won't do anything. In version 9 patch
series I fix this problem by calling free_optinsn_slot(buf, 0) directly.
However, I forgot to mention it in its commit message.
> + return -ERANGE;
> + }
> +
> + /* Copy arch-dep-instance from template. */
> + memcpy(buf, &optprobe_template_entry, TMPL_END_IDX);
> +
> + /* Adjust buffer according to instruction. */
> + BUG_ON(op->kp.ainsn.stack_space < 0);
> + stack_protect += op->kp.ainsn.stack_space;
> +
> + /* Should have been filtered by can_optimize(). */
> + BUG_ON(stack_protect > 255);
> +
> + /* Create a 'sub sp, sp, #<stack_protect>' */
> + code = (unsigned long *)(buf);
> + code[0] = __opcode_to_mem_arm(0xe24dd000 | stack_protect);
> + /* Create a 'add r3, sp, #<stack_protect>' */
> + code[2] = __opcode_to_mem_arm(0xe28d3000 | stack_protect);
> +
> + /* Set probe information */
> + val = (unsigned long)op;
> + memcpy(buf + TMPL_VAL_IDX, &val, sizeof(val));
> +
> + /* Set probe function call */
> + val = (unsigned long)optimized_callback;
> + memcpy(buf + TMPL_CALL_IDX, &val, sizeof(val));
> +
> + flush_icache_range((unsigned long)buf,
> + (unsigned long)buf + TMPL_END_IDX);
> +
> + /* Set op->optinsn.insn means prepared */
> + op->optinsn.insn = (kprobe_opcode_t *)buf;
> + return 0;
> +}
> +
> +void arch_optimize_kprobes(struct list_head *oplist)
> +{
> + struct optimized_kprobe *op, *tmp;
> +
> + list_for_each_entry_safe(op, tmp, oplist, list) {
> + unsigned long insn;
> + WARN_ON(kprobe_disabled(&op->kp));
> +
> + /*
> + * Backup instructions which will be replaced
> + * by jump address
> + */
> + memcpy(op->optinsn.copied_insn, op->kp.addr,
> + RELATIVEJUMP_SIZE);
> +
> + insn = arm_gen_branch((unsigned long)op->kp.addr,
> + (unsigned long)op->optinsn.insn);
> + BUG_ON(insn == 0);
> +
> + /*
> + * Make it a conditional branch if replaced insn
> + * is consitional
> + */
> + insn = (__mem_to_opcode_arm(
> + op->optinsn.copied_insn[0]) & 0xf0000000) |
> + (insn & 0x0fffffff);
> +
> + patch_text(op->kp.addr, insn);
> +
> + list_del_init(&op->list);
> + }
> +}
> +
> +void arch_unoptimize_kprobe(struct optimized_kprobe *op)
> +{
> + arch_arm_kprobe(&op->kp);
> +}
> +
> +/*
> + * Recover original instructions and breakpoints from relative jumps.
> + * Caller must call with locking kprobe_mutex.
> + */
> +void arch_unoptimize_kprobes(struct list_head *oplist,
> + struct list_head *done_list)
> +{
> + struct optimized_kprobe *op, *tmp;
> +
> + list_for_each_entry_safe(op, tmp, oplist, list) {
> + arch_unoptimize_kprobe(op);
> + list_move(&op->list, done_list);
> + }
> +}
> +
> +int arch_within_optimized_kprobe(struct optimized_kprobe *op,
> + unsigned long addr)
> +{
> + return ((unsigned long)op->kp.addr <= addr &&
> + (unsigned long)op->kp.addr + RELATIVEJUMP_SIZE > addr);
> +}
> +
> +void arch_remove_optimized_kprobe(struct optimized_kprobe *op)
> +{
> + __arch_remove_optimized_kprobe(op, 1);
> +}
>
next prev parent reply other threads:[~2014-11-19 7:08 UTC|newest]
Thread overview: 14+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-11-18 6:32 [PATCH v8 0/2] ARM: kprobes: enable OPTPROBES for ARM32 Wang Nan
2014-11-18 6:32 ` Wang Nan
2014-11-18 6:32 ` [PATCH v8 1/2] kprobes: copy ainsn after alloc aggr kprobe Wang Nan
2014-11-18 6:32 ` Wang Nan
2014-11-18 11:38 ` Masami Hiramatsu
2014-11-18 11:38 ` Masami Hiramatsu
2014-11-19 1:53 ` Wang Nan
2014-11-19 1:53 ` Wang Nan
2014-11-18 6:32 ` [PATCH v8 2/2] ARM: kprobes: enable OPTPROBES for ARM 32 Wang Nan
2014-11-18 6:32 ` Wang Nan
2014-11-19 7:08 ` Wang Nan [this message]
2014-11-19 7:08 ` Wang Nan
2014-11-18 19:34 ` [PATCH] kprobes: Pass the original kprobe for preparing optimized kprobe Masami Hiramatsu
2014-11-18 19:34 ` Masami Hiramatsu
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=546C41D5.6020108@huawei.com \
--to=wangnan0@huawei.com \
--cc=linux-arm-kernel@lists.infradead.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.