linux/arch/arm/kernel/kprobes-arm.c

1613 lines
52 KiB
C
Raw Normal View History

ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
/*
* arch/arm/kernel/kprobes-decode.c
*
* Copyright (C) 2006, 2007 Motorola Inc.
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License version 2 as
* published by the Free Software Foundation.
*
* 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.
*/
/*
* We do not have hardware single-stepping on ARM, This
* effort is further complicated by the ARM not having a
* "next PC" register. Instructions that change the PC
* can't be safely single-stepped in a MP environment, so
* we have a lot of work to do:
*
* In the prepare phase:
* *) If it is an instruction that does anything
* with the CPU mode, we reject it for a kprobe.
* (This is out of laziness rather than need. The
* instructions could be simulated.)
*
* *) Otherwise, decode the instruction rewriting its
* registers to take fixed, ordered registers and
* setting a handler for it to run the instruction.
*
* In the execution phase by an instruction's handler:
*
* *) If the PC is written to by the instruction, the
* instruction must be fully simulated in software.
*
* *) Otherwise, a modified form of the instruction is
* directly executed. Its handler calls the
* instruction in insn[0]. In insn[1] is a
* "mov pc, lr" to return.
*
* Before calling, load up the reordered registers
* from the original instruction's registers. If one
* of the original input registers is the PC, compute
* and adjust the appropriate input register.
*
* After call completes, copy the output registers to
* the original instruction's original registers.
*
* We don't use a real breakpoint instruction since that
* would have us in the kernel go from SVC mode to SVC
* mode losing the link register. Instead we use an
* undefined instruction. To simplify processing, the
* undefined instruction used for kprobes must be reserved
* exclusively for kprobes use.
*
* TODO: ifdef out some instruction decoding based on architecture.
*/
#include <linux/kernel.h>
#include <linux/kprobes.h>
#include "kprobes.h"
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
#define sign_extend(x, signbit) ((x) | (0 - ((x) & (1 << (signbit)))))
#define branch_displacement(insn) sign_extend(((insn) & 0xffffff) << 2, 25)
#if __LINUX_ARM_ARCH__ >= 6
#define BLX(reg) "blx "reg" \n\t"
#else
#define BLX(reg) "mov lr, pc \n\t" \
"mov pc, "reg" \n\t"
#endif
#define is_r15(insn, bitpos) (((insn) & (0xf << bitpos)) == (0xf << bitpos))
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
#define PSR_fs (PSR_f|PSR_s)
#define KPROBE_RETURN_INSTRUCTION 0xe1a0f00e /* mov pc, lr */
typedef long (insn_0arg_fn_t)(void);
typedef long (insn_1arg_fn_t)(long);
typedef long (insn_2arg_fn_t)(long, long);
typedef long (insn_3arg_fn_t)(long, long, long);
typedef long (insn_4arg_fn_t)(long, long, long, long);
typedef long long (insn_llret_0arg_fn_t)(void);
typedef long long (insn_llret_3arg_fn_t)(long, long, long);
typedef long long (insn_llret_4arg_fn_t)(long, long, long, long);
union reg_pair {
long long dr;
#ifdef __LITTLE_ENDIAN
struct { long r0, r1; };
#else
struct { long r1, r0; };
#endif
};
/*
* The insnslot_?arg_r[w]flags() functions below are to keep the
* msr -> *fn -> mrs instruction sequences indivisible so that
* the state of the CPSR flags aren't inadvertently modified
* just before or just after the call.
*/
static inline long __kprobes
insnslot_0arg_rflags(long cpsr, insn_0arg_fn_t *fn)
{
register long ret asm("r0");
__asm__ __volatile__ (
"msr cpsr_fs, %[cpsr] \n\t"
"mov lr, pc \n\t"
"mov pc, %[fn] \n\t"
: "=r" (ret)
: [cpsr] "r" (cpsr), [fn] "r" (fn)
: "lr", "cc"
);
return ret;
}
static inline long long __kprobes
insnslot_llret_0arg_rflags(long cpsr, insn_llret_0arg_fn_t *fn)
{
register long ret0 asm("r0");
register long ret1 asm("r1");
union reg_pair fnr;
__asm__ __volatile__ (
"msr cpsr_fs, %[cpsr] \n\t"
"mov lr, pc \n\t"
"mov pc, %[fn] \n\t"
: "=r" (ret0), "=r" (ret1)
: [cpsr] "r" (cpsr), [fn] "r" (fn)
: "lr", "cc"
);
fnr.r0 = ret0;
fnr.r1 = ret1;
return fnr.dr;
}
static inline long __kprobes
insnslot_1arg_rflags(long r0, long cpsr, insn_1arg_fn_t *fn)
{
register long rr0 asm("r0") = r0;
register long ret asm("r0");
__asm__ __volatile__ (
"msr cpsr_fs, %[cpsr] \n\t"
"mov lr, pc \n\t"
"mov pc, %[fn] \n\t"
: "=r" (ret)
: "0" (rr0), [cpsr] "r" (cpsr), [fn] "r" (fn)
: "lr", "cc"
);
return ret;
}
static inline long __kprobes
insnslot_2arg_rflags(long r0, long r1, long cpsr, insn_2arg_fn_t *fn)
{
register long rr0 asm("r0") = r0;
register long rr1 asm("r1") = r1;
register long ret asm("r0");
__asm__ __volatile__ (
"msr cpsr_fs, %[cpsr] \n\t"
"mov lr, pc \n\t"
"mov pc, %[fn] \n\t"
: "=r" (ret)
: "0" (rr0), "r" (rr1),
[cpsr] "r" (cpsr), [fn] "r" (fn)
: "lr", "cc"
);
return ret;
}
static inline long __kprobes
insnslot_3arg_rflags(long r0, long r1, long r2, long cpsr, insn_3arg_fn_t *fn)
{
register long rr0 asm("r0") = r0;
register long rr1 asm("r1") = r1;
register long rr2 asm("r2") = r2;
register long ret asm("r0");
__asm__ __volatile__ (
"msr cpsr_fs, %[cpsr] \n\t"
"mov lr, pc \n\t"
"mov pc, %[fn] \n\t"
: "=r" (ret)
: "0" (rr0), "r" (rr1), "r" (rr2),
[cpsr] "r" (cpsr), [fn] "r" (fn)
: "lr", "cc"
);
return ret;
}
static inline long long __kprobes
insnslot_llret_3arg_rflags(long r0, long r1, long r2, long cpsr,
insn_llret_3arg_fn_t *fn)
{
register long rr0 asm("r0") = r0;
register long rr1 asm("r1") = r1;
register long rr2 asm("r2") = r2;
register long ret0 asm("r0");
register long ret1 asm("r1");
union reg_pair fnr;
__asm__ __volatile__ (
"msr cpsr_fs, %[cpsr] \n\t"
"mov lr, pc \n\t"
"mov pc, %[fn] \n\t"
: "=r" (ret0), "=r" (ret1)
: "0" (rr0), "r" (rr1), "r" (rr2),
[cpsr] "r" (cpsr), [fn] "r" (fn)
: "lr", "cc"
);
fnr.r0 = ret0;
fnr.r1 = ret1;
return fnr.dr;
}
static inline long __kprobes
insnslot_4arg_rflags(long r0, long r1, long r2, long r3, long cpsr,
insn_4arg_fn_t *fn)
{
register long rr0 asm("r0") = r0;
register long rr1 asm("r1") = r1;
register long rr2 asm("r2") = r2;
register long rr3 asm("r3") = r3;
register long ret asm("r0");
__asm__ __volatile__ (
"msr cpsr_fs, %[cpsr] \n\t"
"mov lr, pc \n\t"
"mov pc, %[fn] \n\t"
: "=r" (ret)
: "0" (rr0), "r" (rr1), "r" (rr2), "r" (rr3),
[cpsr] "r" (cpsr), [fn] "r" (fn)
: "lr", "cc"
);
return ret;
}
static inline long __kprobes
insnslot_1arg_rwflags(long r0, long *cpsr, insn_1arg_fn_t *fn)
{
register long rr0 asm("r0") = r0;
register long ret asm("r0");
long oldcpsr = *cpsr;
long newcpsr;
__asm__ __volatile__ (
"msr cpsr_fs, %[oldcpsr] \n\t"
"mov lr, pc \n\t"
"mov pc, %[fn] \n\t"
"mrs %[newcpsr], cpsr \n\t"
: "=r" (ret), [newcpsr] "=r" (newcpsr)
: "0" (rr0), [oldcpsr] "r" (oldcpsr), [fn] "r" (fn)
: "lr", "cc"
);
*cpsr = (oldcpsr & ~PSR_fs) | (newcpsr & PSR_fs);
return ret;
}
static inline long __kprobes
insnslot_2arg_rwflags(long r0, long r1, long *cpsr, insn_2arg_fn_t *fn)
{
register long rr0 asm("r0") = r0;
register long rr1 asm("r1") = r1;
register long ret asm("r0");
long oldcpsr = *cpsr;
long newcpsr;
__asm__ __volatile__ (
"msr cpsr_fs, %[oldcpsr] \n\t"
"mov lr, pc \n\t"
"mov pc, %[fn] \n\t"
"mrs %[newcpsr], cpsr \n\t"
: "=r" (ret), [newcpsr] "=r" (newcpsr)
: "0" (rr0), "r" (rr1), [oldcpsr] "r" (oldcpsr), [fn] "r" (fn)
: "lr", "cc"
);
*cpsr = (oldcpsr & ~PSR_fs) | (newcpsr & PSR_fs);
return ret;
}
static inline long __kprobes
insnslot_3arg_rwflags(long r0, long r1, long r2, long *cpsr,
insn_3arg_fn_t *fn)
{
register long rr0 asm("r0") = r0;
register long rr1 asm("r1") = r1;
register long rr2 asm("r2") = r2;
register long ret asm("r0");
long oldcpsr = *cpsr;
long newcpsr;
__asm__ __volatile__ (
"msr cpsr_fs, %[oldcpsr] \n\t"
"mov lr, pc \n\t"
"mov pc, %[fn] \n\t"
"mrs %[newcpsr], cpsr \n\t"
: "=r" (ret), [newcpsr] "=r" (newcpsr)
: "0" (rr0), "r" (rr1), "r" (rr2),
[oldcpsr] "r" (oldcpsr), [fn] "r" (fn)
: "lr", "cc"
);
*cpsr = (oldcpsr & ~PSR_fs) | (newcpsr & PSR_fs);
return ret;
}
static inline long __kprobes
insnslot_4arg_rwflags(long r0, long r1, long r2, long r3, long *cpsr,
insn_4arg_fn_t *fn)
{
register long rr0 asm("r0") = r0;
register long rr1 asm("r1") = r1;
register long rr2 asm("r2") = r2;
register long rr3 asm("r3") = r3;
register long ret asm("r0");
long oldcpsr = *cpsr;
long newcpsr;
__asm__ __volatile__ (
"msr cpsr_fs, %[oldcpsr] \n\t"
"mov lr, pc \n\t"
"mov pc, %[fn] \n\t"
"mrs %[newcpsr], cpsr \n\t"
: "=r" (ret), [newcpsr] "=r" (newcpsr)
: "0" (rr0), "r" (rr1), "r" (rr2), "r" (rr3),
[oldcpsr] "r" (oldcpsr), [fn] "r" (fn)
: "lr", "cc"
);
*cpsr = (oldcpsr & ~PSR_fs) | (newcpsr & PSR_fs);
return ret;
}
static inline long long __kprobes
insnslot_llret_4arg_rwflags(long r0, long r1, long r2, long r3, long *cpsr,
insn_llret_4arg_fn_t *fn)
{
register long rr0 asm("r0") = r0;
register long rr1 asm("r1") = r1;
register long rr2 asm("r2") = r2;
register long rr3 asm("r3") = r3;
register long ret0 asm("r0");
register long ret1 asm("r1");
long oldcpsr = *cpsr;
long newcpsr;
union reg_pair fnr;
__asm__ __volatile__ (
"msr cpsr_fs, %[oldcpsr] \n\t"
"mov lr, pc \n\t"
"mov pc, %[fn] \n\t"
"mrs %[newcpsr], cpsr \n\t"
: "=r" (ret0), "=r" (ret1), [newcpsr] "=r" (newcpsr)
: "0" (rr0), "r" (rr1), "r" (rr2), "r" (rr3),
[oldcpsr] "r" (oldcpsr), [fn] "r" (fn)
: "lr", "cc"
);
*cpsr = (oldcpsr & ~PSR_fs) | (newcpsr & PSR_fs);
fnr.r0 = ret0;
fnr.r1 = ret1;
return fnr.dr;
}
/*
* To avoid the complications of mimicing single-stepping on a
* processor without a Next-PC or a single-step mode, and to
* avoid having to deal with the side-effects of boosting, we
* simulate or emulate (almost) all ARM instructions.
*
* "Simulation" is where the instruction's behavior is duplicated in
* C code. "Emulation" is where the original instruction is rewritten
* and executed, often by altering its registers.
*
* By having all behavior of the kprobe'd instruction completed before
* returning from the kprobe_handler(), all locks (scheduler and
* interrupt) can safely be released. There is no need for secondary
* breakpoints, no race with MP or preemptable kernels, nor having to
* clean up resources counts at a later time impacting overall system
* performance. By rewriting the instruction, only the minimum registers
* need to be loaded and saved back optimizing performance.
*
* Calling the insnslot_*_rwflags version of a function doesn't hurt
* anything even when the CPSR flags aren't updated by the
* instruction. It's just a little slower in return for saving
* a little space by not having a duplicate function that doesn't
* update the flags. (The same optimization can be said for
* instructions that do or don't perform register writeback)
* Also, instructions can either read the flags, only write the
* flags, or read and write the flags. To save combinations
* rather than for sheer performance, flag functions just assume
* read and write of flags.
*/
static void __kprobes simulate_bbl(struct kprobe *p, struct pt_regs *regs)
{
kprobe_opcode_t insn = p->opcode;
long iaddr = (long)p->addr;
int disp = branch_displacement(insn);
if (insn & (1 << 24))
regs->ARM_lr = iaddr + 4;
regs->ARM_pc = iaddr + 8 + disp;
}
static void __kprobes simulate_blx1(struct kprobe *p, struct pt_regs *regs)
{
kprobe_opcode_t insn = p->opcode;
long iaddr = (long)p->addr;
int disp = branch_displacement(insn);
regs->ARM_lr = iaddr + 4;
regs->ARM_pc = iaddr + 8 + disp + ((insn >> 23) & 0x2);
regs->ARM_cpsr |= PSR_T_BIT;
}
static void __kprobes simulate_blx2bx(struct kprobe *p, struct pt_regs *regs)
{
kprobe_opcode_t insn = p->opcode;
int rm = insn & 0xf;
long rmv = regs->uregs[rm];
if (insn & (1 << 5))
regs->ARM_lr = (long)p->addr + 4;
regs->ARM_pc = rmv & ~0x1;
regs->ARM_cpsr &= ~PSR_T_BIT;
if (rmv & 0x1)
regs->ARM_cpsr |= PSR_T_BIT;
}
static void __kprobes simulate_mrs(struct kprobe *p, struct pt_regs *regs)
{
kprobe_opcode_t insn = p->opcode;
int rd = (insn >> 12) & 0xf;
unsigned long mask = 0xf8ff03df; /* Mask out execution state */
regs->uregs[rd] = regs->ARM_cpsr & mask;
}
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
static void __kprobes simulate_mov_ipsp(struct kprobe *p, struct pt_regs *regs)
{
regs->uregs[12] = regs->uregs[13];
}
static void __kprobes emulate_ldrd(struct kprobe *p, struct pt_regs *regs)
{
insn_2arg_fn_t *i_fn = (insn_2arg_fn_t *)&p->ainsn.insn[0];
kprobe_opcode_t insn = p->opcode;
long ppc = (long)p->addr + 8;
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
int rd = (insn >> 12) & 0xf;
int rn = (insn >> 16) & 0xf;
int rm = insn & 0xf; /* rm may be invalid, don't care. */
long rmv = (rm == 15) ? ppc : regs->uregs[rm];
long rnv = (rn == 15) ? ppc : regs->uregs[rn];
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
/* Not following the C calling convention here, so need asm(). */
__asm__ __volatile__ (
"ldr r0, %[rn] \n\t"
"ldr r1, %[rm] \n\t"
"msr cpsr_fs, %[cpsr]\n\t"
"mov lr, pc \n\t"
"mov pc, %[i_fn] \n\t"
"str r0, %[rn] \n\t" /* in case of writeback */
"str r2, %[rd0] \n\t"
"str r3, %[rd1] \n\t"
: [rn] "+m" (rnv),
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
[rd0] "=m" (regs->uregs[rd]),
[rd1] "=m" (regs->uregs[rd+1])
: [rm] "m" (rmv),
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
[cpsr] "r" (regs->ARM_cpsr),
[i_fn] "r" (i_fn)
: "r0", "r1", "r2", "r3", "lr", "cc"
);
if (is_writeback(insn))
regs->uregs[rn] = rnv;
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
}
static void __kprobes emulate_strd(struct kprobe *p, struct pt_regs *regs)
{
insn_4arg_fn_t *i_fn = (insn_4arg_fn_t *)&p->ainsn.insn[0];
kprobe_opcode_t insn = p->opcode;
long ppc = (long)p->addr + 8;
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
int rd = (insn >> 12) & 0xf;
int rn = (insn >> 16) & 0xf;
int rm = insn & 0xf;
long rnv = (rn == 15) ? ppc : regs->uregs[rn];
/* rm/rmv may be invalid, don't care. */
long rmv = (rm == 15) ? ppc : regs->uregs[rm];
long rnv_wb;
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
rnv_wb = insnslot_4arg_rflags(rnv, rmv, regs->uregs[rd],
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
regs->uregs[rd+1],
regs->ARM_cpsr, i_fn);
if (is_writeback(insn))
regs->uregs[rn] = rnv_wb;
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
}
static void __kprobes emulate_ldr(struct kprobe *p, struct pt_regs *regs)
{
insn_llret_3arg_fn_t *i_fn = (insn_llret_3arg_fn_t *)&p->ainsn.insn[0];
kprobe_opcode_t insn = p->opcode;
long ppc = (long)p->addr + 8;
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
union reg_pair fnr;
int rd = (insn >> 12) & 0xf;
int rn = (insn >> 16) & 0xf;
int rm = insn & 0xf;
long rdv;
long rnv = (rn == 15) ? ppc : regs->uregs[rn];
long rmv = (rm == 15) ? ppc : regs->uregs[rm];
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
long cpsr = regs->ARM_cpsr;
fnr.dr = insnslot_llret_3arg_rflags(rnv, 0, rmv, cpsr, i_fn);
if (rn != 15)
regs->uregs[rn] = fnr.r0; /* Save Rn in case of writeback. */
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
rdv = fnr.r1;
if (rd == 15) {
#if __LINUX_ARM_ARCH__ >= 5
cpsr &= ~PSR_T_BIT;
if (rdv & 0x1)
cpsr |= PSR_T_BIT;
regs->ARM_cpsr = cpsr;
rdv &= ~0x1;
#else
rdv &= ~0x2;
#endif
}
regs->uregs[rd] = rdv;
}
static void __kprobes emulate_str(struct kprobe *p, struct pt_regs *regs)
{
insn_3arg_fn_t *i_fn = (insn_3arg_fn_t *)&p->ainsn.insn[0];
kprobe_opcode_t insn = p->opcode;
long iaddr = (long)p->addr;
int rd = (insn >> 12) & 0xf;
int rn = (insn >> 16) & 0xf;
int rm = insn & 0xf;
long rdv = (rd == 15) ? iaddr + str_pc_offset : regs->uregs[rd];
long rnv = (rn == 15) ? iaddr + 8 : regs->uregs[rn];
long rmv = regs->uregs[rm]; /* rm/rmv may be invalid, don't care. */
long rnv_wb;
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
rnv_wb = insnslot_3arg_rflags(rnv, rdv, rmv, regs->ARM_cpsr, i_fn);
if (rn != 15)
regs->uregs[rn] = rnv_wb; /* Save Rn in case of writeback. */
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
}
static void __kprobes emulate_sat(struct kprobe *p, struct pt_regs *regs)
{
insn_1arg_fn_t *i_fn = (insn_1arg_fn_t *)&p->ainsn.insn[0];
kprobe_opcode_t insn = p->opcode;
int rd = (insn >> 12) & 0xf;
int rm = insn & 0xf;
long rmv = regs->uregs[rm];
/* Writes Q flag */
regs->uregs[rd] = insnslot_1arg_rwflags(rmv, &regs->ARM_cpsr, i_fn);
}
static void __kprobes emulate_sel(struct kprobe *p, struct pt_regs *regs)
{
insn_2arg_fn_t *i_fn = (insn_2arg_fn_t *)&p->ainsn.insn[0];
kprobe_opcode_t insn = p->opcode;
int rd = (insn >> 12) & 0xf;
int rn = (insn >> 16) & 0xf;
int rm = insn & 0xf;
long rnv = regs->uregs[rn];
long rmv = regs->uregs[rm];
/* Reads GE bits */
regs->uregs[rd] = insnslot_2arg_rflags(rnv, rmv, regs->ARM_cpsr, i_fn);
}
static void __kprobes emulate_none(struct kprobe *p, struct pt_regs *regs)
{
insn_0arg_fn_t *i_fn = (insn_0arg_fn_t *)&p->ainsn.insn[0];
insnslot_0arg_rflags(regs->ARM_cpsr, i_fn);
}
static void __kprobes emulate_nop(struct kprobe *p, struct pt_regs *regs)
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
{
}
static void __kprobes
emulate_rd12_modify(struct kprobe *p, struct pt_regs *regs)
{
insn_1arg_fn_t *i_fn = (insn_1arg_fn_t *)&p->ainsn.insn[0];
kprobe_opcode_t insn = p->opcode;
int rd = (insn >> 12) & 0xf;
long rdv = regs->uregs[rd];
regs->uregs[rd] = insnslot_1arg_rflags(rdv, regs->ARM_cpsr, i_fn);
}
static void __kprobes
emulate_rd12rn0_modify(struct kprobe *p, struct pt_regs *regs)
{
insn_2arg_fn_t *i_fn = (insn_2arg_fn_t *)&p->ainsn.insn[0];
kprobe_opcode_t insn = p->opcode;
int rd = (insn >> 12) & 0xf;
int rn = insn & 0xf;
long rdv = regs->uregs[rd];
long rnv = regs->uregs[rn];
regs->uregs[rd] = insnslot_2arg_rflags(rdv, rnv, regs->ARM_cpsr, i_fn);
}
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
static void __kprobes emulate_rd12rm0(struct kprobe *p, struct pt_regs *regs)
{
insn_1arg_fn_t *i_fn = (insn_1arg_fn_t *)&p->ainsn.insn[0];
kprobe_opcode_t insn = p->opcode;
int rd = (insn >> 12) & 0xf;
int rm = insn & 0xf;
long rmv = regs->uregs[rm];
regs->uregs[rd] = insnslot_1arg_rflags(rmv, regs->ARM_cpsr, i_fn);
}
static void __kprobes
emulate_rd12rn16rm0_rwflags(struct kprobe *p, struct pt_regs *regs)
{
insn_2arg_fn_t *i_fn = (insn_2arg_fn_t *)&p->ainsn.insn[0];
kprobe_opcode_t insn = p->opcode;
int rd = (insn >> 12) & 0xf;
int rn = (insn >> 16) & 0xf;
int rm = insn & 0xf;
long rnv = regs->uregs[rn];
long rmv = regs->uregs[rm];
regs->uregs[rd] =
insnslot_2arg_rwflags(rnv, rmv, &regs->ARM_cpsr, i_fn);
}
static void __kprobes
emulate_rd16rn12rs8rm0_rwflags(struct kprobe *p, struct pt_regs *regs)
{
insn_3arg_fn_t *i_fn = (insn_3arg_fn_t *)&p->ainsn.insn[0];
kprobe_opcode_t insn = p->opcode;
int rd = (insn >> 16) & 0xf;
int rn = (insn >> 12) & 0xf;
int rs = (insn >> 8) & 0xf;
int rm = insn & 0xf;
long rnv = regs->uregs[rn];
long rsv = regs->uregs[rs];
long rmv = regs->uregs[rm];
regs->uregs[rd] =
insnslot_3arg_rwflags(rnv, rsv, rmv, &regs->ARM_cpsr, i_fn);
}
static void __kprobes
emulate_rd16rs8rm0_rwflags(struct kprobe *p, struct pt_regs *regs)
{
insn_2arg_fn_t *i_fn = (insn_2arg_fn_t *)&p->ainsn.insn[0];
kprobe_opcode_t insn = p->opcode;
int rd = (insn >> 16) & 0xf;
int rs = (insn >> 8) & 0xf;
int rm = insn & 0xf;
long rsv = regs->uregs[rs];
long rmv = regs->uregs[rm];
regs->uregs[rd] =
insnslot_2arg_rwflags(rsv, rmv, &regs->ARM_cpsr, i_fn);
}
static void __kprobes
emulate_rdhi16rdlo12rs8rm0_rwflags(struct kprobe *p, struct pt_regs *regs)
{
insn_llret_4arg_fn_t *i_fn = (insn_llret_4arg_fn_t *)&p->ainsn.insn[0];
kprobe_opcode_t insn = p->opcode;
union reg_pair fnr;
int rdhi = (insn >> 16) & 0xf;
int rdlo = (insn >> 12) & 0xf;
int rs = (insn >> 8) & 0xf;
int rm = insn & 0xf;
long rsv = regs->uregs[rs];
long rmv = regs->uregs[rm];
fnr.dr = insnslot_llret_4arg_rwflags(regs->uregs[rdhi],
regs->uregs[rdlo], rsv, rmv,
&regs->ARM_cpsr, i_fn);
regs->uregs[rdhi] = fnr.r0;
regs->uregs[rdlo] = fnr.r1;
}
static void __kprobes
emulate_alu_imm_rflags(struct kprobe *p, struct pt_regs *regs)
{
insn_1arg_fn_t *i_fn = (insn_1arg_fn_t *)&p->ainsn.insn[0];
kprobe_opcode_t insn = p->opcode;
int rd = (insn >> 12) & 0xf;
int rn = (insn >> 16) & 0xf;
long rnv = (rn == 15) ? (long)p->addr + 8 : regs->uregs[rn];
regs->uregs[rd] = insnslot_1arg_rflags(rnv, regs->ARM_cpsr, i_fn);
}
static void __kprobes
emulate_alu_imm_rwflags(struct kprobe *p, struct pt_regs *regs)
{
insn_1arg_fn_t *i_fn = (insn_1arg_fn_t *)&p->ainsn.insn[0];
kprobe_opcode_t insn = p->opcode;
int rd = (insn >> 12) & 0xf;
int rn = (insn >> 16) & 0xf;
long rnv = (rn == 15) ? (long)p->addr + 8 : regs->uregs[rn];
regs->uregs[rd] = insnslot_1arg_rwflags(rnv, &regs->ARM_cpsr, i_fn);
}
static void __kprobes
emulate_alu_tests_imm(struct kprobe *p, struct pt_regs *regs)
{
insn_1arg_fn_t *i_fn = (insn_1arg_fn_t *)&p->ainsn.insn[0];
kprobe_opcode_t insn = p->opcode;
int rn = (insn >> 16) & 0xf;
long rnv = (rn == 15) ? (long)p->addr + 8 : regs->uregs[rn];
insnslot_1arg_rwflags(rnv, &regs->ARM_cpsr, i_fn);
}
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
static void __kprobes
emulate_alu_rflags(struct kprobe *p, struct pt_regs *regs)
{
insn_3arg_fn_t *i_fn = (insn_3arg_fn_t *)&p->ainsn.insn[0];
kprobe_opcode_t insn = p->opcode;
long ppc = (long)p->addr + 8;
int rd = (insn >> 12) & 0xf;
int rn = (insn >> 16) & 0xf; /* rn/rnv/rs/rsv may be */
int rs = (insn >> 8) & 0xf; /* invalid, don't care. */
int rm = insn & 0xf;
long rnv = (rn == 15) ? ppc : regs->uregs[rn];
long rmv = (rm == 15) ? ppc : regs->uregs[rm];
long rsv = regs->uregs[rs];
regs->uregs[rd] =
insnslot_3arg_rflags(rnv, rmv, rsv, regs->ARM_cpsr, i_fn);
}
static void __kprobes
emulate_alu_rwflags(struct kprobe *p, struct pt_regs *regs)
{
insn_3arg_fn_t *i_fn = (insn_3arg_fn_t *)&p->ainsn.insn[0];
kprobe_opcode_t insn = p->opcode;
long ppc = (long)p->addr + 8;
int rd = (insn >> 12) & 0xf;
int rn = (insn >> 16) & 0xf; /* rn/rnv/rs/rsv may be */
int rs = (insn >> 8) & 0xf; /* invalid, don't care. */
int rm = insn & 0xf;
long rnv = (rn == 15) ? ppc : regs->uregs[rn];
long rmv = (rm == 15) ? ppc : regs->uregs[rm];
long rsv = regs->uregs[rs];
regs->uregs[rd] =
insnslot_3arg_rwflags(rnv, rmv, rsv, &regs->ARM_cpsr, i_fn);
}
static void __kprobes
emulate_alu_tests(struct kprobe *p, struct pt_regs *regs)
{
insn_3arg_fn_t *i_fn = (insn_3arg_fn_t *)&p->ainsn.insn[0];
kprobe_opcode_t insn = p->opcode;
long ppc = (long)p->addr + 8;
int rn = (insn >> 16) & 0xf;
int rs = (insn >> 8) & 0xf; /* rs/rsv may be invalid, don't care. */
int rm = insn & 0xf;
long rnv = (rn == 15) ? ppc : regs->uregs[rn];
long rmv = (rm == 15) ? ppc : regs->uregs[rm];
long rsv = regs->uregs[rs];
insnslot_3arg_rwflags(rnv, rmv, rsv, &regs->ARM_cpsr, i_fn);
}
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
static enum kprobe_insn __kprobes
prep_emulate_ldr_str(kprobe_opcode_t insn, struct arch_specific_insn *asi)
{
int not_imm = (insn & (1 << 26)) ? (insn & (1 << 25))
: (~insn & (1 << 22));
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
if (is_writeback(insn) && is_r15(insn, 16))
return INSN_REJECTED; /* Writeback to PC */
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
insn &= 0xfff00fff;
insn |= 0x00001000; /* Rn = r0, Rd = r1 */
if (not_imm) {
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
insn &= ~0xf;
insn |= 2; /* Rm = r2 */
}
asi->insn[0] = insn;
asi->insn_handler = (insn & (1 << 20)) ? emulate_ldr : emulate_str;
return INSN_GOOD;
}
static enum kprobe_insn __kprobes
prep_emulate_rd12_modify(kprobe_opcode_t insn, struct arch_specific_insn *asi)
{
if (is_r15(insn, 12))
return INSN_REJECTED; /* Rd is PC */
insn &= 0xffff0fff; /* Rd = r0 */
asi->insn[0] = insn;
asi->insn_handler = emulate_rd12_modify;
return INSN_GOOD;
}
static enum kprobe_insn __kprobes
prep_emulate_rd12rn0_modify(kprobe_opcode_t insn,
struct arch_specific_insn *asi)
{
if (is_r15(insn, 12))
return INSN_REJECTED; /* Rd is PC */
insn &= 0xffff0ff0; /* Rd = r0 */
insn |= 0x00000001; /* Rn = r1 */
asi->insn[0] = insn;
asi->insn_handler = emulate_rd12rn0_modify;
return INSN_GOOD;
}
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
static enum kprobe_insn __kprobes
prep_emulate_rd12rm0(kprobe_opcode_t insn, struct arch_specific_insn *asi)
{
if (is_r15(insn, 12))
return INSN_REJECTED; /* Rd is PC */
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
insn &= 0xffff0ff0; /* Rd = r0, Rm = r0 */
asi->insn[0] = insn;
asi->insn_handler = emulate_rd12rm0;
return INSN_GOOD;
}
static enum kprobe_insn __kprobes
prep_emulate_rd12rn16rm0_wflags(kprobe_opcode_t insn,
struct arch_specific_insn *asi)
{
if (is_r15(insn, 12))
return INSN_REJECTED; /* Rd is PC */
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
insn &= 0xfff00ff0; /* Rd = r0, Rn = r0 */
insn |= 0x00000001; /* Rm = r1 */
asi->insn[0] = insn;
asi->insn_handler = emulate_rd12rn16rm0_rwflags;
return INSN_GOOD;
}
static enum kprobe_insn __kprobes
prep_emulate_rd16rs8rm0_wflags(kprobe_opcode_t insn,
struct arch_specific_insn *asi)
{
if (is_r15(insn, 16))
return INSN_REJECTED; /* Rd is PC */
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
insn &= 0xfff0f0f0; /* Rd = r0, Rs = r0 */
insn |= 0x00000001; /* Rm = r1 */
asi->insn[0] = insn;
asi->insn_handler = emulate_rd16rs8rm0_rwflags;
return INSN_GOOD;
}
static enum kprobe_insn __kprobes
prep_emulate_rd16rn12rs8rm0_wflags(kprobe_opcode_t insn,
struct arch_specific_insn *asi)
{
if (is_r15(insn, 16))
return INSN_REJECTED; /* Rd is PC */
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
insn &= 0xfff000f0; /* Rd = r0, Rn = r0 */
insn |= 0x00000102; /* Rs = r1, Rm = r2 */
asi->insn[0] = insn;
asi->insn_handler = emulate_rd16rn12rs8rm0_rwflags;
return INSN_GOOD;
}
static enum kprobe_insn __kprobes
prep_emulate_rdhi16rdlo12rs8rm0_wflags(kprobe_opcode_t insn,
struct arch_specific_insn *asi)
{
if (is_r15(insn, 16) || is_r15(insn, 12))
return INSN_REJECTED; /* RdHi or RdLo is PC */
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
insn &= 0xfff000f0; /* RdHi = r0, RdLo = r1 */
insn |= 0x00001203; /* Rs = r2, Rm = r3 */
asi->insn[0] = insn;
asi->insn_handler = emulate_rdhi16rdlo12rs8rm0_rwflags;
return INSN_GOOD;
}
static void __kprobes
emulate_ldrdstrd(struct kprobe *p, struct pt_regs *regs)
{
kprobe_opcode_t insn = p->opcode;
unsigned long pc = (unsigned long)p->addr + 8;
int rt = (insn >> 12) & 0xf;
int rn = (insn >> 16) & 0xf;
int rm = insn & 0xf;
register unsigned long rtv asm("r0") = regs->uregs[rt];
register unsigned long rt2v asm("r1") = regs->uregs[rt+1];
register unsigned long rnv asm("r2") = (rn == 15) ? pc
: regs->uregs[rn];
register unsigned long rmv asm("r3") = regs->uregs[rm];
__asm__ __volatile__ (
BLX("%[fn]")
: "=r" (rtv), "=r" (rt2v), "=r" (rnv)
: "0" (rtv), "1" (rt2v), "2" (rnv), "r" (rmv),
[fn] "r" (p->ainsn.insn_fn)
: "lr", "memory", "cc"
);
regs->uregs[rt] = rtv;
regs->uregs[rt+1] = rt2v;
if (is_writeback(insn))
regs->uregs[rn] = rnv;
}
static void __kprobes
emulate_rd12rn16rm0rs8_rwflags(struct kprobe *p, struct pt_regs *regs)
{
kprobe_opcode_t insn = p->opcode;
unsigned long pc = (unsigned long)p->addr + 8;
int rd = (insn >> 12) & 0xf;
int rn = (insn >> 16) & 0xf;
int rm = insn & 0xf;
int rs = (insn >> 8) & 0xf;
register unsigned long rdv asm("r0") = regs->uregs[rd];
register unsigned long rnv asm("r2") = (rn == 15) ? pc
: regs->uregs[rn];
register unsigned long rmv asm("r3") = (rm == 15) ? pc
: regs->uregs[rm];
register unsigned long rsv asm("r1") = regs->uregs[rs];
unsigned long cpsr = regs->ARM_cpsr;
__asm__ __volatile__ (
"msr cpsr_fs, %[cpsr] \n\t"
BLX("%[fn]")
"mrs %[cpsr], cpsr \n\t"
: "=r" (rdv), [cpsr] "=r" (cpsr)
: "0" (rdv), "r" (rnv), "r" (rmv), "r" (rsv),
"1" (cpsr), [fn] "r" (p->ainsn.insn_fn)
: "lr", "memory", "cc"
);
if (rd == 15)
alu_write_pc(rdv, regs);
else
regs->uregs[rd] = rdv;
regs->ARM_cpsr = (regs->ARM_cpsr & ~APSR_MASK) | (cpsr & APSR_MASK);
}
static void __kprobes
emulate_rd12rn16rm0_rwflags_nopc(struct kprobe *p, struct pt_regs *regs)
{
kprobe_opcode_t insn = p->opcode;
int rd = (insn >> 12) & 0xf;
int rn = (insn >> 16) & 0xf;
int rm = insn & 0xf;
register unsigned long rdv asm("r0") = regs->uregs[rd];
register unsigned long rnv asm("r2") = regs->uregs[rn];
register unsigned long rmv asm("r3") = regs->uregs[rm];
unsigned long cpsr = regs->ARM_cpsr;
__asm__ __volatile__ (
"msr cpsr_fs, %[cpsr] \n\t"
BLX("%[fn]")
"mrs %[cpsr], cpsr \n\t"
: "=r" (rdv), [cpsr] "=r" (cpsr)
: "0" (rdv), "r" (rnv), "r" (rmv),
"1" (cpsr), [fn] "r" (p->ainsn.insn_fn)
: "lr", "memory", "cc"
);
regs->uregs[rd] = rdv;
regs->ARM_cpsr = (regs->ARM_cpsr & ~APSR_MASK) | (cpsr & APSR_MASK);
}
static void __kprobes
emulate_rd16rn12rm0rs8_rwflags_nopc(struct kprobe *p, struct pt_regs *regs)
{
kprobe_opcode_t insn = p->opcode;
int rd = (insn >> 16) & 0xf;
int rn = (insn >> 12) & 0xf;
int rm = insn & 0xf;
int rs = (insn >> 8) & 0xf;
register unsigned long rdv asm("r2") = regs->uregs[rd];
register unsigned long rnv asm("r0") = regs->uregs[rn];
register unsigned long rmv asm("r3") = regs->uregs[rm];
register unsigned long rsv asm("r1") = regs->uregs[rs];
unsigned long cpsr = regs->ARM_cpsr;
__asm__ __volatile__ (
"msr cpsr_fs, %[cpsr] \n\t"
BLX("%[fn]")
"mrs %[cpsr], cpsr \n\t"
: "=r" (rdv), [cpsr] "=r" (cpsr)
: "0" (rdv), "r" (rnv), "r" (rmv), "r" (rsv),
"1" (cpsr), [fn] "r" (p->ainsn.insn_fn)
: "lr", "memory", "cc"
);
regs->uregs[rd] = rdv;
regs->ARM_cpsr = (regs->ARM_cpsr & ~APSR_MASK) | (cpsr & APSR_MASK);
}
static void __kprobes
emulate_rd12rm0_noflags_nopc(struct kprobe *p, struct pt_regs *regs)
{
kprobe_opcode_t insn = p->opcode;
int rd = (insn >> 12) & 0xf;
int rm = insn & 0xf;
register unsigned long rdv asm("r0") = regs->uregs[rd];
register unsigned long rmv asm("r3") = regs->uregs[rm];
__asm__ __volatile__ (
BLX("%[fn]")
: "=r" (rdv)
: "0" (rdv), "r" (rmv), [fn] "r" (p->ainsn.insn_fn)
: "lr", "memory", "cc"
);
regs->uregs[rd] = rdv;
}
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
/*
* For the instruction masking and comparisons in all the "space_*"
* functions below, Do _not_ rearrange the order of tests unless
* you're very, very sure of what you are doing. For the sake of
* efficiency, the masks for some tests sometimes assume other test
* have been done prior to them so the number of patterns to test
* for an instruction set can be as broad as possible to reduce the
* number of tests needed.
*/
static const union decode_item arm_1111_table[] = {
/* Unconditional instructions */
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
/* memory hint 1111 0100 x001 xxxx xxxx xxxx xxxx xxxx */
/* PLDI (immediate) 1111 0100 x101 xxxx xxxx xxxx xxxx xxxx */
/* PLDW (immediate) 1111 0101 x001 xxxx xxxx xxxx xxxx xxxx */
/* PLD (immediate) 1111 0101 x101 xxxx xxxx xxxx xxxx xxxx */
DECODE_SIMULATE (0xfe300000, 0xf4100000, kprobe_simulate_nop),
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
/* BLX (immediate) 1111 101x xxxx xxxx xxxx xxxx xxxx xxxx */
DECODE_SIMULATE (0xfe000000, 0xfa000000, simulate_blx1),
/* CPS 1111 0001 0000 xxx0 xxxx xxxx xx0x xxxx */
/* SETEND 1111 0001 0000 0001 xxxx xxxx 0000 xxxx */
/* SRS 1111 100x x1x0 xxxx xxxx xxxx xxxx xxxx */
/* RFE 1111 100x x0x1 xxxx xxxx xxxx xxxx xxxx */
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
/* Coprocessor instructions... */
/* MCRR2 1111 1100 0100 xxxx xxxx xxxx xxxx xxxx */
/* MRRC2 1111 1100 0101 xxxx xxxx xxxx xxxx xxxx */
/* LDC2 1111 110x xxx1 xxxx xxxx xxxx xxxx xxxx */
/* STC2 1111 110x xxx0 xxxx xxxx xxxx xxxx xxxx */
/* CDP2 1111 1110 xxxx xxxx xxxx xxxx xxx0 xxxx */
/* MCR2 1111 1110 xxx0 xxxx xxxx xxxx xxx1 xxxx */
/* MRC2 1111 1110 xxx1 xxxx xxxx xxxx xxx1 xxxx */
/* Other unallocated instructions... */
DECODE_END
};
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
static const union decode_item arm_cccc_0001_0xx0____0xxx_table[] = {
/* Miscellaneous instructions */
/* MRS cpsr cccc 0001 0000 xxxx xxxx xxxx 0000 xxxx */
DECODE_SIMULATEX(0x0ff000f0, 0x01000000, simulate_mrs,
REGS(0, NOPC, 0, 0, 0)),
/* BX cccc 0001 0010 xxxx xxxx xxxx 0001 xxxx */
DECODE_SIMULATE (0x0ff000f0, 0x01200010, simulate_blx2bx),
/* BLX (register) cccc 0001 0010 xxxx xxxx xxxx 0011 xxxx */
DECODE_SIMULATEX(0x0ff000f0, 0x01200030, simulate_blx2bx,
REGS(0, 0, 0, 0, NOPC)),
/* CLZ cccc 0001 0110 xxxx xxxx xxxx 0001 xxxx */
DECODE_EMULATEX (0x0ff000f0, 0x01600010, emulate_rd12rm0_noflags_nopc,
REGS(0, NOPC, 0, 0, NOPC)),
/* QADD cccc 0001 0000 xxxx xxxx xxxx 0101 xxxx */
/* QSUB cccc 0001 0010 xxxx xxxx xxxx 0101 xxxx */
/* QDADD cccc 0001 0100 xxxx xxxx xxxx 0101 xxxx */
/* QDSUB cccc 0001 0110 xxxx xxxx xxxx 0101 xxxx */
DECODE_EMULATEX (0x0f9000f0, 0x01000050, emulate_rd12rn16rm0_rwflags_nopc,
REGS(NOPC, NOPC, 0, 0, NOPC)),
/* BXJ cccc 0001 0010 xxxx xxxx xxxx 0010 xxxx */
/* MSR cccc 0001 0x10 xxxx xxxx xxxx 0000 xxxx */
/* MRS spsr cccc 0001 0100 xxxx xxxx xxxx 0000 xxxx */
/* BKPT 1110 0001 0010 xxxx xxxx xxxx 0111 xxxx */
/* SMC cccc 0001 0110 xxxx xxxx xxxx 0111 xxxx */
/* And unallocated instructions... */
DECODE_END
};
static const union decode_item arm_cccc_0001_0xx0____1xx0_table[] = {
/* Halfword multiply and multiply-accumulate */
/* SMLALxy cccc 0001 0100 xxxx xxxx xxxx 1xx0 xxxx */
DECODE_CUSTOM (0x0ff00090, 0x01400080, prep_emulate_rdhi16rdlo12rs8rm0_wflags),
/* SMULWy cccc 0001 0010 xxxx xxxx xxxx 1x10 xxxx */
DECODE_OR (0x0ff000b0, 0x012000a0),
/* SMULxy cccc 0001 0110 xxxx xxxx xxxx 1xx0 xxxx */
DECODE_EMULATEX (0x0ff00090, 0x01600080, emulate_rd16rn12rm0rs8_rwflags_nopc,
REGS(NOPC, 0, NOPC, 0, NOPC)),
/* SMLAxy cccc 0001 0000 xxxx xxxx xxxx 1xx0 xxxx */
DECODE_OR (0x0ff00090, 0x01000080),
/* SMLAWy cccc 0001 0010 xxxx xxxx xxxx 1x00 xxxx */
DECODE_EMULATEX (0x0ff000b0, 0x01200080, emulate_rd16rn12rm0rs8_rwflags_nopc,
REGS(NOPC, NOPC, NOPC, 0, NOPC)),
DECODE_END
};
static const union decode_item arm_cccc_0000_____1001_table[] = {
/* Multiply and multiply-accumulate */
/* MUL cccc 0000 0000 xxxx xxxx xxxx 1001 xxxx */
/* MULS cccc 0000 0001 xxxx xxxx xxxx 1001 xxxx */
DECODE_EMULATEX (0x0fe000f0, 0x00000090, emulate_rd16rn12rm0rs8_rwflags_nopc,
REGS(NOPC, 0, NOPC, 0, NOPC)),
/* MLA cccc 0000 0010 xxxx xxxx xxxx 1001 xxxx */
/* MLAS cccc 0000 0011 xxxx xxxx xxxx 1001 xxxx */
DECODE_OR (0x0fe000f0, 0x00200090),
/* MLS cccc 0000 0110 xxxx xxxx xxxx 1001 xxxx */
DECODE_EMULATEX (0x0ff000f0, 0x00600090, emulate_rd16rn12rm0rs8_rwflags_nopc,
REGS(NOPC, NOPC, NOPC, 0, NOPC)),
/* UMAAL cccc 0000 0100 xxxx xxxx xxxx 1001 xxxx */
DECODE_OR (0x0ff000f0, 0x00400090),
/* UMULL cccc 0000 1000 xxxx xxxx xxxx 1001 xxxx */
/* UMULLS cccc 0000 1001 xxxx xxxx xxxx 1001 xxxx */
/* UMLAL cccc 0000 1010 xxxx xxxx xxxx 1001 xxxx */
/* UMLALS cccc 0000 1011 xxxx xxxx xxxx 1001 xxxx */
/* SMULL cccc 0000 1100 xxxx xxxx xxxx 1001 xxxx */
/* SMULLS cccc 0000 1101 xxxx xxxx xxxx 1001 xxxx */
/* SMLAL cccc 0000 1110 xxxx xxxx xxxx 1001 xxxx */
/* SMLALS cccc 0000 1111 xxxx xxxx xxxx 1001 xxxx */
DECODE_CUSTOM (0x0f8000f0, 0x00800090, prep_emulate_rdhi16rdlo12rs8rm0_wflags),
DECODE_END
};
static const union decode_item arm_cccc_0001_____1001_table[] = {
/* Synchronization primitives */
/* SMP/SWPB cccc 0001 0x00 xxxx xxxx xxxx 1001 xxxx */
DECODE_EMULATEX (0x0fb000f0, 0x01000090, emulate_rd12rn16rm0_rwflags_nopc,
REGS(NOPC, NOPC, 0, 0, NOPC)),
/* LDREX/STREX{,D,B,H} cccc 0001 1xxx xxxx xxxx xxxx 1001 xxxx */
/* And unallocated instructions... */
DECODE_END
};
static const union decode_item arm_cccc_000x_____1xx1_table[] = {
/* Extra load/store instructions */
/* LDRD/STRD lr,pc,{... cccc 000x x0x0 xxxx 111x xxxx 1101 xxxx */
DECODE_REJECT (0x0e10e0d0, 0x0000e0d0),
/* LDRD (register) cccc 000x x0x0 xxxx xxxx xxxx 1101 xxxx */
/* STRD (register) cccc 000x x0x0 xxxx xxxx xxxx 1111 xxxx */
DECODE_EMULATEX (0x0e5000d0, 0x000000d0, emulate_ldrdstrd,
REGS(NOPCWB, NOPCX, 0, 0, NOPC)),
/* LDRD (immediate) cccc 000x x1x0 xxxx xxxx xxxx 1101 xxxx */
/* STRD (immediate) cccc 000x x1x0 xxxx xxxx xxxx 1111 xxxx */
DECODE_EMULATEX (0x0e5000d0, 0x004000d0, emulate_ldrdstrd,
REGS(NOPCWB, NOPCX, 0, 0, 0)),
/* Reject Rd is PC */
/* TODO: fold this into next entry when it is made a DECODE_EMULATE */
DECODE_REJECT (0x0000f000, 0x0000f000),
/* STRH (register) cccc 000x x0x0 xxxx xxxx xxxx 1011 xxxx */
/* LDRH (register) cccc 000x x0x1 xxxx xxxx xxxx 1011 xxxx */
/* LDRSB (register) cccc 000x x0x1 xxxx xxxx xxxx 1101 xxxx */
/* LDRSH (register) cccc 000x x0x1 xxxx xxxx xxxx 1111 xxxx */
/* STRH (immediate) cccc 000x x1x0 xxxx xxxx xxxx 1011 xxxx */
/* LDRH (immediate) cccc 000x x1x1 xxxx xxxx xxxx 1011 xxxx */
/* LDRSB (immediate) cccc 000x x1x1 xxxx xxxx xxxx 1101 xxxx */
/* LDRSH (immediate) cccc 000x x1x1 xxxx xxxx xxxx 1111 xxxx */
DECODE_CUSTOM (0x0e000090, 0x00000090, prep_emulate_ldr_str),
DECODE_END
};
static const union decode_item arm_cccc_000x_table[] = {
/* Data-processing (register) */
/* <op>S PC, ... cccc 000x xxx1 xxxx 1111 xxxx xxxx xxxx */
DECODE_REJECT (0x0e10f000, 0x0010f000),
/* MOV IP, SP 1110 0001 1010 0000 1100 0000 0000 1101 */
DECODE_SIMULATE (0xffffffff, 0xe1a0c00d, simulate_mov_ipsp),
/* TST (register) cccc 0001 0001 xxxx xxxx xxxx xxx0 xxxx */
/* TEQ (register) cccc 0001 0011 xxxx xxxx xxxx xxx0 xxxx */
/* CMP (register) cccc 0001 0101 xxxx xxxx xxxx xxx0 xxxx */
/* CMN (register) cccc 0001 0111 xxxx xxxx xxxx xxx0 xxxx */
DECODE_EMULATEX (0x0f900010, 0x01100000, emulate_rd12rn16rm0rs8_rwflags,
REGS(ANY, 0, 0, 0, ANY)),
/* MOV (register) cccc 0001 101x xxxx xxxx xxxx xxx0 xxxx */
/* MVN (register) cccc 0001 111x xxxx xxxx xxxx xxx0 xxxx */
DECODE_EMULATEX (0x0fa00010, 0x01a00000, emulate_rd12rn16rm0rs8_rwflags,
REGS(0, ANY, 0, 0, ANY)),
/* AND (register) cccc 0000 000x xxxx xxxx xxxx xxx0 xxxx */
/* EOR (register) cccc 0000 001x xxxx xxxx xxxx xxx0 xxxx */
/* SUB (register) cccc 0000 010x xxxx xxxx xxxx xxx0 xxxx */
/* RSB (register) cccc 0000 011x xxxx xxxx xxxx xxx0 xxxx */
/* ADD (register) cccc 0000 100x xxxx xxxx xxxx xxx0 xxxx */
/* ADC (register) cccc 0000 101x xxxx xxxx xxxx xxx0 xxxx */
/* SBC (register) cccc 0000 110x xxxx xxxx xxxx xxx0 xxxx */
/* RSC (register) cccc 0000 111x xxxx xxxx xxxx xxx0 xxxx */
/* ORR (register) cccc 0001 100x xxxx xxxx xxxx xxx0 xxxx */
/* BIC (register) cccc 0001 110x xxxx xxxx xxxx xxx0 xxxx */
DECODE_EMULATEX (0x0e000010, 0x00000000, emulate_rd12rn16rm0rs8_rwflags,
REGS(ANY, ANY, 0, 0, ANY)),
/* TST (reg-shift reg) cccc 0001 0001 xxxx xxxx xxxx 0xx1 xxxx */
/* TEQ (reg-shift reg) cccc 0001 0011 xxxx xxxx xxxx 0xx1 xxxx */
/* CMP (reg-shift reg) cccc 0001 0101 xxxx xxxx xxxx 0xx1 xxxx */
/* CMN (reg-shift reg) cccc 0001 0111 xxxx xxxx xxxx 0xx1 xxxx */
DECODE_EMULATEX (0x0f900090, 0x01100010, emulate_rd12rn16rm0rs8_rwflags,
REGS(ANY, 0, NOPC, 0, ANY)),
/* MOV (reg-shift reg) cccc 0001 101x xxxx xxxx xxxx 0xx1 xxxx */
/* MVN (reg-shift reg) cccc 0001 111x xxxx xxxx xxxx 0xx1 xxxx */
DECODE_EMULATEX (0x0fa00090, 0x01a00010, emulate_rd12rn16rm0rs8_rwflags,
REGS(0, ANY, NOPC, 0, ANY)),
/* AND (reg-shift reg) cccc 0000 000x xxxx xxxx xxxx 0xx1 xxxx */
/* EOR (reg-shift reg) cccc 0000 001x xxxx xxxx xxxx 0xx1 xxxx */
/* SUB (reg-shift reg) cccc 0000 010x xxxx xxxx xxxx 0xx1 xxxx */
/* RSB (reg-shift reg) cccc 0000 011x xxxx xxxx xxxx 0xx1 xxxx */
/* ADD (reg-shift reg) cccc 0000 100x xxxx xxxx xxxx 0xx1 xxxx */
/* ADC (reg-shift reg) cccc 0000 101x xxxx xxxx xxxx 0xx1 xxxx */
/* SBC (reg-shift reg) cccc 0000 110x xxxx xxxx xxxx 0xx1 xxxx */
/* RSC (reg-shift reg) cccc 0000 111x xxxx xxxx xxxx 0xx1 xxxx */
/* ORR (reg-shift reg) cccc 0001 100x xxxx xxxx xxxx 0xx1 xxxx */
/* BIC (reg-shift reg) cccc 0001 110x xxxx xxxx xxxx 0xx1 xxxx */
DECODE_EMULATEX (0x0e000090, 0x00000010, emulate_rd12rn16rm0rs8_rwflags,
REGS(ANY, ANY, NOPC, 0, ANY)),
DECODE_END
};
static const union decode_item arm_cccc_001x_table[] = {
/* Data-processing (immediate) */
/* MOVW cccc 0011 0000 xxxx xxxx xxxx xxxx xxxx */
/* MOVT cccc 0011 0100 xxxx xxxx xxxx xxxx xxxx */
DECODE_EMULATEX (0x0fb00000, 0x03000000, emulate_rd12rm0_noflags_nopc,
REGS(0, NOPC, 0, 0, 0)),
/* YIELD cccc 0011 0010 0000 xxxx xxxx 0000 0001 */
DECODE_OR (0x0fff00ff, 0x03200001),
/* SEV cccc 0011 0010 0000 xxxx xxxx 0000 0100 */
DECODE_EMULATE (0x0fff00ff, 0x03200004, kprobe_emulate_none),
/* NOP cccc 0011 0010 0000 xxxx xxxx 0000 0000 */
/* WFE cccc 0011 0010 0000 xxxx xxxx 0000 0010 */
/* WFI cccc 0011 0010 0000 xxxx xxxx 0000 0011 */
DECODE_SIMULATE (0x0fff00fc, 0x03200000, kprobe_simulate_nop),
/* DBG cccc 0011 0010 0000 xxxx xxxx ffff xxxx */
/* unallocated hints cccc 0011 0010 0000 xxxx xxxx xxxx xxxx */
/* MSR (immediate) cccc 0011 0x10 xxxx xxxx xxxx xxxx xxxx */
DECODE_REJECT (0x0fb00000, 0x03200000),
/* <op>S PC, ... cccc 001x xxx1 xxxx 1111 xxxx xxxx xxxx */
DECODE_REJECT (0x0e10f000, 0x0210f000),
/* TST (immediate) cccc 0011 0001 xxxx xxxx xxxx xxxx xxxx */
/* TEQ (immediate) cccc 0011 0011 xxxx xxxx xxxx xxxx xxxx */
/* CMP (immediate) cccc 0011 0101 xxxx xxxx xxxx xxxx xxxx */
/* CMN (immediate) cccc 0011 0111 xxxx xxxx xxxx xxxx xxxx */
DECODE_EMULATEX (0x0f900000, 0x03100000, emulate_rd12rn16rm0rs8_rwflags,
REGS(ANY, 0, 0, 0, 0)),
/* MOV (immediate) cccc 0011 101x xxxx xxxx xxxx xxxx xxxx */
/* MVN (immediate) cccc 0011 111x xxxx xxxx xxxx xxxx xxxx */
DECODE_EMULATEX (0x0fa00000, 0x03a00000, emulate_rd12rn16rm0rs8_rwflags,
REGS(0, ANY, 0, 0, 0)),
/* AND (immediate) cccc 0010 000x xxxx xxxx xxxx xxxx xxxx */
/* EOR (immediate) cccc 0010 001x xxxx xxxx xxxx xxxx xxxx */
/* SUB (immediate) cccc 0010 010x xxxx xxxx xxxx xxxx xxxx */
/* RSB (immediate) cccc 0010 011x xxxx xxxx xxxx xxxx xxxx */
/* ADD (immediate) cccc 0010 100x xxxx xxxx xxxx xxxx xxxx */
/* ADC (immediate) cccc 0010 101x xxxx xxxx xxxx xxxx xxxx */
/* SBC (immediate) cccc 0010 110x xxxx xxxx xxxx xxxx xxxx */
/* RSC (immediate) cccc 0010 111x xxxx xxxx xxxx xxxx xxxx */
/* ORR (immediate) cccc 0011 100x xxxx xxxx xxxx xxxx xxxx */
/* BIC (immediate) cccc 0011 110x xxxx xxxx xxxx xxxx xxxx */
DECODE_EMULATEX (0x0e000000, 0x02000000, emulate_rd12rn16rm0rs8_rwflags,
REGS(ANY, ANY, 0, 0, 0)),
DECODE_END
};
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
static const union decode_item arm_cccc_0110_____xxx1_table[] = {
/* Media instructions */
/* SEL cccc 0110 1000 xxxx xxxx xxxx 1011 xxxx */
DECODE_EMULATEX (0x0ff000f0, 0x068000b0, emulate_rd12rn16rm0_rwflags_nopc,
REGS(NOPC, NOPC, 0, 0, NOPC)),
/* SSAT cccc 0110 101x xxxx xxxx xxxx xx01 xxxx */
/* USAT cccc 0110 111x xxxx xxxx xxxx xx01 xxxx */
DECODE_OR(0x0fa00030, 0x06a00010),
/* SSAT16 cccc 0110 1010 xxxx xxxx xxxx 0011 xxxx */
/* USAT16 cccc 0110 1110 xxxx xxxx xxxx 0011 xxxx */
DECODE_EMULATEX (0x0fb000f0, 0x06a00030, emulate_rd12rn16rm0_rwflags_nopc,
REGS(0, NOPC, 0, 0, NOPC)),
/* REV cccc 0110 1011 xxxx xxxx xxxx 0011 xxxx */
/* REV16 cccc 0110 1011 xxxx xxxx xxxx 1011 xxxx */
/* RBIT cccc 0110 1111 xxxx xxxx xxxx 0011 xxxx */
/* REVSH cccc 0110 1111 xxxx xxxx xxxx 1011 xxxx */
DECODE_EMULATEX (0x0fb00070, 0x06b00030, emulate_rd12rm0_noflags_nopc,
REGS(0, NOPC, 0, 0, NOPC)),
/* ??? cccc 0110 0x00 xxxx xxxx xxxx xxx1 xxxx */
DECODE_REJECT (0x0fb00010, 0x06000010),
/* ??? cccc 0110 0xxx xxxx xxxx xxxx 1011 xxxx */
DECODE_REJECT (0x0f8000f0, 0x060000b0),
/* ??? cccc 0110 0xxx xxxx xxxx xxxx 1101 xxxx */
DECODE_REJECT (0x0f8000f0, 0x060000d0),
/* SADD16 cccc 0110 0001 xxxx xxxx xxxx 0001 xxxx */
/* SADDSUBX cccc 0110 0001 xxxx xxxx xxxx 0011 xxxx */
/* SSUBADDX cccc 0110 0001 xxxx xxxx xxxx 0101 xxxx */
/* SSUB16 cccc 0110 0001 xxxx xxxx xxxx 0111 xxxx */
/* SADD8 cccc 0110 0001 xxxx xxxx xxxx 1001 xxxx */
/* SSUB8 cccc 0110 0001 xxxx xxxx xxxx 1111 xxxx */
/* QADD16 cccc 0110 0010 xxxx xxxx xxxx 0001 xxxx */
/* QADDSUBX cccc 0110 0010 xxxx xxxx xxxx 0011 xxxx */
/* QSUBADDX cccc 0110 0010 xxxx xxxx xxxx 0101 xxxx */
/* QSUB16 cccc 0110 0010 xxxx xxxx xxxx 0111 xxxx */
/* QADD8 cccc 0110 0010 xxxx xxxx xxxx 1001 xxxx */
/* QSUB8 cccc 0110 0010 xxxx xxxx xxxx 1111 xxxx */
/* SHADD16 cccc 0110 0011 xxxx xxxx xxxx 0001 xxxx */
/* SHADDSUBX cccc 0110 0011 xxxx xxxx xxxx 0011 xxxx */
/* SHSUBADDX cccc 0110 0011 xxxx xxxx xxxx 0101 xxxx */
/* SHSUB16 cccc 0110 0011 xxxx xxxx xxxx 0111 xxxx */
/* SHADD8 cccc 0110 0011 xxxx xxxx xxxx 1001 xxxx */
/* SHSUB8 cccc 0110 0011 xxxx xxxx xxxx 1111 xxxx */
/* UADD16 cccc 0110 0101 xxxx xxxx xxxx 0001 xxxx */
/* UADDSUBX cccc 0110 0101 xxxx xxxx xxxx 0011 xxxx */
/* USUBADDX cccc 0110 0101 xxxx xxxx xxxx 0101 xxxx */
/* USUB16 cccc 0110 0101 xxxx xxxx xxxx 0111 xxxx */
/* UADD8 cccc 0110 0101 xxxx xxxx xxxx 1001 xxxx */
/* USUB8 cccc 0110 0101 xxxx xxxx xxxx 1111 xxxx */
/* UQADD16 cccc 0110 0110 xxxx xxxx xxxx 0001 xxxx */
/* UQADDSUBX cccc 0110 0110 xxxx xxxx xxxx 0011 xxxx */
/* UQSUBADDX cccc 0110 0110 xxxx xxxx xxxx 0101 xxxx */
/* UQSUB16 cccc 0110 0110 xxxx xxxx xxxx 0111 xxxx */
/* UQADD8 cccc 0110 0110 xxxx xxxx xxxx 1001 xxxx */
/* UQSUB8 cccc 0110 0110 xxxx xxxx xxxx 1111 xxxx */
/* UHADD16 cccc 0110 0111 xxxx xxxx xxxx 0001 xxxx */
/* UHADDSUBX cccc 0110 0111 xxxx xxxx xxxx 0011 xxxx */
/* UHSUBADDX cccc 0110 0111 xxxx xxxx xxxx 0101 xxxx */
/* UHSUB16 cccc 0110 0111 xxxx xxxx xxxx 0111 xxxx */
/* UHADD8 cccc 0110 0111 xxxx xxxx xxxx 1001 xxxx */
/* UHSUB8 cccc 0110 0111 xxxx xxxx xxxx 1111 xxxx */
DECODE_EMULATEX (0x0f800010, 0x06000010, emulate_rd12rn16rm0_rwflags_nopc,
REGS(NOPC, NOPC, 0, 0, NOPC)),
/* PKHBT cccc 0110 1000 xxxx xxxx xxxx x001 xxxx */
/* PKHTB cccc 0110 1000 xxxx xxxx xxxx x101 xxxx */
DECODE_EMULATEX (0x0ff00030, 0x06800010, emulate_rd12rn16rm0_rwflags_nopc,
REGS(NOPC, NOPC, 0, 0, NOPC)),
/* ??? cccc 0110 1001 xxxx xxxx xxxx 0111 xxxx */
/* ??? cccc 0110 1101 xxxx xxxx xxxx 0111 xxxx */
DECODE_REJECT (0x0fb000f0, 0x06900070),
/* SXTB16 cccc 0110 1000 1111 xxxx xxxx 0111 xxxx */
/* SXTB cccc 0110 1010 1111 xxxx xxxx 0111 xxxx */
/* SXTH cccc 0110 1011 1111 xxxx xxxx 0111 xxxx */
/* UXTB16 cccc 0110 1100 1111 xxxx xxxx 0111 xxxx */
/* UXTB cccc 0110 1110 1111 xxxx xxxx 0111 xxxx */
/* UXTH cccc 0110 1111 1111 xxxx xxxx 0111 xxxx */
DECODE_EMULATEX (0x0f8f00f0, 0x068f0070, emulate_rd12rm0_noflags_nopc,
REGS(0, NOPC, 0, 0, NOPC)),
/* SXTAB16 cccc 0110 1000 xxxx xxxx xxxx 0111 xxxx */
/* SXTAB cccc 0110 1010 xxxx xxxx xxxx 0111 xxxx */
/* SXTAH cccc 0110 1011 xxxx xxxx xxxx 0111 xxxx */
/* UXTAB16 cccc 0110 1100 xxxx xxxx xxxx 0111 xxxx */
/* UXTAB cccc 0110 1110 xxxx xxxx xxxx 0111 xxxx */
/* UXTAH cccc 0110 1111 xxxx xxxx xxxx 0111 xxxx */
DECODE_EMULATEX (0x0f8000f0, 0x06800070, emulate_rd12rn16rm0_rwflags_nopc,
REGS(NOPCX, NOPC, 0, 0, NOPC)),
DECODE_END
};
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
static const union decode_item arm_cccc_0111_____xxx1_table[] = {
/* Media instructions */
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
/* UNDEFINED cccc 0111 1111 xxxx xxxx xxxx 1111 xxxx */
DECODE_REJECT (0x0ff000f0, 0x07f000f0),
/* SMLALD cccc 0111 0100 xxxx xxxx xxxx 00x1 xxxx */
/* SMLSLD cccc 0111 0100 xxxx xxxx xxxx 01x1 xxxx */
DECODE_CUSTOM (0x0ff00090, 0x07400010, prep_emulate_rdhi16rdlo12rs8rm0_wflags),
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
/* SMUAD cccc 0111 0000 xxxx 1111 xxxx 00x1 xxxx */
/* SMUSD cccc 0111 0000 xxxx 1111 xxxx 01x1 xxxx */
DECODE_OR (0x0ff0f090, 0x0700f010),
/* SMMUL cccc 0111 0101 xxxx 1111 xxxx 00x1 xxxx */
DECODE_OR (0x0ff0f0d0, 0x0750f010),
/* USAD8 cccc 0111 1000 xxxx 1111 xxxx 0001 xxxx */
DECODE_EMULATEX (0x0ff0f0f0, 0x0780f010, emulate_rd16rn12rm0rs8_rwflags_nopc,
REGS(NOPC, 0, NOPC, 0, NOPC)),
/* SMLAD cccc 0111 0000 xxxx xxxx xxxx 00x1 xxxx */
/* SMLSD cccc 0111 0000 xxxx xxxx xxxx 01x1 xxxx */
DECODE_OR (0x0ff00090, 0x07000010),
/* SMMLA cccc 0111 0101 xxxx xxxx xxxx 00x1 xxxx */
DECODE_OR (0x0ff000d0, 0x07500010),
/* USADA8 cccc 0111 1000 xxxx xxxx xxxx 0001 xxxx */
DECODE_EMULATEX (0x0ff000f0, 0x07800010, emulate_rd16rn12rm0rs8_rwflags_nopc,
REGS(NOPC, NOPCX, NOPC, 0, NOPC)),
/* SMMLS cccc 0111 0101 xxxx xxxx xxxx 11x1 xxxx */
DECODE_EMULATEX (0x0ff000d0, 0x075000d0, emulate_rd16rn12rm0rs8_rwflags_nopc,
REGS(NOPC, NOPC, NOPC, 0, NOPC)),
/* SBFX cccc 0111 101x xxxx xxxx xxxx x101 xxxx */
/* UBFX cccc 0111 111x xxxx xxxx xxxx x101 xxxx */
DECODE_EMULATEX (0x0fa00070, 0x07a00050, emulate_rd12rm0_noflags_nopc,
REGS(0, NOPC, 0, 0, NOPC)),
/* BFC cccc 0111 110x xxxx xxxx xxxx x001 1111 */
DECODE_EMULATEX (0x0fe0007f, 0x07c0001f, emulate_rd12rm0_noflags_nopc,
REGS(0, NOPC, 0, 0, 0)),
/* BFI cccc 0111 110x xxxx xxxx xxxx x001 xxxx */
DECODE_EMULATEX (0x0fe00070, 0x07c00010, emulate_rd12rm0_noflags_nopc,
REGS(0, NOPC, 0, 0, NOPCX)),
DECODE_END
};
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
static const union decode_item arm_cccc_01xx_table[] = {
/* Load/store word and unsigned byte */
/* LDRB/STRB pc,[...] cccc 01xx x0xx xxxx xxxx xxxx xxxx xxxx */
DECODE_REJECT (0x0c40f000, 0x0440f000),
/* LDR cccc 01xx x0x1 xxxx xxxx xxxx xxxx xxxx */
/* LDRB cccc 01xx x1x1 xxxx xxxx xxxx xxxx xxxx */
/* LDRBT cccc 01x0 x111 xxxx xxxx xxxx xxxx xxxx */
/* LDRT cccc 01x0 x011 xxxx xxxx xxxx xxxx xxxx */
/* STR cccc 01xx x0x0 xxxx xxxx xxxx xxxx xxxx */
/* STRB cccc 01xx x1x0 xxxx xxxx xxxx xxxx xxxx */
/* STRBT cccc 01x0 x110 xxxx xxxx xxxx xxxx xxxx */
/* STRT cccc 01x0 x010 xxxx xxxx xxxx xxxx xxxx */
DECODE_CUSTOM (0x0c000000, 0x04000000, prep_emulate_ldr_str),
DECODE_END
};
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
static const union decode_item arm_cccc_100x_table[] = {
/* Block data transfer instructions */
/* LDM cccc 100x x0x1 xxxx xxxx xxxx xxxx xxxx */
/* STM cccc 100x x0x0 xxxx xxxx xxxx xxxx xxxx */
DECODE_CUSTOM (0x0e400000, 0x08000000, kprobe_decode_ldmstm),
/* STM (user registers) cccc 100x x1x0 xxxx xxxx xxxx xxxx xxxx */
/* LDM (user registers) cccc 100x x1x1 xxxx 0xxx xxxx xxxx xxxx */
/* LDM (exception ret) cccc 100x x1x1 xxxx 1xxx xxxx xxxx xxxx */
DECODE_END
};
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
const union decode_item kprobe_decode_arm_table[] = {
/*
* Unconditional instructions
* 1111 xxxx xxxx xxxx xxxx xxxx xxxx xxxx
*/
DECODE_TABLE (0xf0000000, 0xf0000000, arm_1111_table),
/*
* Miscellaneous instructions
* cccc 0001 0xx0 xxxx xxxx xxxx 0xxx xxxx
*/
DECODE_TABLE (0x0f900080, 0x01000000, arm_cccc_0001_0xx0____0xxx_table),
/*
* Halfword multiply and multiply-accumulate
* cccc 0001 0xx0 xxxx xxxx xxxx 1xx0 xxxx
*/
DECODE_TABLE (0x0f900090, 0x01000080, arm_cccc_0001_0xx0____1xx0_table),
/*
* Multiply and multiply-accumulate
* cccc 0000 xxxx xxxx xxxx xxxx 1001 xxxx
*/
DECODE_TABLE (0x0f0000f0, 0x00000090, arm_cccc_0000_____1001_table),
/*
* Synchronization primitives
* cccc 0001 xxxx xxxx xxxx xxxx 1001 xxxx
*/
DECODE_TABLE (0x0f0000f0, 0x01000090, arm_cccc_0001_____1001_table),
/*
* Extra load/store instructions
* cccc 000x xxxx xxxx xxxx xxxx 1xx1 xxxx
*/
DECODE_TABLE (0x0e000090, 0x00000090, arm_cccc_000x_____1xx1_table),
/*
* Data-processing (register)
* cccc 000x xxxx xxxx xxxx xxxx xxx0 xxxx
* Data-processing (register-shifted register)
* cccc 000x xxxx xxxx xxxx xxxx 0xx1 xxxx
*/
DECODE_TABLE (0x0e000000, 0x00000000, arm_cccc_000x_table),
/*
* Data-processing (immediate)
* cccc 001x xxxx xxxx xxxx xxxx xxxx xxxx
*/
DECODE_TABLE (0x0e000000, 0x02000000, arm_cccc_001x_table),
/*
* Media instructions
* cccc 011x xxxx xxxx xxxx xxxx xxx1 xxxx
*/
DECODE_TABLE (0x0f000010, 0x06000010, arm_cccc_0110_____xxx1_table),
DECODE_TABLE (0x0f000010, 0x07000010, arm_cccc_0111_____xxx1_table),
/*
* Load/store word and unsigned byte
* cccc 01xx xxxx xxxx xxxx xxxx xxxx xxxx
*/
DECODE_TABLE (0x0c000000, 0x04000000, arm_cccc_01xx_table),
/*
* Block data transfer instructions
* cccc 100x xxxx xxxx xxxx xxxx xxxx xxxx
*/
DECODE_TABLE (0x0e000000, 0x08000000, arm_cccc_100x_table),
/* B cccc 1010 xxxx xxxx xxxx xxxx xxxx xxxx */
/* BL cccc 1011 xxxx xxxx xxxx xxxx xxxx xxxx */
DECODE_SIMULATE (0x0e000000, 0x0a000000, simulate_bbl),
/*
* Supervisor Call, and coprocessor instructions
*/
/* MCRR cccc 1100 0100 xxxx xxxx xxxx xxxx xxxx */
/* MRRC cccc 1100 0101 xxxx xxxx xxxx xxxx xxxx */
/* LDC cccc 110x xxx1 xxxx xxxx xxxx xxxx xxxx */
/* STC cccc 110x xxx0 xxxx xxxx xxxx xxxx xxxx */
/* CDP cccc 1110 xxxx xxxx xxxx xxxx xxx0 xxxx */
/* MCR cccc 1110 xxx0 xxxx xxxx xxxx xxx1 xxxx */
/* MRC cccc 1110 xxx1 xxxx xxxx xxxx xxx1 xxxx */
/* SVC cccc 1111 xxxx xxxx xxxx xxxx xxxx xxxx */
DECODE_REJECT (0x0c000000, 0x0c000000),
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
DECODE_END
};
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
static void __kprobes arm_singlestep(struct kprobe *p, struct pt_regs *regs)
{
regs->ARM_pc += 4;
p->ainsn.insn_handler(p, regs);
}
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
/* Return:
* INSN_REJECTED If instruction is one not allowed to kprobe,
* INSN_GOOD If instruction is supported and uses instruction slot,
* INSN_GOOD_NO_SLOT If instruction is supported but doesn't use its slot.
*
* For instructions we don't want to kprobe (INSN_REJECTED return result):
* These are generally ones that modify the processor state making
* them "hard" to simulate such as switches processor modes or
* make accesses in alternate modes. Any of these could be simulated
* if the work was put into it, but low return considering they
* should also be very rare.
*/
enum kprobe_insn __kprobes
arm_kprobe_decode_insn(kprobe_opcode_t insn, struct arch_specific_insn *asi)
{
asi->insn_singlestep = arm_singlestep;
asi->insn_check_cc = kprobe_condition_checks[insn>>28];
return kprobe_decode_insn(insn, asi, kprobe_decode_arm_table, false);
ARM kprobes: instruction single-stepping support This is the code implementing instruction single-stepping for kprobes on ARM. To get around the limitation of no Next-PC and no hardware single- stepping, all kprobe'd instructions are split into three camps: simulation, emulation, and rejected. "Simulated" instructions are those instructions which behavior is reproduced by straight C code. "Emulated" instructions are ones that are copied, slightly altered and executed directly in the instruction slot to reproduce their behavior. "Rejected" instructions are ones that could be simulated, but work hasn't been put into simulating them. These instructions should be very rare, if not unencountered, in the kernel. If ever needed, code could be added to simulate them. One might wonder why this and the ptrace singlestep facility are not sharing some code. Both approaches are fundamentally different because the ptrace code regains control after the stepped instruction by installing a breakpoint after the instruction itself, and possibly at the location where the instruction might be branching to, instead of simulating or emulating the target instruction. The ptrace approach isn't suitable for kprobes because the breakpoints would have to be moved back, and the icache flushed, everytime the probe is hit to let normal code execution resume, which would have a significant performance impact. It is also racy on SMP since another CPU could, with the right timing, sail through the probe point without being caught. Because ptrace single-stepping always result in a different process to be scheduled, the concern for performance is much less significant. On the other hand, the kprobes approach isn't (currently) suitable for ptrace because it has no provision for proper user space memory protection and translation, and even if that was implemented, the gain wouldn't be worth the added complexity in the ptrace path compared to the current approach. So, until kprobes does support user space, both kprobes and ptrace are best kept independent and separate. Signed-off-by: Quentin Barnes <qbarnes@gmail.com> Signed-off-by: Abhishek Sagar <sagar.abhishek@gmail.com> Signed-off-by: Nicolas Pitre <nico@marvell.com>
2007-06-12 06:20:10 +08:00
}