From 611b1597680dd4a57896bcca1af0484be463c55e Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Mon, 21 Jul 2008 18:49:56 +0300 Subject: x86: fix mmiotrace 8-bit register decoding When SIL, DIL, BPL or SPL registers were used in MMIO, the datum was extracted from AH, BH, CH, or DH, which are incorrect. Signed-off-by: Pekka Paalanen Cc: "Vegard Nossum" Cc: "Steven Rostedt" Cc: proski@gnu.org Cc: "Pekka Enberg" Signed-off-by: Ingo Molnar --- arch/x86/mm/pf_in.c | 121 ++++++++++++++++++++++++++++++++++++---------------- 1 file changed, 84 insertions(+), 37 deletions(-) (limited to 'arch') diff --git a/arch/x86/mm/pf_in.c b/arch/x86/mm/pf_in.c index efa1911e20ca..df3d5c861cda 100644 --- a/arch/x86/mm/pf_in.c +++ b/arch/x86/mm/pf_in.c @@ -79,25 +79,34 @@ static unsigned int mw32[] = { 0xC7 }; static unsigned int mw64[] = { 0x89, 0x8B }; #endif /* not __i386__ */ -static int skip_prefix(unsigned char *addr, int *shorted, int *enlarged, - int *rexr) +struct prefix_bits { + unsigned shorted:1; + unsigned enlarged:1; + unsigned rexr:1; + unsigned rex:1; +}; + +static int skip_prefix(unsigned char *addr, struct prefix_bits *prf) { int i; unsigned char *p = addr; - *shorted = 0; - *enlarged = 0; - *rexr = 0; + prf->shorted = 0; + prf->enlarged = 0; + prf->rexr = 0; + prf->rex = 0; restart: for (i = 0; i < ARRAY_SIZE(prefix_codes); i++) { if (*p == prefix_codes[i]) { if (*p == 0x66) - *shorted = 1; + prf->shorted = 1; #ifdef __amd64__ if ((*p & 0xf8) == 0x48) - *enlarged = 1; + prf->enlarged = 1; if ((*p & 0xf4) == 0x44) - *rexr = 1; + prf->rexr = 1; + if ((*p & 0xf0) == 0x40) + prf->rex = 1; #endif p++; goto restart; @@ -135,12 +144,12 @@ enum reason_type get_ins_type(unsigned long ins_addr) { unsigned int opcode; unsigned char *p; - int shorted, enlarged, rexr; + struct prefix_bits prf; int i; enum reason_type rv = OTHERS; p = (unsigned char *)ins_addr; - p += skip_prefix(p, &shorted, &enlarged, &rexr); + p += skip_prefix(p, &prf); p += get_opcode(p, &opcode); CHECK_OP_TYPE(opcode, reg_rop, REG_READ); @@ -156,10 +165,11 @@ static unsigned int get_ins_reg_width(unsigned long ins_addr) { unsigned int opcode; unsigned char *p; - int i, shorted, enlarged, rexr; + struct prefix_bits prf; + int i; p = (unsigned char *)ins_addr; - p += skip_prefix(p, &shorted, &enlarged, &rexr); + p += skip_prefix(p, &prf); p += get_opcode(p, &opcode); for (i = 0; i < ARRAY_SIZE(rw8); i++) @@ -168,7 +178,7 @@ static unsigned int get_ins_reg_width(unsigned long ins_addr) for (i = 0; i < ARRAY_SIZE(rw32); i++) if (rw32[i] == opcode) - return (shorted ? 2 : (enlarged ? 8 : 4)); + return prf.shorted ? 2 : (prf.enlarged ? 8 : 4); printk(KERN_ERR "mmiotrace: Unknown opcode 0x%02x\n", opcode); return 0; @@ -178,10 +188,11 @@ unsigned int get_ins_mem_width(unsigned long ins_addr) { unsigned int opcode; unsigned char *p; - int i, shorted, enlarged, rexr; + struct prefix_bits prf; + int i; p = (unsigned char *)ins_addr; - p += skip_prefix(p, &shorted, &enlarged, &rexr); + p += skip_prefix(p, &prf); p += get_opcode(p, &opcode); for (i = 0; i < ARRAY_SIZE(mw8); i++) @@ -194,11 +205,11 @@ unsigned int get_ins_mem_width(unsigned long ins_addr) for (i = 0; i < ARRAY_SIZE(mw32); i++) if (mw32[i] == opcode) - return shorted ? 2 : 4; + return prf.shorted ? 2 : 4; for (i = 0; i < ARRAY_SIZE(mw64); i++) if (mw64[i] == opcode) - return shorted ? 2 : (enlarged ? 8 : 4); + return prf.shorted ? 2 : (prf.enlarged ? 8 : 4); printk(KERN_ERR "mmiotrace: Unknown opcode 0x%02x\n", opcode); return 0; @@ -238,7 +249,7 @@ enum { #endif }; -static unsigned char *get_reg_w8(int no, struct pt_regs *regs) +static unsigned char *get_reg_w8(int no, int rex, struct pt_regs *regs) { unsigned char *rv = NULL; @@ -255,18 +266,6 @@ static unsigned char *get_reg_w8(int no, struct pt_regs *regs) case arg_DL: rv = (unsigned char *)®s->dx; break; - case arg_AH: - rv = 1 + (unsigned char *)®s->ax; - break; - case arg_BH: - rv = 1 + (unsigned char *)®s->bx; - break; - case arg_CH: - rv = 1 + (unsigned char *)®s->cx; - break; - case arg_DH: - rv = 1 + (unsigned char *)®s->dx; - break; #ifdef __amd64__ case arg_R8: rv = (unsigned char *)®s->r8; @@ -294,9 +293,55 @@ static unsigned char *get_reg_w8(int no, struct pt_regs *regs) break; #endif default: - printk(KERN_ERR "mmiotrace: Error reg no# %d\n", no); break; } + + if (rv) + return rv; + + if (rex) { + /* + * If REX prefix exists, access low bytes of SI etc. + * instead of AH etc. + */ + switch (no) { + case arg_SI: + rv = (unsigned char *)®s->si; + break; + case arg_DI: + rv = (unsigned char *)®s->di; + break; + case arg_BP: + rv = (unsigned char *)®s->bp; + break; + case arg_SP: + rv = (unsigned char *)®s->sp; + break; + default: + break; + } + } else { + switch (no) { + case arg_AH: + rv = 1 + (unsigned char *)®s->ax; + break; + case arg_BH: + rv = 1 + (unsigned char *)®s->bx; + break; + case arg_CH: + rv = 1 + (unsigned char *)®s->cx; + break; + case arg_DH: + rv = 1 + (unsigned char *)®s->dx; + break; + default: + break; + } + } + + if (!rv) + printk(KERN_ERR "mmiotrace: Error reg no# %d\n", no); + return rv; } @@ -368,11 +413,12 @@ unsigned long get_ins_reg_val(unsigned long ins_addr, struct pt_regs *regs) unsigned char mod_rm; int reg; unsigned char *p; - int i, shorted, enlarged, rexr; + struct prefix_bits prf; + int i; unsigned long rv; p = (unsigned char *)ins_addr; - p += skip_prefix(p, &shorted, &enlarged, &rexr); + p += skip_prefix(p, &prf); p += get_opcode(p, &opcode); for (i = 0; i < ARRAY_SIZE(reg_rop); i++) if (reg_rop[i] == opcode) { @@ -392,10 +438,10 @@ unsigned long get_ins_reg_val(unsigned long ins_addr, struct pt_regs *regs) do_work: mod_rm = *p; - reg = ((mod_rm >> 3) & 0x7) | (rexr << 3); + reg = ((mod_rm >> 3) & 0x7) | (prf.rexr << 3); switch (get_ins_reg_width(ins_addr)) { case 1: - return *get_reg_w8(reg, regs); + return *get_reg_w8(reg, prf.rex, regs); case 2: return *(unsigned short *)get_reg_w32(reg, regs); @@ -422,11 +468,12 @@ unsigned long get_ins_imm_val(unsigned long ins_addr) unsigned char mod_rm; unsigned char mod; unsigned char *p; - int i, shorted, enlarged, rexr; + struct prefix_bits prf; + int i; unsigned long rv; p = (unsigned char *)ins_addr; - p += skip_prefix(p, &shorted, &enlarged, &rexr); + p += skip_prefix(p, &prf); p += get_opcode(p, &opcode); for (i = 0; i < ARRAY_SIZE(imm_wop); i++) if (imm_wop[i] == opcode) { -- cgit v1.2.3 From 8b1fa1d7b22f386747c7b78b918d4c680c16066f Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Tue, 29 Jul 2008 12:36:02 +0200 Subject: ftrace: mark lapic_wd_event() notrace it can be called in the NMI path: [ 0.645999] calling ftrace_dynamic_init+0x0/0xd6 [ 0.647521] ------------[ cut here ]------------ [ 0.647521] WARNING: at kernel/trace/ftrace.c:348 ftrace_record_ip+0x4e/0x252() [ 0.647521] Modules linked in: [ 0.647521] Pid: 15, comm: kstop1 Not tainted 2.6.27-rc1-tip #22686 [ 0.647521] [ 0.647521] Call Trace: [ 0.647521] [] warn_on_slowpath+0x5d/0x84 [ 0.647521] [] ? lapic_wd_event+0xb/0x5c [ 0.647521] [] ftrace_record_ip+0x4e/0x252 [ 0.647521] [] mcount_call+0x5/0x31 [ 0.647521] [] ? lapic_wd_event+0x10/0x5c [ 0.647521] [] nmi_watchdog_tick+0x19d/0x1ad [ 0.647521] [] default_do_nmi+0x75/0x1e3 [ 0.647521] [] do_nmi+0x5d/0x94 [ 0.647521] [] nmi+0xa2/0xc2 [ 0.647521] [] ? check_bytes_and_report+0x11/0xcc [ 0.647521] <> [] ? mcount_call+0x5/0x31 [ 0.647521] [] check_object+0x61/0x1b0 [ 0.647521] [] __slab_free+0x169/0x2ae [ 0.647521] [] ? __cleanup_sighand+0x25/0x27 [ 0.647521] [] ? __cleanup_sighand+0x25/0x27 [ 0.647521] [] kmem_cache_free+0x85/0xb9 [ 0.647521] [] __cleanup_sighand+0x25/0x27 [ 0.647521] [] release_task+0x256/0x339 [ 0.647521] [] do_exit+0x764/0x7ef [ 0.647521] [] __xchg+0x0/0x38 [ 0.647521] [] ? stop_cpu+0x0/0xb2 [ 0.647521] [] ? stop_cpu+0x0/0xb2 [ 0.647521] [] kthread+0x4e/0x7b [ 0.647521] [] child_rip+0xa/0x11 [ 0.647521] [] ? restore_args+0x0/0x30 [ 0.647521] [] ? native_load_tls+0x14/0x2e [ 0.647521] [] ? kthread+0x0/0x7b [ 0.647521] [] ? child_rip+0x0/0x11 [ 0.647521] [ 0.647521] ---[ end trace 4eaa2a86a8e2da22 ]--- [ 0.672032] initcall ftrace_dynamic_init+0x0/0xd6 returned 0 after 19 msecs also mark it no-kprobes while at it. Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/perfctr-watchdog.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-) (limited to 'arch') diff --git a/arch/x86/kernel/cpu/perfctr-watchdog.c b/arch/x86/kernel/cpu/perfctr-watchdog.c index 6bff382094f5..9abd48b22674 100644 --- a/arch/x86/kernel/cpu/perfctr-watchdog.c +++ b/arch/x86/kernel/cpu/perfctr-watchdog.c @@ -17,6 +17,8 @@ #include #include #include +#include + #include #include @@ -336,7 +338,8 @@ static void single_msr_unreserve(void) release_perfctr_nmi(wd_ops->perfctr); } -static void single_msr_rearm(struct nmi_watchdog_ctlblk *wd, unsigned nmi_hz) +static void __kprobes +single_msr_rearm(struct nmi_watchdog_ctlblk *wd, unsigned nmi_hz) { /* start the cycle over again */ write_watchdog_counter(wd->perfctr_msr, NULL, nmi_hz); @@ -401,7 +404,7 @@ static int setup_p6_watchdog(unsigned nmi_hz) return 1; } -static void p6_rearm(struct nmi_watchdog_ctlblk *wd, unsigned nmi_hz) +static void __kprobes p6_rearm(struct nmi_watchdog_ctlblk *wd, unsigned nmi_hz) { /* * P6 based Pentium M need to re-unmask @@ -605,7 +608,7 @@ static void p4_unreserve(void) release_perfctr_nmi(MSR_P4_IQ_PERFCTR0); } -static void p4_rearm(struct nmi_watchdog_ctlblk *wd, unsigned nmi_hz) +static void __kprobes p4_rearm(struct nmi_watchdog_ctlblk *wd, unsigned nmi_hz) { unsigned dummy; /* @@ -784,7 +787,7 @@ unsigned lapic_adjust_nmi_hz(unsigned hz) return hz; } -int lapic_wd_event(unsigned nmi_hz) +int __kprobes lapic_wd_event(unsigned nmi_hz) { struct nmi_watchdog_ctlblk *wd = &__get_cpu_var(nmi_watchdog_ctlblk); u64 ctr; -- cgit v1.2.3 From e4b2b8866121bd06d5f3d9de0f79a04339ffa252 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 14 Aug 2008 15:45:11 -0400 Subject: ftrace: enable using mcount recording on x86 Enable the use of the __mcount_loc infrastructure on x86_64 and i386. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/Kconfig | 1 + 1 file changed, 1 insertion(+) (limited to 'arch') diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index fc8351f374fd..5a34c5427a07 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -25,6 +25,7 @@ config X86 select HAVE_KPROBES select ARCH_WANT_OPTIONAL_GPIOLIB select HAVE_KRETPROBES + select HAVE_FTRACE_MCOUNT_RECORD select HAVE_DYNAMIC_FTRACE select HAVE_FTRACE select HAVE_KVM if ((X86_32 && !X86_VOYAGER && !X86_VISWS && !X86_NUMAQ) || X86_64) -- cgit v1.2.3 From 0a37605c2261a06d8cafc62dee11374ad676c8c4 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 14 Aug 2008 15:45:12 -0400 Subject: ftrace: x86 mcount stub x86 now sets up the mcount locations through the build and no longer needs to record the ip when the function is executed. This patch changes the initial mcount to simply return. There's no need to do any other work. If the ftrace start up test fails, the original mcount will be what everything will use, so having this as fast as possible is a good thing. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/entry_32.S | 14 -------------- arch/x86/kernel/entry_64.S | 26 -------------------------- arch/x86/kernel/ftrace.c | 14 ++------------ 3 files changed, 2 insertions(+), 52 deletions(-) (limited to 'arch') diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index b21fbfaffe39..4e4269c73bb7 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -1153,20 +1153,6 @@ ENDPROC(xen_failsafe_callback) #ifdef CONFIG_DYNAMIC_FTRACE ENTRY(mcount) - pushl %eax - pushl %ecx - pushl %edx - movl 0xc(%esp), %eax - subl $MCOUNT_INSN_SIZE, %eax - -.globl mcount_call -mcount_call: - call ftrace_stub - - popl %edx - popl %ecx - popl %eax - ret END(mcount) diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index 1db6ce4314e1..09e7145484c5 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -64,32 +64,6 @@ #ifdef CONFIG_FTRACE #ifdef CONFIG_DYNAMIC_FTRACE ENTRY(mcount) - - subq $0x38, %rsp - movq %rax, (%rsp) - movq %rcx, 8(%rsp) - movq %rdx, 16(%rsp) - movq %rsi, 24(%rsp) - movq %rdi, 32(%rsp) - movq %r8, 40(%rsp) - movq %r9, 48(%rsp) - - movq 0x38(%rsp), %rdi - subq $MCOUNT_INSN_SIZE, %rdi - -.globl mcount_call -mcount_call: - call ftrace_stub - - movq 48(%rsp), %r9 - movq 40(%rsp), %r8 - movq 32(%rsp), %rdi - movq 24(%rsp), %rsi - movq 16(%rsp), %rdx - movq 8(%rsp), %rcx - movq (%rsp), %rax - addq $0x38, %rsp - retq END(mcount) diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index ab115cd15fdf..96aadbfedcc6 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -112,18 +112,8 @@ notrace int ftrace_update_ftrace_func(ftrace_func_t func) notrace int ftrace_mcount_set(unsigned long *data) { - unsigned long ip = (long)(&mcount_call); - unsigned long *addr = data; - unsigned char old[MCOUNT_INSN_SIZE], *new; - - /* - * Replace the mcount stub with a pointer to the - * ip recorder function. - */ - memcpy(old, &mcount_call, MCOUNT_INSN_SIZE); - new = ftrace_call_replace(ip, *addr); - *addr = ftrace_modify_code(ip, old, new); - + /* mcount is initialized as a nop */ + *data = 0; return 0; } -- cgit v1.2.3 From 732f3ca7d4ba3c1be8d051d52302ef441ee7748b Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 14 Aug 2008 18:05:05 -0400 Subject: ftrace: use only 5 byte nops for x86 Mathieu Desnoyers revealed a bug in the original code. The nop that is used to relpace the mcount caller can be a two part nop. This runs the risk where a process can be preempted after executing the first nop, but before the second part of the nop. The ftrace code calls kstop_machine to keep multiple CPUs from executing code that is being modified, but it does not protect against a task preempting in the middle of a two part nop. If the above preemption happens and the tracer is enabled, after the kstop_machine runs, all those nops will be calls to the trace function. If the preempted process that was preempted between the two nops is executed again, it will execute half of the call to the trace function, and this might crash the system. This patch instead uses what both the latest Intel and AMD spec suggests. That is the P6_NOP5 sequence of "0x0f 0x1f 0x44 0x00 0x00". Note, some older CPUs and QEMU might fault on this nop, so this nop is executed with fault handling first. If it detects a fault, it will then use the code "0x66 0x66 0x66 0x66 0x90". If that faults, it will then default to a simple "jmp 1f; .byte 0x00 0x00 0x00; 1:". The jmp is not optimal but will do if the first two can not be executed. TODO: Examine the cpuid to determine the nop to use. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 68 +++++++++++++++++++++++++++++++++++++++++++----- 1 file changed, 61 insertions(+), 7 deletions(-) (limited to 'arch') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 96aadbfedcc6..4151c91254e8 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -16,8 +16,8 @@ #include #include -#include #include +#include /* Long is fine, even if it is only 4 bytes ;-) */ @@ -119,13 +119,67 @@ notrace int ftrace_mcount_set(unsigned long *data) int __init ftrace_dyn_arch_init(void *data) { - const unsigned char *const *noptable = find_nop_table(); - - /* This is running in kstop_machine */ - - ftrace_mcount_set(data); + extern const unsigned char ftrace_test_p6nop[]; + extern const unsigned char ftrace_test_nop5[]; + extern const unsigned char ftrace_test_jmp[]; + int faulted = 0; - ftrace_nop = (unsigned long *)noptable[MCOUNT_INSN_SIZE]; + /* + * There is no good nop for all x86 archs. + * We will default to using the P6_NOP5, but first we + * will test to make sure that the nop will actually + * work on this CPU. If it faults, we will then + * go to a lesser efficient 5 byte nop. If that fails + * we then just use a jmp as our nop. This isn't the most + * efficient nop, but we can not use a multi part nop + * since we would then risk being preempted in the middle + * of that nop, and if we enabled tracing then, it might + * cause a system crash. + * + * TODO: check the cpuid to determine the best nop. + */ + asm volatile ( + "jmp ftrace_test_jmp\n" + /* This code needs to stay around */ + ".section .text, \"ax\"\n" + "ftrace_test_jmp:" + "jmp ftrace_test_p6nop\n" + ".byte 0x00,0x00,0x00\n" /* 2 byte jmp + 3 bytes */ + "ftrace_test_p6nop:" + P6_NOP5 + "jmp 1f\n" + "ftrace_test_nop5:" + ".byte 0x66,0x66,0x66,0x66,0x90\n" + "jmp 1f\n" + ".previous\n" + "1:" + ".section .fixup, \"ax\"\n" + "2: movl $1, %0\n" + " jmp ftrace_test_nop5\n" + "3: movl $2, %0\n" + " jmp 1b\n" + ".previous\n" + _ASM_EXTABLE(ftrace_test_p6nop, 2b) + _ASM_EXTABLE(ftrace_test_nop5, 3b) + : "=r"(faulted) : "0" (faulted)); + + switch (faulted) { + case 0: + pr_info("ftrace: converting mcount calls to 0f 1f 44 00 00\n"); + ftrace_nop = (unsigned long *)ftrace_test_p6nop; + break; + case 1: + pr_info("ftrace: converting mcount calls to 66 66 66 66 90\n"); + ftrace_nop = (unsigned long *)ftrace_test_nop5; + break; + case 2: + pr_info("ftrace: converting mcount calls to jmp 1f\n"); + ftrace_nop = (unsigned long *)ftrace_test_jmp; + break; + } + + /* The return code is retured via data */ + *(unsigned long *)data = 0; return 0; } -- cgit v1.2.3 From 6f93fc076a464bfe24e8d4c5fea3f6ca5bdb264d Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 20 Aug 2008 12:55:07 -0400 Subject: ftrace: x86 use copy to and from user functions The modification of code is performed either by kstop_machine, before SMP starts, or on module code before the module is executed. There is no reason to do the modifications from assembly. The copy to and from user functions are sufficient and produces cleaner and easier to read code. Thanks to Benjamin Herrenschmidt for suggesting the idea. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 38 +++++++++++++------------------------- 1 file changed, 13 insertions(+), 25 deletions(-) (limited to 'arch') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 4151c91254e8..082d99642622 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -11,6 +11,7 @@ #include #include +#include #include #include #include @@ -60,11 +61,7 @@ notrace int ftrace_modify_code(unsigned long ip, unsigned char *old_code, unsigned char *new_code) { - unsigned replaced; - unsigned old = *(unsigned *)old_code; /* 4 bytes */ - unsigned new = *(unsigned *)new_code; /* 4 bytes */ - unsigned char newch = new_code[4]; - int faulted = 0; + unsigned char replaced[MCOUNT_INSN_SIZE]; /* * Note: Due to modules and __init, code can @@ -72,29 +69,20 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code, * as well as code changing. * * No real locking needed, this code is run through - * kstop_machine. + * kstop_machine, or before SMP starts. */ - asm volatile ( - "1: lock\n" - " cmpxchg %3, (%2)\n" - " jnz 2f\n" - " movb %b4, 4(%2)\n" - "2:\n" - ".section .fixup, \"ax\"\n" - "3: movl $1, %0\n" - " jmp 2b\n" - ".previous\n" - _ASM_EXTABLE(1b, 3b) - : "=r"(faulted), "=a"(replaced) - : "r"(ip), "r"(new), "c"(newch), - "0"(faulted), "a"(old) - : "memory"); - sync_core(); + if (__copy_from_user(replaced, (char __user *)ip, MCOUNT_INSN_SIZE)) + return 1; - if (replaced != old && replaced != new) - faulted = 2; + if (memcmp(replaced, old_code, MCOUNT_INSN_SIZE) != 0) + return 2; - return faulted; + WARN_ON_ONCE(__copy_to_user((char __user *)ip, new_code, + MCOUNT_INSN_SIZE)); + + sync_core(); + + return 0; } notrace int ftrace_update_ftrace_func(ftrace_func_t func) -- cgit v1.2.3 From bbe5c7830c6dbde58726d44ec0337bc8b2d95d37 Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Tue, 16 Sep 2008 21:54:16 +0300 Subject: x86 mmiotrace: fix a rare memory leak Signed-off-by: Pekka Paalanen Acked-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/mm/mmio-mod.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'arch') diff --git a/arch/x86/mm/mmio-mod.c b/arch/x86/mm/mmio-mod.c index 635b50e85581..754bd1eaf4f6 100644 --- a/arch/x86/mm/mmio-mod.c +++ b/arch/x86/mm/mmio-mod.c @@ -307,8 +307,10 @@ static void ioremap_trace_core(resource_size_t offset, unsigned long size, map.map_id = trace->id; spin_lock_irq(&trace_lock); - if (!is_enabled()) + if (!is_enabled()) { + kfree(trace); goto not_enabled; + } mmio_trace_mapping(&map); list_add_tail(&trace->list, &trace_list); -- cgit v1.2.3 From 9e57fb35d711331a9b1410c5c56ebeb3733428a0 Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Tue, 16 Sep 2008 22:00:34 +0300 Subject: x86 mmiotrace: implement mmiotrace_printk() Offer mmiotrace users a function to inject markers from inside the kernel. This depends on the trace_vprintk() patch. Signed-off-by: Pekka Paalanen Acked-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/mm/mmio-mod.c | 19 ++++++++++++++++++- arch/x86/mm/testmmiotrace.c | 4 ++++ include/linux/mmiotrace.h | 17 +++++++++++++++-- kernel/trace/trace_mmiotrace.c | 5 +++++ 4 files changed, 42 insertions(+), 3 deletions(-) (limited to 'arch') diff --git a/arch/x86/mm/mmio-mod.c b/arch/x86/mm/mmio-mod.c index 754bd1eaf4f6..5e2e2e72ee80 100644 --- a/arch/x86/mm/mmio-mod.c +++ b/arch/x86/mm/mmio-mod.c @@ -75,7 +75,7 @@ static LIST_HEAD(trace_list); /* struct remap_trace */ * and trace_lock. * - Routines depending on is_enabled() must take trace_lock. * - trace_list users must hold trace_lock. - * - is_enabled() guarantees that mmio_trace_record is allowed. + * - is_enabled() guarantees that mmio_trace_{rw,mapping} are allowed. * - pre/post callbacks assume the effect of is_enabled() being true. */ @@ -379,6 +379,23 @@ void mmiotrace_iounmap(volatile void __iomem *addr) iounmap_trace_core(addr); } +int mmiotrace_printk(const char *fmt, ...) +{ + int ret = 0; + va_list args; + unsigned long flags; + va_start(args, fmt); + + spin_lock_irqsave(&trace_lock, flags); + if (is_enabled()) + ret = mmio_trace_printk(fmt, args); + spin_unlock_irqrestore(&trace_lock, flags); + + va_end(args); + return ret; +} +EXPORT_SYMBOL(mmiotrace_printk); + static void clear_trace_list(void) { struct remap_trace *trace; diff --git a/arch/x86/mm/testmmiotrace.c b/arch/x86/mm/testmmiotrace.c index d877c5b423ef..ab50a8d7402c 100644 --- a/arch/x86/mm/testmmiotrace.c +++ b/arch/x86/mm/testmmiotrace.c @@ -3,6 +3,7 @@ */ #include #include +#include #define MODULE_NAME "testmmiotrace" @@ -13,6 +14,7 @@ MODULE_PARM_DESC(mmio_address, "Start address of the mapping of 16 kB."); static void do_write_test(void __iomem *p) { unsigned int i; + mmiotrace_printk("Write test.\n"); for (i = 0; i < 256; i++) iowrite8(i, p + i); for (i = 1024; i < (5 * 1024); i += 2) @@ -24,6 +26,7 @@ static void do_write_test(void __iomem *p) static void do_read_test(void __iomem *p) { unsigned int i; + mmiotrace_printk("Read test.\n"); for (i = 0; i < 256; i++) ioread8(p + i); for (i = 1024; i < (5 * 1024); i += 2) @@ -39,6 +42,7 @@ static void do_test(void) pr_err(MODULE_NAME ": could not ioremap, aborting.\n"); return; } + mmiotrace_printk("ioremap returned %p.\n", p); do_write_test(p); do_read_test(p); iounmap(p); diff --git a/include/linux/mmiotrace.h b/include/linux/mmiotrace.h index 61d19e1b7a0b..60cc3bf5c538 100644 --- a/include/linux/mmiotrace.h +++ b/include/linux/mmiotrace.h @@ -34,11 +34,15 @@ extern void unregister_kmmio_probe(struct kmmio_probe *p); /* Called from page fault handler. */ extern int kmmio_handler(struct pt_regs *regs, unsigned long addr); -/* Called from ioremap.c */ #ifdef CONFIG_MMIOTRACE +/* Called from ioremap.c */ extern void mmiotrace_ioremap(resource_size_t offset, unsigned long size, void __iomem *addr); extern void mmiotrace_iounmap(volatile void __iomem *addr); + +/* For anyone to insert markers. Remember trailing newline. */ +extern int mmiotrace_printk(const char *fmt, ...) + __attribute__ ((format (printf, 1, 2))); #else static inline void mmiotrace_ioremap(resource_size_t offset, unsigned long size, void __iomem *addr) @@ -48,7 +52,15 @@ static inline void mmiotrace_ioremap(resource_size_t offset, static inline void mmiotrace_iounmap(volatile void __iomem *addr) { } -#endif /* CONFIG_MMIOTRACE_HOOKS */ + +static inline int mmiotrace_printk(const char *fmt, ...) + __attribute__ ((format (printf, 1, 0))); + +static inline int mmiotrace_printk(const char *fmt, ...) +{ + return 0; +} +#endif /* CONFIG_MMIOTRACE */ enum mm_io_opcode { MMIO_READ = 0x1, /* struct mmiotrace_rw */ @@ -81,5 +93,6 @@ extern void enable_mmiotrace(void); extern void disable_mmiotrace(void); extern void mmio_trace_rw(struct mmiotrace_rw *rw); extern void mmio_trace_mapping(struct mmiotrace_map *map); +extern int mmio_trace_printk(const char *fmt, va_list args); #endif /* MMIOTRACE_H */ diff --git a/kernel/trace/trace_mmiotrace.c b/kernel/trace/trace_mmiotrace.c index ef02747b26d9..767d1faf56e5 100644 --- a/kernel/trace/trace_mmiotrace.c +++ b/kernel/trace/trace_mmiotrace.c @@ -335,3 +335,8 @@ void mmio_trace_mapping(struct mmiotrace_map *map) __trace_mmiotrace_map(tr, data, map); preempt_enable(); } + +int mmio_trace_printk(const char *fmt, va_list args) +{ + return trace_vprintk(0, fmt, args); +} -- cgit v1.2.3 From 4427414170a63331a9cc36b9598502c5cdfe453b Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Tue, 16 Sep 2008 22:03:56 +0300 Subject: mmiotrace: remove left-over marker cruft Signed-off-by: Pekka Paalanen Acked-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/mm/mmio-mod.c | 64 ----------------------------------------------- include/linux/mmiotrace.h | 3 +-- 2 files changed, 1 insertion(+), 66 deletions(-) (limited to 'arch') diff --git a/arch/x86/mm/mmio-mod.c b/arch/x86/mm/mmio-mod.c index 5e2e2e72ee80..2c4baa88f2cb 100644 --- a/arch/x86/mm/mmio-mod.c +++ b/arch/x86/mm/mmio-mod.c @@ -56,13 +56,6 @@ struct remap_trace { static DEFINE_PER_CPU(struct trap_reason, pf_reason); static DEFINE_PER_CPU(struct mmiotrace_rw, cpu_trace); -#if 0 /* XXX: no way gather this info anymore */ -/* Access to this is not per-cpu. */ -static DEFINE_PER_CPU(atomic_t, dropped); -#endif - -static struct dentry *marker_file; - static DEFINE_MUTEX(mmiotrace_mutex); static DEFINE_SPINLOCK(trace_lock); static atomic_t mmiotrace_enabled; @@ -97,44 +90,6 @@ static bool is_enabled(void) return atomic_read(&mmiotrace_enabled); } -#if 0 /* XXX: needs rewrite */ -/* - * Write callback for the debugfs entry: - * Read a marker and write it to the mmio trace log - */ -static ssize_t write_marker(struct file *file, const char __user *buffer, - size_t count, loff_t *ppos) -{ - char *event = NULL; - struct mm_io_header *headp; - ssize_t len = (count > 65535) ? 65535 : count; - - event = kzalloc(sizeof(*headp) + len, GFP_KERNEL); - if (!event) - return -ENOMEM; - - headp = (struct mm_io_header *)event; - headp->type = MMIO_MAGIC | (MMIO_MARKER << MMIO_OPCODE_SHIFT); - headp->data_len = len; - - if (copy_from_user(event + sizeof(*headp), buffer, len)) { - kfree(event); - return -EFAULT; - } - - spin_lock_irq(&trace_lock); -#if 0 /* XXX: convert this to use tracing */ - if (is_enabled()) - relay_write(chan, event, sizeof(*headp) + len); - else -#endif - len = -EINVAL; - spin_unlock_irq(&trace_lock); - kfree(event); - return len; -} -#endif - static void print_pte(unsigned long address) { unsigned int level; @@ -481,26 +436,12 @@ static void leave_uniprocessor(void) } #endif -#if 0 /* XXX: out of order */ -static struct file_operations fops_marker = { - .owner = THIS_MODULE, - .write = write_marker -}; -#endif - void enable_mmiotrace(void) { mutex_lock(&mmiotrace_mutex); if (is_enabled()) goto out; -#if 0 /* XXX: tracing does not support text entries */ - marker_file = debugfs_create_file("marker", 0660, dir, NULL, - &fops_marker); - if (!marker_file) - pr_err(NAME "marker file creation failed.\n"); -#endif - if (nommiotrace) pr_info(NAME "MMIO tracing disabled.\n"); enter_uniprocessor(); @@ -525,11 +466,6 @@ void disable_mmiotrace(void) clear_trace_list(); /* guarantees: no more kmmio callbacks */ leave_uniprocessor(); - if (marker_file) { - debugfs_remove(marker_file); - marker_file = NULL; - } - pr_info(NAME "disabled.\n"); out: mutex_unlock(&mmiotrace_mutex); diff --git a/include/linux/mmiotrace.h b/include/linux/mmiotrace.h index 60cc3bf5c538..139d7c88d9c9 100644 --- a/include/linux/mmiotrace.h +++ b/include/linux/mmiotrace.h @@ -67,8 +67,7 @@ enum mm_io_opcode { MMIO_WRITE = 0x2, /* struct mmiotrace_rw */ MMIO_PROBE = 0x3, /* struct mmiotrace_map */ MMIO_UNPROBE = 0x4, /* struct mmiotrace_map */ - MMIO_MARKER = 0x5, /* raw char data */ - MMIO_UNKNOWN_OP = 0x6, /* struct mmiotrace_rw */ + MMIO_UNKNOWN_OP = 0x5, /* struct mmiotrace_rw */ }; struct mmiotrace_rw { -- cgit v1.2.3 From 37a52f5ef120b93734bb2461744512b55695f69c Mon Sep 17 00:00:00 2001 From: Harvey Harrison Date: Tue, 23 Sep 2008 15:05:46 -0700 Subject: x86: suppress trivial sparse signedness warnings Could just as easily change the three casts to cast to the correct type...this patch changes the type of ftrace_nop instead. Supresses sparse warnings: arch/x86/kernel/ftrace.c:157:14: warning: incorrect type in assignment (different signedness) arch/x86/kernel/ftrace.c:157:14: expected long *static [toplevel] ftrace_nop arch/x86/kernel/ftrace.c:157:14: got unsigned long * arch/x86/kernel/ftrace.c:161:14: warning: incorrect type in assignment (different signedness) arch/x86/kernel/ftrace.c:161:14: expected long *static [toplevel] ftrace_nop arch/x86/kernel/ftrace.c:161:14: got unsigned long * arch/x86/kernel/ftrace.c:165:14: warning: incorrect type in assignment (different signedness) arch/x86/kernel/ftrace.c:165:14: expected long *static [toplevel] ftrace_nop arch/x86/kernel/ftrace.c:165:14: got unsigned long * Signed-off-by: Harvey Harrison Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 082d99642622..66d900248fc2 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -22,7 +22,7 @@ /* Long is fine, even if it is only 4 bytes ;-) */ -static long *ftrace_nop; +static unsigned long *ftrace_nop; union ftrace_code_union { char code[MCOUNT_INSN_SIZE]; -- cgit v1.2.3 From ac2b86fdef5b44f194eefaa6b7b6aea9423d1bc2 Mon Sep 17 00:00:00 2001 From: Frédéric Weisbecker Date: Wed, 24 Sep 2008 16:31:56 +0100 Subject: x86/ftrace: use uaccess in atomic context With latest -tip I get this bug: [ 49.439988] in_atomic():0, irqs_disabled():1 [ 49.440118] INFO: lockdep is turned off. [ 49.440118] Pid: 2814, comm: modprobe Tainted: G W 2.6.27-rc7 #4 [ 49.440118] [] __might_sleep+0xe1/0x120 [ 49.440118] [] ftrace_modify_code+0x2a/0xd0 [ 49.440118] [] ? ftrace_test_p6nop+0x0/0xa [ 49.440118] [] __ftrace_update_code+0xfe/0x2f0 [ 49.440118] [] ? ftrace_test_p6nop+0x0/0xa [ 49.440118] [] ftrace_convert_nops+0x50/0x80 [ 49.440118] [] ftrace_init_module+0x16/0x20 [ 49.440118] [] load_module+0x185b/0x1d30 [ 49.440118] [] ? find_get_page+0x0/0xf0 [ 49.440118] [] ? sprintf+0x0/0x30 [ 49.440118] [] ? mutex_lock_interruptible_nested+0x1f2/0x350 [ 49.440118] [] sys_init_module+0x53/0x1b0 [ 49.440118] [] ? do_page_fault+0x0/0x740 [ 49.440118] [] syscall_call+0x7/0xb [ 49.440118] ======================= It is because ftrace_modify_code() calls copy_to_user and copy_from_user. These functions have been inserted after guessing that there couldn't be any race condition but copy_[to/from]_user might sleep and __ftrace_update_code is called with local_irq_saved. These function have been inserted since this commit: d5e92e8978fd2574e415dc2792c5eb592978243d: "ftrace: x86 use copy from user function" Signed-off-by: Frederic Weisbecker Acked-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'arch') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 66d900248fc2..222507e8157b 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -71,13 +71,13 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code, * No real locking needed, this code is run through * kstop_machine, or before SMP starts. */ - if (__copy_from_user(replaced, (char __user *)ip, MCOUNT_INSN_SIZE)) + if (__copy_from_user_inatomic(replaced, (char __user *)ip, MCOUNT_INSN_SIZE)) return 1; if (memcmp(replaced, old_code, MCOUNT_INSN_SIZE) != 0) return 2; - WARN_ON_ONCE(__copy_to_user((char __user *)ip, new_code, + WARN_ON_ONCE(__copy_to_user_inatomic((char __user *)ip, new_code, MCOUNT_INSN_SIZE)); sync_core(); -- cgit v1.2.3 From 5b9261d93e5fa3db4995d5b77b5ed365166e001c Mon Sep 17 00:00:00 2001 From: Mathieu Desnoyers Date: Mon, 29 Sep 2008 11:11:47 -0400 Subject: sputrace: use marker_synchronize_unregister() We need a marker_synchronize_unregister() before the end of exit() to make sure every probe callers have exited the non preemptible section and thus are not executing the probe code anymore. Signed-off-by: Mathieu Desnoyers Acked-by: Jeremy Kerr Signed-off-by: Ingo Molnar --- arch/powerpc/platforms/cell/spufs/sputrace.c | 1 + 1 file changed, 1 insertion(+) (limited to 'arch') diff --git a/arch/powerpc/platforms/cell/spufs/sputrace.c b/arch/powerpc/platforms/cell/spufs/sputrace.c index 92d20e993ede..2ece399f2862 100644 --- a/arch/powerpc/platforms/cell/spufs/sputrace.c +++ b/arch/powerpc/platforms/cell/spufs/sputrace.c @@ -232,6 +232,7 @@ static void __exit sputrace_exit(void) remove_proc_entry("sputrace", NULL); kfree(sputrace_log); + marker_synchronize_unregister(); } module_init(sputrace_init); -- cgit v1.2.3 From 8b27386a9ce9c7f0f8702cff7565a46802ad57d1 Mon Sep 17 00:00:00 2001 From: Anders Kaseorg Date: Thu, 9 Oct 2008 22:19:08 -0400 Subject: ftrace: make ftrace_test_p6nop disassembler-friendly Commit 4c3dc21b136f8cb4b72afee16c3ba7e961656c0b in tip introduced the 5-byte NOP ftrace_test_p6nop: jmp . + 5 .byte 0x00, 0x00, 0x00 This is not friendly to disassemblers because an odd number of 0x00s ends in the middle of an instruction boundary. This changes the 0x00s to 1-byte NOPs (0x90). Signed-off-by: Anders Kaseorg Acked-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'arch') diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 222507e8157b..d073d981a730 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -132,7 +132,9 @@ int __init ftrace_dyn_arch_init(void *data) ".section .text, \"ax\"\n" "ftrace_test_jmp:" "jmp ftrace_test_p6nop\n" - ".byte 0x00,0x00,0x00\n" /* 2 byte jmp + 3 bytes */ + "nop\n" + "nop\n" + "nop\n" /* 2 byte jmp + 3 bytes */ "ftrace_test_p6nop:" P6_NOP5 "jmp 1f\n" @@ -161,7 +163,7 @@ int __init ftrace_dyn_arch_init(void *data) ftrace_nop = (unsigned long *)ftrace_test_nop5; break; case 2: - pr_info("ftrace: converting mcount calls to jmp 1f\n"); + pr_info("ftrace: converting mcount calls to jmp . + 5\n"); ftrace_nop = (unsigned long *)ftrace_test_jmp; break; } -- cgit v1.2.3