diff options
Diffstat (limited to 'arch')
48 files changed, 2630 insertions, 1793 deletions
diff --git a/arch/powerpc/include/asm/ftrace.h b/arch/powerpc/include/asm/ftrace.h index b298f7a..e5f2ae8 100644 --- a/arch/powerpc/include/asm/ftrace.h +++ b/arch/powerpc/include/asm/ftrace.h @@ -7,7 +7,19 @@ #ifndef __ASSEMBLY__ extern void _mcount(void); -#endif + +#ifdef CONFIG_DYNAMIC_FTRACE +static inline unsigned long ftrace_call_adjust(unsigned long addr) +{ + /* reloction of mcount call site is the same as the address */ + return addr; +} + +struct dyn_arch_ftrace { + struct module *mod; +}; +#endif /* CONFIG_DYNAMIC_FTRACE */ +#endif /* __ASSEMBLY__ */ #endif diff --git a/arch/powerpc/include/asm/module.h b/arch/powerpc/include/asm/module.h index e5f14b1..0845488 100644 --- a/arch/powerpc/include/asm/module.h +++ b/arch/powerpc/include/asm/module.h @@ -34,11 +34,19 @@ struct mod_arch_specific { #ifdef __powerpc64__ unsigned int stubs_section; /* Index of stubs section in module */ unsigned int toc_section; /* What section is the TOC? */ -#else +#ifdef CONFIG_DYNAMIC_FTRACE + unsigned long toc; + unsigned long tramp; +#endif + +#else /* powerpc64 */ /* Indices of PLT sections within module. */ unsigned int core_plt_section; unsigned int init_plt_section; +#ifdef CONFIG_DYNAMIC_FTRACE + unsigned long tramp; #endif +#endif /* powerpc64 */ /* List of BUG addresses, source line numbers and filenames */ struct list_head bug_list; @@ -68,6 +76,12 @@ struct mod_arch_specific { # endif /* MODULE */ #endif +#ifdef CONFIG_DYNAMIC_FTRACE +# ifdef MODULE + asm(".section .ftrace.tramp,\"ax\",@nobits; .align 3; .previous"); +# endif /* MODULE */ +#endif + struct exception_table_entry; void sort_ex_table(struct exception_table_entry *start, diff --git a/arch/powerpc/kernel/Makefile b/arch/powerpc/kernel/Makefile index 92673b4..d17edb4 100644 --- a/arch/powerpc/kernel/Makefile +++ b/arch/powerpc/kernel/Makefile @@ -17,6 +17,7 @@ ifdef CONFIG_FUNCTION_TRACER CFLAGS_REMOVE_cputable.o = -pg -mno-sched-epilog CFLAGS_REMOVE_prom_init.o = -pg -mno-sched-epilog CFLAGS_REMOVE_btext.o = -pg -mno-sched-epilog +CFLAGS_REMOVE_prom.o = -pg -mno-sched-epilog ifdef CONFIG_DYNAMIC_FTRACE # dynamic ftrace setup. diff --git a/arch/powerpc/kernel/entry_32.S b/arch/powerpc/kernel/entry_32.S index 7ecc0d1..6f7eb7e 100644 --- a/arch/powerpc/kernel/entry_32.S +++ b/arch/powerpc/kernel/entry_32.S @@ -1162,39 +1162,17 @@ machine_check_in_rtas: #ifdef CONFIG_DYNAMIC_FTRACE _GLOBAL(mcount) _GLOBAL(_mcount) - stwu r1,-48(r1) - stw r3, 12(r1) - stw r4, 16(r1) - stw r5, 20(r1) - stw r6, 24(r1) - mflr r3 - stw r7, 28(r1) - mfcr r5 - stw r8, 32(r1) - stw r9, 36(r1) - stw r10,40(r1) - stw r3, 44(r1) - stw r5, 8(r1) - subi r3, r3, MCOUNT_INSN_SIZE - .globl mcount_call -mcount_call: - bl ftrace_stub - nop - lwz r6, 8(r1) - lwz r0, 44(r1) - lwz r3, 12(r1) + /* + * It is required that _mcount on PPC32 must preserve the + * link register. But we have r0 to play with. We use r0 + * to push the return address back to the caller of mcount + * into the ctr register, restore the link register and + * then jump back using the ctr register. + */ + mflr r0 mtctr r0 - lwz r4, 16(r1) - mtcr r6 - lwz r5, 20(r1) - lwz r6, 24(r1) - lwz r0, 52(r1) - lwz r7, 28(r1) - lwz r8, 32(r1) + lwz r0, 4(r1) mtlr r0 - lwz r9, 36(r1) - lwz r10,40(r1) - addi r1, r1, 48 bctr _GLOBAL(ftrace_caller) diff --git a/arch/powerpc/kernel/entry_64.S b/arch/powerpc/kernel/entry_64.S index e0bcf93..383ed6e 100644 --- a/arch/powerpc/kernel/entry_64.S +++ b/arch/powerpc/kernel/entry_64.S @@ -894,18 +894,6 @@ _GLOBAL(enter_prom) #ifdef CONFIG_DYNAMIC_FTRACE _GLOBAL(mcount) _GLOBAL(_mcount) - /* Taken from output of objdump from lib64/glibc */ - mflr r3 - stdu r1, -112(r1) - std r3, 128(r1) - subi r3, r3, MCOUNT_INSN_SIZE - .globl mcount_call -mcount_call: - bl ftrace_stub - nop - ld r0, 128(r1) - mtlr r0 - addi r1, r1, 112 blr _GLOBAL(ftrace_caller) diff --git a/arch/powerpc/kernel/ftrace.c b/arch/powerpc/kernel/ftrace.c index f4b006e..5355244 100644 --- a/arch/powerpc/kernel/ftrace.c +++ b/arch/powerpc/kernel/ftrace.c @@ -9,22 +9,30 @@ #include <linux/spinlock.h> #include <linux/hardirq.h> +#include <linux/uaccess.h> +#include <linux/module.h> #include <linux/ftrace.h> #include <linux/percpu.h> #include <linux/init.h> #include <linux/list.h> #include <asm/cacheflush.h> +#include <asm/code-patching.h> #include <asm/ftrace.h> +#if 0 +#define DEBUGP printk +#else +#define DEBUGP(fmt , ...) do { } while (0) +#endif -static unsigned int ftrace_nop = 0x60000000; +static unsigned int ftrace_nop = PPC_NOP_INSTR; #ifdef CONFIG_PPC32 # define GET_ADDR(addr) addr #else /* PowerPC64's functions are data that points to the functions */ -# define GET_ADDR(addr) *(unsigned long *)addr +# define GET_ADDR(addr) (*(unsigned long *)addr) #endif @@ -33,12 +41,12 @@ static unsigned int ftrace_calc_offset(long ip, long addr) return (int)(addr - ip); } -unsigned char *ftrace_nop_replace(void) +static unsigned char *ftrace_nop_replace(void) { return (char *)&ftrace_nop; } -unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) +static unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) { static unsigned int op; @@ -68,49 +76,422 @@ unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) # define _ASM_PTR " .long " #endif -int +static int ftrace_modify_code(unsigned long ip, unsigned char *old_code, unsigned char *new_code) { - unsigned replaced; - unsigned old = *(unsigned *)old_code; - unsigned new = *(unsigned *)new_code; - int faulted = 0; + unsigned char replaced[MCOUNT_INSN_SIZE]; /* * Note: Due to modules and __init, code can * disappear and change, we need to protect against faulting - * as well as code changing. + * as well as code changing. We do this by using the + * probe_kernel_* functions. * * No real locking needed, this code is run through - * kstop_machine. + * kstop_machine, or before SMP starts. */ - asm volatile ( - "1: lwz %1, 0(%2)\n" - " cmpw %1, %5\n" - " bne 2f\n" - " stwu %3, 0(%2)\n" - "2:\n" - ".section .fixup, \"ax\"\n" - "3: li %0, 1\n" - " b 2b\n" - ".previous\n" - ".section __ex_table,\"a\"\n" - _ASM_ALIGN "\n" - _ASM_PTR "1b, 3b\n" - ".previous" - : "=r"(faulted), "=r"(replaced) - : "r"(ip), "r"(new), - "0"(faulted), "r"(old) - : "memory"); - - if (replaced != old && replaced != new) - faulted = 2; - - if (!faulted) - flush_icache_range(ip, ip + 8); - - return faulted; + + /* read the text we want to modify */ + if (probe_kernel_read(replaced, (void *)ip, MCOUNT_INSN_SIZE)) + return -EFAULT; + + /* Make sure it is what we expect it to be */ + if (memcmp(replaced, old_code, MCOUNT_INSN_SIZE) != 0) + return -EINVAL; + + /* replace the text with the new text */ + if (probe_kernel_write((void *)ip, new_code, MCOUNT_INSN_SIZE)) + return -EPERM; + + flush_icache_range(ip, ip + 8); + + return 0; +} + +/* + * Helper functions that are the same for both PPC64 and PPC32. + */ +static int test_24bit_addr(unsigned long ip, unsigned long addr) +{ + + /* use the create_branch to verify that this offset can be branched */ + return create_branch((unsigned int *)ip, addr, 0); +} + +static int is_bl_op(unsigned int op) +{ + return (op & 0xfc000003) == 0x48000001; +} + +static unsigned long find_bl_target(unsigned long ip, unsigned int op) +{ + static int offset; + + offset = (op & 0x03fffffc); + /* make it signed */ + if (offset & 0x02000000) + offset |= 0xfe000000; + + return ip + (long)offset; +} + +#ifdef CONFIG_PPC64 +static int +__ftrace_make_nop(struct module *mod, + struct dyn_ftrace *rec, unsigned long addr) +{ + unsigned int op; + unsigned int jmp[5]; + unsigned long ptr; + unsigned long ip = rec->ip; + unsigned long tramp; + int offset; + + /* read where this goes */ + if (probe_kernel_read(&op, (void *)ip, sizeof(int))) + return -EFAULT; + + /* Make sure that that this is still a 24bit jump */ + if (!is_bl_op(op)) { + printk(KERN_ERR "Not expected bl: opcode is %x\n", op); + return -EINVAL; + } + + /* lets find where the pointer goes */ + tramp = find_bl_target(ip, op); + + /* + * On PPC64 the trampoline looks like: + * 0x3d, 0x82, 0x00, 0x00, addis r12,r2, <high> + * 0x39, 0x8c, 0x00, 0x00, addi r12,r12, <low> + * Where the bytes 2,3,6 and 7 make up the 32bit offset + * to the TOC that holds the pointer. + * to jump to. + * 0xf8, 0x41, 0x00, 0x28, std r2,40(r1) + * 0xe9, 0x6c, 0x00, 0x20, ld r11,32(r12) + * The actually address is 32 bytes from the offset + * into the TOC. + * 0xe8, 0x4c, 0x00, 0x28, ld r2,40(r12) + */ + + DEBUGP("ip:%lx jumps to %lx r2: %lx", ip, tramp, mod->arch.toc); + + /* Find where the trampoline jumps to */ + if (probe_kernel_read(jmp, (void *)tramp, sizeof(jmp))) { + printk(KERN_ERR "Failed to read %lx\n", tramp); + return -EFAULT; + } + + DEBUGP(" %08x %08x", jmp[0], jmp[1]); + + /* verify that this is what we expect it to be */ + if (((jmp[0] & 0xffff0000) != 0x3d820000) || + ((jmp[1] & 0xffff0000) != 0x398c0000) || + (jmp[2] != 0xf8410028) || + (jmp[3] != 0xe96c0020) || + (jmp[4] != 0xe84c0028)) { + printk(KERN_ERR "Not a trampoline\n"); + return -EINVAL; + } + + offset = (unsigned)((unsigned short)jmp[0]) << 16 | + (unsigned)((unsigned short)jmp[1]); + + DEBUGP(" %x ", offset); + + /* get the address this jumps too */ + tramp = mod->arch.toc + offset + 32; + DEBUGP("toc: %lx", tramp); + + if (probe_kernel_read(jmp, (void *)tramp, 8)) { + printk(KERN_ERR "Failed to read %lx\n", tramp); + return -EFAULT; + } + + DEBUGP(" %08x %08x\n", jmp[0], jmp[1]); + + ptr = ((unsigned long)jmp[0] << 32) + jmp[1]; + + /* This should match what was called */ + if (ptr != GET_ADDR(addr)) { + printk(KERN_ERR "addr does not match %lx\n", ptr); + return -EINVAL; + } + + /* + * We want to nop the line, but the next line is + * 0xe8, 0x41, 0x00, 0x28 ld r2,40(r1) + * This needs to be turned to a nop too. + */ + if (probe_kernel_read(&op, (void *)(ip+4), MCOUNT_INSN_SIZE)) + return -EFAULT; + + if (op != 0xe8410028) { + printk(KERN_ERR "Next line is not ld! (%08x)\n", op); + return -EINVAL; + } + + /* + * Milton Miller pointed out that we can not blindly do nops. + * If a task was preempted when calling a trace function, + * the nops will remove the way to restore the TOC in r2 + * and the r2 TOC will get corrupted. + */ + + /* + * Replace: + * bl <tramp> <==== will be replaced with "b 1f" + * ld r2,40(r1) + * 1: + */ + op = 0x48000008; /* b +8 */ + + if (probe_kernel_write((void *)ip, &op, MCOUNT_INSN_SIZE)) + return -EPERM; + + + flush_icache_range(ip, ip + 8); + + return 0; +} + +#else /* !PPC64 */ +static int +__ftrace_make_nop(struct module *mod, + struct dyn_ftrace *rec, unsigned long addr) +{ + unsigned int op; + unsigned int jmp[4]; + unsigned long ip = rec->ip; + unsigned long tramp; + + if (probe_kernel_read(&op, (void *)ip, MCOUNT_INSN_SIZE)) + return -EFAULT; + + /* Make sure that that this is still a 24bit jump */ + if (!is_bl_op(op)) { + printk(KERN_ERR "Not expected bl: opcode is %x\n", op); + return -EINVAL; + } + + /* lets find where the pointer goes */ + tramp = find_bl_target(ip, op); + + /* + * On PPC32 the trampoline looks like: + * 0x3d, 0x60, 0x00, 0x00 lis r11,sym@ha + * 0x39, 0x6b, 0x00, 0x00 addi r11,r11,sym@l + * 0x7d, 0x69, 0x03, 0xa6 mtctr r11 + * 0x4e, 0x80, 0x04, 0x20 bctr + */ + + DEBUGP("ip:%lx jumps to %lx", ip, tramp); + + /* Find where the trampoline jumps to */ + if (probe_kernel_read(jmp, (void *)tramp, sizeof(jmp))) { + printk(KERN_ERR "Failed to read %lx\n", tramp); + return -EFAULT; + } + + DEBUGP(" %08x %08x ", jmp[0], jmp[1]); + + /* verify that this is what we expect it to be */ + if (((jmp[0] & 0xffff0000) != 0x3d600000) || + ((jmp[1] & 0xffff0000) != 0x396b0000) || + (jmp[2] != 0x7d6903a6) || + (jmp[3] != 0x4e800420)) { + printk(KERN_ERR "Not a trampoline\n"); + return -EINVAL; + } + + tramp = (jmp[1] & 0xffff) | + ((jmp[0] & 0xffff) << 16); + if (tramp & 0x8000) + tramp -= 0x10000; + + DEBUGP(" %x ", tramp); + + if (tramp != addr) { + printk(KERN_ERR + "Trampoline location %08lx does not match addr\n", + tramp); + return -EINVAL; + } + + op = PPC_NOP_INSTR; + + if (probe_kernel_write((void *)ip, &op, MCOUNT_INSN_SIZE)) + return -EPERM; + + flush_icache_range(ip, ip + 8); + + return 0; +} +#endif /* PPC64 */ + +int ftrace_make_nop(struct module *mod, + struct dyn_ftrace *rec, unsigned long addr) +{ + unsigned char *old, *new; + unsigned long ip = rec->ip; + + /* + * If the calling address is more that 24 bits away, + * then we had to use a trampoline to make the call. + * Otherwise just update the call site. + */ + if (test_24bit_addr(ip, addr)) { + /* within range */ + old = ftrace_call_replace(ip, addr); + new = ftrace_nop_replace(); + return ftrace_modify_code(ip, old, new); + } + + /* + * Out of range jumps are called from modules. + * We should either already have a pointer to the module + * or it has been passed in. + */ + if (!rec->arch.mod) { + if (!mod) { + printk(KERN_ERR "No module loaded addr=%lx\n", + addr); + return -EFAULT; + } + rec->arch.mod = mod; + } else if (mod) { + if (mod != rec->arch.mod) { + printk(KERN_ERR + "Record mod %p not equal to passed in mod %p\n", + rec->arch.mod, mod); + return -EINVAL; + } + /* nothing to do if mod == rec->arch.mod */ + } else + mod = rec->arch.mod; + + return __ftrace_make_nop(mod, rec, addr); + +} + +#ifdef CONFIG_PPC64 +static int +__ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr) +{ + unsigned int op[2]; + unsigned long ip = rec->ip; + + /* read where this goes */ + if (probe_kernel_read(op, (void *)ip, MCOUNT_INSN_SIZE * 2)) + return -EFAULT; + + /* + * It should be pointing to two nops or + * b +8; ld r2,40(r1) + */ + if (((op[0] != 0x48000008) || (op[1] != 0xe8410028)) && + ((op[0] != PPC_NOP_INSTR) || (op[1] != PPC_NOP_INSTR))) { + printk(KERN_ERR "Expected NOPs but have %x %x\n", op[0], op[1]); + return -EINVAL; + } + + /* If we never set up a trampoline to ftrace_caller, then bail */ + if (!rec->arch.mod->arch.tramp) { + printk(KERN_ERR "No ftrace trampoline\n"); + return -EINVAL; + } + + /* create the branch to the trampoline */ + op[0] = create_branch((unsigned int *)ip, + rec->arch.mod->arch.tramp, BRANCH_SET_LINK); + if (!op[0]) { + printk(KERN_ERR "REL24 out of range!\n"); + return -EINVAL; + } + + /* ld r2,40(r1) */ + op[1] = 0xe8410028; + + DEBUGP("write to %lx\n", rec->ip); + + if (probe_kernel_write((void *)ip, op, MCOUNT_INSN_SIZE * 2)) + return -EPERM; + + flush_icache_range(ip, ip + 8); + + return 0; +} +#else +static int +__ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr) +{ + unsigned int op; + unsigned long ip = rec->ip; + + /* read where this goes */ + if (probe_kernel_read(&op, (void *)ip, MCOUNT_INSN_SIZE)) + return -EFAULT; + + /* It should be pointing to a nop */ + if (op != PPC_NOP_INSTR) { + printk(KERN_ERR "Expected NOP but have %x\n", op); + return -EINVAL; + } + + /* If we never set up a trampoline to ftrace_caller, then bail */ + if (!rec->arch.mod->arch.tramp) { + printk(KERN_ERR "No ftrace trampoline\n"); + return -EINVAL; + } + + /* create the branch to the trampoline */ + op = create_branch((unsigned int *)ip, + rec->arch.mod->arch.tramp, BRANCH_SET_LINK); + if (!op) { + printk(KERN_ERR "REL24 out of range!\n"); + return -EINVAL; + } + + DEBUGP("write to %lx\n", rec->ip); + + if (probe_kernel_write((void *)ip, &op, MCOUNT_INSN_SIZE)) + return -EPERM; + + flush_icache_range(ip, ip + 8); + + return 0; +} +#endif /* CONFIG_PPC64 */ + +int ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr) +{ + unsigned char *old, *new; + unsigned long ip = rec->ip; + + /* + * If the calling address is more that 24 bits away, + * then we had to use a trampoline to make the call. + * Otherwise just update the call site. + */ + if (test_24bit_addr(ip, addr)) { + /* within range */ + old = ftrace_nop_replace(); + new = ftrace_call_replace(ip, addr); + return ftrace_modify_code(ip, old, new); + } + + /* + * Out of range jumps are called from modules. + * Being that we are converting from nop, it had better + * already have a module defined. + */ + if (!rec->arch.mod) { + printk(KERN_ERR "No module loaded\n"); + return -EINVAL; + } + + return __ftrace_make_call(rec, addr); } int ftrace_update_ftrace_func(ftrace_func_t func) @@ -128,10 +509,10 @@ int ftrace_update_ftrace_func(ftrace_func_t func) int __init ftrace_dyn_arch_init(void *data) { - /* This is running in kstop_machine */ + /* caller expects data to be zero */ + unsigned long *p = data; - ftrace_mcount_set(data); + *p = 0; return 0; } - diff --git a/arch/powerpc/kernel/idle.c b/arch/powerpc/kernel/idle.c index 31982d0..88d9c1d 100644 --- a/arch/powerpc/kernel/idle.c +++ b/arch/powerpc/kernel/idle.c @@ -69,10 +69,15 @@ void cpu_idle(void) smp_mb(); local_irq_disable(); + /* Don't trace irqs off for idle */ + stop_critical_timings(); + /* check again after disabling irqs */ if (!need_resched() && !cpu_should_die()) ppc_md.power_save(); + start_critical_timings(); + local_irq_enable(); set_thread_flag(TIF_POLLING_NRFLAG); diff --git a/arch/powerpc/kernel/module_32.c b/arch/powerpc/kernel/module_32.c index 2df91a0..f832773 100644 --- a/arch/powerpc/kernel/module_32.c +++ b/arch/powerpc/kernel/module_32.c @@ -22,6 +22,7 @@ #include <linux/fs.h> #include <linux/string.h> #include <linux/kernel.h> +#include <linux/ftrace.h> #include <linux/cache.h> #include <linux/bug.h> #include <linux/sort.h> @@ -53,6 +54,9 @@ static unsigned int count_relocs(const Elf32_Rela *rela, unsigned int num) r_addend = rela[i].r_addend; } +#ifdef CONFIG_DYNAMIC_FTRACE + _count_relocs++; /* add one for ftrace_caller */ +#endif return _count_relocs; } @@ -306,5 +310,11 @@ int apply_relocate_add(Elf32_Shdr *sechdrs, return -ENOEXEC; } } +#ifdef CONFIG_DYNAMIC_FTRACE + module->arch.tramp = + do_plt_call(module->module_core, + (unsigned long)ftrace_caller, + sechdrs, module); +#endif return 0; } diff --git a/arch/powerpc/kernel/module_64.c b/arch/powerpc/kernel/module_64.c index 1af2377..8992b03 100644 --- a/arch/powerpc/kernel/module_64.c +++ b/arch/powerpc/kernel/module_64.c @@ -20,6 +20,7 @@ #include <linux/moduleloader.h> #include <linux/err.h> #include <linux/vmalloc.h> +#include <linux/ftrace.h> #include <linux/bug.h> #include <asm/module.h> #include <asm/firmware.h> @@ -163,6 +164,11 @@ static unsigned long get_stubs_size(const Elf64_Ehdr *hdr, } } +#ifdef CONFIG_DYNAMIC_FTRACE + /* make the trampoline to the ftrace_caller */ + relocs++; +#endif + DEBUGP("Looks like a total of %lu stubs, max\n", relocs); return relocs * sizeof(struct ppc64_stub_entry); } @@ -441,5 +447,12 @@ int apply_relocate_add(Elf64_Shdr *sechdrs, } } +#ifdef CONFIG_DYNAMIC_FTRACE + me->arch.toc = my_r2(sechdrs, me); + me->arch.tramp = stub_for_addr(sechdrs, + (unsigned long)ftrace_caller, + me); +#endif + return 0; } diff --git a/arch/powerpc/lib/Makefile b/arch/powerpc/lib/Makefile index d69912c..8db3527 100644 --- a/arch/powerpc/lib/Makefile +++ b/arch/powerpc/lib/Makefile @@ -6,6 +6,9 @@ ifeq ($(CONFIG_PPC64),y) EXTRA_CFLAGS += -mno-minimal-toc endif +CFLAGS_REMOVE_code-patching.o = -pg +CFLAGS_REMOVE_feature-fixups.o = -pg + obj-y := string.o alloc.o \ checksum_$(CONFIG_WORD_SIZE).o obj-$(CONFIG_PPC32) += div64.o copy_32.o crtsavres.o diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index ac22bb7..45c86fb 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -29,11 +29,14 @@ config X86 select HAVE_FTRACE_MCOUNT_RECORD select HAVE_DYNAMIC_FTRACE select HAVE_FUNCTION_TRACER + select HAVE_FUNCTION_GRAPH_TRACER + select HAVE_FUNCTION_TRACE_MCOUNT_TEST select HAVE_KVM if ((X86_32 && !X86_VOYAGER && !X86_VISWS && !X86_NUMAQ) || X86_64) select HAVE_ARCH_KGDB if !X86_VOYAGER select HAVE_ARCH_TRACEHOOK select HAVE_GENERIC_DMA_COHERENT if X86_32 select HAVE_EFFICIENT_UNALIGNED_ACCESS + select USER_STACKTRACE_SUPPORT config ARCH_DEFCONFIG string diff --git a/arch/x86/Kconfig.cpu b/arch/x86/Kconfig.cpu index 8e99073..85a7857 100644 --- a/arch/x86/Kconfig.cpu +++ b/arch/x86/Kconfig.cpu @@ -515,12 +515,12 @@ config CPU_SUP_UMC_32 config X86_DS def_bool X86_PTRACE_BTS depends on X86_DEBUGCTLMSR + select HAVE_HW_BRANCH_TRACER config X86_PTRACE_BTS bool "Branch Trace Store" default y depends on X86_DEBUGCTLMSR - depends on BROKEN help This adds a ptrace interface to the hardware's branch trace store. diff --git a/arch/x86/Kconfig.debug b/arch/x86/Kconfig.debug index 2a3dfbd..fa013f5 100644 --- a/arch/x86/Kconfig.debug +++ b/arch/x86/Kconfig.debug @@ -186,14 +186,10 @@ config IOMMU_LEAK Add a simple leak tracer to the IOMMU code. This is useful when you are debugging a buggy device driver that leaks IOMMU mappings. -config MMIOTRACE_HOOKS - bool - config MMIOTRACE bool "Memory mapped IO tracing" depends on DEBUG_KERNEL && PCI select TRACING - select MMIOTRACE_HOOKS help Mmiotrace traces Memory Mapped I/O access and is meant for debugging and reverse engineering. It is called from the ioremap diff --git a/arch/x86/include/asm/cpufeature.h b/arch/x86/include/asm/cpufeature.h index cfdf8c2..ea408dc 100644 --- a/arch/x86/include/asm/cpufeature.h +++ b/arch/x86/include/asm/cpufeature.h @@ -80,7 +80,6 @@ #define X86_FEATURE_UP (3*32+ 9) /* smp kernel running on up */ #define X86_FEATURE_FXSAVE_LEAK (3*32+10) /* "" FXSAVE leaks FOP/FIP/FOP */ #define X86_FEATURE_ARCH_PERFMON (3*32+11) /* Intel Architectural PerfMon */ -#define X86_FEATURE_NOPL (3*32+20) /* The NOPL (0F 1F) instructions */ #define X86_FEATURE_PEBS (3*32+12) /* Precise-Event Based Sampling */ #define X86_FEATURE_BTS (3*32+13) /* Branch Trace Store */ #define X86_FEATURE_SYSCALL32 (3*32+14) /* "" syscall in ia32 userspace */ @@ -92,6 +91,8 @@ #define X86_FEATURE_NOPL (3*32+20) /* The NOPL (0F 1F) instructions */ #define X86_FEATURE_AMDC1E (3*32+21) /* AMD C1E detected */ #define X86_FEATURE_XTOPOLOGY (3*32+22) /* cpu topology enum extensions */ +#define X86_FEATURE_TSC_RELIABLE (3*32+23) /* TSC is known to be reliable */ +#define X86_FEATURE_NONSTOP_TSC (3*32+24) /* TSC does not stop in C states */ /* Intel-defined CPU features, CPUID level 0x00000001 (ecx), word 4 */ #define X86_FEATURE_XMM3 (4*32+ 0) /* "pni" SSE-3 */ @@ -117,6 +118,7 @@ #define X86_FEATURE_XSAVE (4*32+26) /* XSAVE/XRSTOR/XSETBV/XGETBV */ #define X86_FEATURE_OSXSAVE (4*32+27) /* "" XSAVE enabled in the OS */ #define X86_FEATURE_AVX (4*32+28) /* Advanced Vector Extensions */ +#define X86_FEATURE_HYPERVISOR (4*32+31) /* Running on a hypervisor */ /* VIA/Cyrix/Centaur-defined CPU features, CPUID level 0xC0000001, word 5 */ #define X86_FEATURE_XSTORE (5*32+ 2) /* "rng" RNG present (xstore) */ @@ -237,6 +239,7 @@ extern const char * const x86_power_flags[32]; #define cpu_has_xmm4_2 boot_cpu_has(X86_FEATURE_XMM4_2) #define cpu_has_x2apic boot_cpu_has(X86_FEATURE_X2APIC) #define cpu_has_xsave boot_cpu_has(X86_FEATURE_XSAVE) +#define cpu_has_hypervisor boot_cpu_has(X86_FEATURE_HYPERVISOR) #if defined(CONFIG_X86_INVLPG) || defined(CONFIG_X86_64) # define cpu_has_invlpg 1 diff --git a/arch/x86/include/asm/ds.h b/arch/x86/include/asm/ds.h index a950084..a8f672b 100644 --- a/arch/x86/include/asm/ds.h +++ b/arch/x86/include/asm/ds.h @@ -6,14 +6,13 @@ * precise-event based sampling (PEBS). * * It manages: - * - per-thread and per-cpu allocation of BTS and PEBS - * - buffer memory allocation (optional) - * - buffer overflow handling + * - DS and BTS hardware configuration + * - buffer overflow handling (to be done) * - buffer access * - * It assumes: - * - get_task_struct on all parameter tasks - * - current is allowed to trace parameter tasks + * It does not do: + * - security checking (is the caller allowed to trace the task) + * - buffer allocation (memory accounting) * * * Copyright (C) 2007-2008 Intel Corporation. @@ -26,11 +25,51 @@ #include <linux/types.h> #include <linux/init.h> +#include <linux/err.h> #ifdef CONFIG_X86_DS struct task_struct; +struct ds_context; +struct ds_tracer; +struct bts_tracer; +struct pebs_tracer; + +typedef void (*bts_ovfl_callback_t)(struct bts_tracer *); +typedef void (*pebs_ovfl_callback_t)(struct pebs_tracer *); + + +/* + * A list of features plus corresponding macros to talk about them in + * the ds_request function's flags parameter. + * + * We use the enum to index an array of corresponding control bits; + * we use the macro to index a flags bit-vector. + */ +enum ds_feature { + dsf_bts = 0, + dsf_bts_kernel, +#define BTS_KERNEL (1 << dsf_bts_kernel) + /* trace kernel-mode branches */ + + dsf_bts_user, +#define BTS_USER (1 << dsf_bts_user) + /* trace user-mode branches */ + + dsf_bts_overflow, + dsf_bts_max, + dsf_pebs = dsf_bts_max, + + dsf_pebs_max, + dsf_ctl_max = dsf_pebs_max, + dsf_bts_timestamps = dsf_ctl_max, +#define BTS_TIMESTAMPS (1 << dsf_bts_timestamps) + /* add timestamps into BTS trace */ + +#define BTS_USER_FLAGS (BTS_KERNEL | BTS_USER | BTS_TIMESTAMPS) +}; + /* * Request BTS or PEBS @@ -38,163 +77,169 @@ struct task_struct; * Due to alignement constraints, the actual buffer may be slightly * smaller than the requested or provided buffer. * - * Returns 0 on success; -Eerrno otherwise + * Returns a pointer to a tracer structure on success, or + * ERR_PTR(errcode) on failure. + * + * The interrupt threshold is independent from the overflow callback + * to allow users to use their own overflow interrupt handling mechanism. * * task: the task to request recording for; * NULL for per-cpu recording on the current cpu * base: the base pointer for the (non-pageable) buffer; - * NULL if buffer allocation requested - * size: the size of the requested or provided buffer + * size: the size of the provided buffer in bytes * ovfl: pointer to a function to be called on buffer overflow; * NULL if cyclic buffer requested + * th: the interrupt threshold in records from the end of the buffer; + * -1 if no interrupt threshold is requested. + * flags: a bit-mask of the above flags */ -typedef void (*ds_ovfl_callback_t)(struct task_struct *); -extern int ds_request_bts(struct task_struct *task, void *base, size_t size, - ds_ovfl_callback_t ovfl); -extern int ds_request_pebs(struct task_struct *task, void *base, size_t size, - ds_ovfl_callback_t ovfl); +extern struct bts_tracer *ds_request_bts(struct task_struct *task, + void *base, size_t size, + bts_ovfl_callback_t ovfl, + size_t th, unsigned int flags); +extern struct pebs_tracer *ds_request_pebs(struct task_struct *task, + void *base, size_t size, + pebs_ovfl_callback_t ovfl, + size_t th, unsigned int flags); /* * Release BTS or PEBS resources + * Suspend and resume BTS or PEBS tracing * - * Frees buffers allocated on ds_request. - * - * Returns 0 on success; -Eerrno otherwise - * - * task: the task to release resources for; - * NULL to release resources for the current cpu + * tracer: the tracer handle returned from ds_request_~() */ -extern int ds_release_bts(struct task_struct *task); -extern int ds_release_pebs(struct task_struct *task); +extern void ds_release_bts(struct bts_tracer *tracer); +extern void ds_suspend_bts(struct bts_tracer *tracer); +extern void ds_resume_bts(struct bts_tracer *tracer); +extern void ds_release_pebs(struct pebs_tracer *tracer); +extern void ds_suspend_pebs(struct pebs_tracer *tracer); +extern void ds_resume_pebs(struct pebs_tracer *tracer); -/* - * Return the (array) index of the write pointer. - * (assuming an array of BTS/PEBS records) - * - * Returns -Eerrno on error - * - * task: the task to access; - * NULL to access the current cpu - * pos (out): if not NULL, will hold the result - */ -extern int ds_get_bts_index(struct task_struct *task, size_t *pos); -extern int ds_get_pebs_index(struct task_struct *task, size_t *pos); /* - * Return the (array) index one record beyond the end of the array. - * (assuming an array of BTS/PEBS records) + * The raw DS buffer state as it is used for BTS and PEBS recording. * - * Returns -Eerrno on error - * - * task: the task to access; - * NULL to access the current cpu - * pos (out): if not NULL, will hold the result + * This is the low-level, arch-dependent interface for working + * directly on the raw trace data. */ -extern int ds_get_bts_end(struct task_struct *task, size_t *pos); -extern int ds_get_pebs_end(struct task_struct *task, size_t *pos); +struct ds_trace { + /* the number of bts/pebs records */ + size_t n; + /* the size of a bts/pebs record in bytes */ + size_t size; + /* pointers into the raw buffer: + - to the first entry */ + void *begin; + /* - one beyond the last entry */ + void *end; + /* - one beyond the newest entry */ + void *top; + /* - the interrupt threshold */ + void *ith; + /* flags given on ds_request() */ + unsigned int flags; +}; /* - * Provide a pointer to the BTS/PEBS record at parameter index. - * (assuming an array of BTS/PEBS records) - * - * The pointer points directly into the buffer. The user is - * responsible for copying the record. - * - * Returns the size of a single record on success; -Eerrno on error - * - * task: the task to access; - * NULL to access the current cpu - * index: the index of the requested record - * record (out): pointer to the requested record + * An arch-independent view on branch trace data. */ -extern int ds_access_bts(struct task_struct *task, - size_t index, const void **record); -extern int ds_access_pebs(struct task_struct *task, - size_t index, const void **record); +enum bts_qualifier { + bts_invalid, +#define BTS_INVALID bts_invalid + + bts_branch, +#define BTS_BRANCH bts_branch + + bts_task_arrives, +#define BTS_TASK_ARRIVES bts_task_arrives + + bts_task_departs, +#define BTS_TASK_DEPARTS bts_task_departs + + bts_qual_bit_size = 4, + bts_qual_max = (1 << bts_qual_bit_size), +}; + +struct bts_struct { + __u64 qualifier; + union { + /* BTS_BRANCH */ + struct { + __u64 from; + __u64 to; + } lbr; + /* BTS_TASK_ARRIVES or BTS_TASK_DEPARTS */ + struct { + __u64 jiffies; + pid_t pid; + } timestamp; + } variant; +}; -/* - * Write one or more BTS/PEBS records at the write pointer index and - * advance the write pointer. - * - * If size is not a multiple of the record size, trailing bytes are - * zeroed out. - * - * May result in one or more overflow notifications. - * - * If called during overflow handling, that is, with index >= - * interrupt threshold, the write will wrap around. - * - * An overflow notification is given if and when the interrupt - * threshold is reached during or after the write. - * - * Returns the number of bytes written or -Eerrno. - * - * task: the task to access; - * NULL to access the current cpu - * buffer: the buffer to write - * size: the size of the buffer - */ -extern int ds_write_bts(struct task_struct *task, - const void *buffer, size_t size); -extern int ds_write_pebs(struct task_struct *task, - const void *buffer, size_t size); /* - * Same as ds_write_bts/pebs, but omit ownership checks. + * The BTS state. * - * This is needed to have some other task than the owner of the - * BTS/PEBS buffer or the parameter task itself write into the - * respective buffer. + * This gives access to the raw DS state and adds functions to provide + * an arch-independent view of the BTS data. */ -extern int ds_unchecked_write_bts(struct task_struct *task, - const void *buffer, size_t size); -extern int ds_unchecked_write_pebs(struct task_struct *task, - const void *buffer, size_t size); +struct bts_trace { + struct ds_trace ds; + + int (*read)(struct bts_tracer *tracer, const void *at, + struct bts_struct *out); + int (*write)(struct bts_tracer *tracer, const struct bts_struct *in); +}; + /* - * Reset the write pointer of the BTS/PEBS buffer. + * The PEBS state. * - * Returns 0 on success; -Eerrno on error - * - * task: the task to access; - * NULL to access the current cpu + * This gives access to the raw DS state and the PEBS-specific counter + * reset value. */ -extern int ds_reset_bts(struct task_struct *task); -extern int ds_reset_pebs(struct task_struct *task); +struct pebs_trace { + struct ds_trace ds; + + /* the PEBS reset value */ + unsigned long long reset_value; +}; + /* - * Clear the BTS/PEBS buffer and reset the write pointer. - * The entire buffer will be zeroed out. + * Read the BTS or PEBS trace. * - * Returns 0 on success; -Eerrno on error + * Returns a view on the trace collected for the parameter tracer. + * + * The view remains valid as long as the traced task is not running or + * the tracer is suspended. + * Writes into the trace buffer are not reflected. * - * task: the task to access; - * NULL to access the current cpu + * tracer: the tracer handle returned from ds_request_~() */ -extern int ds_clear_bts(struct task_struct *task); -extern int ds_clear_pebs(struct task_struct *task); +extern const struct bts_trace *ds_read_bts(struct bts_tracer *tracer); +extern const struct pebs_trace *ds_read_pebs(struct pebs_tracer *tracer); + /* - * Provide the PEBS counter reset value. + * Reset the write pointer of the BTS/PEBS buffer. * * Returns 0 on success; -Eerrno on error * - * task: the task to access; - * NULL to access the current cpu - * value (out): the counter reset value + * tracer: the tracer handle returned from ds_request_~() */ -extern int ds_get_pebs_reset(struct task_struct *task, u64 *value); +extern int ds_reset_bts(struct bts_tracer *tracer); +extern int ds_reset_pebs(struct pebs_tracer *tracer); /* * Set the PEBS counter reset value. * * Returns 0 on success; -Eerrno on error * - * task: the task to access; - * NULL to access the current cpu + * tracer: the tracer handle returned from ds_request_pebs() * value: the new counter reset value */ -extern int ds_set_pebs_reset(struct task_struct *task, u64 value); +extern int ds_set_pebs_reset(struct pebs_tracer *tracer, u64 value); /* * Initialization @@ -202,39 +247,26 @@ extern int ds_set_pebs_reset(struct task_struct *task, u64 value); struct cpuinfo_x86; extern void __cpuinit ds_init_intel(struct cpuinfo_x86 *); - - /* - * The DS context - part of struct thread_struct. + * Context switch work */ -struct ds_context { - /* pointer to the DS configuration; goes into MSR_IA32_DS_AREA */ - unsigned char *ds; - /* the owner of the BTS and PEBS configuration, respectively */ - struct task_struct *owner[2]; - /* buffer overflow notification function for BTS and PEBS */ - ds_ovfl_callback_t callback[2]; - /* the original buffer address */ - void *buffer[2]; - /* the number of allocated pages for on-request allocated buffers */ - unsigned int pages[2]; - /* use count */ - unsigned long count; - /* a pointer to the context location inside the thread_struct - * or the per_cpu context array */ - struct ds_context **this; - /* a pointer to the task owning this context, or NULL, if the - * context is owned by a cpu */ - struct task_struct *task; -}; +extern void ds_switch_to(struct task_struct *prev, struct task_struct *next); -/* called by exit_thread() to free leftover contexts */ -extern void ds_free(struct ds_context *context); +/* + * Task clone/init and cleanup work + */ +extern void ds_copy_thread(struct task_struct *tsk, struct task_struct *father); +extern void ds_exit_thread(struct task_struct *tsk); #else /* CONFIG_X86_DS */ struct cpuinfo_x86; static inline void __cpuinit ds_init_intel(struct cpuinfo_x86 *ignored) {} +static inline void ds_switch_to(struct task_struct *prev, + struct task_struct *next) {} +static inline void ds_copy_thread(struct task_struct *tsk, + struct task_struct *father) {} +static inline void ds_exit_thread(struct task_struct *tsk) {} #endif /* CONFIG_X86_DS */ #endif /* _ASM_X86_DS_H */ diff --git a/arch/x86/include/asm/ftrace.h b/arch/x86/include/asm/ftrace.h index 9e8bc29..b55b4a7 100644 --- a/arch/x86/include/asm/ftrace.h +++ b/arch/x86/include/asm/ftrace.h @@ -1,6 +1,33 @@ #ifndef _ASM_X86_FTRACE_H #define _ASM_X86_FTRACE_H +#ifdef __ASSEMBLY__ + + .macro MCOUNT_SAVE_FRAME + /* taken from glibc */ + 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) + .endm + + .macro MCOUNT_RESTORE_FRAME + 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 + .endm + +#endif + #ifdef CONFIG_FUNCTION_TRACER #define MCOUNT_ADDR ((long)(mcount)) #define MCOUNT_INSN_SIZE 5 /* sizeof mcount call */ @@ -17,8 +44,40 @@ static inline unsigned long ftrace_call_adjust(unsigned long addr) */ return addr - 1; } -#endif +#ifdef CONFIG_DYNAMIC_FTRACE + +struct dyn_arch_ftrace { + /* No extra data needed for x86 */ +}; + +#endif /* CONFIG_DYNAMIC_FTRACE */ +#endif /* __ASSEMBLY__ */ #endif /* CONFIG_FUNCTION_TRACER */ +#ifdef CONFIG_FUNCTION_GRAPH_TRACER + +#ifndef __ASSEMBLY__ + +/* + * Stack of return addresses for functions + * of a thread. + * Used in struct thread_info + */ +struct ftrace_ret_stack { + unsigned long ret; + unsigned long func; + unsigned long long calltime; +}; + +/* + * Primary handler of a function return. + * It relays on ftrace_return_to_handler. + * Defined in entry_32/64.S + */ +extern void return_to_handler(void); + +#endif /* __ASSEMBLY__ */ +#endif /* CONFIG_FUNCTION_GRAPH_TRACER */ + #endif /* _ASM_X86_FTRACE_H */ diff --git a/arch/x86/include/asm/msr.h b/arch/x86/include/asm/msr.h index c2a812e..b8a1799 100644 --- a/arch/x86/include/asm/msr.h +++ b/arch/x86/include/asm/msr.h @@ -85,7 +85,8 @@ static inline void native_write_msr(unsigned int msr, asm volatile("wrmsr" : : "c" (msr), "a"(low), "d" (high) : "memory"); } -static inline int native_write_msr_safe(unsigned int msr, +/* Can be uninlined because referenced by paravirt */ +notrace static inline int native_write_msr_safe(unsigned int msr, unsigned low, unsigned high) { int err; diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h index 5ca01e3..aa5914f 100644 --- a/arch/x86/include/asm/processor.h +++ b/arch/x86/include/asm/processor.h @@ -752,6 +752,19 @@ extern void switch_to_new_gdt(void); extern void cpu_init(void); extern void init_gdt(int cpu); +static inline unsigned long get_debugctlmsr(void) +{ + unsigned long debugctlmsr = 0; + +#ifndef CONFIG_X86_DEBUGCTLMSR + if (boot_cpu_data.x86 < 6) + return 0; +#endif + rdmsrl(MSR_IA32_DEBUGCTLMSR, debugctlmsr); + + return debugctlmsr; +} + static inline void update_debugctlmsr(unsigned long debugctlmsr) { #ifndef CONFIG_X86_DEBUGCTLMSR diff --git a/arch/x86/include/asm/ptrace.h b/arch/x86/include/asm/ptrace.h index eefb059..6d34d95 100644 --- a/arch/x86/include/asm/ptrace.h +++ b/arch/x86/include/asm/ptrace.h @@ -6,7 +6,6 @@ #include <asm/processor-flags.h> #ifdef __KERNEL__ -#include <asm/ds.h> /* the DS BTS struct is used for ptrace too */ #include <asm/segment.h> #endif @@ -128,34 +127,6 @@ struct pt_regs { #endif /* !__i386__ */ -#ifdef CONFIG_X86_PTRACE_BTS -/* a branch trace record entry - * - * In order to unify the interface between various processor versions, - * we use the below data structure for all processors. - */ -enum bts_qualifier { - BTS_INVALID = 0, - BTS_BRANCH, - BTS_TASK_ARRIVES, - BTS_TASK_DEPARTS -}; - -struct bts_struct { - __u64 qualifier; - union { - /* BTS_BRANCH */ - struct { - __u64 from_ip; - __u64 to_ip; - } lbr; - /* BTS_TASK_ARRIVES or - BTS_TASK_DEPARTS */ - __u64 jiffies; - } variant; -}; -#endif /* CONFIG_X86_PTRACE_BTS */ - #ifdef __KERNEL__ #include <linux/init.h> @@ -163,13 +134,6 @@ struct bts_struct { struct cpuinfo_x86; struct task_struct; -#ifdef CONFIG_X86_PTRACE_BTS -extern void __cpuinit ptrace_bts_init_intel(struct cpuinfo_x86 *); -extern void ptrace_bts_take_timestamp(struct task_struct *, enum bts_qualifier); -#else -#define ptrace_bts_init_intel(config) do {} while (0) -#endif /* CONFIG_X86_PTRACE_BTS */ - extern unsigned long profile_pc(struct pt_regs *regs); extern unsigned long @@ -271,6 +235,13 @@ extern int do_get_thread_area(struct task_struct *p, int idx, extern int do_set_thread_area(struct task_struct *p, int idx, struct user_desc __user *info, int can_allocate); +extern void x86_ptrace_untrace(struct task_struct *); +extern void x86_ptrace_fork(struct task_struct *child, + unsigned long clone_flags); + +#define arch_ptrace_untrace(tsk) x86_ptrace_untrace(tsk) +#define arch_ptrace_fork(child, flags) x86_ptrace_fork(child, flags) + #endif /* __KERNEL__ */ #endif /* !__ASSEMBLY__ */ diff --git a/arch/x86/include/asm/thread_info.h b/arch/x86/include/asm/thread_info.h index e44d379..bf8113d 100644 --- a/arch/x86/include/asm/thread_info.h +++ b/arch/x86/include/asm/thread_info.h @@ -20,6 +20,8 @@ struct task_struct; struct exec_domain; #include <asm/processor.h> +#include <asm/ftrace.h> +#include <asm/atomic.h> struct thread_info { struct task_struct *task; /* main task structure */ @@ -91,7 +93,6 @@ struct thread_info { #define TIF_FORCED_TF 24 /* true if TF in eflags artificially */ #define TIF_DEBUGCTLMSR 25 /* uses thread_struct.debugctlmsr */ #define TIF_DS_AREA_MSR 26 /* uses thread_struct.ds_area_msr */ -#define TIF_BTS_TRACE_TS 27 /* record scheduling event timestamps */ #define _TIF_SYSCALL_TRACE (1 << TIF_SYSCALL_TRACE) #define _TIF_NOTIFY_RESUME (1 << TIF_NOTIFY_RESUME) @@ -113,7 +114,6 @@ struct thread_info { #define _TIF_FORCED_TF (1 << TIF_FORCED_TF) #define _TIF_DEBUGCTLMSR (1 << TIF_DEBUGCTLMSR) #define _TIF_DS_AREA_MSR (1 << TIF_DS_AREA_MSR) -#define _TIF_BTS_TRACE_TS (1 << TIF_BTS_TRACE_TS) /* work to do in syscall_trace_enter() */ #define _TIF_WORK_SYSCALL_ENTRY \ @@ -139,8 +139,7 @@ struct thread_info { /* flags to check in __switch_to() */ #define _TIF_WORK_CTXSW \ - (_TIF_IO_BITMAP|_TIF_DEBUGCTLMSR|_TIF_DS_AREA_MSR|_TIF_BTS_TRACE_TS| \ - _TIF_NOTSC) + (_TIF_IO_BITMAP|_TIF_DEBUGCTLMSR|_TIF_DS_AREA_MSR|_TIF_NOTSC) #define _TIF_WORK_CTXSW_PREV _TIF_WORK_CTXSW #define _TIF_WORK_CTXSW_NEXT (_TIF_WORK_CTXSW|_TIF_DEBUG) diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index b62a766..1cad931 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -25,7 +25,7 @@ CFLAGS_tsc.o := $(nostackp) obj-y := process_$(BITS).o signal_$(BITS).o entry_$(BITS).o obj-y += traps.o irq.o irq_$(BITS).o dumpstack_$(BITS).o -obj-y += time_$(BITS).o ioport.o ldt.o +obj-y += time_$(BITS).o ioport.o ldt.o dumpstack.o obj-y += setup.o i8259.o irqinit_$(BITS).o setup_percpu.o obj-$(CONFIG_X86_VISWS) += visws_quirks.o obj-$(CONFIG_X86_32) += probe_roms_32.o @@ -65,6 +65,7 @@ obj-$(CONFIG_X86_LOCAL_APIC) += apic.o nmi.o obj-$(CONFIG_X86_IO_APIC) += io_apic.o obj-$(CONFIG_X86_REBOOTFIXUPS) += reboot_fixups_32.o obj-$(CONFIG_DYNAMIC_FTRACE) += ftrace.o +obj-$(CONFIG_FUNCTION_GRAPH_TRACER) += ftrace.o obj-$(CONFIG_KEXEC) += machine_kexec_$(BITS).o obj-$(CONFIG_KEXEC) += relocate_kernel_$(BITS).o crash.o obj-$(CONFIG_CRASH_DUMP) += crash_dump_$(BITS).o diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index 16f9487..b946ac1 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -30,6 +30,7 @@ #include <linux/module.h> #include <linux/dmi.h> #include <linux/dmar.h> +#include <linux/ftrace.h> #include <asm/atomic.h> #include <asm/smp.h> @@ -800,7 +801,7 @@ static void local_apic_timer_interrupt(void) * [ if a single-CPU system runs an SMP kernel then we call the local * interrupt as well. Thus we cannot inline the local irq ... ] */ -void smp_apic_timer_interrupt(struct pt_regs *regs) +void __irq_entry smp_apic_timer_interrupt(struct pt_regs *regs) { struct pt_regs *old_regs = set_irq_regs(regs); diff --git a/arch/x86/kernel/cpu/Makefile b/arch/x86/kernel/cpu/Makefile index 82ec607..4ae495a 100644 --- a/arch/x86/kernel/cpu/Makefile +++ b/arch/x86/kernel/cpu/Makefile @@ -2,6 +2,11 @@ # Makefile for x86-compatible CPU details and quirks # +# Don't trace early stages of a secondary CPU boot +ifdef CONFIG_FUNCTION_TRACER +CFLAGS_REMOVE_common.o = -pg +endif + obj-y := intel_cacheinfo.o addon_cpuid_features.o obj-y += proc.o capflags.o powerflags.o common.o diff --git a/arch/x86/kernel/cpu/addon_cpuid_features.c b/arch/x86/kernel/cpu/addon_cpuid_features.c index ef8f831..2cf2363 100644 --- a/arch/x86/kernel/cpu/addon_cpuid_features.c +++ b/arch/x86/kernel/cpu/addon_cpuid_features.c @@ -120,9 +120,17 @@ void __cpuinit detect_extended_topology(struct cpuinfo_x86 *c) c->cpu_core_id = phys_pkg_id(c->initial_apicid, ht_mask_width) & core_select_mask; c->phys_proc_id = phys_pkg_id(c->initial_apicid, core_plus_mask_width); + /* + * Reinit the apicid, now that we have extended initial_apicid. + */ + c->apicid = phys_pkg_id(c->initial_apicid, 0); #else c->cpu_core_id = phys_pkg_id(ht_mask_width) & core_select_mask; c->phys_proc_id = phys_pkg_id(core_plus_mask_width); + /* + * Reinit the apicid, now that we have extended initial_apicid. + */ + c->apicid = phys_pkg_id(0); #endif c->x86_max_cores = (core_level_siblings / smp_num_siblings); diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c index 8f1e31d..7c878f6 100644 --- a/arch/x86/kernel/cpu/amd.c +++ b/arch/x86/kernel/cpu/amd.c @@ -283,9 +283,14 @@ static void __cpuinit early_init_amd(struct cpuinfo_x86 *c) { early_init_amd_mc(c); - /* c->x86_power is 8000_0007 edx. Bit 8 is constant TSC */ - if (c->x86_power & (1<<8)) + /* + * c->x86_power is 8000_0007 edx. Bit 8 is TSC runs at constant rate + * with P/T states and does not stop in deep C-states + */ + if (c->x86_power & (1 << 8)) { set_cpu_cap(c, X86_FEATURE_CONSTANT_TSC); + set_cpu_cap(c, X86_FEATURE_NONSTOP_TSC); + } #ifdef CONFIG_X86_64 set_cpu_cap(c, X86_FEATURE_SYSCALL32); diff --git a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c index 8e48c5d..88ea02d 100644 --- a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c +++ b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c @@ -33,6 +33,7 @@ #include <linux/cpufreq.h> #include <linux/compiler.h> #include <linux/dmi.h> +#include <linux/ftrace.h> #include <linux/acpi.h> #include <acpi/processor.h> @@ -391,6 +392,7 @@ static int acpi_cpufreq_target(struct cpufreq_policy *policy, unsigned int next_perf_state = 0; /* Index into perf table */ unsigned int i; int result = 0; + struct power_trace it; dprintk("acpi_cpufreq_target %d (%d)\n", target_freq, policy->cpu); @@ -427,6 +429,8 @@ static int acpi_cpufreq_target(struct cpufreq_policy *policy, } } + trace_power_mark(&it, POWER_PSTATE, next_perf_state); + switch (data->cpu_feature) { case SYSTEM_INTEL_MSR_CAPABLE: cmd.type = SYSTEM_INTEL_MSR_CAPABLE; diff --git a/arch/x86/kernel/cpu/intel.c b/arch/x86/kernel/cpu/intel.c index cce0b61..8ea6929 100644 --- a/arch/x86/kernel/cpu/intel.c +++ b/arch/x86/kernel/cpu/intel.c @@ -11,7 +11,6 @@ #include <asm/pgtable.h> #include <asm/msr.h> #include <asm/uaccess.h> -#include <asm/ptrace.h> #include <asm/ds.h> #include <asm/bugs.h> @@ -41,6 +40,16 @@ static void __cpuinit early_init_intel(struct cpuinfo_x86 *c) if (c->x86 == 15 && c->x86_cache_alignment == 64) c->x86_cache_alignment = 128; #endif + + /* + * c->x86_power is 8000_0007 edx. Bit 8 is TSC runs at constant rate + * with P/T states and does not stop in deep C-states + */ + if (c->x86_power & (1 << 8)) { + set_cpu_cap(c, X86_FEATURE_CONSTANT_TSC); + set_cpu_cap(c, X86_FEATURE_NONSTOP_TSC); + } + } #ifdef CONFIG_X86_32 @@ -242,6 +251,13 @@ static void __cpuinit init_intel(struct cpuinfo_x86 *c) intel_workarounds(c); + /* + * Detect the extended topology information if available. This + * will reinitialise the initial_apicid which will be used + * in init_intel_cacheinfo() + */ + detect_extended_topology(c); + l2 = init_intel_cacheinfo(c); if (c->cpuid_level > 9) { unsigned eax = cpuid_eax(10); @@ -307,13 +323,8 @@ static void __cpuinit init_intel(struct cpuinfo_x86 *c) set_cpu_cap(c, X86_FEATURE_P4); if (c->x86 == 6) set_cpu_cap(c, X86_FEATURE_P3); - - if (cpu_has_bts) - ptrace_bts_init_intel(c); - #endif - detect_extended_topology(c); if (!cpu_has(c, X86_FEATURE_XTOPOLOGY)) { /* * let's use the legacy cpuid vector 0x1 and 0x4 for topology diff --git a/arch/x86/kernel/ds.c b/arch/x86/kernel/ds.c index a2d1176..da91701 100644 --- a/arch/x86/kernel/ds.c +++ b/arch/x86/kernel/ds.c @@ -6,14 +6,13 @@ * precise-event based sampling (PEBS). * * It manages: - * - per-thread and per-cpu allocation of BTS and PEBS - * - buffer memory allocation (optional) - * - buffer overflow handling + * - DS and BTS hardware configuration + * - buffer overflow handling (to be done) * - buffer access * - * It assumes: - * - get_task_struct on all parameter tasks - * - current is allowed to trace parameter tasks + * It does not do: + * - security checking (is the caller allowed to trace the task) + * - buffer allocation (memory accounting) * * * Copyright (C) 2007-2008 Intel Corporation. @@ -28,22 +27,69 @@ #include <linux/slab.h> #include <linux/sched.h> #include <linux/mm.h> +#include <linux/kernel.h> /* * The configuration for a particular DS hardware implementation. */ struct ds_configuration { - /* the size of the DS structure in bytes */ - unsigned char sizeof_ds; - /* the size of one pointer-typed field in the DS structure in bytes; - this covers the first 8 fields related to buffer management. */ + /* the name of the configuration */ + const char *name; + /* the size of one pointer-typed field in the DS structure and + in the BTS and PEBS buffers in bytes; + this covers the first 8 DS fields related to buffer management. */ unsigned char sizeof_field; /* the size of a BTS/PEBS record in bytes */ unsigned char sizeof_rec[2]; + /* a series of bit-masks to control various features indexed + * by enum ds_feature */ + unsigned long ctl[dsf_ctl_max]; }; -static struct ds_configuration ds_cfg; +static DEFINE_PER_CPU(struct ds_configuration, ds_cfg_array); +#define ds_cfg per_cpu(ds_cfg_array, smp_processor_id()) + +#define MAX_SIZEOF_DS (12 * 8) /* maximal size of a DS configuration */ +#define MAX_SIZEOF_BTS (3 * 8) /* maximal size of a BTS record */ +#define DS_ALIGNMENT (1 << 3) /* BTS and PEBS buffer alignment */ + +#define BTS_CONTROL \ + (ds_cfg.ctl[dsf_bts] | ds_cfg.ctl[dsf_bts_kernel] | ds_cfg.ctl[dsf_bts_user] |\ + ds_cfg.ctl[dsf_bts_overflow]) + + +/* + * A BTS or PEBS tracer. + * + * This holds the configuration of the tracer and serves as a handle + * to identify tracers. + */ +struct ds_tracer { + /* the DS context (partially) owned by this tracer */ + struct ds_context *context; + /* the buffer provided on ds_request() and its size in bytes */ + void *buffer; + size_t size; +}; + +struct bts_tracer { + /* the common DS part */ + struct ds_tracer ds; + /* the trace including the DS configuration */ + struct bts_trace trace; + /* buffer overflow notification function */ + bts_ovfl_callback_t ovfl; +}; + +struct pebs_tracer { + /* the common DS part */ + struct ds_tracer ds; + /* the trace including the DS configuration */ + struct pebs_trace trace; + /* buffer overflow notification function */ + pebs_ovfl_callback_t ovfl; +}; /* * Debug Store (DS) save area configuration (see Intel64 and IA32 @@ -109,32 +155,9 @@ static inline void ds_set(unsigned char *base, enum ds_qualifier qual, /* - * Locking is done only for allocating BTS or PEBS resources and for - * guarding context and buffer memory allocation. - * - * Most functions require the current task to own the ds context part - * they are going to access. All the locking is done when validating - * access to the context. + * Locking is done only for allocating BTS or PEBS resources. */ -static spinlock_t ds_lock = __SPIN_LOCK_UNLOCKED(ds_lock); - -/* - * Validate that the current task is allowed to access the BTS/PEBS - * buffer of the parameter task. - * - * Returns 0, if access is granted; -Eerrno, otherwise. - */ -static inline int ds_validate_access(struct ds_context *context, - enum ds_qualifier qual) -{ - if (!context) - return -EPERM; - - if (context->owner[qual] == current) - return 0; - - return -EPERM; -} +static DEFINE_SPINLOCK(ds_lock); /* @@ -150,27 +173,32 @@ static inline int ds_validate_access(struct ds_context *context, * >0 number of per-thread tracers * <0 number of per-cpu tracers * - * The below functions to get and put tracers and to check the - * allocation type require the ds_lock to be held by the caller. - * * Tracers essentially gives the number of ds contexts for a certain * type of allocation. */ -static long tracers; +static atomic_t tracers = ATOMIC_INIT(0); static inline void get_tracer(struct task_struct *task) { - tracers += (task ? 1 : -1); + if (task) + atomic_inc(&tracers); + else + atomic_dec(&tracers); } static inline void put_tracer(struct task_struct *task) { - tracers -= (task ? 1 : -1); + if (task) + atomic_dec(&tracers); + else + atomic_inc(&tracers); } static inline int check_tracer(struct task_struct *task) { - return (task ? (tracers >= 0) : (tracers <= 0)); + return task ? + (atomic_read(&tracers) >= 0) : + (atomic_read(&tracers) <= 0); } @@ -183,99 +211,70 @@ static inline int check_tracer(struct task_struct *task) * * Contexts are use-counted. They are allocated on first access and * deallocated when the last user puts the context. - * - * We distinguish between an allocating and a non-allocating get of a - * context: - * - the allocating get is used for requesting BTS/PEBS resources. It - * requires the caller to hold the global ds_lock. - * - the non-allocating get is used for all other cases. A - * non-existing context indicates an error. It acquires and releases - * the ds_lock itself for obtaining the context. - * - * A context and its DS configuration are allocated and deallocated - * together. A context always has a DS configuration of the - * appropriate size. - */ -static DEFINE_PER_CPU(struct ds_context *, system_context); - -#define this_system_context per_cpu(system_context, smp_processor_id()) - -/* - * Returns the pointer to the parameter task's context or to the - * system-wide context, if task is NULL. - * - * Increases the use count of the returned context, if not NULL. */ -static inline struct ds_context *ds_get_context(struct task_struct *task) -{ - struct ds_context *context; - unsigned long irq; +struct ds_context { + /* pointer to the DS configuration; goes into MSR_IA32_DS_AREA */ + unsigned char ds[MAX_SIZEOF_DS]; + /* the owner of the BTS and PEBS configuration, respectively */ + struct bts_tracer *bts_master; + struct pebs_tracer *pebs_master; + /* use count */ + unsigned long count; + /* a pointer to the context location inside the thread_struct + * or the per_cpu context array */ + struct ds_context **this; + /* a pointer to the task owning this context, or NULL, if the + * context is owned by a cpu */ + struct task_struct *task; +}; - spin_lock_irqsave(&ds_lock, irq); +static DEFINE_PER_CPU(struct ds_context *, system_context_array); - context = (task ? task->thread.ds_ctx : this_system_context); - if (context) - context->count++; +#define system_context per_cpu(system_context_array, smp_processor_id()) - spin_unlock_irqrestore(&ds_lock, irq); - - return context; -} -/* - * Same as ds_get_context, but allocates the context and it's DS - * structure, if necessary; returns NULL; if out of memory. - */ -static inline struct ds_context *ds_alloc_context(struct task_struct *task) +static inline struct ds_context *ds_get_context(struct task_struct *task) { struct ds_context **p_context = - (task ? &task->thread.ds_ctx : &this_system_context); - struct ds_context *context = *p_context; + (task ? &task->thread.ds_ctx : &system_context); + struct ds_context *context = NULL; + struct ds_context *new_context = NULL; unsigned long irq; - if (!context) { - context = kzalloc(sizeof(*context), GFP_KERNEL); - if (!context) - return NULL; - - context->ds = kzalloc(ds_cfg.sizeof_ds, GFP_KERNEL); - if (!context->ds) { - kfree(context); - return NULL; - } + /* Chances are small that we already have a context. */ + new_context = kzalloc(sizeof(*new_context), GFP_KERNEL); + if (!new_context) + return NULL; - spin_lock_irqsave(&ds_lock, irq); + spin_lock_irqsave(&ds_lock, irq); - if (*p_context) { - kfree(context->ds); - kfree(context); + context = *p_context; + if (!context) { + context = new_context; - context = *p_context; - } else { - *p_context = context; + context->this = p_context; + context->task = task; + context->count = 0; - context->this = p_context; - context->task = task; + if (task) + set_tsk_thread_flag(task, TIF_DS_AREA_MSR); - if (task) - set_tsk_thread_flag(task, TIF_DS_AREA_MSR); + if (!task || (task == current)) + wrmsrl(MSR_IA32_DS_AREA, (unsigned long)context->ds); - if (!task || (task == current)) - wrmsrl(MSR_IA32_DS_AREA, - (unsigned long)context->ds); - } - spin_unlock_irqrestore(&ds_lock, irq); + *p_context = context; } context->count++; + spin_unlock_irqrestore(&ds_lock, irq); + + if (context != new_context) + kfree(new_context); + return context; } -/* - * Decreases the use count of the parameter context, if not NULL. - * Deallocates the context, if the use count reaches zero. - */ static inline void ds_put_context(struct ds_context *context) { unsigned long irq; @@ -285,8 +284,10 @@ static inline void ds_put_context(struct ds_context *context) spin_lock_irqsave(&ds_lock, irq); - if (--context->count) - goto out; + if (--context->count) { + spin_unlock_irqrestore(&ds_lock, irq); + return; + } *(context->this) = NULL; @@ -296,135 +297,263 @@ static inline void ds_put_context(struct ds_context *context) if (!context->task || (context->task == current)) wrmsrl(MSR_IA32_DS_AREA, 0); - put_tracer(context->task); + spin_unlock_irqrestore(&ds_lock, irq); - /* free any leftover buffers from tracers that did not - * deallocate them properly. */ - kfree(context->buffer[ds_bts]); - kfree(context->buffer[ds_pebs]); - kfree(context->ds); kfree(context); - out: - spin_unlock_irqrestore(&ds_lock, irq); } /* - * Handle a buffer overflow + * Call the tracer's callback on a buffer overflow. * - * task: the task whose buffers are overflowing; - * NULL for a buffer overflow on the current cpu * context: the ds context * qual: the buffer type */ -static void ds_overflow(struct task_struct *task, struct ds_context *context, - enum ds_qualifier qual) +static void ds_overflow(struct ds_context *context, enum ds_qualifier qual) { - if (!context) - return; - - if (context->callback[qual]) - (*context->callback[qual])(task); - - /* todo: do some more overflow handling */ + switch (qual) { + case ds_bts: + if (context->bts_master && + context->bts_master->ovfl) + context->bts_master->ovfl(context->bts_master); + break; + case ds_pebs: + if (context->pebs_master && + context->pebs_master->ovfl) + context->pebs_master->ovfl(context->pebs_master); + break; + } } /* - * Allocate a non-pageable buffer of the parameter size. - * Checks the memory and the locked memory rlimit. + * Write raw data into the BTS or PEBS buffer. * - * Returns the buffer, if successful; - * NULL, if out of memory or rlimit exceeded. + * The remainder of any partially written record is zeroed out. * - * size: the requested buffer size in bytes - * pages (out): if not NULL, contains the number of pages reserved + * context: the DS context + * qual: the buffer type + * record: the data to write + * size: the size of the data */ -static inline void *ds_allocate_buffer(size_t size, unsigned int *pages) +static int ds_write(struct ds_context *context, enum ds_qualifier qual, + const void *record, size_t size) { - unsigned long rlim, vm, pgsz; - void *buffer; + int bytes_written = 0; - pgsz = PAGE_ALIGN(size) >> PAGE_SHIFT; + if (!record) + return -EINVAL; - rlim = current->signal->rlim[RLIMIT_AS].rlim_cur >> PAGE_SHIFT; - vm = current->mm->total_vm + pgsz; - if (rlim < vm) - return NULL; + while (size) { + unsigned long base, index, end, write_end, int_th; + unsigned long write_size, adj_write_size; - rlim = current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur >> PAGE_SHIFT; - vm = current->mm->locked_vm + pgsz; - if (rlim < vm) - return NULL; + /* + * write as much as possible without producing an + * overflow interrupt. + * + * interrupt_threshold must either be + * - bigger than absolute_maximum or + * - point to a record between buffer_base and absolute_maximum + * + * index points to a valid record. + */ + base = ds_get(context->ds, qual, ds_buffer_base); + index = ds_get(context->ds, qual, ds_index); + end = ds_get(context->ds, qual, ds_absolute_maximum); + int_th = ds_get(context->ds, qual, ds_interrupt_threshold); - buffer = kzalloc(size, GFP_KERNEL); - if (!buffer) - return NULL; + write_end = min(end, int_th); - current->mm->total_vm += pgsz; - current->mm->locked_vm += pgsz; + /* if we are already beyond the interrupt threshold, + * we fill the entire buffer */ + if (write_end <= index) + write_end = end; - if (pages) - *pages = pgsz; + if (write_end <= index) + break; + + write_size = min((unsigned long) size, write_end - index); + memcpy((void *)index, record, write_size); - return buffer; + record = (const char *)record + write_size; + size -= write_size; + bytes_written += write_size; + + adj_write_size = write_size / ds_cfg.sizeof_rec[qual]; + adj_write_size *= ds_cfg.sizeof_rec[qual]; + + /* zero out trailing bytes */ + memset((char *)index + write_size, 0, + adj_write_size - write_size); + index += adj_write_size; + + if (index >= end) + index = base; + ds_set(context->ds, qual, ds_index, index); + + if (index >= int_th) + ds_overflow(context, qual); + } + + return bytes_written; } -static int ds_request(struct task_struct *task, void *base, size_t size, - ds_ovfl_callback_t ovfl, enum ds_qualifier qual) + +/* + * Branch Trace Store (BTS) uses the following format. Different + * architectures vary in the size of those fields. + * - source linear address + * - destination linear address + * - flags + * + * Later architectures use 64bit pointers throughout, whereas earlier + * architectures use 32bit pointers in 32bit mode. + * + * We compute the base address for the first 8 fields based on: + * - the field size stored in the DS configuration + * - the relative field position + * + * In order to store additional information in the BTS buffer, we use + * a special source address to indicate that the record requires + * special interpretation. + * + * Netburst indicated via a bit in the flags field whether the branch + * was predicted; this is ignored. + * + * We use two levels of abstraction: + * - the raw data level defined here + * - an arch-independent level defined in ds.h + */ + +enum bts_field { + bts_from, + bts_to, + bts_flags, + + bts_qual = bts_from, + bts_jiffies = bts_to, + bts_pid = bts_flags, + + bts_qual_mask = (bts_qual_max - 1), + bts_escape = ((unsigned long)-1 & ~bts_qual_mask) +}; + +static inline unsigned long bts_get(const char *base, enum bts_field field) { - struct ds_context *context; - unsigned long buffer, adj; - const unsigned long alignment = (1 << 3); - unsigned long irq; - int error = 0; + base += (ds_cfg.sizeof_field * field); + return *(unsigned long *)base; +} + +static inline void bts_set(char *base, enum bts_field field, unsigned long val) +{ + base += (ds_cfg.sizeof_field * field);; + (*(unsigned long *)base) = val; +} - if (!ds_cfg.sizeof_ds) - return -EOPNOTSUPP; - /* we require some space to do alignment adjustments below */ - if (size < (alignment + ds_cfg.sizeof_rec[qual])) +/* + * The raw BTS data is architecture dependent. + * + * For higher-level users, we give an arch-independent view. + * - ds.h defines struct bts_struct + * - bts_read translates one raw bts record into a bts_struct + * - bts_write translates one bts_struct into the raw format and + * writes it into the top of the parameter tracer's buffer. + * + * return: bytes read/written on success; -Eerrno, otherwise + */ +static int bts_read(struct bts_tracer *tracer, const void *at, + struct bts_struct *out) +{ + if (!tracer) return -EINVAL; - /* buffer overflow notification is not yet implemented */ - if (ovfl) - return -EOPNOTSUPP; + if (at < tracer->trace.ds.begin) + return -EINVAL; + if (tracer->trace.ds.end < (at + tracer->trace.ds.size)) + return -EINVAL; - context = ds_alloc_context(task); - if (!context) - return -ENOMEM; + memset(out, 0, sizeof(*out)); + if ((bts_get(at, bts_qual) & ~bts_qual_mask) == bts_escape) { + out->qualifier = (bts_get(at, bts_qual) & bts_qual_mask); + out->variant.timestamp.jiffies = bts_get(at, bts_jiffies); + out->variant.timestamp.pid = bts_get(at, bts_pid); + } else { + out->qualifier = bts_branch; + out->variant.lbr.from = bts_get(at, bts_from); + out->variant.lbr.to = bts_get(at, bts_to); + + if (!out->variant.lbr.from && !out->variant.lbr.to) + out->qualifier = bts_invalid; + } - spin_lock_irqsave(&ds_lock, irq); + return ds_cfg.sizeof_rec[ds_bts]; +} - error = -EPERM; - if (!check_tracer(task)) - goto out_unlock; +static int bts_write(struct bts_tracer *tracer, const struct bts_struct *in) +{ + unsigned char raw[MAX_SIZEOF_BTS]; - get_tracer(task); + if (!tracer) + return -EINVAL; - error = -EALREADY; - if (context->owner[qual] == current) - goto out_put_tracer; - error = -EPERM; - if (context->owner[qual] != NULL) - goto out_put_tracer; - context->owner[qual] = current; + if (MAX_SIZEOF_BTS < ds_cfg.sizeof_rec[ds_bts]) + return -EOVERFLOW; - spin_unlock_irqrestore(&ds_lock, irq); + switch (in->qualifier) { + case bts_invalid: + bts_set(raw, bts_from, 0); + bts_set(raw, bts_to, 0); + bts_set(raw, bts_flags, 0); + break; + case bts_branch: + bts_set(raw, bts_from, in->variant.lbr.from); + bts_set(raw, bts_to, in->variant.lbr.to); + bts_set(raw, bts_flags, 0); + break; + case bts_task_arrives: + case bts_task_departs: + bts_set(raw, bts_qual, (bts_escape | in->qualifier)); + bts_set(raw, bts_jiffies, in->variant.timestamp.jiffies); + bts_set(raw, bts_pid, in->variant.timestamp.pid); + break; + default: + return -EINVAL; + } + return ds_write(tracer->ds.context, ds_bts, raw, + ds_cfg.sizeof_rec[ds_bts]); +} - error = -ENOMEM; - if (!base) { - base = ds_allocate_buffer(size, &context->pages[qual]); - if (!base) - goto out_release; - context->buffer[qual] = base; - } - error = 0; +static void ds_write_config(struct ds_context *context, + struct ds_trace *cfg, enum ds_qualifier qual) +{ + unsigned char *ds = context->ds; + + ds_set(ds, qual, ds_buffer_base, (unsigned long)cfg->begin); + ds_set(ds, qual, ds_index, (unsigned long)cfg->top); + ds_set(ds, qual, ds_absolute_maximum, (unsigned long)cfg->end); + ds_set(ds, qual, ds_interrupt_threshold, (unsigned long)cfg->ith); +} + +static void ds_read_config(struct ds_context *context, + struct ds_trace *cfg, enum ds_qualifier qual) +{ + unsigned char *ds = context->ds; - context->callback[qual] = ovfl; + cfg->begin = (void *)ds_get(ds, qual, ds_buffer_base); + cfg->top = (void *)ds_get(ds, qual, ds_index); + cfg->end = (void *)ds_get(ds, qual, ds_absolute_maximum); + cfg->ith = (void *)ds_get(ds, qual, ds_interrupt_threshold); +} + +static void ds_init_ds_trace(struct ds_trace *trace, enum ds_qualifier qual, + void *base, size_t size, size_t ith, + unsigned int flags) { + unsigned long buffer, adj; /* adjust the buffer address and size to meet alignment * constraints: @@ -436,410 +565,383 @@ static int ds_request(struct task_struct *task, void *base, size_t size, */ buffer = (unsigned long)base; - adj = ALIGN(buffer, alignment) - buffer; + adj = ALIGN(buffer, DS_ALIGNMENT) - buffer; buffer += adj; size -= adj; - size /= ds_cfg.sizeof_rec[qual]; - size *= ds_cfg.sizeof_rec[qual]; - - ds_set(context->ds, qual, ds_buffer_base, buffer); - ds_set(context->ds, qual, ds_index, buffer); - ds_set(context->ds, qual, ds_absolute_maximum, buffer + size); + trace->n = size / ds_cfg.sizeof_rec[qual]; + trace->size = ds_cfg.sizeof_rec[qual]; - if (ovfl) { - /* todo: select a suitable interrupt threshold */ - } else - ds_set(context->ds, qual, - ds_interrupt_threshold, buffer + size + 1); + size = (trace->n * trace->size); - /* we keep the context until ds_release */ - return error; - - out_release: - context->owner[qual] = NULL; - ds_put_context(context); - put_tracer(task); - return error; - - out_put_tracer: - spin_unlock_irqrestore(&ds_lock, irq); - ds_put_context(context); - put_tracer(task); - return error; + trace->begin = (void *)buffer; + trace->top = trace->begin; + trace->end = (void *)(buffer + size); + /* The value for 'no threshold' is -1, which will set the + * threshold outside of the buffer, just like we want it. + */ + trace->ith = (void *)(buffer + size - ith); - out_unlock: - spin_unlock_irqrestore(&ds_lock, irq); - ds_put_context(context); - return error; + trace->flags = flags; } -int ds_request_bts(struct task_struct *task, void *base, size_t size, - ds_ovfl_callback_t ovfl) -{ - return ds_request(task, base, size, ovfl, ds_bts); -} -int ds_request_pebs(struct task_struct *task, void *base, size_t size, - ds_ovfl_callback_t ovfl) -{ - return ds_request(task, base, size, ovfl, ds_pebs); -} - -static int ds_release(struct task_struct *task, enum ds_qualifier qual) +static int ds_request(struct ds_tracer *tracer, struct ds_trace *trace, + enum ds_qualifier qual, struct task_struct *task, + void *base, size_t size, size_t th, unsigned int flags) { struct ds_context *context; int error; - context = ds_get_context(task); - error = ds_validate_access(context, qual); - if (error < 0) + error = -EINVAL; + if (!base) goto out; - kfree(context->buffer[qual]); - context->buffer[qual] = NULL; - - current->mm->total_vm -= context->pages[qual]; - current->mm->locked_vm -= context->pages[qual]; - context->pages[qual] = 0; - context->owner[qual] = NULL; - - /* - * we put the context twice: - * once for the ds_get_context - * once for the corresponding ds_request - */ - ds_put_context(context); - out: - ds_put_context(context); - return error; -} + /* we require some space to do alignment adjustments below */ + error = -EINVAL; + if (size < (DS_ALIGNMENT + ds_cfg.sizeof_rec[qual])) + goto out; -int ds_release_bts(struct task_struct *task) -{ - return ds_release(task, ds_bts); -} + if (th != (size_t)-1) { + th *= ds_cfg.sizeof_rec[qual]; -int ds_release_pebs(struct task_struct *task) -{ - return ds_release(task, ds_pebs); -} + error = -EINVAL; + if (size <= th) + goto out; + } -static int ds_get_index(struct task_struct *task, size_t *pos, - enum ds_qualifier qual) -{ - struct ds_context *context; - unsigned long base, index; - int error; + tracer->buffer = base; + tracer->size = size; + error = -ENOMEM; context = ds_get_context(task); - error = ds_validate_access(context, qual); - if (error < 0) + if (!context) goto out; + tracer->context = context; - base = ds_get(context->ds, qual, ds_buffer_base); - index = ds_get(context->ds, qual, ds_index); + ds_init_ds_trace(trace, qual, base, size, th, flags); - error = ((index - base) / ds_cfg.sizeof_rec[qual]); - if (pos) - *pos = error; + error = 0; out: - ds_put_context(context); return error; } -int ds_get_bts_index(struct task_struct *task, size_t *pos) -{ - return ds_get_index(task, pos, ds_bts); -} - -int ds_get_pebs_index(struct task_struct *task, size_t *pos) +struct bts_tracer *ds_request_bts(struct task_struct *task, + void *base, size_t size, + bts_ovfl_callback_t ovfl, size_t th, + unsigned int flags) { - return ds_get_index(task, pos, ds_pebs); -} - -static int ds_get_end(struct task_struct *task, size_t *pos, - enum ds_qualifier qual) -{ - struct ds_context *context; - unsigned long base, end; + struct bts_tracer *tracer; + unsigned long irq; int error; - context = ds_get_context(task); - error = ds_validate_access(context, qual); - if (error < 0) + error = -EOPNOTSUPP; + if (!ds_cfg.ctl[dsf_bts]) goto out; - base = ds_get(context->ds, qual, ds_buffer_base); - end = ds_get(context->ds, qual, ds_absolute_maximum); + /* buffer overflow notification is not yet implemented */ + error = -EOPNOTSUPP; + if (ovfl) + goto out; - error = ((end - base) / ds_cfg.sizeof_rec[qual]); - if (pos) - *pos = error; - out: - ds_put_context(context); - return error; -} + error = -ENOMEM; + tracer = kzalloc(sizeof(*tracer), GFP_KERNEL); + if (!tracer) + goto out; + tracer->ovfl = ovfl; -int ds_get_bts_end(struct task_struct *task, size_t *pos) -{ - return ds_get_end(task, pos, ds_bts); -} + error = ds_request(&tracer->ds, &tracer->trace.ds, + ds_bts, task, base, size, th, flags); + if (error < 0) + goto out_tracer; -int ds_get_pebs_end(struct task_struct *task, size_t *pos) -{ - return ds_get_end(task, pos, ds_pebs); -} -static int ds_access(struct task_struct *task, size_t index, - const void **record, enum ds_qualifier qual) -{ - struct ds_context *context; - unsigned long base, idx; - int error; + spin_lock_irqsave(&ds_lock, irq); - if (!record) - return -EINVAL; + error = -EPERM; + if (!check_tracer(task)) + goto out_unlock; + get_tracer(task); - context = ds_get_context(task); - error = ds_validate_access(context, qual); - if (error < 0) - goto out; + error = -EPERM; + if (tracer->ds.context->bts_master) + goto out_put_tracer; + tracer->ds.context->bts_master = tracer; - base = ds_get(context->ds, qual, ds_buffer_base); - idx = base + (index * ds_cfg.sizeof_rec[qual]); + spin_unlock_irqrestore(&ds_lock, irq); - error = -EINVAL; - if (idx > ds_get(context->ds, qual, ds_absolute_maximum)) - goto out; - *record = (const void *)idx; - error = ds_cfg.sizeof_rec[qual]; - out: - ds_put_context(context); - return error; -} + tracer->trace.read = bts_read; + tracer->trace.write = bts_write; -int ds_access_bts(struct task_struct *task, size_t index, const void **record) -{ - return ds_access(task, index, record, ds_bts); -} + ds_write_config(tracer->ds.context, &tracer->trace.ds, ds_bts); + ds_resume_bts(tracer); -int ds_access_pebs(struct task_struct *task, size_t index, const void **record) -{ - return ds_access(task, index, record, ds_pebs); + return tracer; + + out_put_tracer: + put_tracer(task); + out_unlock: + spin_unlock_irqrestore(&ds_lock, irq); + ds_put_context(tracer->ds.context); + out_tracer: + kfree(tracer); + out: + return ERR_PTR(error); } -static int ds_write(struct task_struct *task, const void *record, size_t size, - enum ds_qualifier qual, int force) +struct pebs_tracer *ds_request_pebs(struct task_struct *task, + void *base, size_t size, + pebs_ovfl_callback_t ovfl, size_t th, + unsigned int flags) { - struct ds_context *context; + struct pebs_tracer *tracer; + unsigned long irq; int error; - if (!record) - return -EINVAL; + /* buffer overflow notification is not yet implemented */ + error = -EOPNOTSUPP; + if (ovfl) + goto out; - error = -EPERM; - context = ds_get_context(task); - if (!context) + error = -ENOMEM; + tracer = kzalloc(sizeof(*tracer), GFP_KERNEL); + if (!tracer) goto out; + tracer->ovfl = ovfl; - if (!force) { - error = ds_validate_access(context, qual); - if (error < 0) - goto out; - } + error = ds_request(&tracer->ds, &tracer->trace.ds, + ds_pebs, task, base, size, th, flags); + if (error < 0) + goto out_tracer; - error = 0; - while (size) { - unsigned long base, index, end, write_end, int_th; - unsigned long write_size, adj_write_size; + spin_lock_irqsave(&ds_lock, irq); - /* - * write as much as possible without producing an - * overflow interrupt. - * - * interrupt_threshold must either be - * - bigger than absolute_maximum or - * - point to a record between buffer_base and absolute_maximum - * - * index points to a valid record. - */ - base = ds_get(context->ds, qual, ds_buffer_base); - index = ds_get(context->ds, qual, ds_index); - end = ds_get(context->ds, qual, ds_absolute_maximum); - int_th = ds_get(context->ds, qual, ds_interrupt_threshold); + error = -EPERM; + if (!check_tracer(task)) + goto out_unlock; + get_tracer(task); - write_end = min(end, int_th); + error = -EPERM; + if (tracer->ds.context->pebs_master) + goto out_put_tracer; + tracer->ds.context->pebs_master = tracer; - /* if we are already beyond the interrupt threshold, - * we fill the entire buffer */ - if (write_end <= index) - write_end = end; + spin_unlock_irqrestore(&ds_lock, irq); - if (write_end <= index) - goto out; + ds_write_config(tracer->ds.context, &tracer->trace.ds, ds_bts); + ds_resume_pebs(tracer); - write_size = min((unsigned long) size, write_end - index); - memcpy((void *)index, record, write_size); + return tracer; - record = (const char *)record + write_size; - size -= write_size; - error += write_size; + out_put_tracer: + put_tracer(task); + out_unlock: + spin_unlock_irqrestore(&ds_lock, irq); + ds_put_context(tracer->ds.context); + out_tracer: + kfree(tracer); + out: + return ERR_PTR(error); +} - adj_write_size = write_size / ds_cfg.sizeof_rec[qual]; - adj_write_size *= ds_cfg.sizeof_rec[qual]; +void ds_release_bts(struct bts_tracer *tracer) +{ + if (!tracer) + return; - /* zero out trailing bytes */ - memset((char *)index + write_size, 0, - adj_write_size - write_size); - index += adj_write_size; + ds_suspend_bts(tracer); - if (index >= end) - index = base; - ds_set(context->ds, qual, ds_index, index); + WARN_ON_ONCE(tracer->ds.context->bts_master != tracer); + tracer->ds.context->bts_master = NULL; - if (index >= int_th) - ds_overflow(task, context, qual); - } + put_tracer(tracer->ds.context->task); + ds_put_context(tracer->ds.context); - out: - ds_put_context(context); - return error; + kfree(tracer); } -int ds_write_bts(struct task_struct *task, const void *record, size_t size) +void ds_suspend_bts(struct bts_tracer *tracer) { - return ds_write(task, record, size, ds_bts, /* force = */ 0); -} + struct task_struct *task; -int ds_write_pebs(struct task_struct *task, const void *record, size_t size) -{ - return ds_write(task, record, size, ds_pebs, /* force = */ 0); -} + if (!tracer) + return; -int ds_unchecked_write_bts(struct task_struct *task, - const void *record, size_t size) -{ - return ds_write(task, record, size, ds_bts, /* force = */ 1); -} + task = tracer->ds.context->task; -int ds_unchecked_write_pebs(struct task_struct *task, - const void *record, size_t size) -{ - return ds_write(task, record, size, ds_pebs, /* force = */ 1); + if (!task || (task == current)) + update_debugctlmsr(get_debugctlmsr() & ~BTS_CONTROL); + + if (task) { + task->thread.debugctlmsr &= ~BTS_CONTROL; + + if (!task->thread.debugctlmsr) + clear_tsk_thread_flag(task, TIF_DEBUGCTLMSR); + } } -static int ds_reset_or_clear(struct task_struct *task, - enum ds_qualifier qual, int clear) +void ds_resume_bts(struct bts_tracer *tracer) { - struct ds_context *context; - unsigned long base, end; - int error; + struct task_struct *task; + unsigned long control; - context = ds_get_context(task); - error = ds_validate_access(context, qual); - if (error < 0) - goto out; + if (!tracer) + return; - base = ds_get(context->ds, qual, ds_buffer_base); - end = ds_get(context->ds, qual, ds_absolute_maximum); + task = tracer->ds.context->task; - if (clear) - memset((void *)base, 0, end - base); + control = ds_cfg.ctl[dsf_bts]; + if (!(tracer->trace.ds.flags & BTS_KERNEL)) + control |= ds_cfg.ctl[dsf_bts_kernel]; + if (!(tracer->trace.ds.flags & BTS_USER)) + control |= ds_cfg.ctl[dsf_bts_user]; - ds_set(context->ds, qual, ds_index, base); + if (task) { + task->thread.debugctlmsr |= control; + set_tsk_thread_flag(task, TIF_DEBUGCTLMSR); + } - error = 0; - out: - ds_put_context(context); - return error; + if (!task || (task == current)) + update_debugctlmsr(get_debugctlmsr() | control); } -int ds_reset_bts(struct task_struct *task) +void ds_release_pebs(struct pebs_tracer *tracer) { - return ds_reset_or_clear(task, ds_bts, /* clear = */ 0); + if (!tracer) + return; + + ds_suspend_pebs(tracer); + + WARN_ON_ONCE(tracer->ds.context->pebs_master != tracer); + tracer->ds.context->pebs_master = NULL; + + put_tracer(tracer->ds.context->task); + ds_put_context(tracer->ds.context); + + kfree(tracer); } -int ds_reset_pebs(struct task_struct *task) +void ds_suspend_pebs(struct pebs_tracer *tracer) { - return ds_reset_or_clear(task, ds_pebs, /* clear = */ 0); + } -int ds_clear_bts(struct task_struct *task) +void ds_resume_pebs(struct pebs_tracer *tracer) { - return ds_reset_or_clear(task, ds_bts, /* clear = */ 1); + } -int ds_clear_pebs(struct task_struct *task) +const struct bts_trace *ds_read_bts(struct bts_tracer *tracer) { - return ds_reset_or_clear(task, ds_pebs, /* clear = */ 1); + if (!tracer) + return NULL; + + ds_read_config(tracer->ds.context, &tracer->trace.ds, ds_bts); + return &tracer->trace; } -int ds_get_pebs_reset(struct task_struct *task, u64 *value) +const struct pebs_trace *ds_read_pebs(struct pebs_tracer *tracer) { - struct ds_context *context; - int error; + if (!tracer) + return NULL; + + ds_read_config(tracer->ds.context, &tracer->trace.ds, ds_pebs); + tracer->trace.reset_value = + *(u64 *)(tracer->ds.context->ds + (ds_cfg.sizeof_field * 8)); - if (!value) + return &tracer->trace; +} + +int ds_reset_bts(struct bts_tracer *tracer) +{ + if (!tracer) return -EINVAL; - context = ds_get_context(task); - error = ds_validate_access(context, ds_pebs); - if (error < 0) - goto out; + tracer->trace.ds.top = tracer->trace.ds.begin; - *value = *(u64 *)(context->ds + (ds_cfg.sizeof_field * 8)); + ds_set(tracer->ds.context->ds, ds_bts, ds_index, + (unsigned long)tracer->trace.ds.top); - error = 0; - out: - ds_put_context(context); - return error; + return 0; } -int ds_set_pebs_reset(struct task_struct *task, u64 value) +int ds_reset_pebs(struct pebs_tracer *tracer) { - struct ds_context *context; - int error; + if (!tracer) + return -EINVAL; - context = ds_get_context(task); - error = ds_validate_access(context, ds_pebs); - if (error < 0) - goto out; + tracer->trace.ds.top = tracer->trace.ds.begin; - *(u64 *)(context->ds + (ds_cfg.sizeof_field * 8)) = value; + ds_set(tracer->ds.context->ds, ds_bts, ds_index, + (unsigned long)tracer->trace.ds.top); - error = 0; - out: - ds_put_context(context); - return error; + return 0; +} + +int ds_set_pebs_reset(struct pebs_tracer *tracer, u64 value) +{ + if (!tracer) + return -EINVAL; + + *(u64 *)(tracer->ds.context->ds + (ds_cfg.sizeof_field * 8)) = value; + + return 0; } -static const struct ds_configuration ds_cfg_var = { - .sizeof_ds = sizeof(long) * 12, - .sizeof_field = sizeof(long), - .sizeof_rec[ds_bts] = sizeof(long) * 3, +static const struct ds_configuration ds_cfg_netburst = { + .name = "netburst", + .ctl[dsf_bts] = (1 << 2) | (1 << 3), + .ctl[dsf_bts_kernel] = (1 << 5), + .ctl[dsf_bts_user] = (1 << 6), + + .sizeof_field = sizeof(long), + .sizeof_rec[ds_bts] = sizeof(long) * 3, #ifdef __i386__ - .sizeof_rec[ds_pebs] = sizeof(long) * 10 + .sizeof_rec[ds_pebs] = sizeof(long) * 10, #else - .sizeof_rec[ds_pebs] = sizeof(long) * 18 + .sizeof_rec[ds_pebs] = sizeof(long) * 18, #endif }; -static const struct ds_configuration ds_cfg_64 = { - .sizeof_ds = 8 * 12, - .sizeof_field = 8, - .sizeof_rec[ds_bts] = 8 * 3, +static const struct ds_configuration ds_cfg_pentium_m = { + .name = "pentium m", + .ctl[dsf_bts] = (1 << 6) | (1 << 7), + + .sizeof_field = sizeof(long), + .sizeof_rec[ds_bts] = sizeof(long) * 3, #ifdef __i386__ - .sizeof_rec[ds_pebs] = 8 * 10 + .sizeof_rec[ds_pebs] = sizeof(long) * 10, #else - .sizeof_rec[ds_pebs] = 8 * 18 + .sizeof_rec[ds_pebs] = sizeof(long) * 18, #endif }; +static const struct ds_configuration ds_cfg_core2 = { + .name = "core 2", + .ctl[dsf_bts] = (1 << 6) | (1 << 7), + .ctl[dsf_bts_kernel] = (1 << 9), + .ctl[dsf_bts_user] = (1 << 10), + + .sizeof_field = 8, + .sizeof_rec[ds_bts] = 8 * 3, + .sizeof_rec[ds_pebs] = 8 * 18, +}; -static inline void +static void ds_configure(const struct ds_configuration *cfg) { + memset(&ds_cfg, 0, sizeof(ds_cfg)); ds_cfg = *cfg; + + printk(KERN_INFO "[ds] using %s configuration\n", ds_cfg.name); + + if (!cpu_has_bts) { + ds_cfg.ctl[dsf_bts] = 0; + printk(KERN_INFO "[ds] bts not available\n"); + } + if (!cpu_has_pebs) + printk(KERN_INFO "[ds] pebs not available\n"); + + WARN_ON_ONCE(MAX_SIZEOF_DS < (12 * ds_cfg.sizeof_field)); } void __cpuinit ds_init_intel(struct cpuinfo_x86 *c) @@ -847,16 +949,15 @@ void __cpuinit ds_init_intel(struct cpuinfo_x86 *c) switch (c->x86) { case 0x6: switch (c->x86_model) { + case 0 ... 0xC: + /* sorry, don't know about them */ + break; case 0xD: case 0xE: /* Pentium M */ - ds_configure(&ds_cfg_var); + ds_configure(&ds_cfg_pentium_m); break; - case 0xF: /* Core2 */ - case 0x1C: /* Atom */ - ds_configure(&ds_cfg_64); - break; - default: - /* sorry, don't know about them */ + default: /* Core2, Atom, ... */ + ds_configure(&ds_cfg_core2); break; } break; @@ -865,7 +966,7 @@ void __cpuinit ds_init_intel(struct cpuinfo_x86 *c) case 0x0: case 0x1: case 0x2: /* Netburst */ - ds_configure(&ds_cfg_var); + ds_configure(&ds_cfg_netburst); break; default: /* sorry, don't know about them */ @@ -878,12 +979,52 @@ void __cpuinit ds_init_intel(struct cpuinfo_x86 *c) } } -void ds_free(struct ds_context *context) +/* + * Change the DS configuration from tracing prev to tracing next. + */ +void ds_switch_to(struct task_struct *prev, struct task_struct *next) +{ + struct ds_context *prev_ctx = prev->thread.ds_ctx; + struct ds_context *next_ctx = next->thread.ds_ctx; + + if (prev_ctx) { + update_debugctlmsr(0); + + if (prev_ctx->bts_master && + (prev_ctx->bts_master->trace.ds.flags & BTS_TIMESTAMPS)) { + struct bts_struct ts = { + .qualifier = bts_task_departs, + .variant.timestamp.jiffies = jiffies_64, + .variant.timestamp.pid = prev->pid + }; + bts_write(prev_ctx->bts_master, &ts); + } + } + + if (next_ctx) { + if (next_ctx->bts_master && + (next_ctx->bts_master->trace.ds.flags & BTS_TIMESTAMPS)) { + struct bts_struct ts = { + .qualifier = bts_task_arrives, + .variant.timestamp.jiffies = jiffies_64, + .variant.timestamp.pid = next->pid + }; + bts_write(next_ctx->bts_master, &ts); + } + + wrmsrl(MSR_IA32_DS_AREA, (unsigned long)next_ctx->ds); + } + + update_debugctlmsr(next->thread.debugctlmsr); +} + +void ds_copy_thread(struct task_struct *tsk, struct task_struct *father) +{ + clear_tsk_thread_flag(tsk, TIF_DS_AREA_MSR); + tsk->thread.ds_ctx = NULL; +} + +void ds_exit_thread(struct task_struct *tsk) { - /* This is called when the task owning the parameter context - * is dying. There should not be any user of that context left - * to disturb us, anymore. */ - unsigned long leftovers = context->count; - while (leftovers--) - ds_put_context(context); + WARN_ON(tsk->thread.ds_ctx); } diff --git a/arch/x86/kernel/dumpstack.c b/arch/x86/kernel/dumpstack.c new file mode 100644 index 0000000..6b1f6f6 --- /dev/null +++ b/arch/x86/kernel/dumpstack.c @@ -0,0 +1,351 @@ +/* + * Copyright (C) 1991, 1992 Linus Torvalds + * Copyright (C) 2000, 2001, 2002 Andi Kleen, SuSE Labs + */ +#include <linux/kallsyms.h> +#include <linux/kprobes.h> +#include <linux/uaccess.h> +#include <linux/utsname.h> +#include <linux/hardirq.h> +#include <linux/kdebug.h> +#include <linux/module.h> +#include <linux/ptrace.h> +#include <linux/kexec.h> +#include <linux/bug.h> +#include <linux/nmi.h> +#include <linux/sysfs.h> + +#include <asm/stacktrace.h> + +#include "dumpstack.h" + +int panic_on_unrecovered_nmi; +unsigned int code_bytes = 64; +int kstack_depth_to_print = 3 * STACKSLOTS_PER_LINE; +static int die_counter; + +void printk_address(unsigned long address, int reliable) +{ + printk(" [<%p>] %s%pS\n", (void *) address, + reliable ? "" : "? ", (void *) address); +} + +#ifdef CONFIG_FUNCTION_GRAPH_TRACER +static void +print_ftrace_graph_addr(unsigned long addr, void *data, + const struct stacktrace_ops *ops, + struct thread_info *tinfo, int *graph) +{ + struct task_struct *task = tinfo->task; + unsigned long ret_addr; + int index = task->curr_ret_stack; + + if (addr != (unsigned long)return_to_handler) + return; + + if (!task->ret_stack || index < *graph) + return; + + index -= *graph; + ret_addr = task->ret_stack[index].ret; + + ops->address(data, ret_addr, 1); + + (*graph)++; +} +#else +static inline void +print_ftrace_graph_addr(unsigned long addr, void *data, + const struct stacktrace_ops *ops, + struct thread_info *tinfo, int *graph) +{ } +#endif + +/* + * x86-64 can have up to three kernel stacks: + * process stack + * interrupt stack + * severe exception (double fault, nmi, stack fault, debug, mce) hardware stack + */ + +static inline int valid_stack_ptr(struct thread_info *tinfo, + void *p, unsigned int size, void *end) +{ + void *t = tinfo; + if (end) { + if (p < end && p >= (end-THREAD_SIZE)) + return 1; + else + return 0; + } + return p > t && p < t + THREAD_SIZE - size; +} + +unsigned long +print_context_stack(struct thread_info *tinfo, + unsigned long *stack, unsigned long bp, + const struct stacktrace_ops *ops, void *data, + unsigned long *end, int *graph) +{ + struct stack_frame *frame = (struct stack_frame *)bp; + + while (valid_stack_ptr(tinfo, stack, sizeof(*stack), end)) { + unsigned long addr; + + addr = *stack; + if (__kernel_text_address(addr)) { + if ((unsigned long) stack == bp + sizeof(long)) { + ops->address(data, addr, 1); + frame = frame->next_frame; + bp = (unsigned long) frame; + } else { + ops->address(data, addr, bp == 0); + } + print_ftrace_graph_addr(addr, data, ops, tinfo, graph); + } + stack++; + } + return bp; +} + + +static void +print_trace_warning_symbol(void *data, char *msg, unsigned long symbol) +{ + printk(data); + print_symbol(msg, symbol); + printk("\n"); +} + +static void print_trace_warning(void *data, char *msg) +{ + printk("%s%s\n", (char *)data, msg); +} + +static int print_trace_stack(void *data, char *name) +{ + printk("%s <%s> ", (char *)data, name); + return 0; +} + +/* + * Print one address/symbol entries per line. + */ +static void print_trace_address(void *data, unsigned long addr, int reliable) +{ + touch_nmi_watchdog(); + printk(data); + printk_address(addr, reliable); +} + +static const struct stacktrace_ops print_trace_ops = { + .warning = print_trace_warning, + .warning_symbol = print_trace_warning_symbol, + .stack = print_trace_stack, + .address = print_trace_address, +}; + +void +show_trace_log_lvl(struct task_struct *task, struct pt_regs *regs, + unsigned long *stack, unsigned long bp, char *log_lvl) +{ + printk("%sCall Trace:\n", log_lvl); + dump_trace(task, regs, stack, bp, &print_trace_ops, log_lvl); +} + +void show_trace(struct task_struct *task, struct pt_regs *regs, + unsigned long *stack, unsigned long bp) +{ + show_trace_log_lvl(task, regs, stack, bp, ""); +} + +void show_stack(struct task_struct *task, unsigned long *sp) +{ + show_stack_log_lvl(task, NULL, sp, 0, ""); +} + +/* + * The architecture-independent dump_stack generator + */ +void dump_stack(void) +{ + unsigned long bp = 0; + unsigned long stack; + +#ifdef CONFIG_FRAME_POINTER + if (!bp) + get_bp(bp); +#endif + + printk("Pid: %d, comm: %.20s %s %s %.*s\n", + current->pid, current->comm, print_tainted(), + init_utsname()->release, + (int)strcspn(init_utsname()->version, " "), + init_utsname()->version); + show_trace(NULL, NULL, &stack, bp); +} +EXPORT_SYMBOL(dump_stack); + +static raw_spinlock_t die_lock = __RAW_SPIN_LOCK_UNLOCKED; +static int die_owner = -1; +static unsigned int die_nest_count; + +unsigned __kprobes long oops_begin(void) +{ + int cpu; + unsigned long flags; + + oops_enter(); + + /* racy, but better than risking deadlock. */ + raw_local_irq_save(flags); + cpu = smp_processor_id(); + if (!__raw_spin_trylock(&die_lock)) { + if (cpu == die_owner) + /* nested oops. should stop eventually */; + else + __raw_spin_lock(&die_lock); + } + die_nest_count++; + die_owner = cpu; + console_verbose(); + bust_spinlocks(1); + return flags; +} + +void __kprobes oops_end(unsigned long flags, struct pt_regs *regs, int signr) +{ + if (regs && kexec_should_crash(current)) + crash_kexec(regs); + + bust_spinlocks(0); + die_owner = -1; + add_taint(TAINT_DIE); + die_nest_count--; + if (!die_nest_count) + /* Nest count reaches zero, release the lock. */ + __raw_spin_unlock(&die_lock); + raw_local_irq_restore(flags); + oops_exit(); + + if (!signr) + return; + if (in_interrupt()) + panic("Fatal exception in interrupt"); + if (panic_on_oops) + panic("Fatal exception"); + do_exit(signr); +} + +int __kprobes __die(const char *str, struct pt_regs *regs, long err) +{ +#ifdef CONFIG_X86_32 + unsigned short ss; + unsigned long sp; +#endif + printk(KERN_EMERG "%s: %04lx [#%d] ", str, err & 0xffff, ++die_counter); +#ifdef CONFIG_PREEMPT + printk("PREEMPT "); +#endif +#ifdef CONFIG_SMP + printk("SMP "); +#endif +#ifdef CONFIG_DEBUG_PAGEALLOC + printk("DEBUG_PAGEALLOC"); +#endif + printk("\n"); + sysfs_printk_last_file(); + if (notify_die(DIE_OOPS, str, regs, err, + current->thread.trap_no, SIGSEGV) == NOTIFY_STOP) + return 1; + + show_registers(regs); +#ifdef CONFIG_X86_32 + sp = (unsigned long) (®s->sp); + savesegment(ss, ss); + if (user_mode(regs)) { + sp = regs->sp; + ss = regs->ss & 0xffff; + } + printk(KERN_EMERG "EIP: [<%08lx>] ", regs->ip); + print_symbol("%s", regs->ip); + printk(" SS:ESP %04x:%08lx\n", ss, sp); +#else + /* Executive summary in case the oops scrolled away */ + printk(KERN_ALERT "RIP "); + printk_address(regs->ip, 1); + printk(" RSP <%016lx>\n", regs->sp); +#endif + return 0; +} + +/* + * This is gone through when something in the kernel has done something bad + * and is about to be terminated: + */ +void die(const char *str, struct pt_regs *regs, long err) +{ + unsigned long flags = oops_begin(); + int sig = SIGSEGV; + + if (!user_mode_vm(regs)) + report_bug(regs->ip, regs); + + if (__die(str, regs, err)) + sig = 0; + oops_end(flags, regs, sig); +} + +void notrace __kprobes +die_nmi(char *str, struct pt_regs *regs, int do_panic) +{ + unsigned long flags; + + if (notify_die(DIE_NMIWATCHDOG, str, regs, 0, 2, SIGINT) == NOTIFY_STOP) + return; + + /* + * We are in trouble anyway, lets at least try + * to get a message out. + */ + flags = oops_begin(); + printk(KERN_EMERG "%s", str); + printk(" on CPU%d, ip %08lx, registers:\n", + smp_processor_id(), regs->ip); + show_registers(regs); + oops_end(flags, regs, 0); + if (do_panic || panic_on_oops) + panic("Non maskable interrupt"); + nmi_exit(); + local_irq_enable(); + do_exit(SIGBUS); +} + +static int __init oops_setup(char *s) +{ + if (!s) + return -EINVAL; + if (!strcmp(s, "panic")) + panic_on_oops = 1; + return 0; +} +early_param("oops", oops_setup); + +static int __init kstack_setup(char *s) +{ + if (!s) + return -EINVAL; + kstack_depth_to_print = simple_strtoul(s, NULL, 0); + return 0; +} +early_param("kstack", kstack_setup); + +static int __init code_bytes_setup(char *s) +{ + code_bytes = simple_strtoul(s, NULL, 0); + if (code_bytes > 8192) + code_bytes = 8192; + + return 1; +} +__setup("code_bytes=", code_bytes_setup); diff --git a/arch/x86/kernel/dumpstack.h b/arch/x86/kernel/dumpstack.h new file mode 100644 index 0000000..da87590b --- /dev/null +++ b/arch/x86/kernel/dumpstack.h @@ -0,0 +1,39 @@ +/* + * Copyright (C) 1991, 1992 Linus Torvalds + * Copyright (C) 2000, 2001, 2002 Andi Kleen, SuSE Labs + */ + +#ifndef DUMPSTACK_H +#define DUMPSTACK_H + +#ifdef CONFIG_X86_32 +#define STACKSLOTS_PER_LINE 8 +#define get_bp(bp) asm("movl %%ebp, %0" : "=r" (bp) :) +#else +#define STACKSLOTS_PER_LINE 4 +#define get_bp(bp) asm("movq %%rbp, %0" : "=r" (bp) :) +#endif + +extern unsigned long +print_context_stack(struct thread_info *tinfo, + unsigned long *stack, unsigned long bp, + const struct stacktrace_ops *ops, void *data, + unsigned long *end, int *graph); + +extern void +show_trace_log_lvl(struct task_struct *task, struct pt_regs *regs, + unsigned long *stack, unsigned long bp, char *log_lvl); + +extern void +show_stack_log_lvl(struct task_struct *task, struct pt_regs *regs, + unsigned long *sp, unsigned long bp, char *log_lvl); + +extern unsigned int code_bytes; +extern int kstack_depth_to_print; + +/* The form of the top of the frame on the stack */ +struct stack_frame { + struct stack_frame *next_frame; + unsigned long return_address; +}; +#endif diff --git a/arch/x86/kernel/dumpstack_32.c b/arch/x86/kernel/dumpstack_32.c index b361475..d593cd1 100644 --- a/arch/x86/kernel/dumpstack_32.c +++ b/arch/x86/kernel/dumpstack_32.c @@ -17,69 +17,14 @@ #include <asm/stacktrace.h> -#define STACKSLOTS_PER_LINE 8 -#define get_bp(bp) asm("movl %%ebp, %0" : "=r" (bp) :) - -int panic_on_unrecovered_nmi; -int kstack_depth_to_print = 3 * STACKSLOTS_PER_LINE; -static unsigned int code_bytes = 64; -static int die_counter; - -void printk_address(unsigned long address, int reliable) -{ - printk(" [<%p>] %s%pS\n", (void *) address, - reliable ? "" : "? ", (void *) address); -} - -static inline int valid_stack_ptr(struct thread_info *tinfo, - void *p, unsigned int size, void *end) -{ - void *t = tinfo; - if (end) { - if (p < end && p >= (end-THREAD_SIZE)) - return 1; - else - return 0; - } - return p > t && p < t + THREAD_SIZE - size; -} - -/* The form of the top of the frame on the stack */ -struct stack_frame { - struct stack_frame *next_frame; - unsigned long return_address; -}; - -static inline unsigned long -print_context_stack(struct thread_info *tinfo, - unsigned long *stack, unsigned long bp, - const struct stacktrace_ops *ops, void *data, - unsigned long *end) -{ - struct stack_frame *frame = (struct stack_frame *)bp; - - while (valid_stack_ptr(tinfo, stack, sizeof(*stack), end)) { - unsigned long addr; - - addr = *stack; - if (__kernel_text_address(addr)) { - if ((unsigned long) stack == bp + sizeof(long)) { - ops->address(data, addr, 1); - frame = frame->next_frame; - bp = (unsigned long) frame; - } else { - ops->address(data, addr, bp == 0); - } - } - stack++; - } - return bp; -} +#include "dumpstack.h" void dump_trace(struct task_struct *task, struct pt_regs *regs, unsigned long *stack, unsigned long bp, const struct stacktrace_ops *ops, void *data) { + int graph = 0; + if (!task) task = current; @@ -107,7 +52,8 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs, context = (struct thread_info *) ((unsigned long)stack & (~(THREAD_SIZE - 1))); - bp = print_context_stack(context, stack, bp, ops, data, NULL); + bp = print_context_stack(context, stack, bp, ops, + data, NULL, &graph); stack = (unsigned long *)context->previous_esp; if (!stack) @@ -119,57 +65,7 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs, } EXPORT_SYMBOL(dump_trace); -static void -print_trace_warning_symbol(void *data, char *msg, unsigned long symbol) -{ - printk(data); - print_symbol(msg, symbol); - printk("\n"); -} - -static void print_trace_warning(void *data, char *msg) -{ - printk("%s%s\n", (char *)data, msg); -} - -static int print_trace_stack(void *data, char *name) -{ - printk("%s <%s> ", (char *)data, name); - return 0; -} - -/* - * Print one address/symbol entries per line. - */ -static void print_trace_address(void *data, unsigned long addr, int reliable) -{ - touch_nmi_watchdog(); - printk(data); - printk_address(addr, reliable); -} - -static const struct stacktrace_ops print_trace_ops = { - .warning = print_trace_warning, - .warning_symbol = print_trace_warning_symbol, - .stack = print_trace_stack, - .address = print_trace_address, -}; - -static void -show_trace_log_lvl(struct task_struct *task, struct pt_regs *regs, - unsigned long *stack, unsigned long bp, char *log_lvl) -{ - printk("%sCall Trace:\n", log_lvl); - dump_trace(task, regs, stack, bp, &print_trace_ops, log_lvl); -} - -void show_trace(struct task_struct *task, struct pt_regs *regs, - unsigned long *stack, unsigned long bp) -{ - show_trace_log_lvl(task, regs, stack, bp, ""); -} - -static void +void show_stack_log_lvl(struct task_struct *task, struct pt_regs *regs, unsigned long *sp, unsigned long bp, char *log_lvl) { @@ -196,33 +92,6 @@ show_stack_log_lvl(struct task_struct *task, struct pt_regs *regs, show_trace_log_lvl(task, regs, sp, bp, log_lvl); } -void show_stack(struct task_struct *task, unsigned long *sp) -{ - show_stack_log_lvl(task, NULL, sp, 0, ""); -} - -/* - * The architecture-independent dump_stack generator - */ -void dump_stack(void) -{ - unsigned long bp = 0; - unsigned long stack; - -#ifdef CONFIG_FRAME_POINTER - if (!bp) - get_bp(bp); -#endif - - printk("Pid: %d, comm: %.20s %s %s %.*s\n", - current->pid, current->comm, print_tainted(), - init_utsname()->release, - (int)strcspn(init_utsname()->version, " "), - init_utsname()->version); - show_trace(NULL, NULL, &stack, bp); -} - -EXPORT_SYMBOL(dump_stack); void show_registers(struct pt_regs *regs) { @@ -283,167 +152,3 @@ int is_valid_bugaddr(unsigned long ip) return ud2 == 0x0b0f; } -static raw_spinlock_t die_lock = __RAW_SPIN_LOCK_UNLOCKED; -static int die_owner = -1; -static unsigned int die_nest_count; - -unsigned __kprobes long oops_begin(void) -{ - unsigned long flags; - - oops_enter(); - - if (die_owner != raw_smp_processor_id()) { - console_verbose(); - raw_local_irq_save(flags); - __raw_spin_lock(&die_lock); - die_owner = smp_processor_id(); - die_nest_count = 0; - bust_spinlocks(1); - } else { - raw_local_irq_save(flags); - } - die_nest_count++; - return flags; -} - -void __kprobes oops_end(unsigned long flags, struct pt_regs *regs, int signr) -{ - bust_spinlocks(0); - die_owner = -1; - add_taint(TAINT_DIE); - __raw_spin_unlock(&die_lock); - raw_local_irq_restore(flags); - - if (!regs) - return; - - if (kexec_should_crash(current)) - crash_kexec(regs); - if (in_interrupt()) - panic("Fatal exception in interrupt"); - if (panic_on_oops) - panic("Fatal exception"); - oops_exit(); - do_exit(signr); -} - -int __kprobes __die(const char *str, struct pt_regs *regs, long err) -{ - unsigned short ss; - unsigned long sp; - - printk(KERN_EMERG "%s: %04lx [#%d] ", str, err & 0xffff, ++die_counter); -#ifdef CONFIG_PREEMPT - printk("PREEMPT "); -#endif -#ifdef CONFIG_SMP - printk("SMP "); -#endif -#ifdef CONFIG_DEBUG_PAGEALLOC - printk("DEBUG_PAGEALLOC"); -#endif - printk("\n"); - sysfs_printk_last_file(); - if (notify_die(DIE_OOPS, str, regs, err, - current->thread.trap_no, SIGSEGV) == NOTIFY_STOP) - return 1; - - show_registers(regs); - /* Executive summary in case the oops scrolled away */ - sp = (unsigned long) (®s->sp); - savesegment(ss, ss); - if (user_mode(regs)) { - sp = regs->sp; - ss = regs->ss & 0xffff; - } - printk(KERN_EMERG "EIP: [<%08lx>] ", regs->ip); - print_symbol("%s", regs->ip); - printk(" SS:ESP %04x:%08lx\n", ss, sp); - return 0; -} - -/* - * This is gone through when something in the kernel has done something bad - * and is about to be terminated: - */ -void die(const char *str, struct pt_regs *regs, long err) -{ - unsigned long flags = oops_begin(); - - if (die_nest_count < 3) { - report_bug(regs->ip, regs); - - if (__die(str, regs, err)) - regs = NULL; - } else { - printk(KERN_EMERG "Recursive die() failure, output suppressed\n"); - } - - oops_end(flags, regs, SIGSEGV); -} - -static DEFINE_SPINLOCK(nmi_print_lock); - -void notrace __kprobes -die_nmi(char *str, struct pt_regs *regs, int do_panic) -{ - if (notify_die(DIE_NMIWATCHDOG, str, regs, 0, 2, SIGINT) == NOTIFY_STOP) - return; - - spin_lock(&nmi_print_lock); - /* - * We are in trouble anyway, lets at least try - * to get a message out: - */ - bust_spinlocks(1); - printk(KERN_EMERG "%s", str); - printk(" on CPU%d, ip %08lx, registers:\n", - smp_processor_id(), regs->ip); - show_registers(regs); - if (do_panic) - panic("Non maskable interrupt"); - console_silent(); - spin_unlock(&nmi_print_lock); - - /* - * If we are in kernel we are probably nested up pretty bad - * and might aswell get out now while we still can: - */ - if (!user_mode_vm(regs)) { - current->thread.trap_no = 2; - crash_kexec(regs); - } - - bust_spinlocks(0); - do_exit(SIGSEGV); -} - -static int __init oops_setup(char *s) -{ - if (!s) - return -EINVAL; - if (!strcmp(s, "panic")) - panic_on_oops = 1; - return 0; -} -early_param("oops", oops_setup); - -static int __init kstack_setup(char *s) -{ - if (!s) - return -EINVAL; - kstack_depth_to_print = simple_strtoul(s, NULL, 0); - return 0; -} -early_param("kstack", kstack_setup); - -static int __init code_bytes_setup(char *s) -{ - code_bytes = simple_strtoul(s, NULL, 0); - if (code_bytes > 8192) - code_bytes = 8192; - - return 1; -} -__setup("code_bytes=", code_bytes_setup); diff --git a/arch/x86/kernel/dumpstack_64.c b/arch/x86/kernel/dumpstack_64.c index 96a5db7..c302d07 100644 --- a/arch/x86/kernel/dumpstack_64.c +++ b/arch/x86/kernel/dumpstack_64.c @@ -17,19 +17,7 @@ #include <asm/stacktrace.h> -#define STACKSLOTS_PER_LINE 4 -#define get_bp(bp) asm("movq %%rbp, %0" : "=r" (bp) :) - -int panic_on_unrecovered_nmi; -int kstack_depth_to_print = 3 * STACKSLOTS_PER_LINE; -static unsigned int code_bytes = 64; -static int die_counter; - -void printk_address(unsigned long address, int reliable) -{ - printk(" [<%p>] %s%pS\n", (void *) address, - reliable ? "" : "? ", (void *) address); -} +#include "dumpstack.h" static unsigned long *in_exception_stack(unsigned cpu, unsigned long stack, unsigned *usedp, char **idp) @@ -113,51 +101,6 @@ static unsigned long *in_exception_stack(unsigned cpu, unsigned long stack, * severe exception (double fault, nmi, stack fault, debug, mce) hardware stack */ -static inline int valid_stack_ptr(struct thread_info *tinfo, - void *p, unsigned int size, void *end) -{ - void *t = tinfo; - if (end) { - if (p < end && p >= (end-THREAD_SIZE)) - return 1; - else - return 0; - } - return p > t && p < t + THREAD_SIZE - size; -} - -/* The form of the top of the frame on the stack */ -struct stack_frame { - struct stack_frame *next_frame; - unsigned long return_address; -}; - -static inline unsigned long -print_context_stack(struct thread_info *tinfo, - unsigned long *stack, unsigned long bp, - const struct stacktrace_ops *ops, void *data, - unsigned long *end) -{ - struct stack_frame *frame = (struct stack_frame *)bp; - - while (valid_stack_ptr(tinfo, stack, sizeof(*stack), end)) { - unsigned long addr; - - addr = *stack; - if (__kernel_text_address(addr)) { - if ((unsigned long) stack == bp + sizeof(long)) { - ops->address(data, addr, 1); - frame = frame->next_frame; - bp = (unsigned long) frame; - } else { - ops->address(data, addr, bp == 0); - } - } - stack++; - } - return bp; -} - void dump_trace(struct task_struct *task, struct pt_regs *regs, unsigned long *stack, unsigned long bp, const struct stacktrace_ops *ops, void *data) @@ -166,6 +109,7 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs, unsigned long *irqstack_end = (unsigned long *)cpu_pda(cpu)->irqstackptr; unsigned used = 0; struct thread_info *tinfo; + int graph = 0; if (!task) task = current; @@ -206,7 +150,7 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs, break; bp = print_context_stack(tinfo, stack, bp, ops, - data, estack_end); + data, estack_end, &graph); ops->stack(data, "<EOE>"); /* * We link to the next stack via the @@ -225,7 +169,7 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs, if (ops->stack(data, "IRQ") < 0) break; bp = print_context_stack(tinfo, stack, bp, - ops, data, irqstack_end); + ops, data, irqstack_end, &graph); /* * We link to the next stack (which would be * the process stack normally) the last @@ -243,62 +187,12 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs, /* * This handles the process stack: */ - bp = print_context_stack(tinfo, stack, bp, ops, data, NULL); + bp = print_context_stack(tinfo, stack, bp, ops, data, NULL, &graph); put_cpu(); } EXPORT_SYMBOL(dump_trace); -static void -print_trace_warning_symbol(void *data, char *msg, unsigned long symbol) -{ - printk(data); - print_symbol(msg, symbol); - printk("\n"); -} - -static void print_trace_warning(void *data, char *msg) -{ - printk("%s%s\n", (char *)data, msg); -} - -static int print_trace_stack(void *data, char *name) -{ - printk("%s <%s> ", (char *)data, name); - return 0; -} - -/* - * Print one address/symbol entries per line. - */ -static void print_trace_address(void *data, unsigned long addr, int reliable) -{ - touch_nmi_watchdog(); - printk(data); - printk_address(addr, reliable); -} - -static const struct stacktrace_ops print_trace_ops = { - .warning = print_trace_warning, - .warning_symbol = print_trace_warning_symbol, - .stack = print_trace_stack, - .address = print_trace_address, -}; - -static void -show_trace_log_lvl(struct task_struct *task, struct pt_regs *regs, - unsigned long *stack, unsigned long bp, char *log_lvl) -{ - printk("%sCall Trace:\n", log_lvl); - dump_trace(task, regs, stack, bp, &print_trace_ops, log_lvl); -} - -void show_trace(struct task_struct *task, struct pt_regs *regs, - unsigned long *stack, unsigned long bp) -{ - show_trace_log_lvl(task, regs, stack, bp, ""); -} - -static void +void show_stack_log_lvl(struct task_struct *task, struct pt_regs *regs, unsigned long *sp, unsigned long bp, char *log_lvl) { @@ -342,33 +236,6 @@ show_stack_log_lvl(struct task_struct *task, struct pt_regs *regs, show_trace_log_lvl(task, regs, sp, bp, log_lvl); } -void show_stack(struct task_struct *task, unsigned long *sp) -{ - show_stack_log_lvl(task, NULL, sp, 0, ""); -} - -/* - * The architecture-independent dump_stack generator - */ -void dump_stack(void) -{ - unsigned long bp = 0; - unsigned long stack; - -#ifdef CONFIG_FRAME_POINTER - if (!bp) - get_bp(bp); -#endif - - printk("Pid: %d, comm: %.20s %s %s %.*s\n", - current->pid, current->comm, print_tainted(), - init_utsname()->release, - (int)strcspn(init_utsname()->version, " "), - init_utsname()->version); - show_trace(NULL, NULL, &stack, bp); -} -EXPORT_SYMBOL(dump_stack); - void show_registers(struct pt_regs *regs) { int i; @@ -429,147 +296,3 @@ int is_valid_bugaddr(unsigned long ip) return ud2 == 0x0b0f; } -static raw_spinlock_t die_lock = __RAW_SPIN_LOCK_UNLOCKED; -static int die_owner = -1; -static unsigned int die_nest_count; - -unsigned __kprobes long oops_begin(void) -{ - int cpu; - unsigned long flags; - - oops_enter(); - - /* racy, but better than risking deadlock. */ - raw_local_irq_save(flags); - cpu = smp_processor_id(); - if (!__raw_spin_trylock(&die_lock)) { - if (cpu == die_owner) - /* nested oops. should stop eventually */; - else - __raw_spin_lock(&die_lock); - } - die_nest_count++; - die_owner = cpu; - console_verbose(); - bust_spinlocks(1); - return flags; -} - -void __kprobes oops_end(unsigned long flags, struct pt_regs *regs, int signr) -{ - die_owner = -1; - bust_spinlocks(0); - die_nest_count--; - if (!die_nest_count) - /* Nest count reaches zero, release the lock. */ - __raw_spin_unlock(&die_lock); - raw_local_irq_restore(flags); - if (!regs) { - oops_exit(); - return; - } - if (in_interrupt()) - panic("Fatal exception in interrupt"); - if (panic_on_oops) - panic("Fatal exception"); - oops_exit(); - do_exit(signr); -} - -int __kprobes __die(const char *str, struct pt_regs *regs, long err) -{ - printk(KERN_EMERG "%s: %04lx [#%d] ", str, err & 0xffff, ++die_counter); -#ifdef CONFIG_PREEMPT - printk("PREEMPT "); -#endif -#ifdef CONFIG_SMP - printk("SMP "); -#endif -#ifdef CONFIG_DEBUG_PAGEALLOC - printk("DEBUG_PAGEALLOC"); -#endif - printk("\n"); - sysfs_printk_last_file(); - if (notify_die(DIE_OOPS, str, regs, err, - current->thread.trap_no, SIGSEGV) == NOTIFY_STOP) - return 1; - - show_registers(regs); - add_taint(TAINT_DIE); - /* Executive summary in case the oops scrolled away */ - printk(KERN_ALERT "RIP "); - printk_address(regs->ip, 1); - printk(" RSP <%016lx>\n", regs->sp); - if (kexec_should_crash(current)) - crash_kexec(regs); - return 0; -} - -void die(const char *str, struct pt_regs *regs, long err) -{ - unsigned long flags = oops_begin(); - - if (!user_mode(regs)) - report_bug(regs->ip, regs); - - if (__die(str, regs, err)) - regs = NULL; - oops_end(flags, regs, SIGSEGV); -} - -notrace __kprobes void -die_nmi(char *str, struct pt_regs *regs, int do_panic) -{ - unsigned long flags; - - if (notify_die(DIE_NMIWATCHDOG, str, regs, 0, 2, SIGINT) == NOTIFY_STOP) - return; - - flags = oops_begin(); - /* - * We are in trouble anyway, lets at least try - * to get a message out. - */ - printk(KERN_EMERG "%s", str); - printk(" on CPU%d, ip %08lx, registers:\n", - smp_processor_id(), regs->ip); - show_registers(regs); - if (kexec_should_crash(current)) - crash_kexec(regs); - if (do_panic || panic_on_oops) - panic("Non maskable interrupt"); - oops_end(flags, NULL, SIGBUS); - nmi_exit(); - local_irq_enable(); - do_exit(SIGBUS); -} - -static int __init oops_setup(char *s) -{ - if (!s) - return -EINVAL; - if (!strcmp(s, "panic")) - panic_on_oops = 1; - return 0; -} -early_param("oops", oops_setup); - -static int __init kstack_setup(char *s) -{ - if (!s) - return -EINVAL; - kstack_depth_to_print = simple_strtoul(s, NULL, 0); - return 0; -} -early_param("kstack", kstack_setup); - -static int __init code_bytes_setup(char *s) -{ - code_bytes = simple_strtoul(s, NULL, 0); - if (code_bytes > 8192) - code_bytes = 8192; - - return 1; -} -__setup("code_bytes=", code_bytes_setup); diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index 28b597e..43ceb3f 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -1157,6 +1157,9 @@ ENTRY(mcount) END(mcount) ENTRY(ftrace_caller) + cmpl $0, function_trace_stop + jne ftrace_stub + pushl %eax pushl %ecx pushl %edx @@ -1171,6 +1174,11 @@ ftrace_call: popl %edx popl %ecx popl %eax +#ifdef CONFIG_FUNCTION_GRAPH_TRACER +.globl ftrace_graph_call +ftrace_graph_call: + jmp ftrace_stub +#endif .globl ftrace_stub ftrace_stub: @@ -1180,8 +1188,18 @@ END(ftrace_caller) #else /* ! CONFIG_DYNAMIC_FTRACE */ ENTRY(mcount) + cmpl $0, function_trace_stop + jne ftrace_stub + cmpl $ftrace_stub, ftrace_trace_function jnz trace +#ifdef CONFIG_FUNCTION_GRAPH_TRACER + cmpl $ftrace_stub, ftrace_graph_return + jnz ftrace_graph_caller + + cmpl $ftrace_graph_entry_stub, ftrace_graph_entry + jnz ftrace_graph_caller +#endif .globl ftrace_stub ftrace_stub: ret @@ -1200,12 +1218,43 @@ trace: popl %edx popl %ecx popl %eax - jmp ftrace_stub END(mcount) #endif /* CONFIG_DYNAMIC_FTRACE */ #endif /* CONFIG_FUNCTION_TRACER */ +#ifdef CONFIG_FUNCTION_GRAPH_TRACER +ENTRY(ftrace_graph_caller) + cmpl $0, function_trace_stop + jne ftrace_stub + + pushl %eax + pushl %ecx + pushl %edx + movl 0xc(%esp), %edx + lea 0x4(%ebp), %eax + subl $MCOUNT_INSN_SIZE, %edx + call prepare_ftrace_return + popl %edx + popl %ecx + popl %eax + ret +END(ftrace_graph_caller) + +.globl return_to_handler +return_to_handler: + pushl $0 + pushl %eax + pushl %ecx + pushl %edx + call ftrace_return_to_handler + movl %eax, 0xc(%esp) + popl %edx + popl %ecx + popl %eax + ret +#endif + .section .rodata,"a" #include "syscall_table_32.S" diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index b86f332..303dd84 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -68,16 +68,10 @@ ENTRY(mcount) END(mcount) ENTRY(ftrace_caller) + cmpl $0, function_trace_stop + jne ftrace_stub - /* taken from glibc */ - 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) + MCOUNT_SAVE_FRAME movq 0x38(%rsp), %rdi movq 8(%rbp), %rsi @@ -87,14 +81,13 @@ ENTRY(ftrace_caller) ftrace_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 + MCOUNT_RESTORE_FRAME + +#ifdef CONFIG_FUNCTION_GRAPH_TRACER +.globl ftrace_graph_call +ftrace_graph_call: + jmp ftrace_stub +#endif .globl ftrace_stub ftrace_stub: @@ -103,15 +96,63 @@ END(ftrace_caller) #else /* ! CONFIG_DYNAMIC_FTRACE */ ENTRY(mcount) + cmpl $0, function_trace_stop + jne ftrace_stub + cmpq $ftrace_stub, ftrace_trace_function jnz trace + +#ifdef CONFIG_FUNCTION_GRAPH_TRACER + cmpq $ftrace_stub, ftrace_graph_return + jnz ftrace_graph_caller + + cmpq $ftrace_graph_entry_stub, ftrace_graph_entry + jnz ftrace_graph_caller +#endif + .globl ftrace_stub ftrace_stub: retq trace: - /* taken from glibc */ - subq $0x38, %rsp + MCOUNT_SAVE_FRAME + + movq 0x38(%rsp), %rdi + movq 8(%rbp), %rsi + subq $MCOUNT_INSN_SIZE, %rdi + + call *ftrace_trace_function + + MCOUNT_RESTORE_FRAME + + jmp ftrace_stub +END(mcount) +#endif /* CONFIG_DYNAMIC_FTRACE */ +#endif /* CONFIG_FUNCTION_TRACER */ + +#ifdef CONFIG_FUNCTION_GRAPH_TRACER +ENTRY(ftrace_graph_caller) + cmpl $0, function_trace_stop + jne ftrace_stub + + MCOUNT_SAVE_FRAME + + leaq 8(%rbp), %rdi + movq 0x38(%rsp), %rsi + subq $MCOUNT_INSN_SIZE, %rsi + + call prepare_ftrace_return + + MCOUNT_RESTORE_FRAME + + retq +END(ftrace_graph_caller) + + +.globl return_to_handler +return_to_handler: + subq $80, %rsp + movq %rax, (%rsp) movq %rcx, 8(%rsp) movq %rdx, 16(%rsp) @@ -119,13 +160,14 @@ trace: movq %rdi, 32(%rsp) movq %r8, 40(%rsp) movq %r9, 48(%rsp) + movq %r10, 56(%rsp) + movq %r11, 64(%rsp) - movq 0x38(%rsp), %rdi - movq 8(%rbp), %rsi - subq $MCOUNT_INSN_SIZE, %rdi - - call *ftrace_trace_function + call ftrace_return_to_handler + movq %rax, 72(%rsp) + movq 64(%rsp), %r11 + movq 56(%rsp), %r10 movq 48(%rsp), %r9 movq 40(%rsp), %r8 movq 32(%rsp), %rdi @@ -133,12 +175,10 @@ trace: movq 16(%rsp), %rdx movq 8(%rsp), %rcx movq (%rsp), %rax - addq $0x38, %rsp + addq $72, %rsp + retq +#endif - jmp ftrace_stub -END(mcount) -#endif /* CONFIG_DYNAMIC_FTRACE */ -#endif /* CONFIG_FUNCTION_TRACER */ #ifndef CONFIG_PREEMPT #define retint_kernel retint_restore_args diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 50ea0ac..1b43086 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -14,14 +14,17 @@ #include <linux/uaccess.h> #include <linux/ftrace.h> #include <linux/percpu.h> +#include <linux/sched.h> #include <linux/init.h> #include <linux/list.h> #include <asm/ftrace.h> +#include <linux/ftrace.h> #include <asm/nops.h> +#include <asm/nmi.h> -static unsigned char ftrace_nop[MCOUNT_INSN_SIZE]; +#ifdef CONFIG_DYNAMIC_FTRACE union ftrace_code_union { char code[MCOUNT_INSN_SIZE]; @@ -31,18 +34,12 @@ union ftrace_code_union { } __attribute__((packed)); }; - static int ftrace_calc_offset(long ip, long addr) { return (int)(addr - ip); } -unsigned char *ftrace_nop_replace(void) -{ - return ftrace_nop; -} - -unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) +static unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) { static union ftrace_code_union calc; @@ -56,7 +53,142 @@ unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) return calc.code; } -int +/* + * Modifying code must take extra care. On an SMP machine, if + * the code being modified is also being executed on another CPU + * that CPU will have undefined results and possibly take a GPF. + * We use kstop_machine to stop other CPUS from exectuing code. + * But this does not stop NMIs from happening. We still need + * to protect against that. We separate out the modification of + * the code to take care of this. + * + * Two buffers are added: An IP buffer and a "code" buffer. + * + * 1) Put the instruction pointer into the IP buffer + * and the new code into the "code" buffer. + * 2) Set a flag that says we are modifying code + * 3) Wait for any running NMIs to finish. + * 4) Write the code + * 5) clear the flag. + * 6) Wait for any running NMIs to finish. + * + * If an NMI is executed, the first thing it does is to call + * "ftrace_nmi_enter". This will check if the flag is set to write + * and if it is, it will write what is in the IP and "code" buffers. + * + * The trick is, it does not matter if everyone is writing the same + * content to the code location. Also, if a CPU is executing code + * it is OK to write to that code location if the contents being written + * are the same as what exists. + */ + +static atomic_t in_nmi = ATOMIC_INIT(0); +static int mod_code_status; /* holds return value of text write */ +static int mod_code_write; /* set when NMI should do the write */ +static void *mod_code_ip; /* holds the IP to write to */ +static void *mod_code_newcode; /* holds the text to write to the IP */ + +static unsigned nmi_wait_count; +static atomic_t nmi_update_count = ATOMIC_INIT(0); + +int ftrace_arch_read_dyn_info(char *buf, int size) +{ + int r; + + r = snprintf(buf, size, "%u %u", + nmi_wait_count, + atomic_read(&nmi_update_count)); + return r; +} + +static void ftrace_mod_code(void) +{ + /* + * Yes, more than one CPU process can be writing to mod_code_status. + * (and the code itself) + * But if one were to fail, then they all should, and if one were + * to succeed, then they all should. + */ + mod_code_status = probe_kernel_write(mod_code_ip, mod_code_newcode, + MCOUNT_INSN_SIZE); +} + +void ftrace_nmi_enter(void) +{ + atomic_inc(&in_nmi); + /* Must have in_nmi seen before reading write flag */ + smp_mb(); + if (mod_code_write) { + ftrace_mod_code(); + atomic_inc(&nmi_update_count); + } +} + +void ftrace_nmi_exit(void) +{ + /* Finish all executions before clearing in_nmi */ + smp_wmb(); + atomic_dec(&in_nmi); +} + +static void wait_for_nmi(void) +{ + int waited = 0; + + while (atomic_read(&in_nmi)) { + waited = 1; + cpu_relax(); + } + + if (waited) + nmi_wait_count++; +} + +static int +do_ftrace_mod_code(unsigned long ip, void *new_code) +{ + mod_code_ip = (void *)ip; + mod_code_newcode = new_code; + + /* The buffers need to be visible before we let NMIs write them */ + smp_wmb(); + + mod_code_write = 1; + + /* Make sure write bit is visible before we wait on NMIs */ + smp_mb(); + + wait_for_nmi(); + + /* Make sure all running NMIs have finished before we write the code */ + smp_mb(); + + ftrace_mod_code(); + + /* Make sure the write happens before clearing the bit */ + smp_wmb(); + + mod_code_write = 0; + + /* make sure NMIs see the cleared bit */ + smp_mb(); + + wait_for_nmi(); + + return mod_code_status; +} + + + + +static unsigned char ftrace_nop[MCOUNT_INSN_SIZE]; + +static unsigned char *ftrace_nop_replace(void) +{ + return ftrace_nop; +} + +static int ftrace_modify_code(unsigned long ip, unsigned char *old_code, unsigned char *new_code) { @@ -81,7 +213,7 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code, return -EINVAL; /* replace the text with the new text */ - if (probe_kernel_write((void *)ip, new_code, MCOUNT_INSN_SIZE)) + if (do_ftrace_mod_code(ip, new_code)) return -EPERM; sync_core(); @@ -89,6 +221,29 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code, return 0; } +int ftrace_make_nop(struct module *mod, + struct dyn_ftrace *rec, unsigned long addr) +{ + unsigned char *new, *old; + unsigned long ip = rec->ip; + + old = ftrace_call_replace(ip, addr); + new = ftrace_nop_replace(); + + return ftrace_modify_code(rec->ip, old, new); +} + +int ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr) +{ + unsigned char *new, *old; + unsigned long ip = rec->ip; + + old = ftrace_nop_replace(); + new = ftrace_call_replace(ip, addr); + + return ftrace_modify_code(rec->ip, old, new); +} + int ftrace_update_ftrace_func(ftrace_func_t func) { unsigned long ip = (unsigned long)(&ftrace_call); @@ -165,3 +320,218 @@ int __init ftrace_dyn_arch_init(void *data) return 0; } +#endif + +#ifdef CONFIG_FUNCTION_GRAPH_TRACER + +#ifdef CONFIG_DYNAMIC_FTRACE +extern void ftrace_graph_call(void); + +static int ftrace_mod_jmp(unsigned long ip, + int old_offset, int new_offset) +{ + unsigned char code[MCOUNT_INSN_SIZE]; + + if (probe_kernel_read(code, (void *)ip, MCOUNT_INSN_SIZE)) + return -EFAULT; + + if (code[0] != 0xe9 || old_offset != *(int *)(&code[1])) + return -EINVAL; + + *(int *)(&code[1]) = new_offset; + + if (do_ftrace_mod_code(ip, &code)) + return -EPERM; + + return 0; +} + +int ftrace_enable_ftrace_graph_caller(void) +{ + unsigned long ip = (unsigned long)(&ftrace_graph_call); + int old_offset, new_offset; + + old_offset = (unsigned long)(&ftrace_stub) - (ip + MCOUNT_INSN_SIZE); + new_offset = (unsigned long)(&ftrace_graph_caller) - (ip + MCOUNT_INSN_SIZE); + + return ftrace_mod_jmp(ip, old_offset, new_offset); +} + +int ftrace_disable_ftrace_graph_caller(void) +{ + unsigned long ip = (unsigned long)(&ftrace_graph_call); + int old_offset, new_offset; + + old_offset = (unsigned long)(&ftrace_graph_caller) - (ip + MCOUNT_INSN_SIZE); + new_offset = (unsigned long)(&ftrace_stub) - (ip + MCOUNT_INSN_SIZE); + + return ftrace_mod_jmp(ip, old_offset, new_offset); +} + +#else /* CONFIG_DYNAMIC_FTRACE */ + +/* + * These functions are picked from those used on + * this page for dynamic ftrace. They have been + * simplified to ignore all traces in NMI context. + */ +static atomic_t in_nmi; + +void ftrace_nmi_enter(void) +{ + atomic_inc(&in_nmi); +} + +void ftrace_nmi_exit(void) +{ + atomic_dec(&in_nmi); +} + +#endif /* !CONFIG_DYNAMIC_FTRACE */ + +/* Add a function return address to the trace stack on thread info.*/ +static int push_return_trace(unsigned long ret, unsigned long long time, + unsigned long func, int *depth) +{ + int index; + + if (!current->ret_stack) + return -EBUSY; + + /* The return trace stack is full */ + if (current->curr_ret_stack == FTRACE_RETFUNC_DEPTH - 1) { + atomic_inc(¤t->trace_overrun); + return -EBUSY; + } + + index = ++current->curr_ret_stack; + barrier(); + current->ret_stack[index].ret = ret; + current->ret_stack[index].func = func; + current->ret_stack[index].calltime = time; + *depth = index; + + return 0; +} + +/* Retrieve a function return address to the trace stack on thread info.*/ +static void pop_return_trace(struct ftrace_graph_ret *trace, unsigned long *ret) +{ + int index; + + index = current->curr_ret_stack; + + if (unlikely(index < 0)) { + ftrace_graph_stop(); + WARN_ON(1); + /* Might as well panic, otherwise we have no where to go */ + *ret = (unsigned long)panic; + return; + } + + *ret = current->ret_stack[index].ret; + trace->func = current->ret_stack[index].func; + trace->calltime = current->ret_stack[index].calltime; + trace->overrun = atomic_read(¤t->trace_overrun); + trace->depth = index; + barrier(); + current->curr_ret_stack--; + +} + +/* + * Send the trace to the ring-buffer. + * @return the original return address. + */ +unsigned long ftrace_return_to_handler(void) +{ + struct ftrace_graph_ret trace; + unsigned long ret; + + pop_return_trace(&trace, &ret); + trace.rettime = cpu_clock(raw_smp_processor_id()); + ftrace_graph_return(&trace); + + if (unlikely(!ret)) { + ftrace_graph_stop(); + WARN_ON(1); + /* Might as well panic. What else to do? */ + ret = (unsigned long)panic; + } + + return ret; +} + +/* + * Hook the return address and push it in the stack of return addrs + * in current thread info. + */ +void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) +{ + unsigned long old; + unsigned long long calltime; + int faulted; + struct ftrace_graph_ent trace; + unsigned long return_hooker = (unsigned long) + &return_to_handler; + + /* Nmi's are currently unsupported */ + if (unlikely(atomic_read(&in_nmi))) + return; + + if (unlikely(atomic_read(¤t->tracing_graph_pause))) + return; + + /* + * Protect against fault, even if it shouldn't + * happen. This tool is too much intrusive to + * ignore such a protection. + */ + asm volatile( + "1: " _ASM_MOV " (%[parent_old]), %[old]\n" + "2: " _ASM_MOV " %[return_hooker], (%[parent_replaced])\n" + " movl $0, %[faulted]\n" + + ".section .fixup, \"ax\"\n" + "3: movl $1, %[faulted]\n" + ".previous\n" + + _ASM_EXTABLE(1b, 3b) + _ASM_EXTABLE(2b, 3b) + + : [parent_replaced] "=r" (parent), [old] "=r" (old), + [faulted] "=r" (faulted) + : [parent_old] "0" (parent), [return_hooker] "r" (return_hooker) + : "memory" + ); + + if (unlikely(faulted)) { + ftrace_graph_stop(); + WARN_ON(1); + return; + } + + if (unlikely(!__kernel_text_address(old))) { + ftrace_graph_stop(); + *parent = old; + WARN_ON(1); + return; + } + + calltime = cpu_clock(raw_smp_processor_id()); + + if (push_return_trace(old, calltime, + self_addr, &trace.depth) == -EBUSY) { + *parent = old; + return; + } + + trace.func = self_addr; + + /* Only trace if the calling function expects to */ + if (!ftrace_graph_entry(&trace)) { + current->curr_ret_stack--; + *parent = old; + } +} +#endif /* CONFIG_FUNCTION_GRAPH_TRACER */ diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index 60eb84e..11c65e8 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -13,6 +13,7 @@ #include <linux/seq_file.h> #include <linux/module.h> #include <linux/delay.h> +#include <linux/ftrace.h> #include <asm/uaccess.h> #include <asm/io_apic.h> #include <asm/idle.h> @@ -47,7 +48,7 @@ static inline void stack_overflow_check(struct pt_regs *regs) * SMP cross-CPU interrupts have their own specific * handlers). */ -asmlinkage unsigned int do_IRQ(struct pt_regs *regs) +asmlinkage unsigned int __irq_entry do_IRQ(struct pt_regs *regs) { struct pt_regs *old_regs = set_irq_regs(regs); struct irq_desc *desc; diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index c622772..cff9a50 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -7,6 +7,7 @@ #include <linux/module.h> #include <linux/pm.h> #include <linux/clockchips.h> +#include <linux/ftrace.h> #include <asm/system.h> unsigned long idle_halt; @@ -100,6 +101,9 @@ static inline int hlt_use_halt(void) void default_idle(void) { if (hlt_use_halt()) { + struct power_trace it; + + trace_power_start(&it, POWER_CSTATE, 1); current_thread_info()->status &= ~TS_POLLING; /* * TS_POLLING-cleared state must be visible before we @@ -112,6 +116,7 @@ void default_idle(void) else local_irq_enable(); current_thread_info()->status |= TS_POLLING; + trace_power_end(&it); } else { local_irq_enable(); /* loop is done by the caller */ @@ -154,24 +159,31 @@ EXPORT_SYMBOL_GPL(cpu_idle_wait); */ void mwait_idle_with_hints(unsigned long ax, unsigned long cx) { + struct power_trace it; + + trace_power_start(&it, POWER_CSTATE, (ax>>4)+1); if (!need_resched()) { __monitor((void *)¤t_thread_info()->flags, 0, 0); smp_mb(); if (!need_resched()) __mwait(ax, cx); } + trace_power_end(&it); } /* Default MONITOR/MWAIT with no hints, used for default C1 state */ static void mwait_idle(void) { + struct power_trace it; if (!need_resched()) { + trace_power_start(&it, POWER_CSTATE, 1); __monitor((void *)¤t_thread_info()->flags, 0, 0); smp_mb(); if (!need_resched()) __sti_mwait(0, 0); else local_irq_enable(); + trace_power_end(&it); } else local_irq_enable(); } @@ -183,9 +195,13 @@ static void mwait_idle(void) */ static void poll_idle(void) { + struct power_trace it; + + trace_power_start(&it, POWER_CSTATE, 0); local_irq_enable(); while (!need_resched()) cpu_relax(); + trace_power_end(&it); } /* @@ -270,7 +286,7 @@ static void c1e_idle(void) rdmsr(MSR_K8_INT_PENDING_MSG, lo, hi); if (lo & K8_INTP_C1E_ACTIVE_MASK) { c1e_detected = 1; - if (!boot_cpu_has(X86_FEATURE_CONSTANT_TSC)) + if (!boot_cpu_has(X86_FEATURE_NONSTOP_TSC)) mark_tsc_unstable("TSC halt in AMD C1E"); printk(KERN_INFO "System has AMD C1E enabled\n"); set_cpu_cap(&boot_cpu_data, X86_FEATURE_AMDC1E); diff --git a/arch/x86/kernel/process_32.c b/arch/x86/kernel/process_32.c index 0a1302f..3ba155d 100644 --- a/arch/x86/kernel/process_32.c +++ b/arch/x86/kernel/process_32.c @@ -38,6 +38,7 @@ #include <linux/percpu.h> #include <linux/prctl.h> #include <linux/dmi.h> +#include <linux/ftrace.h> #include <asm/uaccess.h> #include <asm/pgtable.h> @@ -59,6 +60,7 @@ #include <asm/idle.h> #include <asm/syscalls.h> #include <asm/smp.h> +#include <asm/ds.h> asmlinkage void ret_from_fork(void) __asm__("ret_from_fork"); @@ -250,14 +252,8 @@ void exit_thread(void) tss->x86_tss.io_bitmap_base = INVALID_IO_BITMAP_OFFSET; put_cpu(); } -#ifdef CONFIG_X86_DS - /* Free any DS contexts that have not been properly released. */ - if (unlikely(current->thread.ds_ctx)) { - /* we clear debugctl to make sure DS is not used. */ - update_debugctlmsr(0); - ds_free(current->thread.ds_ctx); - } -#endif /* CONFIG_X86_DS */ + + ds_exit_thread(current); } void flush_thread(void) @@ -339,6 +335,12 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long sp, kfree(p->thread.io_bitmap_ptr); p->thread.io_bitmap_max = 0; } + + ds_copy_thread(p, current); + + clear_tsk_thread_flag(p, TIF_DEBUGCTLMSR); + p->thread.debugctlmsr = 0; + return err; } @@ -419,48 +421,19 @@ int set_tsc_mode(unsigned int val) return 0; } -#ifdef CONFIG_X86_DS -static int update_debugctl(struct thread_struct *prev, - struct thread_struct *next, unsigned long debugctl) -{ - unsigned long ds_prev = 0; - unsigned long ds_next = 0; - - if (prev->ds_ctx) - ds_prev = (unsigned long)prev->ds_ctx->ds; - if (next->ds_ctx) - ds_next = (unsigned long)next->ds_ctx->ds; - - if (ds_next != ds_prev) { - /* we clear debugctl to make sure DS - * is not in use when we change it */ - debugctl = 0; - update_debugctlmsr(0); - wrmsr(MSR_IA32_DS_AREA, ds_next, 0); - } - return debugctl; -} -#else -static int update_debugctl(struct thread_struct *prev, - struct thread_struct *next, unsigned long debugctl) -{ - return debugctl; -} -#endif /* CONFIG_X86_DS */ - static noinline void __switch_to_xtra(struct task_struct *prev_p, struct task_struct *next_p, struct tss_struct *tss) { struct thread_struct *prev, *next; - unsigned long debugctl; prev = &prev_p->thread; next = &next_p->thread; - debugctl = update_debugctl(prev, next, prev->debugctlmsr); - - if (next->debugctlmsr != debugctl) + if (test_tsk_thread_flag(next_p, TIF_DS_AREA_MSR) || + test_tsk_thread_flag(prev_p, TIF_DS_AREA_MSR)) + ds_switch_to(prev_p, next_p); + else if (next->debugctlmsr != prev->debugctlmsr) update_debugctlmsr(next->debugctlmsr); if (test_tsk_thread_flag(next_p, TIF_DEBUG)) { @@ -482,15 +455,6 @@ __switch_to_xtra(struct task_struct *prev_p, struct task_struct *next_p, hard_enable_TSC(); } -#ifdef CONFIG_X86_PTRACE_BTS - if (test_tsk_thread_flag(prev_p, TIF_BTS_TRACE_TS)) - ptrace_bts_take_timestamp(prev_p, BTS_TASK_DEPARTS); - - if (test_tsk_thread_flag(next_p, TIF_BTS_TRACE_TS)) - ptrace_bts_take_timestamp(next_p, BTS_TASK_ARRIVES); -#endif /* CONFIG_X86_PTRACE_BTS */ - - if (!test_tsk_thread_flag(next_p, TIF_IO_BITMAP)) { /* * Disable the bitmap via an invalid offset. We still cache @@ -548,7 +512,8 @@ __switch_to_xtra(struct task_struct *prev_p, struct task_struct *next_p, * the task-switch, and shows up in ret_from_fork in entry.S, * for example. */ -struct task_struct * __switch_to(struct task_struct *prev_p, struct task_struct *next_p) +__notrace_funcgraph struct task_struct * +__switch_to(struct task_struct *prev_p, struct task_struct *next_p) { struct thread_struct *prev = &prev_p->thread, *next = &next_p->thread; diff --git a/arch/x86/kernel/process_64.c b/arch/x86/kernel/process_64.c index c958120f..416fb92 100644 --- a/arch/x86/kernel/process_64.c +++ b/arch/x86/kernel/process_64.c @@ -39,6 +39,7 @@ #include <linux/prctl.h> #include <linux/uaccess.h> #include <linux/io.h> +#include <linux/ftrace.h> #include <asm/pgtable.h> #include <asm/system.h> @@ -52,6 +53,7 @@ #include <asm/ia32.h> #include <asm/idle.h> #include <asm/syscalls.h> +#include <asm/ds.h> asmlinkage extern void ret_from_fork(void); @@ -235,14 +237,8 @@ void exit_thread(void) t->io_bitmap_max = 0; put_cpu(); } -#ifdef CONFIG_X86_DS - /* Free any DS contexts that have not been properly released. */ - if (unlikely(t->ds_ctx)) { - /* we clear debugctl to make sure DS is not used. */ - update_debugctlmsr(0); - ds_free(t->ds_ctx); - } -#endif /* CONFIG_X86_DS */ + + ds_exit_thread(current); } void flush_thread(void) @@ -372,6 +368,12 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long sp, if (err) goto out; } + + ds_copy_thread(p, me); + + clear_tsk_thread_flag(p, TIF_DEBUGCTLMSR); + p->thread.debugctlmsr = 0; + err = 0; out: if (err && p->thread.io_bitmap_ptr) { @@ -470,35 +472,14 @@ static inline void __switch_to_xtra(struct task_struct *prev_p, struct tss_struct *tss) { struct thread_struct *prev, *next; - unsigned long debugctl; prev = &prev_p->thread, next = &next_p->thread; - debugctl = prev->debugctlmsr; - -#ifdef CONFIG_X86_DS - { - unsigned long ds_prev = 0, ds_next = 0; - - if (prev->ds_ctx) - ds_prev = (unsigned long)prev->ds_ctx->ds; - if (next->ds_ctx) - ds_next = (unsigned long)next->ds_ctx->ds; - - if (ds_next != ds_prev) { - /* - * We clear debugctl to make sure DS - * is not in use when we change it: - */ - debugctl = 0; - update_debugctlmsr(0); - wrmsrl(MSR_IA32_DS_AREA, ds_next); - } - } -#endif /* CONFIG_X86_DS */ - - if (next->debugctlmsr != debugctl) + if (test_tsk_thread_flag(next_p, TIF_DS_AREA_MSR) || + test_tsk_thread_flag(prev_p, TIF_DS_AREA_MSR)) + ds_switch_to(prev_p, next_p); + else if (next->debugctlmsr != prev->debugctlmsr) update_debugctlmsr(next->debugctlmsr); if (test_tsk_thread_flag(next_p, TIF_DEBUG)) { @@ -533,14 +514,6 @@ static inline void __switch_to_xtra(struct task_struct *prev_p, */ memset(tss->io_bitmap, 0xff, prev->io_bitmap_max); } - -#ifdef CONFIG_X86_PTRACE_BTS - if (test_tsk_thread_flag(prev_p, TIF_BTS_TRACE_TS)) - ptrace_bts_take_timestamp(prev_p, BTS_TASK_DEPARTS); - - if (test_tsk_thread_flag(next_p, TIF_BTS_TRACE_TS)) - ptrace_bts_take_timestamp(next_p, BTS_TASK_ARRIVES); -#endif /* CONFIG_X86_PTRACE_BTS */ } /* @@ -551,8 +524,9 @@ static inline void __switch_to_xtra(struct task_struct *prev_p, * - could test fs/gs bitsliced * * Kprobes not supported here. Set the probe on schedule instead. + * Function graph tracer not supported too. */ -struct task_struct * +__notrace_funcgraph struct task_struct * __switch_to(struct task_struct *prev_p, struct task_struct *next_p) { struct thread_struct *prev = &prev_p->thread; diff --git a/arch/x86/kernel/ptrace.c b/arch/x86/kernel/ptrace.c index 0a6d8c1..0a5df5f 100644 --- a/arch/x86/kernel/ptrace.c +++ b/arch/x86/kernel/ptrace.c @@ -581,158 +581,91 @@ static int ioperm_get(struct task_struct *target, } #ifdef CONFIG_X86_PTRACE_BTS -/* - * The configuration for a particular BTS hardware implementation. - */ -struct bts_configuration { - /* the size of a BTS record in bytes; at most BTS_MAX_RECORD_SIZE */ - unsigned char sizeof_bts; - /* the size of a field in the BTS record in bytes */ - unsigned char sizeof_field; - /* a bitmask to enable/disable BTS in DEBUGCTL MSR */ - unsigned long debugctl_mask; -}; -static struct bts_configuration bts_cfg; - -#define BTS_MAX_RECORD_SIZE (8 * 3) - - -/* - * Branch Trace Store (BTS) uses the following format. Different - * architectures vary in the size of those fields. - * - source linear address - * - destination linear address - * - flags - * - * Later architectures use 64bit pointers throughout, whereas earlier - * architectures use 32bit pointers in 32bit mode. - * - * We compute the base address for the first 8 fields based on: - * - the field size stored in the DS configuration - * - the relative field position - * - * In order to store additional information in the BTS buffer, we use - * a special source address to indicate that the record requires - * special interpretation. - * - * Netburst indicated via a bit in the flags field whether the branch - * was predicted; this is ignored. - */ - -enum bts_field { - bts_from = 0, - bts_to, - bts_flags, - - bts_escape = (unsigned long)-1, - bts_qual = bts_to, - bts_jiffies = bts_flags -}; - -static inline unsigned long bts_get(const char *base, enum bts_field field) -{ - base += (bts_cfg.sizeof_field * field); - return *(unsigned long *)base; -} - -static inline void bts_set(char *base, enum bts_field field, unsigned long val) -{ - base += (bts_cfg.sizeof_field * field);; - (*(unsigned long *)base) = val; -} - -/* - * Translate a BTS record from the raw format into the bts_struct format - * - * out (out): bts_struct interpretation - * raw: raw BTS record - */ -static void ptrace_bts_translate_record(struct bts_struct *out, const void *raw) -{ - memset(out, 0, sizeof(*out)); - if (bts_get(raw, bts_from) == bts_escape) { - out->qualifier = bts_get(raw, bts_qual); - out->variant.jiffies = bts_get(raw, bts_jiffies); - } else { - out->qualifier = BTS_BRANCH; - out->variant.lbr.from_ip = bts_get(raw, bts_from); - out->variant.lbr.to_ip = bts_get(raw, bts_to); - } -} - static int ptrace_bts_read_record(struct task_struct *child, size_t index, struct bts_struct __user *out) { - struct bts_struct ret; - const void *bts_record; - size_t bts_index, bts_end; + const struct bts_trace *trace; + struct bts_struct bts; + const unsigned char *at; int error; - error = ds_get_bts_end(child, &bts_end); - if (error < 0) - return error; - - if (bts_end <= index) - return -EINVAL; + trace = ds_read_bts(child->bts); + if (!trace) + return -EPERM; - error = ds_get_bts_index(child, &bts_index); - if (error < 0) - return error; + at = trace->ds.top - ((index + 1) * trace->ds.size); + if ((void *)at < trace->ds.begin) + at += (trace->ds.n * trace->ds.size); - /* translate the ptrace bts index into the ds bts index */ - bts_index += bts_end - (index + 1); - if (bts_end <= bts_index) - bts_index -= bts_end; + if (!trace->read) + return -EOPNOTSUPP; - error = ds_access_bts(child, bts_index, &bts_record); + error = trace->read(child->bts, at, &bts); if (error < 0) return error; - ptrace_bts_translate_record(&ret, bts_record); - - if (copy_to_user(out, &ret, sizeof(ret))) + if (copy_to_user(out, &bts, sizeof(bts))) return -EFAULT; - return sizeof(ret); + return sizeof(bts); } static int ptrace_bts_drain(struct task_struct *child, long size, struct bts_struct __user *out) { - struct bts_struct ret; - const unsigned char *raw; - size_t end, i; - int error; + const struct bts_trace *trace; + const unsigned char *at; + int error, drained = 0; - error = ds_get_bts_index(child, &end); - if (error < 0) - return error; + trace = ds_read_bts(child->bts); + if (!trace) + return -EPERM; - if (size < (end * sizeof(struct bts_struct))) + if (!trace->read) + return -EOPNOTSUPP; + + if (size < (trace->ds.top - trace->ds.begin)) return -EIO; - error = ds_access_bts(child, 0, (const void **)&raw); - if (error < 0) - return error; + for (at = trace->ds.begin; (void *)at < trace->ds.top; + out++, drained++, at += trace->ds.size) { + struct bts_struct bts; + int error; - for (i = 0; i < end; i++, out++, raw += bts_cfg.sizeof_bts) { - ptrace_bts_translate_record(&ret, raw); + error = trace->read(child->bts, at, &bts); + if (error < 0) + return error; - if (copy_to_user(out, &ret, sizeof(ret))) + if (copy_to_user(out, &bts, sizeof(bts))) return -EFAULT; } - error = ds_clear_bts(child); + memset(trace->ds.begin, 0, trace->ds.n * trace->ds.size); + + error = ds_reset_bts(child->bts); if (error < 0) return error; - return end; + return drained; } -static void ptrace_bts_ovfl(struct task_struct *child) +static int ptrace_bts_allocate_buffer(struct task_struct *child, size_t size) { - send_sig(child->thread.bts_ovfl_signal, child, 0); + child->bts_buffer = alloc_locked_buffer(size); + if (!child->bts_buffer) + return -ENOMEM; + + child->bts_size = size; + + return 0; +} + +static void ptrace_bts_free_buffer(struct task_struct *child) +{ + free_locked_buffer(child->bts_buffer, child->bts_size); + child->bts_buffer = NULL; + child->bts_size = 0; } static int ptrace_bts_config(struct task_struct *child, @@ -740,114 +673,86 @@ static int ptrace_bts_config(struct task_struct *child, const struct ptrace_bts_config __user *ucfg) { struct ptrace_bts_config cfg; - int error = 0; - - error = -EOPNOTSUPP; - if (!bts_cfg.sizeof_bts) - goto errout; + unsigned int flags = 0; - error = -EIO; if (cfg_size < sizeof(cfg)) - goto errout; + return -EIO; - error = -EFAULT; if (copy_from_user(&cfg, ucfg, sizeof(cfg))) - goto errout; + return -EFAULT; - error = -EINVAL; - if ((cfg.flags & PTRACE_BTS_O_SIGNAL) && - !(cfg.flags & PTRACE_BTS_O_ALLOC)) - goto errout; + if (child->bts) { + ds_release_bts(child->bts); + child->bts = NULL; + } - if (cfg.flags & PTRACE_BTS_O_ALLOC) { - ds_ovfl_callback_t ovfl = NULL; - unsigned int sig = 0; + if (cfg.flags & PTRACE_BTS_O_SIGNAL) { + if (!cfg.signal) + return -EINVAL; - /* we ignore the error in case we were not tracing child */ - (void)ds_release_bts(child); + return -EOPNOTSUPP; - if (cfg.flags & PTRACE_BTS_O_SIGNAL) { - if (!cfg.signal) - goto errout; + child->thread.bts_ovfl_signal = cfg.signal; + } - sig = cfg.signal; - ovfl = ptrace_bts_ovfl; - } + if ((cfg.flags & PTRACE_BTS_O_ALLOC) && + (cfg.size != child->bts_size)) { + int error; - error = ds_request_bts(child, /* base = */ NULL, cfg.size, ovfl); - if (error < 0) - goto errout; + ptrace_bts_free_buffer(child); - child->thread.bts_ovfl_signal = sig; + error = ptrace_bts_allocate_buffer(child, cfg.size); + if (error < 0) + return error; } - error = -EINVAL; - if (!child->thread.ds_ctx && cfg.flags) - goto errout; - if (cfg.flags & PTRACE_BTS_O_TRACE) - child->thread.debugctlmsr |= bts_cfg.debugctl_mask; - else - child->thread.debugctlmsr &= ~bts_cfg.debugctl_mask; + flags |= BTS_USER; if (cfg.flags & PTRACE_BTS_O_SCHED) - set_tsk_thread_flag(child, TIF_BTS_TRACE_TS); - else - clear_tsk_thread_flag(child, TIF_BTS_TRACE_TS); + flags |= BTS_TIMESTAMPS; - error = sizeof(cfg); + child->bts = ds_request_bts(child, child->bts_buffer, child->bts_size, + /* ovfl = */ NULL, /* th = */ (size_t)-1, + flags); + if (IS_ERR(child->bts)) { + int error = PTR_ERR(child->bts); -out: - if (child->thread.debugctlmsr) - set_tsk_thread_flag(child, TIF_DEBUGCTLMSR); - else - clear_tsk_thread_flag(child, TIF_DEBUGCTLMSR); + ptrace_bts_free_buffer(child); + child->bts = NULL; - return error; + return error; + } -errout: - child->thread.debugctlmsr &= ~bts_cfg.debugctl_mask; - clear_tsk_thread_flag(child, TIF_BTS_TRACE_TS); - goto out; + return sizeof(cfg); } static int ptrace_bts_status(struct task_struct *child, long cfg_size, struct ptrace_bts_config __user *ucfg) { + const struct bts_trace *trace; struct ptrace_bts_config cfg; - size_t end; - const void *base, *max; - int error; if (cfg_size < sizeof(cfg)) return -EIO; - error = ds_get_bts_end(child, &end); - if (error < 0) - return error; - - error = ds_access_bts(child, /* index = */ 0, &base); - if (error < 0) - return error; - - error = ds_access_bts(child, /* index = */ end, &max); - if (error < 0) - return error; + trace = ds_read_bts(child->bts); + if (!trace) + return -EPERM; memset(&cfg, 0, sizeof(cfg)); - cfg.size = (max - base); + cfg.size = trace->ds.end - trace->ds.begin; cfg.signal = child->thread.bts_ovfl_signal; cfg.bts_size = sizeof(struct bts_struct); if (cfg.signal) cfg.flags |= PTRACE_BTS_O_SIGNAL; - if (test_tsk_thread_flag(child, TIF_DEBUGCTLMSR) && - child->thread.debugctlmsr & bts_cfg.debugctl_mask) + if (trace->ds.flags & BTS_USER) cfg.flags |= PTRACE_BTS_O_TRACE; - if (test_tsk_thread_flag(child, TIF_BTS_TRACE_TS)) + if (trace->ds.flags & BTS_TIMESTAMPS) cfg.flags |= PTRACE_BTS_O_SCHED; if (copy_to_user(ucfg, &cfg, sizeof(cfg))) @@ -856,110 +761,77 @@ static int ptrace_bts_status(struct task_struct *child, return sizeof(cfg); } -static int ptrace_bts_write_record(struct task_struct *child, - const struct bts_struct *in) +static int ptrace_bts_clear(struct task_struct *child) { - unsigned char bts_record[BTS_MAX_RECORD_SIZE]; + const struct bts_trace *trace; - BUG_ON(BTS_MAX_RECORD_SIZE < bts_cfg.sizeof_bts); + trace = ds_read_bts(child->bts); + if (!trace) + return -EPERM; - memset(bts_record, 0, bts_cfg.sizeof_bts); - switch (in->qualifier) { - case BTS_INVALID: - break; + memset(trace->ds.begin, 0, trace->ds.n * trace->ds.size); - case BTS_BRANCH: - bts_set(bts_record, bts_from, in->variant.lbr.from_ip); - bts_set(bts_record, bts_to, in->variant.lbr.to_ip); - break; + return ds_reset_bts(child->bts); +} - case BTS_TASK_ARRIVES: - case BTS_TASK_DEPARTS: - bts_set(bts_record, bts_from, bts_escape); - bts_set(bts_record, bts_qual, in->qualifier); - bts_set(bts_record, bts_jiffies, in->variant.jiffies); - break; +static int ptrace_bts_size(struct task_struct *child) +{ + const struct bts_trace *trace; - default: - return -EINVAL; - } + trace = ds_read_bts(child->bts); + if (!trace) + return -EPERM; - /* The writing task will be the switched-to task on a context - * switch. It needs to write into the switched-from task's BTS - * buffer. */ - return ds_unchecked_write_bts(child, bts_record, bts_cfg.sizeof_bts); + return (trace->ds.top - trace->ds.begin) / trace->ds.size; } -void ptrace_bts_take_timestamp(struct task_struct *tsk, - enum bts_qualifier qualifier) +static void ptrace_bts_fork(struct task_struct *tsk) { - struct bts_struct rec = { - .qualifier = qualifier, - .variant.jiffies = jiffies_64 - }; - - ptrace_bts_write_record(tsk, &rec); + tsk->bts = NULL; + tsk->bts_buffer = NULL; + tsk->bts_size = 0; + tsk->thread.bts_ovfl_signal = 0; } -static const struct bts_configuration bts_cfg_netburst = { - .sizeof_bts = sizeof(long) * 3, - .sizeof_field = sizeof(long), - .debugctl_mask = (1<<2)|(1<<3)|(1<<5) -}; +static void ptrace_bts_untrace(struct task_struct *child) +{ + if (unlikely(child->bts)) { + ds_release_bts(child->bts); + child->bts = NULL; + + /* We cannot update total_vm and locked_vm since + child's mm is already gone. But we can reclaim the + memory. */ + kfree(child->bts_buffer); + child->bts_buffer = NULL; + child->bts_size = 0; + } +} -static const struct bts_configuration bts_cfg_pentium_m = { - .sizeof_bts = sizeof(long) * 3, - .sizeof_field = sizeof(long), - .debugctl_mask = (1<<6)|(1<<7) -}; +static void ptrace_bts_detach(struct task_struct *child) +{ + if (unlikely(child->bts)) { + ds_release_bts(child->bts); + child->bts = NULL; -static const struct bts_configuration bts_cfg_core2 = { - .sizeof_bts = 8 * 3, - .sizeof_field = 8, - .debugctl_mask = (1<<6)|(1<<7)|(1<<9) -}; + ptrace_bts_free_buffer(child); + } +} +#else +static inline void ptrace_bts_fork(struct task_struct *tsk) {} +static inline void ptrace_bts_detach(struct task_struct *child) {} +static inline void ptrace_bts_untrace(struct task_struct *child) {} +#endif /* CONFIG_X86_PTRACE_BTS */ -static inline void bts_configure(const struct bts_configuration *cfg) +void x86_ptrace_fork(struct task_struct *child, unsigned long clone_flags) { - bts_cfg = *cfg; + ptrace_bts_fork(child); } -void __cpuinit ptrace_bts_init_intel(struct cpuinfo_x86 *c) +void x86_ptrace_untrace(struct task_struct *child) { - switch (c->x86) { - case 0x6: - switch (c->x86_model) { - case 0xD: - case 0xE: /* Pentium M */ - bts_configure(&bts_cfg_pentium_m); - break; - case 0xF: /* Core2 */ - case 0x1C: /* Atom */ - bts_configure(&bts_cfg_core2); - break; - default: - /* sorry, don't know about them */ - break; - } - break; - case 0xF: - switch (c->x86_model) { - case 0x0: - case 0x1: - case 0x2: /* Netburst */ - bts_configure(&bts_cfg_netburst); - break; - default: - /* sorry, don't know about them */ - break; - } - break; - default: - /* sorry, don't know about them */ - break; - } + ptrace_bts_untrace(child); } -#endif /* CONFIG_X86_PTRACE_BTS */ /* * Called by kernel/ptrace.c when detaching.. @@ -972,15 +844,7 @@ void ptrace_disable(struct task_struct *child) #ifdef TIF_SYSCALL_EMU clear_tsk_thread_flag(child, TIF_SYSCALL_EMU); #endif -#ifdef CONFIG_X86_PTRACE_BTS - (void)ds_release_bts(child); - - child->thread.debugctlmsr &= ~bts_cfg.debugctl_mask; - if (!child->thread.debugctlmsr) - clear_tsk_thread_flag(child, TIF_DEBUGCTLMSR); - - clear_tsk_thread_flag(child, TIF_BTS_TRACE_TS); -#endif /* CONFIG_X86_PTRACE_BTS */ + ptrace_bts_detach(child); } #if defined CONFIG_X86_32 || defined CONFIG_IA32_EMULATION @@ -1112,7 +976,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data) break; case PTRACE_BTS_SIZE: - ret = ds_get_bts_index(child, /* pos = */ NULL); + ret = ptrace_bts_size(child); break; case PTRACE_BTS_GET: @@ -1121,7 +985,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data) break; case PTRACE_BTS_CLEAR: - ret = ds_clear_bts(child); + ret = ptrace_bts_clear(child); break; case PTRACE_BTS_DRAIN: @@ -1384,6 +1248,14 @@ long compat_arch_ptrace(struct task_struct *child, compat_long_t request, case PTRACE_GET_THREAD_AREA: case PTRACE_SET_THREAD_AREA: +#ifdef CONFIG_X86_PTRACE_BTS + case PTRACE_BTS_CONFIG: + case PTRACE_BTS_STATUS: + case PTRACE_BTS_SIZE: + case PTRACE_BTS_GET: + case PTRACE_BTS_CLEAR: + case PTRACE_BTS_DRAIN: +#endif /* CONFIG_X86_PTRACE_BTS */ return arch_ptrace(child, request, addr, data); default: diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index f71f96f..f6174d2 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -287,7 +287,7 @@ static int __cpuinitdata unsafe_smp; /* * Activate a secondary processor. */ -static void __cpuinit start_secondary(void *unused) +notrace static void __cpuinit start_secondary(void *unused) { /* * Don't put *anything* before cpu_init(), SMP booting is too diff --git a/arch/x86/kernel/stacktrace.c b/arch/x86/kernel/stacktrace.c index a03e7f6..10786af 100644 --- a/arch/x86/kernel/stacktrace.c +++ b/arch/x86/kernel/stacktrace.c @@ -6,6 +6,7 @@ #include <linux/sched.h> #include <linux/stacktrace.h> #include <linux/module.h> +#include <linux/uaccess.h> #include <asm/stacktrace.h> static void save_stack_warning(void *data, char *msg) @@ -83,3 +84,66 @@ void save_stack_trace_tsk(struct task_struct *tsk, struct stack_trace *trace) trace->entries[trace->nr_entries++] = ULONG_MAX; } EXPORT_SYMBOL_GPL(save_stack_trace_tsk); + +/* Userspace stacktrace - based on kernel/trace/trace_sysprof.c */ + +struct stack_frame { + const void __user *next_fp; + unsigned long ret_addr; +}; + +static int copy_stack_frame(const void __user *fp, struct stack_frame *frame) +{ + int ret; + + if (!access_ok(VERIFY_READ, fp, sizeof(*frame))) + return 0; + + ret = 1; + pagefault_disable(); + if (__copy_from_user_inatomic(frame, fp, sizeof(*frame))) + ret = 0; + pagefault_enable(); + + return ret; +} + +static inline void __save_stack_trace_user(struct stack_trace *trace) +{ + const struct pt_regs *regs = task_pt_regs(current); + const void __user *fp = (const void __user *)regs->bp; + + if (trace->nr_entries < trace->max_entries) + trace->entries[trace->nr_entries++] = regs->ip; + + while (trace->nr_entries < trace->max_entries) { + struct stack_frame frame; + + frame.next_fp = NULL; + frame.ret_addr = 0; + if (!copy_stack_frame(fp, &frame)) + break; + if ((unsigned long)fp < regs->sp) + break; + if (frame.ret_addr) { + trace->entries[trace->nr_entries++] = + frame.ret_addr; + } + if (fp == frame.next_fp) + break; + fp = frame.next_fp; + } +} + +void save_stack_trace_user(struct stack_trace *trace) +{ + /* + * Trace user stack if we are not a kernel thread + */ + if (current->mm) { + __save_stack_trace_user(trace); + } + if (trace->nr_entries < trace->max_entries) + trace->entries[trace->nr_entries++] = ULONG_MAX; +} + diff --git a/arch/x86/kernel/vmlinux_32.lds.S b/arch/x86/kernel/vmlinux_32.lds.S index a9b8560..82c6755 100644 --- a/arch/x86/kernel/vmlinux_32.lds.S +++ b/arch/x86/kernel/vmlinux_32.lds.S @@ -44,6 +44,7 @@ SECTIONS SCHED_TEXT LOCK_TEXT KPROBES_TEXT + IRQENTRY_TEXT *(.fixup) *(.gnu.warning) _etext = .; /* End of text section */ diff --git a/arch/x86/kernel/vmlinux_64.lds.S b/arch/x86/kernel/vmlinux_64.lds.S index 46e0544..1a614c0 100644 --- a/arch/x86/kernel/vmlinux_64.lds.S +++ b/arch/x86/kernel/vmlinux_64.lds.S @@ -35,6 +35,7 @@ SECTIONS SCHED_TEXT LOCK_TEXT KPROBES_TEXT + IRQENTRY_TEXT *(.fixup) *(.gnu.warning) _etext = .; /* End of text section */ diff --git a/arch/x86/kernel/vsyscall_64.c b/arch/x86/kernel/vsyscall_64.c index 0b8b6690..6f3d3d4 100644 --- a/arch/x86/kernel/vsyscall_64.c +++ b/arch/x86/kernel/vsyscall_64.c @@ -17,6 +17,9 @@ * want per guest time just set the kernel.vsyscall64 sysctl to 0. */ +/* Disable profiling for userspace code: */ +#define DISABLE_BRANCH_PROFILING + #include <linux/time.h> #include <linux/init.h> #include <linux/kernel.h> diff --git a/arch/x86/mm/Makefile b/arch/x86/mm/Makefile index fea4565..d8cc96a2 100644 --- a/arch/x86/mm/Makefile +++ b/arch/x86/mm/Makefile @@ -8,9 +8,8 @@ obj-$(CONFIG_X86_PTDUMP) += dump_pagetables.o obj-$(CONFIG_HIGHMEM) += highmem_32.o -obj-$(CONFIG_MMIOTRACE_HOOKS) += kmmio.o obj-$(CONFIG_MMIOTRACE) += mmiotrace.o -mmiotrace-y := pf_in.o mmio-mod.o +mmiotrace-y := kmmio.o pf_in.o mmio-mod.o obj-$(CONFIG_MMIOTRACE_TEST) += testmmiotrace.o obj-$(CONFIG_NUMA) += numa_$(BITS).o diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c index 31e8730..21e996a 100644 --- a/arch/x86/mm/fault.c +++ b/arch/x86/mm/fault.c @@ -53,7 +53,7 @@ static inline int kmmio_fault(struct pt_regs *regs, unsigned long addr) { -#ifdef CONFIG_MMIOTRACE_HOOKS +#ifdef CONFIG_MMIOTRACE if (unlikely(is_kmmio_active())) if (kmmio_handler(regs, addr) == 1) return -1; @@ -413,6 +413,7 @@ static noinline void pgtable_bad(unsigned long address, struct pt_regs *regs, unsigned long error_code) { unsigned long flags = oops_begin(); + int sig = SIGKILL; struct task_struct *tsk; printk(KERN_ALERT "%s: Corrupted page table at address %lx\n", @@ -423,8 +424,8 @@ static noinline void pgtable_bad(unsigned long address, struct pt_regs *regs, tsk->thread.trap_no = 14; tsk->thread.error_code = error_code; if (__die("Bad pagetable", regs, error_code)) - regs = NULL; - oops_end(flags, regs, SIGKILL); + sig = 0; + oops_end(flags, regs, sig); } #endif @@ -590,6 +591,7 @@ void __kprobes do_page_fault(struct pt_regs *regs, unsigned long error_code) int fault; #ifdef CONFIG_X86_64 unsigned long flags; + int sig; #endif tsk = current; @@ -849,11 +851,12 @@ no_context: bust_spinlocks(0); do_exit(SIGKILL); #else + sig = SIGKILL; if (__die("Oops", regs, error_code)) - regs = NULL; + sig = 0; /* Executive summary in case the body of the oops scrolled away */ printk(KERN_EMERG "CR2: %016lx\n", address); - oops_end(flags, regs, SIGKILL); + oops_end(flags, regs, sig); #endif /* diff --git a/arch/x86/vdso/vclock_gettime.c b/arch/x86/vdso/vclock_gettime.c index 1ef0f90..d9d3582 100644 --- a/arch/x86/vdso/vclock_gettime.c +++ b/arch/x86/vdso/vclock_gettime.c @@ -9,6 +9,9 @@ * Also alternative() doesn't work. */ +/* Disable profiling for userspace code: */ +#define DISABLE_BRANCH_PROFILING + #include <linux/kernel.h> #include <linux/posix-timers.h> #include <linux/time.h> |