From f984b51e0779a6dd30feedc41404013ca54e5d05 Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Mon, 12 May 2008 21:20:57 +0200 Subject: ftrace: add mmiotrace plugin On Sat, 22 Mar 2008 13:07:47 +0100 Ingo Molnar wrote: > > > i'd suggest the following: pull x86.git and sched-devel.git into a > > > single tree [the two will combine without rejects]. Then try to add a > > > kernel/tracing/trace_mmiotrace.c ftrace plugin. The trace_sysprof.c > > > plugin might be a good example. > > > > I did this and now I have mmiotrace enabled/disabled via the tracing > > framework (what do we call this, since ftrace is one of the tracers?). > > cool! could you send the patches for that? (even if they are not fully > functional yet) Patch attached in the end. Nice to see how much code disappeared. I tried to mark all the features I had to break with XXX-comments. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/Makefile | 1 + kernel/trace/trace_mmiotrace.c | 84 ++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 85 insertions(+) create mode 100644 kernel/trace/trace_mmiotrace.c (limited to 'kernel') diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile index d9efbbfa2bdf..c44a7dce9086 100644 --- a/kernel/trace/Makefile +++ b/kernel/trace/Makefile @@ -18,5 +18,6 @@ obj-$(CONFIG_FTRACE) += trace_functions.o obj-$(CONFIG_IRQSOFF_TRACER) += trace_irqsoff.o obj-$(CONFIG_PREEMPT_TRACER) += trace_irqsoff.o obj-$(CONFIG_SCHED_TRACER) += trace_sched_wakeup.o +obj-$(CONFIG_MMIOTRACE) += trace_mmiotrace.o libftrace-y := ftrace.o diff --git a/kernel/trace/trace_mmiotrace.c b/kernel/trace/trace_mmiotrace.c new file mode 100644 index 000000000000..e4dd03cc5aa6 --- /dev/null +++ b/kernel/trace/trace_mmiotrace.c @@ -0,0 +1,84 @@ +/* + * Memory mapped I/O tracing + * + * Copyright (C) 2008 Pekka Paalanen + */ + +#define DEBUG 1 + +#include +#include + +#include "trace.h" + +extern void +__trace_special(void *__tr, void *__data, + unsigned long arg1, unsigned long arg2, unsigned long arg3); + +static struct trace_array *mmio_trace_array; + + +static void mmio_trace_init(struct trace_array *tr) +{ + pr_debug("in %s\n", __func__); + mmio_trace_array = tr; + if (tr->ctrl) + enable_mmiotrace(); +} + +static void mmio_trace_reset(struct trace_array *tr) +{ + pr_debug("in %s\n", __func__); + if (tr->ctrl) + disable_mmiotrace(); +} + +static void mmio_trace_ctrl_update(struct trace_array *tr) +{ + pr_debug("in %s\n", __func__); + if (tr->ctrl) + enable_mmiotrace(); + else + disable_mmiotrace(); +} + +static struct tracer mmio_tracer __read_mostly = +{ + .name = "mmiotrace", + .init = mmio_trace_init, + .reset = mmio_trace_reset, + .ctrl_update = mmio_trace_ctrl_update, +}; + +__init static int init_mmio_trace(void) +{ + int ret = init_mmiotrace(); + if (ret) + return ret; + return register_tracer(&mmio_tracer); +} +device_initcall(init_mmio_trace); + +void mmio_trace_record(u32 type, unsigned long addr, unsigned long arg) +{ + struct trace_array *tr = mmio_trace_array; + struct trace_array_cpu *data = tr->data[smp_processor_id()]; + + if (!current || current->pid == 0) { + /* + * XXX: This is a problem. We need to able to record, no + * matter what. tracing_generic_entry_update() would crash. + */ + static unsigned limit; + if (limit++ < 12) + pr_err("Error in %s: no current.\n", __func__); + return; + } + if (!tr || !data) { + static unsigned limit; + if (limit++ < 12) + pr_err("%s: no tr or data\n", __func__); + return; + } + __trace_special(tr, data, type, addr, arg); +} -- cgit v1.2.3 From bd8ac686c73c7e925fcfe0b02dc4e7b947127864 Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Mon, 12 May 2008 21:20:57 +0200 Subject: ftrace: mmiotrace, updates here is a patch that makes mmiotrace work almost well within the tracing framework. The patch applies on top of my previous patch. I have my own output formatting in place now. Summary of changes: - fix the NULL dereference that was due to not calling tracing_reset() - add print_line() callback into struct tracer - implement print_line() for mmiotrace, producing up-to-spec text - add my output header, but that is not really called in the right place - rewrote the main structs in mmiotrace - added two new trace entry types: TRACE_MMIO_RW and TRACE_MMIO_MAP - made some functions in trace.c non-static - check current==NULL in tracing_generic_entry_update() - fix(?) comparison in trace_seq_printf() Things seem to work fine except a few issues. Markers (text lines injected into mmiotrace log) are missing, I did not feel hacking them in before we have variable length entries. My output header is printed only for 'trace' file, but not 'trace_pipe'. For some reason, despite my quick fix, iter->trace is NULL in print_trace_line() when called from 'trace_pipe' file, which means I don't get proper output formatting. I only tried by loading nouveau.ko, which just detects the card, and that is traced fine. I didn't try further. Map, two reads and unmap. Works perfectly. I am missing the information about overflows, I'd prefer to have a counter for lost events. I didn't try, but I guess currently there is no way of knowning when it overflows? So, not too far from being fully operational, it seems :-) And looking at the diffstat, there also is some 700-900 lines of user space code that just became obsolete. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- arch/x86/Kconfig.debug | 2 +- arch/x86/kernel/mmiotrace/mmio-mod.c | 140 ++++++++++---------------------- include/linux/mmiotrace.h | 85 ++++++-------------- kernel/trace/trace.c | 34 ++++++++ kernel/trace/trace.h | 14 ++++ kernel/trace/trace_mmiotrace.c | 151 ++++++++++++++++++++++++++++------- 6 files changed, 238 insertions(+), 188 deletions(-) (limited to 'kernel') diff --git a/arch/x86/Kconfig.debug b/arch/x86/Kconfig.debug index 7e4b8494078e..1d6de0d67f99 100644 --- a/arch/x86/Kconfig.debug +++ b/arch/x86/Kconfig.debug @@ -173,7 +173,7 @@ config MMIOTRACE_HOOKS config MMIOTRACE bool "Memory mapped IO tracing" - depends on DEBUG_KERNEL && RELAY + depends on DEBUG_KERNEL select TRACING select MMIOTRACE_HOOKS default y diff --git a/arch/x86/kernel/mmiotrace/mmio-mod.c b/arch/x86/kernel/mmiotrace/mmio-mod.c index c7a67d7e482b..62abc281a512 100644 --- a/arch/x86/kernel/mmiotrace/mmio-mod.c +++ b/arch/x86/kernel/mmiotrace/mmio-mod.c @@ -37,11 +37,6 @@ #define NAME "mmiotrace: " -/* This app's relay channel files will appear in /debug/mmio-trace */ -static const char APP_DIR[] = "mmio-trace"; -/* the marker injection file in /debug/APP_DIR */ -static const char MARKER_FILE[] = "mmio-marker"; - struct trap_reason { unsigned long addr; unsigned long ip; @@ -56,18 +51,15 @@ struct remap_trace { unsigned long id; }; -static const size_t subbuf_size = 256*1024; - /* Accessed per-cpu. */ static DEFINE_PER_CPU(struct trap_reason, pf_reason); -static DEFINE_PER_CPU(struct mm_io_header_rw, cpu_trace); +static DEFINE_PER_CPU(struct mmiotrace_rw, cpu_trace); #if 0 /* XXX: no way gather this info anymore */ /* Access to this is not per-cpu. */ static DEFINE_PER_CPU(atomic_t, dropped); #endif -static struct dentry *dir; static struct dentry *marker_file; static DEFINE_MUTEX(mmiotrace_mutex); @@ -82,24 +74,21 @@ static LIST_HEAD(trace_list); /* struct remap_trace */ * and trace_lock. * - Routines depending on is_enabled() must take trace_lock. * - trace_list users must hold trace_lock. - * - is_enabled() guarantees that chan is valid. + * - is_enabled() guarantees that mmio_trace_record is allowed. * - pre/post callbacks assume the effect of is_enabled() being true. */ /* module parameters */ -static unsigned int n_subbufs = 32*4; static unsigned long filter_offset; static int nommiotrace; static int ISA_trace; static int trace_pc; -module_param(n_subbufs, uint, 0); module_param(filter_offset, ulong, 0); module_param(nommiotrace, bool, 0); module_param(ISA_trace, bool, 0); module_param(trace_pc, bool, 0); -MODULE_PARM_DESC(n_subbufs, "Number of 256kB buffers, default 128."); MODULE_PARM_DESC(filter_offset, "Start address of traced mappings."); MODULE_PARM_DESC(nommiotrace, "Disable actual MMIO tracing."); MODULE_PARM_DESC(ISA_trace, "Do not exclude the low ISA range."); @@ -110,6 +99,7 @@ static bool is_enabled(void) return atomic_read(&mmiotrace_enabled); } +#if 0 /* XXX: needs rewrite */ /* * Write callback for the debugfs entry: * Read a marker and write it to the mmio trace log @@ -145,6 +135,7 @@ static ssize_t write_marker(struct file *file, const char __user *buffer, kfree(event); return len; } +#endif static void print_pte(unsigned long address) { @@ -198,9 +189,10 @@ static void pre(struct kmmio_probe *p, struct pt_regs *regs, unsigned long addr) { struct trap_reason *my_reason = &get_cpu_var(pf_reason); - struct mm_io_header_rw *my_trace = &get_cpu_var(cpu_trace); + struct mmiotrace_rw *my_trace = &get_cpu_var(cpu_trace); const unsigned long instptr = instruction_pointer(regs); const enum reason_type type = get_ins_type(instptr); + struct remap_trace *trace = p->user_data; /* it doesn't make sense to have more than one active trace per cpu */ if (my_reason->active_traces) @@ -212,23 +204,17 @@ static void pre(struct kmmio_probe *p, struct pt_regs *regs, my_reason->addr = addr; my_reason->ip = instptr; - my_trace->header.type = MMIO_MAGIC; - my_trace->header.pid = 0; - my_trace->header.data_len = sizeof(struct mm_io_rw); - my_trace->rw.address = addr; - /* - * struct remap_trace *trace = p->user_data; - * phys = addr - trace->probe.addr + trace->phys; - */ + my_trace->phys = addr - trace->probe.addr + trace->phys; + my_trace->map_id = trace->id; /* * Only record the program counter when requested. * It may taint clean-room reverse engineering. */ if (trace_pc) - my_trace->rw.pc = instptr; + my_trace->pc = instptr; else - my_trace->rw.pc = 0; + my_trace->pc = 0; /* * XXX: the timestamp recorded will be *after* the tracing has been @@ -238,28 +224,25 @@ static void pre(struct kmmio_probe *p, struct pt_regs *regs, switch (type) { case REG_READ: - my_trace->header.type |= - (MMIO_READ << MMIO_OPCODE_SHIFT) | - (get_ins_mem_width(instptr) << MMIO_WIDTH_SHIFT); + my_trace->opcode = MMIO_READ; + my_trace->width = get_ins_mem_width(instptr); break; case REG_WRITE: - my_trace->header.type |= - (MMIO_WRITE << MMIO_OPCODE_SHIFT) | - (get_ins_mem_width(instptr) << MMIO_WIDTH_SHIFT); - my_trace->rw.value = get_ins_reg_val(instptr, regs); + my_trace->opcode = MMIO_WRITE; + my_trace->width = get_ins_mem_width(instptr); + my_trace->value = get_ins_reg_val(instptr, regs); break; case IMM_WRITE: - my_trace->header.type |= - (MMIO_WRITE << MMIO_OPCODE_SHIFT) | - (get_ins_mem_width(instptr) << MMIO_WIDTH_SHIFT); - my_trace->rw.value = get_ins_imm_val(instptr); + my_trace->opcode = MMIO_WRITE; + my_trace->width = get_ins_mem_width(instptr); + my_trace->value = get_ins_imm_val(instptr); break; default: { unsigned char *ip = (unsigned char *)instptr; - my_trace->header.type |= - (MMIO_UNKNOWN_OP << MMIO_OPCODE_SHIFT); - my_trace->rw.value = (*ip) << 16 | *(ip + 1) << 8 | + my_trace->opcode = MMIO_UNKNOWN_OP; + my_trace->width = 0; + my_trace->value = (*ip) << 16 | *(ip + 1) << 8 | *(ip + 2); } } @@ -271,7 +254,7 @@ static void post(struct kmmio_probe *p, unsigned long condition, struct pt_regs *regs) { struct trap_reason *my_reason = &get_cpu_var(pf_reason); - struct mm_io_header_rw *my_trace = &get_cpu_var(cpu_trace); + struct mmiotrace_rw *my_trace = &get_cpu_var(cpu_trace); /* this should always return the active_trace count to 0 */ my_reason->active_traces--; @@ -282,20 +265,13 @@ static void post(struct kmmio_probe *p, unsigned long condition, switch (my_reason->type) { case REG_READ: - my_trace->rw.value = get_ins_reg_val(my_reason->ip, regs); + my_trace->value = get_ins_reg_val(my_reason->ip, regs); break; default: break; } - /* - * XXX: Several required values are ignored: - * - mapping id - * - program counter - * Also the address should be physical, not virtual. - */ - mmio_trace_record(my_trace->header.type, my_trace->rw.address, - my_trace->rw.value); + mmio_trace_rw(my_trace); put_cpu_var(cpu_trace); put_cpu_var(pf_reason); } @@ -305,21 +281,11 @@ static void ioremap_trace_core(unsigned long offset, unsigned long size, { static atomic_t next_id; struct remap_trace *trace = kmalloc(sizeof(*trace), GFP_KERNEL); - struct mm_io_header_map event = { - .header = { - .type = MMIO_MAGIC | - (MMIO_PROBE << MMIO_OPCODE_SHIFT), - .sec = 0, - .nsec = 0, - .pid = 0, - .data_len = sizeof(struct mm_io_map) - }, - .map = { - .phys = offset, - .addr = (unsigned long)addr, - .len = size, - .pc = 0 - } + struct mmiotrace_map map = { + .phys = offset, + .virt = (unsigned long)addr, + .len = size, + .opcode = MMIO_PROBE }; if (!trace) { @@ -338,15 +304,13 @@ static void ioremap_trace_core(unsigned long offset, unsigned long size, .phys = offset, .id = atomic_inc_return(&next_id) }; + map.map_id = trace->id; spin_lock_irq(&trace_lock); if (!is_enabled()) goto not_enabled; - /* - * XXX: Insufficient data recorded! - */ - mmio_trace_record(event.header.type, event.map.addr, event.map.len); + mmio_trace_mapping(&map); list_add_tail(&trace->list, &trace_list); if (!nommiotrace) register_kmmio_probe(&trace->probe); @@ -369,21 +333,11 @@ mmiotrace_ioremap(unsigned long offset, unsigned long size, void __iomem *addr) static void iounmap_trace_core(volatile void __iomem *addr) { - struct mm_io_header_map event = { - .header = { - .type = MMIO_MAGIC | - (MMIO_UNPROBE << MMIO_OPCODE_SHIFT), - .sec = 0, - .nsec = 0, - .pid = 0, - .data_len = sizeof(struct mm_io_map) - }, - .map = { - .phys = 0, - .addr = (unsigned long)addr, - .len = 0, - .pc = 0 - } + struct mmiotrace_map map = { + .phys = 0, + .virt = (unsigned long)addr, + .len = 0, + .opcode = MMIO_UNPROBE }; struct remap_trace *trace; struct remap_trace *tmp; @@ -404,8 +358,8 @@ static void iounmap_trace_core(volatile void __iomem *addr) break; } } - mmio_trace_record(event.header.type, event.map.addr, - found_trace ? found_trace->id : -1); + map.map_id = (found_trace) ? found_trace->id : -1; + mmio_trace_mapping(&map); not_enabled: spin_unlock_irq(&trace_lock); @@ -448,10 +402,12 @@ static void clear_trace_list(void) } } +#if 0 /* XXX: out of order */ static struct file_operations fops_marker = { .owner = THIS_MODULE, .write = write_marker }; +#endif void enable_mmiotrace(void) { @@ -464,9 +420,9 @@ void enable_mmiotrace(void) #if 0 /* XXX: tracing does not support text entries */ marker_file = debugfs_create_file("marker", 0660, dir, NULL, &fops_marker); -#endif if (!marker_file) pr_err(NAME "marker file creation failed.\n"); +#endif if (nommiotrace) pr_info(NAME "MMIO tracing disabled.\n"); @@ -502,17 +458,3 @@ void disable_mmiotrace(void) out: mutex_unlock(&mmiotrace_mutex); } - -int __init init_mmiotrace(void) -{ - pr_debug(NAME "load...\n"); - if (n_subbufs < 2) - return -EINVAL; - - dir = debugfs_create_dir(APP_DIR, NULL); - if (!dir) { - pr_err(NAME "Couldn't create relay app directory.\n"); - return -ENOMEM; - } - return 0; -} diff --git a/include/linux/mmiotrace.h b/include/linux/mmiotrace.h index 579b3b06c90e..c88a9c197d22 100644 --- a/include/linux/mmiotrace.h +++ b/include/linux/mmiotrace.h @@ -54,73 +54,38 @@ static inline void mmiotrace_iounmap(volatile void __iomem *addr) } #endif /* CONFIG_MMIOTRACE_HOOKS */ -/* in kernel/trace/trace_mmiotrace.c */ -extern int __init init_mmiotrace(void); -extern void enable_mmiotrace(void); -extern void disable_mmiotrace(void); -extern void mmio_trace_record(u32 type, unsigned long addr, unsigned long arg); - -#endif /* __KERNEL__ */ - - -/* - * If you change anything here, you must bump MMIO_VERSION. - * This is the relay data format for user space. - */ -#define MMIO_VERSION 0x04 - -/* mm_io_header.type */ -#define MMIO_OPCODE_MASK 0xff -#define MMIO_OPCODE_SHIFT 0 -#define MMIO_WIDTH_MASK 0xff00 -#define MMIO_WIDTH_SHIFT 8 -#define MMIO_MAGIC (0x6f000000 | (MMIO_VERSION<<16)) -#define MMIO_MAGIC_MASK 0xffff0000 - -enum mm_io_opcode { /* payload type: */ - MMIO_READ = 0x1, /* struct mm_io_rw */ - MMIO_WRITE = 0x2, /* struct mm_io_rw */ - MMIO_PROBE = 0x3, /* struct mm_io_map */ - MMIO_UNPROBE = 0x4, /* struct mm_io_map */ +enum mm_io_opcode { + MMIO_READ = 0x1, /* struct mmiotrace_rw */ + MMIO_WRITE = 0x2, /* struct mmiotrace_rw */ + MMIO_PROBE = 0x3, /* struct mmiotrace_map */ + MMIO_UNPROBE = 0x4, /* struct mmiotrace_map */ MMIO_MARKER = 0x5, /* raw char data */ - MMIO_UNKNOWN_OP = 0x6, /* struct mm_io_rw */ + MMIO_UNKNOWN_OP = 0x6, /* struct mmiotrace_rw */ }; -struct mm_io_header { - __u32 type; /* see MMIO_* macros above */ - __u32 sec; /* timestamp */ - __u32 nsec; - __u32 pid; /* PID of the process, or 0 for kernel core */ - __u16 data_len; /* length of the following payload */ +struct mmiotrace_rw { + unsigned long phys; /* PCI address of register */ + unsigned long value; + unsigned long pc; /* optional program counter */ + int map_id; + unsigned char opcode; /* one of MMIO_{READ,WRITE,UNKNOWN_OP} */ + unsigned char width; /* size of register access in bytes */ }; -struct mm_io_rw { - __u64 address; /* virtual address of register */ - __u64 value; - __u64 pc; /* optional program counter */ +struct mmiotrace_map { + unsigned long phys; /* base address in PCI space */ + unsigned long virt; /* base virtual address */ + unsigned long len; /* mapping size */ + int map_id; + unsigned char opcode; /* MMIO_PROBE or MMIO_UNPROBE */ }; -struct mm_io_map { - __u64 phys; /* base address in PCI space */ - __u64 addr; /* base virtual address */ - __u64 len; /* mapping size */ - __u64 pc; /* optional program counter */ -}; - - -/* - * These structures are used to allow a single relay_write() - * call to write a full packet. - */ - -struct mm_io_header_rw { - struct mm_io_header header; - struct mm_io_rw rw; -} __attribute__((packed)); +/* in kernel/trace/trace_mmiotrace.c */ +extern void enable_mmiotrace(void); +extern void disable_mmiotrace(void); +extern void mmio_trace_rw(struct mmiotrace_rw *rw); +extern void mmio_trace_mapping(struct mmiotrace_map *map); -struct mm_io_header_map { - struct mm_io_header header; - struct mm_io_map map; -} __attribute__((packed)); +#endif /* __KERNEL__ */ #endif /* MMIOTRACE_H */ diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 3271916ff033..d14fe49e9638 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -831,6 +831,40 @@ ftrace(struct trace_array *tr, struct trace_array_cpu *data, trace_function(tr, data, ip, parent_ip, flags); } +#ifdef CONFIG_MMIOTRACE +void __trace_mmiotrace_rw(struct trace_array *tr, struct trace_array_cpu *data, + struct mmiotrace_rw *rw) +{ + struct trace_entry *entry; + unsigned long irq_flags; + + spin_lock_irqsave(&data->lock, irq_flags); + entry = tracing_get_trace_entry(tr, data); + tracing_generic_entry_update(entry, 0); + entry->type = TRACE_MMIO_RW; + entry->mmiorw = *rw; + spin_unlock_irqrestore(&data->lock, irq_flags); + + trace_wake_up(); +} + +void __trace_mmiotrace_map(struct trace_array *tr, struct trace_array_cpu *data, + struct mmiotrace_map *map) +{ + struct trace_entry *entry; + unsigned long irq_flags; + + spin_lock_irqsave(&data->lock, irq_flags); + entry = tracing_get_trace_entry(tr, data); + tracing_generic_entry_update(entry, 0); + entry->type = TRACE_MMIO_MAP; + entry->mmiomap = *map; + spin_unlock_irqrestore(&data->lock, irq_flags); + + trace_wake_up(); +} +#endif + void __trace_stack(struct trace_array *tr, struct trace_array_cpu *data, unsigned long flags, diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index c460e85e94ed..0ef9ef74c806 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -5,6 +5,7 @@ #include #include #include +#include enum trace_type { __TRACE_FIRST_TYPE = 0, @@ -14,6 +15,8 @@ enum trace_type { TRACE_WAKE, TRACE_STACK, TRACE_SPECIAL, + TRACE_MMIO_RW, + TRACE_MMIO_MAP, __TRACE_LAST_TYPE }; @@ -75,6 +78,8 @@ struct trace_entry { struct ctx_switch_entry ctx; struct special_entry special; struct stack_entry stack; + struct mmiotrace_rw mmiorw; + struct mmiotrace_map mmiomap; }; }; @@ -255,6 +260,15 @@ extern unsigned long ftrace_update_tot_cnt; extern int DYN_FTRACE_TEST_NAME(void); #endif +#ifdef CONFIG_MMIOTRACE +extern void __trace_mmiotrace_rw(struct trace_array *tr, + struct trace_array_cpu *data, + struct mmiotrace_rw *rw); +extern void __trace_mmiotrace_map(struct trace_array *tr, + struct trace_array_cpu *data, + struct mmiotrace_map *map); +#endif + #ifdef CONFIG_FTRACE_STARTUP_TEST #ifdef CONFIG_FTRACE extern int trace_selftest_startup_function(struct tracer *trace, diff --git a/kernel/trace/trace_mmiotrace.c b/kernel/trace/trace_mmiotrace.c index e4dd03cc5aa6..3a12b1ad0c63 100644 --- a/kernel/trace/trace_mmiotrace.c +++ b/kernel/trace/trace_mmiotrace.c @@ -11,19 +11,26 @@ #include "trace.h" -extern void -__trace_special(void *__tr, void *__data, - unsigned long arg1, unsigned long arg2, unsigned long arg3); - static struct trace_array *mmio_trace_array; +static void mmio_reset_data(struct trace_array *tr) +{ + int cpu; + + tr->time_start = ftrace_now(tr->cpu); + + for_each_online_cpu(cpu) + tracing_reset(tr->data[cpu]); +} static void mmio_trace_init(struct trace_array *tr) { pr_debug("in %s\n", __func__); mmio_trace_array = tr; - if (tr->ctrl) + if (tr->ctrl) { + mmio_reset_data(tr); enable_mmiotrace(); + } } static void mmio_trace_reset(struct trace_array *tr) @@ -31,15 +38,110 @@ static void mmio_trace_reset(struct trace_array *tr) pr_debug("in %s\n", __func__); if (tr->ctrl) disable_mmiotrace(); + mmio_reset_data(tr); + mmio_trace_array = NULL; } static void mmio_trace_ctrl_update(struct trace_array *tr) { pr_debug("in %s\n", __func__); - if (tr->ctrl) + if (tr->ctrl) { + mmio_reset_data(tr); enable_mmiotrace(); - else + } else { disable_mmiotrace(); + } +} + +/* XXX: This is not called for trace_pipe file! */ +void mmio_print_header(struct trace_iterator *iter) +{ + struct trace_seq *s = &iter->seq; + trace_seq_printf(s, "VERSION broken 20070824\n"); + /* TODO: print /proc/bus/pci/devices contents as PCIDEV lines */ +} + +static int mmio_print_rw(struct trace_iterator *iter) +{ + struct trace_entry *entry = iter->ent; + struct mmiotrace_rw *rw = &entry->mmiorw; + struct trace_seq *s = &iter->seq; + unsigned long long t = ns2usecs(entry->t); + unsigned long usec_rem = do_div(t, 1000000ULL); + unsigned secs = (unsigned long)t; + int ret = 1; + + switch (entry->mmiorw.opcode) { + case MMIO_READ: + ret = trace_seq_printf(s, + "R %d %lu.%06lu %d 0x%lx 0x%lx 0x%lx %d\n", + rw->width, secs, usec_rem, rw->map_id, rw->phys, + rw->value, rw->pc, entry->pid); + break; + case MMIO_WRITE: + ret = trace_seq_printf(s, + "W %d %lu.%06lu %d 0x%lx 0x%lx 0x%lx %d\n", + rw->width, secs, usec_rem, rw->map_id, rw->phys, + rw->value, rw->pc, entry->pid); + break; + case MMIO_UNKNOWN_OP: + ret = trace_seq_printf(s, + "UNKNOWN %lu.%06lu %d 0x%lx %02x,%02x,%02x 0x%lx %d\n", + secs, usec_rem, rw->map_id, rw->phys, + (rw->value >> 16) & 0xff, (rw->value >> 8) & 0xff, + (rw->value >> 0) & 0xff, rw->pc, entry->pid); + break; + default: + ret = trace_seq_printf(s, "rw what?\n"); + break; + } + if (ret) + return 1; + return 0; +} + +static int mmio_print_map(struct trace_iterator *iter) +{ + struct trace_entry *entry = iter->ent; + struct mmiotrace_map *m = &entry->mmiomap; + struct trace_seq *s = &iter->seq; + unsigned long long t = ns2usecs(entry->t); + unsigned long usec_rem = do_div(t, 1000000ULL); + unsigned secs = (unsigned long)t; + int ret = 1; + + switch (entry->mmiorw.opcode) { + case MMIO_PROBE: + ret = trace_seq_printf(s, + "MAP %lu.%06lu %d 0x%lx 0x%lx 0x%lx 0x%lx %d\n", + secs, usec_rem, m->map_id, m->phys, m->virt, m->len, + 0UL, entry->pid); + break; + case MMIO_UNPROBE: + ret = trace_seq_printf(s, + "UNMAP %lu.%06lu %d 0x%lx %d\n", + secs, usec_rem, m->map_id, 0UL, entry->pid); + break; + default: + ret = trace_seq_printf(s, "map what?\n"); + break; + } + if (ret) + return 1; + return 0; +} + +/* return 0 to abort printing without consuming current entry in pipe mode */ +static int mmio_print_line(struct trace_iterator *iter) +{ + switch (iter->ent->type) { + case TRACE_MMIO_RW: + return mmio_print_rw(iter); + case TRACE_MMIO_MAP: + return mmio_print_map(iter); + default: + return 1; /* ignore unknown entries */ + } } static struct tracer mmio_tracer __read_mostly = @@ -47,38 +149,31 @@ static struct tracer mmio_tracer __read_mostly = .name = "mmiotrace", .init = mmio_trace_init, .reset = mmio_trace_reset, + .open = mmio_print_header, .ctrl_update = mmio_trace_ctrl_update, + .print_line = mmio_print_line, }; __init static int init_mmio_trace(void) { - int ret = init_mmiotrace(); - if (ret) - return ret; return register_tracer(&mmio_tracer); } device_initcall(init_mmio_trace); -void mmio_trace_record(u32 type, unsigned long addr, unsigned long arg) +void mmio_trace_rw(struct mmiotrace_rw *rw) { struct trace_array *tr = mmio_trace_array; struct trace_array_cpu *data = tr->data[smp_processor_id()]; + __trace_mmiotrace_rw(tr, data, rw); +} - if (!current || current->pid == 0) { - /* - * XXX: This is a problem. We need to able to record, no - * matter what. tracing_generic_entry_update() would crash. - */ - static unsigned limit; - if (limit++ < 12) - pr_err("Error in %s: no current.\n", __func__); - return; - } - if (!tr || !data) { - static unsigned limit; - if (limit++ < 12) - pr_err("%s: no tr or data\n", __func__); - return; - } - __trace_special(tr, data, type, addr, arg); +void mmio_trace_mapping(struct mmiotrace_map *map) +{ + struct trace_array *tr = mmio_trace_array; + struct trace_array_cpu *data; + + preempt_disable(); + data = tr->data[smp_processor_id()]; + __trace_mmiotrace_map(tr, data, map); + preempt_enable(); } -- cgit v1.2.3 From 138295373ccf7625fcb0218dfea114837983bc39 Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Mon, 12 May 2008 21:20:58 +0200 Subject: ftrace: mmiotrace update, #2 another weekend, another patch. This should apply on top of my previous patch from March 23rd. Summary of changes: - Print PCI device list in output header - work around recursive probe hits on SMP - refactor dis/arm_kmmio_fault_page() and add check for page levels - remove un/reference_kmmio(), the die notifier hook is registered permanently into the list - explicitly check for single stepping in die notifier callback I have tested this version on my UP Athlon64 desktop with Nouveau, and SMP Core 2 Duo laptop with the proprietary nvidia driver. Both systems are 64-bit. One previously unknown bug crept into daylight: the ftrace framework's output routines print the first entry last after buffer has wrapped around. The most important regressions compared to non-ftrace mmiotrace at this time are: - failure of trace_pipe file - illegal lines in output file - unaware of losing data due to buffer full Personally I'd like to see these three solved before submitting to mainline. Other issues may come up once we know when we lose events. Signed-off-by: Pekka Paalanen Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- arch/x86/kernel/mmiotrace/kmmio.c | 186 ++++++++++++++--------------------- arch/x86/kernel/mmiotrace/mmio-mod.c | 3 - include/linux/mmiotrace.h | 2 - kernel/trace/trace_mmiotrace.c | 47 ++++++++- 4 files changed, 120 insertions(+), 118 deletions(-) (limited to 'kernel') diff --git a/arch/x86/kernel/mmiotrace/kmmio.c b/arch/x86/kernel/mmiotrace/kmmio.c index efb467933087..cd0d95fe4fe6 100644 --- a/arch/x86/kernel/mmiotrace/kmmio.c +++ b/arch/x86/kernel/mmiotrace/kmmio.c @@ -5,15 +5,12 @@ * 2008 Pekka Paalanen */ -#include #include #include #include #include #include -#include #include -#include #include #include #include @@ -22,10 +19,9 @@ #include #include #include -#include #include -#include - +#include +#include #include #define KMMIO_PAGE_HASH_BITS 4 @@ -57,14 +53,9 @@ struct kmmio_context { int active; }; -static int kmmio_die_notifier(struct notifier_block *nb, unsigned long val, - void *args); - -static DEFINE_MUTEX(kmmio_init_mutex); static DEFINE_SPINLOCK(kmmio_lock); -/* These are protected by kmmio_lock */ -static int kmmio_initialized; +/* Protected by kmmio_lock */ unsigned int kmmio_count; /* Read-protected by RCU, write-protected by kmmio_lock. */ @@ -79,60 +70,6 @@ static struct list_head *kmmio_page_list(unsigned long page) /* Accessed per-cpu */ static DEFINE_PER_CPU(struct kmmio_context, kmmio_ctx); -/* protected by kmmio_init_mutex */ -static struct notifier_block nb_die = { - .notifier_call = kmmio_die_notifier -}; - -/** - * Makes sure kmmio is initialized and usable. - * This must be called before any other kmmio function defined here. - * May sleep. - */ -void reference_kmmio(void) -{ - mutex_lock(&kmmio_init_mutex); - spin_lock_irq(&kmmio_lock); - if (!kmmio_initialized) { - int i; - for (i = 0; i < KMMIO_PAGE_TABLE_SIZE; i++) - INIT_LIST_HEAD(&kmmio_page_table[i]); - if (register_die_notifier(&nb_die)) - BUG(); - } - kmmio_initialized++; - spin_unlock_irq(&kmmio_lock); - mutex_unlock(&kmmio_init_mutex); -} -EXPORT_SYMBOL_GPL(reference_kmmio); - -/** - * Clean up kmmio after use. This must be called for every call to - * reference_kmmio(). All probes registered after the corresponding - * reference_kmmio() must have been unregistered when calling this. - * May sleep. - */ -void unreference_kmmio(void) -{ - bool unreg = false; - - mutex_lock(&kmmio_init_mutex); - spin_lock_irq(&kmmio_lock); - - if (kmmio_initialized == 1) { - BUG_ON(is_kmmio_active()); - unreg = true; - } - kmmio_initialized--; - BUG_ON(kmmio_initialized < 0); - spin_unlock_irq(&kmmio_lock); - - if (unreg) - unregister_die_notifier(&nb_die); /* calls sync_rcu() */ - mutex_unlock(&kmmio_init_mutex); -} -EXPORT_SYMBOL(unreference_kmmio); - /* * this is basically a dynamic stabbing problem: * Could use the existing prio tree code or @@ -167,58 +104,56 @@ static struct kmmio_fault_page *get_kmmio_fault_page(unsigned long page) return NULL; } -/** Mark the given page as not present. Access to it will trigger a fault. */ -static void arm_kmmio_fault_page(unsigned long page, int *page_level) +static void set_page_present(unsigned long addr, bool present, int *pglevel) { - unsigned long address = page & PAGE_MASK; + pteval_t pteval; + pmdval_t pmdval; int level; - pte_t *pte = lookup_address(address, &level); + pmd_t *pmd; + pte_t *pte = lookup_address(addr, &level); if (!pte) { - pr_err("kmmio: Error in %s: no pte for page 0x%08lx\n", - __func__, page); + pr_err("kmmio: no pte for page 0x%08lx\n", addr); return; } - if (level == PG_LEVEL_2M) { - pmd_t *pmd = (pmd_t *)pte; - set_pmd(pmd, __pmd(pmd_val(*pmd) & ~_PAGE_PRESENT)); - } else { - /* PG_LEVEL_4K */ - set_pte(pte, __pte(pte_val(*pte) & ~_PAGE_PRESENT)); + if (pglevel) + *pglevel = level; + + switch (level) { + case PG_LEVEL_2M: + pmd = (pmd_t *)pte; + pmdval = pmd_val(*pmd) & ~_PAGE_PRESENT; + if (present) + pmdval |= _PAGE_PRESENT; + set_pmd(pmd, __pmd(pmdval)); + break; + + case PG_LEVEL_4K: + pteval = pte_val(*pte) & ~_PAGE_PRESENT; + if (present) + pteval |= _PAGE_PRESENT; + set_pte_atomic(pte, __pte(pteval)); + break; + + default: + pr_err("kmmio: unexpected page level 0x%x.\n", level); + return; } - if (page_level) - *page_level = level; + __flush_tlb_one(addr); +} - __flush_tlb_one(page); +/** Mark the given page as not present. Access to it will trigger a fault. */ +static void arm_kmmio_fault_page(unsigned long page, int *page_level) +{ + set_page_present(page & PAGE_MASK, false, page_level); } /** Mark the given page as present. */ static void disarm_kmmio_fault_page(unsigned long page, int *page_level) { - unsigned long address = page & PAGE_MASK; - int level; - pte_t *pte = lookup_address(address, &level); - - if (!pte) { - pr_err("kmmio: Error in %s: no pte for page 0x%08lx\n", - __func__, page); - return; - } - - if (level == PG_LEVEL_2M) { - pmd_t *pmd = (pmd_t *)pte; - set_pmd(pmd, __pmd(pmd_val(*pmd) | _PAGE_PRESENT)); - } else { - /* PG_LEVEL_4K */ - set_pte(pte, __pte(pte_val(*pte) | _PAGE_PRESENT)); - } - - if (page_level) - *page_level = level; - - __flush_tlb_one(page); + set_page_present(page & PAGE_MASK, true, page_level); } /* @@ -240,6 +175,7 @@ int kmmio_handler(struct pt_regs *regs, unsigned long addr) { struct kmmio_context *ctx; struct kmmio_fault_page *faultpage; + int ret = 0; /* default to fault not handled */ /* * Preemption is now disabled to prevent process switch during @@ -257,21 +193,35 @@ int kmmio_handler(struct pt_regs *regs, unsigned long addr) /* * Either this page fault is not caused by kmmio, or * another CPU just pulled the kmmio probe from under - * our feet. In the latter case all hell breaks loose. + * our feet. The latter case should not be possible. */ goto no_kmmio; } ctx = &get_cpu_var(kmmio_ctx); if (ctx->active) { + disarm_kmmio_fault_page(faultpage->page, NULL); + if (addr == ctx->addr) { + /* + * On SMP we sometimes get recursive probe hits on the + * same address. Context is already saved, fall out. + */ + pr_debug("kmmio: duplicate probe hit on CPU %d, for " + "address 0x%08lx.\n", + smp_processor_id(), addr); + ret = 1; + goto no_kmmio_ctx; + } /* * Prevent overwriting already in-flight context. - * If this page fault really was due to kmmio trap, - * all hell breaks loose. + * This should not happen, let's hope disarming at least + * prevents a panic. */ pr_emerg("kmmio: recursive probe hit on CPU %d, " "for address 0x%08lx. Ignoring.\n", smp_processor_id(), addr); + pr_emerg("kmmio: previous hit was at 0x%08lx.\n", + ctx->addr); goto no_kmmio_ctx; } ctx->active++; @@ -302,14 +252,14 @@ int kmmio_handler(struct pt_regs *regs, unsigned long addr) */ put_cpu_var(kmmio_ctx); - return 1; + return 1; /* fault handled */ no_kmmio_ctx: put_cpu_var(kmmio_ctx); no_kmmio: rcu_read_unlock(); preempt_enable_no_resched(); - return 0; /* page fault not handled by kmmio */ + return ret; } /* @@ -322,8 +272,11 @@ static int post_kmmio_handler(unsigned long condition, struct pt_regs *regs) int ret = 0; struct kmmio_context *ctx = &get_cpu_var(kmmio_ctx); - if (!ctx->active) + if (!ctx->active) { + pr_debug("kmmio: spurious debug trap on CPU %d.\n", + smp_processor_id()); goto out; + } if (ctx->probe && ctx->probe->post_handler) ctx->probe->post_handler(ctx->probe, condition, regs); @@ -525,9 +478,22 @@ static int kmmio_die_notifier(struct notifier_block *nb, unsigned long val, { struct die_args *arg = args; - if (val == DIE_DEBUG) + if (val == DIE_DEBUG && (arg->err & DR_STEP)) if (post_kmmio_handler(arg->err, arg->regs) == 1) return NOTIFY_STOP; return NOTIFY_DONE; } + +static struct notifier_block nb_die = { + .notifier_call = kmmio_die_notifier +}; + +static int __init init_kmmio(void) +{ + int i; + for (i = 0; i < KMMIO_PAGE_TABLE_SIZE; i++) + INIT_LIST_HEAD(&kmmio_page_table[i]); + return register_die_notifier(&nb_die); +} +fs_initcall(init_kmmio); /* should be before device_initcall() */ diff --git a/arch/x86/kernel/mmiotrace/mmio-mod.c b/arch/x86/kernel/mmiotrace/mmio-mod.c index 62abc281a512..8256546d49bf 100644 --- a/arch/x86/kernel/mmiotrace/mmio-mod.c +++ b/arch/x86/kernel/mmiotrace/mmio-mod.c @@ -415,8 +415,6 @@ void enable_mmiotrace(void) if (is_enabled()) goto out; - reference_kmmio(); - #if 0 /* XXX: tracing does not support text entries */ marker_file = debugfs_create_file("marker", 0660, dir, NULL, &fops_marker); @@ -448,7 +446,6 @@ void disable_mmiotrace(void) spin_unlock_irq(&trace_lock); clear_trace_list(); /* guarantees: no more kmmio callbacks */ - unreference_kmmio(); if (marker_file) { debugfs_remove(marker_file); marker_file = NULL; diff --git a/include/linux/mmiotrace.h b/include/linux/mmiotrace.h index c88a9c197d22..dd6b64b160fc 100644 --- a/include/linux/mmiotrace.h +++ b/include/linux/mmiotrace.h @@ -31,8 +31,6 @@ static inline int is_kmmio_active(void) return kmmio_count; } -extern void reference_kmmio(void); -extern void unreference_kmmio(void); extern int register_kmmio_probe(struct kmmio_probe *p); extern void unregister_kmmio_probe(struct kmmio_probe *p); diff --git a/kernel/trace/trace_mmiotrace.c b/kernel/trace/trace_mmiotrace.c index 3a12b1ad0c63..361472b5788c 100644 --- a/kernel/trace/trace_mmiotrace.c +++ b/kernel/trace/trace_mmiotrace.c @@ -8,6 +8,7 @@ #include #include +#include #include "trace.h" @@ -53,12 +54,52 @@ static void mmio_trace_ctrl_update(struct trace_array *tr) } } +static int mmio_print_pcidev(struct trace_seq *s, const struct pci_dev *dev) +{ + int ret = 0; + int i; + resource_size_t start, end; + const struct pci_driver *drv = pci_dev_driver(dev); + + /* XXX: incomplete checks for trace_seq_printf() return value */ + ret += trace_seq_printf(s, "PCIDEV %02x%02x %04x%04x %x", + dev->bus->number, dev->devfn, + dev->vendor, dev->device, dev->irq); + /* + * XXX: is pci_resource_to_user() appropriate, since we are + * supposed to interpret the __ioremap() phys_addr argument based on + * these printed values? + */ + for (i = 0; i < 7; i++) { + pci_resource_to_user(dev, i, &dev->resource[i], &start, &end); + ret += trace_seq_printf(s, " %llx", + (unsigned long long)(start | + (dev->resource[i].flags & PCI_REGION_FLAG_MASK))); + } + for (i = 0; i < 7; i++) { + pci_resource_to_user(dev, i, &dev->resource[i], &start, &end); + ret += trace_seq_printf(s, " %llx", + dev->resource[i].start < dev->resource[i].end ? + (unsigned long long)(end - start) + 1 : 0); + } + if (drv) + ret += trace_seq_printf(s, " %s\n", drv->name); + else + ret += trace_seq_printf(s, " \n"); + return ret; +} + /* XXX: This is not called for trace_pipe file! */ -void mmio_print_header(struct trace_iterator *iter) +static void mmio_print_header(struct trace_iterator *iter) { struct trace_seq *s = &iter->seq; - trace_seq_printf(s, "VERSION broken 20070824\n"); - /* TODO: print /proc/bus/pci/devices contents as PCIDEV lines */ + struct pci_dev *dev = NULL; + + trace_seq_printf(s, "VERSION 20070824\n"); + + for_each_pci_dev(dev) + mmio_print_pcidev(s, dev); + /* XXX: return value? What if header is very long? */ } static int mmio_print_rw(struct trace_iterator *iter) -- cgit v1.2.3 From 801a175bf601f9a9d5e86e92dee9adeeb6625da8 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:58 +0200 Subject: mmiotrace: ftrace fix Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 16 ++++++++++++---- 1 file changed, 12 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index d14fe49e9638..4dcc4e85c5d6 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -838,12 +838,16 @@ void __trace_mmiotrace_rw(struct trace_array *tr, struct trace_array_cpu *data, struct trace_entry *entry; unsigned long irq_flags; - spin_lock_irqsave(&data->lock, irq_flags); + raw_local_irq_save(irq_flags); + __raw_spin_lock(&data->lock); + entry = tracing_get_trace_entry(tr, data); tracing_generic_entry_update(entry, 0); entry->type = TRACE_MMIO_RW; entry->mmiorw = *rw; - spin_unlock_irqrestore(&data->lock, irq_flags); + + __raw_spin_unlock(&data->lock); + raw_local_irq_restore(irq_flags); trace_wake_up(); } @@ -854,12 +858,16 @@ void __trace_mmiotrace_map(struct trace_array *tr, struct trace_array_cpu *data, struct trace_entry *entry; unsigned long irq_flags; - spin_lock_irqsave(&data->lock, irq_flags); + raw_local_irq_save(irq_flags); + __raw_spin_lock(&data->lock); + entry = tracing_get_trace_entry(tr, data); tracing_generic_entry_update(entry, 0); entry->type = TRACE_MMIO_MAP; entry->mmiomap = *map; - spin_unlock_irqrestore(&data->lock, irq_flags); + + __raw_spin_unlock(&data->lock); + raw_local_irq_restore(irq_flags); trace_wake_up(); } -- cgit v1.2.3 From 736ca61fa81874b3fee205a593251b1869d0bcf1 Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Mon, 12 May 2008 21:20:59 +0200 Subject: x86 mmiotrace: Do not print bogus pid Non-zero pid indicates the MMIO access originated in user space. We do not catch that kind of accesses yet, so always print zero for now. Signed-off-by: Pekka Paalanen Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace_mmiotrace.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_mmiotrace.c b/kernel/trace/trace_mmiotrace.c index 361472b5788c..79be4a18ec1e 100644 --- a/kernel/trace/trace_mmiotrace.c +++ b/kernel/trace/trace_mmiotrace.c @@ -117,20 +117,20 @@ static int mmio_print_rw(struct trace_iterator *iter) ret = trace_seq_printf(s, "R %d %lu.%06lu %d 0x%lx 0x%lx 0x%lx %d\n", rw->width, secs, usec_rem, rw->map_id, rw->phys, - rw->value, rw->pc, entry->pid); + rw->value, rw->pc, 0); break; case MMIO_WRITE: ret = trace_seq_printf(s, "W %d %lu.%06lu %d 0x%lx 0x%lx 0x%lx %d\n", rw->width, secs, usec_rem, rw->map_id, rw->phys, - rw->value, rw->pc, entry->pid); + rw->value, rw->pc, 0); break; case MMIO_UNKNOWN_OP: ret = trace_seq_printf(s, "UNKNOWN %lu.%06lu %d 0x%lx %02x,%02x,%02x 0x%lx %d\n", secs, usec_rem, rw->map_id, rw->phys, (rw->value >> 16) & 0xff, (rw->value >> 8) & 0xff, - (rw->value >> 0) & 0xff, rw->pc, entry->pid); + (rw->value >> 0) & 0xff, rw->pc, 0); break; default: ret = trace_seq_printf(s, "rw what?\n"); -- cgit v1.2.3 From d0a7e8ca5b996d36219e6fc002907291c8ee677b Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Mon, 12 May 2008 21:21:02 +0200 Subject: mmiotrace: print header using the read hook. Now the header is printed only for `trace_pipe' file. Signed-off-by: Pekka Paalanen Signed-off-by: Ingo Molnar --- kernel/trace/trace_mmiotrace.c | 60 +++++++++++++++++++++++++++++++++++++----- 1 file changed, 53 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_mmiotrace.c b/kernel/trace/trace_mmiotrace.c index 79be4a18ec1e..6d2edbdde939 100644 --- a/kernel/trace/trace_mmiotrace.c +++ b/kernel/trace/trace_mmiotrace.c @@ -12,6 +12,10 @@ #include "trace.h" +struct header_iter { + struct pci_dev *dev; +}; + static struct trace_array *mmio_trace_array; static void mmio_reset_data(struct trace_array *tr) @@ -89,17 +93,57 @@ static int mmio_print_pcidev(struct trace_seq *s, const struct pci_dev *dev) return ret; } -/* XXX: This is not called for trace_pipe file! */ -static void mmio_print_header(struct trace_iterator *iter) +static void destroy_header_iter(struct header_iter *hiter) +{ + if (!hiter) + return; + pci_dev_put(hiter->dev); + kfree(hiter); +} + +static void mmio_pipe_open(struct trace_iterator *iter) { + struct header_iter *hiter; struct trace_seq *s = &iter->seq; - struct pci_dev *dev = NULL; trace_seq_printf(s, "VERSION 20070824\n"); - for_each_pci_dev(dev) - mmio_print_pcidev(s, dev); - /* XXX: return value? What if header is very long? */ + hiter = kzalloc(sizeof(*hiter), GFP_KERNEL); + if (!hiter) + return; + + hiter->dev = pci_get_device(PCI_ANY_ID, PCI_ANY_ID, NULL); + iter->private = hiter; +} + +/* XXX: This is not called when the pipe is closed! */ +static void mmio_close(struct trace_iterator *iter) +{ + struct header_iter *hiter = iter->private; + destroy_header_iter(hiter); + iter->private = NULL; +} + +static ssize_t mmio_read(struct trace_iterator *iter, struct file *filp, + char __user *ubuf, size_t cnt, loff_t *ppos) +{ + ssize_t ret; + struct header_iter *hiter = iter->private; + struct trace_seq *s = &iter->seq; + + if (!hiter) + return 0; + + mmio_print_pcidev(s, hiter->dev); + hiter->dev = pci_get_device(PCI_ANY_ID, PCI_ANY_ID, hiter->dev); + + if (!hiter->dev) { + destroy_header_iter(hiter); + iter->private = NULL; + } + + ret = trace_seq_to_user(s, ubuf, cnt); + return (ret == -EBUSY) ? 0 : ret; } static int mmio_print_rw(struct trace_iterator *iter) @@ -190,7 +234,9 @@ static struct tracer mmio_tracer __read_mostly = .name = "mmiotrace", .init = mmio_trace_init, .reset = mmio_trace_reset, - .open = mmio_print_header, + .pipe_open = mmio_pipe_open, + .close = mmio_close, + .read = mmio_read, .ctrl_update = mmio_trace_ctrl_update, .print_line = mmio_print_line, }; -- cgit v1.2.3 From 2039238b79b51a50f8477f53f33750e1c3fc146a Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Mon, 12 May 2008 21:21:02 +0200 Subject: mmiotrace: print overrun counts Signed-off-by: Pekka Paalanen Signed-off-by: Ingo Molnar --- kernel/trace/trace_mmiotrace.c | 25 +++++++++++++++++++++++++ 1 file changed, 25 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/trace_mmiotrace.c b/kernel/trace/trace_mmiotrace.c index 6d2edbdde939..d0f649a2203d 100644 --- a/kernel/trace/trace_mmiotrace.c +++ b/kernel/trace/trace_mmiotrace.c @@ -17,11 +17,13 @@ struct header_iter { }; static struct trace_array *mmio_trace_array; +static bool overrun_detected; static void mmio_reset_data(struct trace_array *tr) { int cpu; + overrun_detected = false; tr->time_start = ftrace_now(tr->cpu); for_each_online_cpu(cpu) @@ -124,12 +126,34 @@ static void mmio_close(struct trace_iterator *iter) iter->private = NULL; } +static unsigned long count_overruns(struct trace_iterator *iter) +{ + int cpu; + unsigned long cnt = 0; + for_each_online_cpu(cpu) { + cnt += iter->overrun[cpu]; + iter->overrun[cpu] = 0; + } + return cnt; +} + static ssize_t mmio_read(struct trace_iterator *iter, struct file *filp, char __user *ubuf, size_t cnt, loff_t *ppos) { ssize_t ret; struct header_iter *hiter = iter->private; struct trace_seq *s = &iter->seq; + unsigned long n; + + n = count_overruns(iter); + if (n) { + /* XXX: This is later than where events were lost. */ + trace_seq_printf(s, "MARK 0.000000 Lost %lu events.\n", n); + if (!overrun_detected) + pr_warning("mmiotrace has lost events.\n"); + overrun_detected = true; + goto print_out; + } if (!hiter) return 0; @@ -142,6 +166,7 @@ static ssize_t mmio_read(struct trace_iterator *iter, struct file *filp, iter->private = NULL; } +print_out: ret = trace_seq_to_user(s, ubuf, cnt); return (ret == -EBUSY) ? 0 : ret; } -- cgit v1.2.3 From e0fd5c2fa188311667267c02a702ae699a9fc2bd Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Mon, 12 May 2008 21:21:02 +0200 Subject: mmiotrace: do not print bogus pid for maps either Signed-off-by: Pekka Paalanen Signed-off-by: Ingo Molnar --- kernel/trace/trace_mmiotrace.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_mmiotrace.c b/kernel/trace/trace_mmiotrace.c index d0f649a2203d..3c1dacdc2d85 100644 --- a/kernel/trace/trace_mmiotrace.c +++ b/kernel/trace/trace_mmiotrace.c @@ -225,12 +225,12 @@ static int mmio_print_map(struct trace_iterator *iter) ret = trace_seq_printf(s, "MAP %lu.%06lu %d 0x%lx 0x%lx 0x%lx 0x%lx %d\n", secs, usec_rem, m->map_id, m->phys, m->virt, m->len, - 0UL, entry->pid); + 0UL, 0); break; case MMIO_UNPROBE: ret = trace_seq_printf(s, "UNMAP %lu.%06lu %d 0x%lx %d\n", - secs, usec_rem, m->map_id, 0UL, entry->pid); + secs, usec_rem, m->map_id, 0UL, 0); break; default: ret = trace_seq_printf(s, "map what?\n"); -- cgit v1.2.3 From dee310d0adf41019aca476052ac3085ff286d9be Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Mon, 12 May 2008 21:21:03 +0200 Subject: x86 mmiotrace: use resource_size_t for phys addresses Signed-off-by: Pekka Paalanen Signed-off-by: Ingo Molnar --- arch/x86/mm/mmio-mod.c | 11 ++++++----- include/linux/mmiotrace.h | 14 +++++++------- kernel/trace/trace_mmiotrace.c | 20 ++++++++++++-------- 3 files changed, 25 insertions(+), 20 deletions(-) (limited to 'kernel') diff --git a/arch/x86/mm/mmio-mod.c b/arch/x86/mm/mmio-mod.c index 278998c1998f..3b04a0126121 100644 --- a/arch/x86/mm/mmio-mod.c +++ b/arch/x86/mm/mmio-mod.c @@ -48,7 +48,7 @@ struct trap_reason { struct remap_trace { struct list_head list; struct kmmio_probe probe; - unsigned long phys; + resource_size_t phys; unsigned long id; }; @@ -275,7 +275,7 @@ static void post(struct kmmio_probe *p, unsigned long condition, put_cpu_var(pf_reason); } -static void ioremap_trace_core(unsigned long offset, unsigned long size, +static void ioremap_trace_core(resource_size_t offset, unsigned long size, void __iomem *addr) { static atomic_t next_id; @@ -319,13 +319,14 @@ not_enabled: spin_unlock_irq(&trace_lock); } -void -mmiotrace_ioremap(unsigned long offset, unsigned long size, void __iomem *addr) +void mmiotrace_ioremap(resource_size_t offset, unsigned long size, + void __iomem *addr) { if (!is_enabled()) /* recheck and proper locking in *_core() */ return; - pr_debug(NAME "ioremap_*(0x%lx, 0x%lx) = %p\n", offset, size, addr); + pr_debug(NAME "ioremap_*(0x%llx, 0x%lx) = %p\n", + (unsigned long long)offset, size, addr); if ((filter_offset) && (offset != filter_offset)) return; ioremap_trace_core(offset, size, addr); diff --git a/include/linux/mmiotrace.h b/include/linux/mmiotrace.h index de8e91258da7..5cbbc374e945 100644 --- a/include/linux/mmiotrace.h +++ b/include/linux/mmiotrace.h @@ -2,7 +2,6 @@ #define MMIOTRACE_H #include - #include struct kmmio_probe; @@ -37,14 +36,15 @@ extern int kmmio_handler(struct pt_regs *regs, unsigned long addr); /* Called from ioremap.c */ #ifdef CONFIG_MMIOTRACE -extern void -mmiotrace_ioremap(unsigned long offset, unsigned long size, void __iomem *addr); +extern void mmiotrace_ioremap(resource_size_t offset, unsigned long size, + void __iomem *addr); extern void mmiotrace_iounmap(volatile void __iomem *addr); #else -static inline void -mmiotrace_ioremap(unsigned long offset, unsigned long size, void __iomem *addr) +static inline void mmiotrace_ioremap(resource_size_t offset, + unsigned long size, void __iomem *addr) { } + static inline void mmiotrace_iounmap(volatile void __iomem *addr) { } @@ -60,7 +60,7 @@ enum mm_io_opcode { }; struct mmiotrace_rw { - unsigned long phys; /* PCI address of register */ + resource_size_t phys; /* PCI address of register */ unsigned long value; unsigned long pc; /* optional program counter */ int map_id; @@ -69,7 +69,7 @@ struct mmiotrace_rw { }; struct mmiotrace_map { - unsigned long phys; /* base address in PCI space */ + resource_size_t phys; /* base address in PCI space */ unsigned long virt; /* base virtual address */ unsigned long len; /* mapping size */ int map_id; diff --git a/kernel/trace/trace_mmiotrace.c b/kernel/trace/trace_mmiotrace.c index 3c1dacdc2d85..b13dc19dcbb4 100644 --- a/kernel/trace/trace_mmiotrace.c +++ b/kernel/trace/trace_mmiotrace.c @@ -184,20 +184,23 @@ static int mmio_print_rw(struct trace_iterator *iter) switch (entry->mmiorw.opcode) { case MMIO_READ: ret = trace_seq_printf(s, - "R %d %lu.%06lu %d 0x%lx 0x%lx 0x%lx %d\n", - rw->width, secs, usec_rem, rw->map_id, rw->phys, + "R %d %lu.%06lu %d 0x%llx 0x%lx 0x%lx %d\n", + rw->width, secs, usec_rem, rw->map_id, + (unsigned long long)rw->phys, rw->value, rw->pc, 0); break; case MMIO_WRITE: ret = trace_seq_printf(s, - "W %d %lu.%06lu %d 0x%lx 0x%lx 0x%lx %d\n", - rw->width, secs, usec_rem, rw->map_id, rw->phys, + "W %d %lu.%06lu %d 0x%llx 0x%lx 0x%lx %d\n", + rw->width, secs, usec_rem, rw->map_id, + (unsigned long long)rw->phys, rw->value, rw->pc, 0); break; case MMIO_UNKNOWN_OP: ret = trace_seq_printf(s, - "UNKNOWN %lu.%06lu %d 0x%lx %02x,%02x,%02x 0x%lx %d\n", - secs, usec_rem, rw->map_id, rw->phys, + "UNKNOWN %lu.%06lu %d 0x%llx %02x,%02x,%02x 0x%lx %d\n", + secs, usec_rem, rw->map_id, + (unsigned long long)rw->phys, (rw->value >> 16) & 0xff, (rw->value >> 8) & 0xff, (rw->value >> 0) & 0xff, rw->pc, 0); break; @@ -223,8 +226,9 @@ static int mmio_print_map(struct trace_iterator *iter) switch (entry->mmiorw.opcode) { case MMIO_PROBE: ret = trace_seq_printf(s, - "MAP %lu.%06lu %d 0x%lx 0x%lx 0x%lx 0x%lx %d\n", - secs, usec_rem, m->map_id, m->phys, m->virt, m->len, + "MAP %lu.%06lu %d 0x%llx 0x%lx 0x%lx 0x%lx %d\n", + secs, usec_rem, m->map_id, + (unsigned long long)m->phys, m->virt, m->len, 0UL, 0); break; case MMIO_UNPROBE: -- cgit v1.2.3