From 0bb943c7a2136716757a263f604d26309fd98042 Mon Sep 17 00:00:00 2001 From: Julia Lawall Date: Fri, 14 Nov 2008 19:05:31 +0100 Subject: tracing: kernel/trace/trace.c: introduce missing kfree() Impact: fix memory leak Error handling code following a kzalloc should free the allocated data. The semantic match that finds the problem is as follows: (http://www.emn.fr/x-info/coccinelle/) // @r exists@ local idexpression x; statement S; expression E; identifier f,l; position p1,p2; expression *ptr != NULL; @@ ( if ((x@p1 = \(kmalloc\|kzalloc\|kcalloc\)(...)) == NULL) S | x@p1 = \(kmalloc\|kzalloc\|kcalloc\)(...); ... if (x == NULL) S ) <... when != x when != if (...) { <+...x...+> } x->f = E ...> ( return \(0\|<+...x...+>\|ptr\); | return@p2 ...; ) @script:python@ p1 << r.p1; p2 << r.p2; @@ print "* file: %s kmalloc %s return %s" % (p1[0].file,p1[0].line,p2[0].line) // Signed-off-by: Julia Lawall Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- kernel/trace/trace.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel/trace') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 697eda36b86..d86e3252f30 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -1936,6 +1936,7 @@ __tracing_open(struct inode *inode, struct file *file, int *ret) ring_buffer_read_finish(iter->buffer_iter[cpu]); } mutex_unlock(&trace_types_lock); + kfree(iter); return ERR_PTR(-ENOMEM); } -- cgit v1.2.3 From 641d2f63cfe24539e154efa2f932937934c27dde Mon Sep 17 00:00:00 2001 From: Vegard Nossum Date: Tue, 18 Nov 2008 19:22:13 +0100 Subject: trace: introduce missing mutex_unlock() Impact: fix tracing buffer mutex leak in case of allocation failure This error was spotted by this semantic patch: http://www.emn.fr/x-info/coccinelle/mut.html It looks correct as far as I can tell. Please review. Signed-off-by: Vegard Nossum Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- kernel/trace/ring_buffer.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel/trace') diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index 036456cbb4f..f780e9552f9 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -617,6 +617,7 @@ int ring_buffer_resize(struct ring_buffer *buffer, unsigned long size) list_del_init(&page->list); free_buffer_page(page); } + mutex_unlock(&buffer->mutex); return -ENOMEM; } -- cgit v1.2.3 From a22506347d038a66506c6f57e9b97104128e280d Mon Sep 17 00:00:00 2001 From: Heiko Carstens Date: Tue, 18 Nov 2008 18:06:35 +0100 Subject: ftrace: preemptoff selftest not working Impact: fix preemptoff and preemptirqsoff tracer self-tests I was wondering why the preemptoff and preemptirqsoff tracer selftests don't work on s390. After all its just that they get called from non-preemptible context: kernel_init() will execute all initcalls, however the first line in kernel_init() is lock_kernel(), which causes the preempt_count to be increased. Any later calls to add_preempt_count() (especially those from the selftests) will therefore not result in a call to trace_preempt_off() since the check below in add_preempt_count() will be false: if (preempt_count() == val) trace_preempt_off(CALLER_ADDR0, get_parent_ip(CALLER_ADDR1)); Hence the trace buffer will be empty. Fix this by releasing the BKL during the self-tests. Signed-off-by: Heiko Carstens Acked-by: Steven Rostedt Signed-off-by: Ingo Molnar --- kernel/trace/trace.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'kernel/trace') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 396fda034e3..16892121cb7 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -532,6 +532,13 @@ int register_tracer(struct tracer *type) } #ifdef CONFIG_FTRACE_STARTUP_TEST + /* + * When this gets called we hold the BKL which means that preemption + * is disabled. Various trace selftests however need to disable + * and enable preemption for successful tests. So we drop the BKL here + * and grab it after the tests again. + */ + unlock_kernel(); if (type->selftest) { struct tracer *saved_tracer = current_trace; struct trace_array *tr = &global_trace; @@ -562,6 +569,7 @@ int register_tracer(struct tracer *type) } printk(KERN_CONT "PASSED\n"); } + lock_kernel(); #endif type->next = trace_types; -- cgit v1.2.3 From f10ed36ec1118c6f9523cd7e53cb0aadb53efe9f Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Fri, 7 Nov 2008 22:36:02 -0500 Subject: ftrace: fix set_ftrace_filter Impact: fix of output of set_ftrace_filter The commit "ftrace: do not show freed records in available_filter_functions" Removed a bit too much from the set_ftrace_filter code, where we now see all functions in the set_ftrace_filter file even when we set a filter. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- kernel/trace/ftrace.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'kernel/trace') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 4a39d24568c..dcac7418f68 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -738,6 +738,9 @@ t_next(struct seq_file *m, void *v, loff_t *pos) ((iter->flags & FTRACE_ITER_FAILURES) && !(rec->flags & FTRACE_FL_FAILED)) || + ((iter->flags & FTRACE_ITER_FILTER) && + !(rec->flags & FTRACE_FL_FILTER)) || + ((iter->flags & FTRACE_ITER_NOTRACE) && !(rec->flags & FTRACE_FL_NOTRACE))) { rec = NULL; -- cgit v1.2.3 From 820432783190b4096499e38a4a4d7095c511913d Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Tue, 18 Nov 2008 23:57:14 -0500 Subject: ftrace: make filtered functions effective on setting Impact: fix filter selection to apply when set It can be confusing when the set_filter_functions is set (or cleared) and the functions being recorded by the dynamic tracer does not match. This patch causes the code to be updated if the function tracer is enabled and the filter is changed. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- kernel/trace/ftrace.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel/trace') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index dcac7418f68..5cbddb59e99 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -1189,7 +1189,7 @@ ftrace_regex_release(struct inode *inode, struct file *file, int enable) mutex_lock(&ftrace_sysctl_lock); mutex_lock(&ftrace_start_lock); - if (iter->filtered && ftrace_start && ftrace_enabled) + if (ftrace_start && ftrace_enabled) ftrace_run_update_code(FTRACE_ENABLE_CALLS); mutex_unlock(&ftrace_start_lock); mutex_unlock(&ftrace_sysctl_lock); -- cgit v1.2.3 From 32464779a1b8c15e9aa9aa0306b2f735080df9d8 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Tue, 18 Nov 2008 20:33:02 -0500 Subject: ftrace: fix dyn ftrace filter selection Impact: clean up and fix for dyn ftrace filter selection The previous logic of the dynamic ftrace selection of enabling or disabling functions was complex and incorrect. This patch simplifies the code and corrects the usage. This simplification also makes the code more robust. Here is the correct logic: Given a function that can be traced by dynamic ftrace: If the function is not to be traced, disable it if it was enabled. (this is if the function is in the set_ftrace_notrace file) (filter is on if there exists any functions in set_ftrace_filter file) If the filter is on, and we are enabling functions: If the function is in set_ftrace_filter, enable it if it is not already enabled. If the function is not in set_ftrace_filter, disable it if it is not already disabled. Otherwise, if the filter is off and we are enabling function tracing: Enable the function if it is not already enabled. Otherwise, if we are disabling function tracing: Disable the function if it is not already disabled. This code now sets or clears the ENABLED flag in the record, and at the end it will enable the function if the flag is set, or disable the function if the flag is cleared. The parameters for the function that does the above logic is also simplified. Instead of passing in confusing "new" and "old" where they might be swapped if the "enabled" flag is not set. The old logic even had one of the above always NULL and had to be filled in. The new logic simply passes in one parameter called "nop". A "call" is calculated in the code, and at the end of the logic, when we know we need to either disable or enable the function, we can then use the "nop" and "call" properly. This code is more robust than the previous version. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- kernel/trace/ftrace.c | 108 +++++++++++++++++++++++--------------------------- 1 file changed, 50 insertions(+), 58 deletions(-) (limited to 'kernel/trace') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 5cbddb59e99..fdaab04a028 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -327,96 +327,89 @@ ftrace_record_ip(unsigned long ip) static int __ftrace_replace_code(struct dyn_ftrace *rec, - unsigned char *old, unsigned char *new, int enable) + unsigned char *nop, int enable) { unsigned long ip, fl; + unsigned char *call, *old, *new; ip = rec->ip; - if (ftrace_filtered && enable) { + /* + * If this record is not to be traced and + * it is not enabled then do nothing. + * + * If this record is not to be traced and + * it is enabled then disabled it. + * + */ + if (rec->flags & FTRACE_FL_NOTRACE) { + if (rec->flags & FTRACE_FL_ENABLED) + rec->flags &= ~FTRACE_FL_ENABLED; + else + return 0; + + } else if (ftrace_filtered && enable) { /* - * If filtering is on: - * - * If this record is set to be filtered and - * is enabled then do nothing. - * - * If this record is set to be filtered and - * it is not enabled, enable it. - * - * If this record is not set to be filtered - * and it is not enabled do nothing. - * - * If this record is set not to trace then - * do nothing. - * - * If this record is set not to trace and - * it is enabled then disable it. - * - * If this record is not set to be filtered and - * it is enabled, disable it. + * Filtering is on: */ - fl = rec->flags & (FTRACE_FL_FILTER | FTRACE_FL_NOTRACE | - FTRACE_FL_ENABLED); + fl = rec->flags & (FTRACE_FL_FILTER | FTRACE_FL_ENABLED); - if ((fl == (FTRACE_FL_FILTER | FTRACE_FL_ENABLED)) || - (fl == (FTRACE_FL_FILTER | FTRACE_FL_NOTRACE)) || - !fl || (fl == FTRACE_FL_NOTRACE)) + /* Record is filtered and enabled, do nothing */ + if (fl == (FTRACE_FL_FILTER | FTRACE_FL_ENABLED)) return 0; - /* - * If it is enabled disable it, - * otherwise enable it! - */ - if (fl & FTRACE_FL_ENABLED) { - /* swap new and old */ - new = old; - old = ftrace_call_replace(ip, FTRACE_ADDR); + /* Record is not filtered and is not enabled do nothing */ + if (!fl) + return 0; + + /* Record is not filtered but enabled, disable it */ + if (fl == FTRACE_FL_ENABLED) rec->flags &= ~FTRACE_FL_ENABLED; - } else { - new = ftrace_call_replace(ip, FTRACE_ADDR); + else + /* Otherwise record is filtered but not enabled, enable it */ rec->flags |= FTRACE_FL_ENABLED; - } } else { + /* Disable or not filtered */ if (enable) { - /* - * If this record is set not to trace and is - * not enabled, do nothing. - */ - fl = rec->flags & (FTRACE_FL_NOTRACE | FTRACE_FL_ENABLED); - if (fl == FTRACE_FL_NOTRACE) - return 0; - - new = ftrace_call_replace(ip, FTRACE_ADDR); - } else - old = ftrace_call_replace(ip, FTRACE_ADDR); - - if (enable) { + /* if record is enabled, do nothing */ if (rec->flags & FTRACE_FL_ENABLED) return 0; + rec->flags |= FTRACE_FL_ENABLED; + } else { + + /* if record is not enabled do nothing */ if (!(rec->flags & FTRACE_FL_ENABLED)) return 0; + rec->flags &= ~FTRACE_FL_ENABLED; } } + call = ftrace_call_replace(ip, FTRACE_ADDR); + + if (rec->flags & FTRACE_FL_ENABLED) { + old = nop; + new = call; + } else { + old = call; + new = nop; + } + return ftrace_modify_code(ip, old, new); } static void ftrace_replace_code(int enable) { int i, failed; - unsigned char *new = NULL, *old = NULL; + unsigned char *nop = NULL; struct dyn_ftrace *rec; struct ftrace_page *pg; - if (enable) - old = ftrace_nop_replace(); - else - new = ftrace_nop_replace(); + nop = ftrace_nop_replace(); for (pg = ftrace_pages_start; pg; pg = pg->next) { for (i = 0; i < pg->index; i++) { @@ -434,7 +427,7 @@ static void ftrace_replace_code(int enable) unfreeze_record(rec); } - failed = __ftrace_replace_code(rec, old, new, enable); + failed = __ftrace_replace_code(rec, nop, enable); if (failed && (rec->flags & FTRACE_FL_CONVERTED)) { rec->flags |= FTRACE_FL_FAILED; if ((system_state == SYSTEM_BOOTING) || @@ -538,8 +531,7 @@ static void ftrace_startup(void) mutex_lock(&ftrace_start_lock); ftrace_start++; - if (ftrace_start == 1) - command |= FTRACE_ENABLE_CALLS; + command |= FTRACE_ENABLE_CALLS; if (saved_ftrace_func != ftrace_trace_function) { saved_ftrace_func = ftrace_trace_function; -- cgit v1.2.3 From 86fa2f60674540df0b34f5c547ed0c1cf3a8f212 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Wed, 19 Nov 2008 10:00:15 +0100 Subject: ftrace: fix selftest locking Impact: fix self-test boot crash Self-test failure forgot to re-lock the BKL - crashing the next initcall: Testing tracer irqsoff: .. no entries found ..FAILED! initcall init_irqsoff_tracer+0x0/0x11 returned 0 after 3906 usecs calling init_mmio_trace+0x0/0xf @ 1 ------------[ cut here ]------------ Kernel BUG at c0c0a915 [verbose debug info unavailable] invalid opcode: 0000 [#1] SMP DEBUG_PAGEALLOC last sysfs file: Pid: 1, comm: swapper Not tainted (2.6.28-rc5-tip #53704) EIP: 0060:[] EFLAGS: 00010286 CPU: 1 EIP is at unlock_kernel+0x10/0x2b EAX: ffffffff EBX: 00000000 ECX: 00000000 EDX: f7030000 ESI: c12da19c EDI: 00000000 EBP: f7039f54 ESP: f7039f54 DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068 Process swapper (pid: 1, ti=f7038000 task=f7030000 task.ti=f7038000) Stack: f7039f6c c0164d30 c013fed8 a7d8d7b4 00000000 00000000 f7039f74 c12fb78a f7039fd0 c0101132 c12fb77d 00000000 6f727200 6f632072 2d206564 c1002031 0000000f f7039fa2 f7039fb0 3531b171 00000000 00000000 0000002f c12ca480 Call Trace: [] ? register_tracer+0x66/0x13f [] ? ktime_get+0x19/0x1b [] ? init_mmio_trace+0xd/0xf [] ? do_one_initcall+0x4a/0x111 [] ? init_mmio_trace+0x0/0xf [] ? init_irq_proc+0x46/0x59 [] ? kernel_init+0x104/0x152 [] ? kernel_init+0x0/0x152 [] ? kernel_thread_helper+0x7/0x10 Code: 58 14 43 75 0a b8 00 9b 2d c1 e8 51 43 7a ff 64 a1 00 a0 37 c1 89 58 14 5b 5d c3 55 64 8b 15 00 a0 37 c1 83 7a 14 00 89 e5 79 04 <0f> 0b eb fe 8b 42 14 48 85 c0 89 42 14 79 0a b8 00 9b 2d c1 e8 EIP: [] unlock_kernel+0x10/0x2b SS:ESP 0068:f7039f54 ---[ end trace a7919e7f17c0a725 ]--- Kernel panic - not syncing: Attempted to kill init! So clean up the flow a bit. Signed-off-by: Ingo Molnar --- kernel/trace/trace.c | 25 +++++++++++++------------ 1 file changed, 13 insertions(+), 12 deletions(-) (limited to 'kernel/trace') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 16892121cb7..24b6238884f 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -520,7 +520,15 @@ int register_tracer(struct tracer *type) return -1; } + /* + * When this gets called we hold the BKL which means that + * preemption is disabled. Various trace selftests however + * need to disable and enable preemption for successful tests. + * So we drop the BKL here and grab it after the tests again. + */ + unlock_kernel(); mutex_lock(&trace_types_lock); + for (t = trace_types; t; t = t->next) { if (strcmp(type->name, t->name) == 0) { /* already found */ @@ -532,13 +540,6 @@ int register_tracer(struct tracer *type) } #ifdef CONFIG_FTRACE_STARTUP_TEST - /* - * When this gets called we hold the BKL which means that preemption - * is disabled. Various trace selftests however need to disable - * and enable preemption for successful tests. So we drop the BKL here - * and grab it after the tests again. - */ - unlock_kernel(); if (type->selftest) { struct tracer *saved_tracer = current_trace; struct trace_array *tr = &global_trace; @@ -550,9 +551,9 @@ int register_tracer(struct tracer *type) * internal tracing to verify that everything is in order. * If we fail, we do not register this tracer. */ - for_each_tracing_cpu(i) { + for_each_tracing_cpu(i) tracing_reset(tr, i); - } + current_trace = type; /* the test is responsible for initializing and enabling */ pr_info("Testing tracer %s: ", type->name); @@ -564,12 +565,11 @@ int register_tracer(struct tracer *type) goto out; } /* Only reset on passing, to avoid touching corrupted buffers */ - for_each_tracing_cpu(i) { + for_each_tracing_cpu(i) tracing_reset(tr, i); - } + printk(KERN_CONT "PASSED\n"); } - lock_kernel(); #endif type->next = trace_types; @@ -580,6 +580,7 @@ int register_tracer(struct tracer *type) out: mutex_unlock(&trace_types_lock); + lock_kernel(); return ret; } -- cgit v1.2.3