From e85c5e9792b942381ad92ccd0ff745b6d408a91f Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 20 Feb 2025 15:20:14 -0500 Subject: [PATCH 01/16] selftests/ftrace: Update fprobe test to check enabled_functions file A few bugs were found in the fprobe accounting logic along with it using the function graph infrastructure. Update the fprobe selftest to catch those bugs in case they or something similar shows up in the future. The test now checks the enabled_functions file which shows all the functions attached to ftrace or fgraph. When enabling a fprobe, make sure that its corresponding function is also added to that file. Also add two more fprobes to enable to make sure that the fprobe logic works properly with multiple probes. Cc: Mark Rutland Cc: Mathieu Desnoyers Cc: Andrew Morton Cc: Sven Schnelle Cc: Vasily Gorbik Cc: Alexander Gordeev Link: https://lore.kernel.org/20250220202055.733001756@goodmis.org Acked-by: Masami Hiramatsu (Google) Tested-by: Heiko Carstens Signed-off-by: Steven Rostedt (Google) --- .../test.d/dynevent/add_remove_fprobe.tc | 54 +++++++++++++++++++ 1 file changed, 54 insertions(+) diff --git a/tools/testing/selftests/ftrace/test.d/dynevent/add_remove_fprobe.tc b/tools/testing/selftests/ftrace/test.d/dynevent/add_remove_fprobe.tc index dc25bcf4f9e2..449f9d8be746 100644 --- a/tools/testing/selftests/ftrace/test.d/dynevent/add_remove_fprobe.tc +++ b/tools/testing/selftests/ftrace/test.d/dynevent/add_remove_fprobe.tc @@ -7,12 +7,38 @@ echo 0 > events/enable echo > dynamic_events PLACE=$FUNCTION_FORK +PLACE2="kmem_cache_free" +PLACE3="schedule_timeout" echo "f:myevent1 $PLACE" >> dynamic_events + +# Make sure the event is attached and is the only one +grep -q $PLACE enabled_functions +cnt=`cat enabled_functions | wc -l` +if [ $cnt -ne 1 ]; then + exit_fail +fi + echo "f:myevent2 $PLACE%return" >> dynamic_events +# It should till be the only attached function +cnt=`cat enabled_functions | wc -l` +if [ $cnt -ne 1 ]; then + exit_fail +fi + +# add another event +echo "f:myevent3 $PLACE2" >> dynamic_events + +grep -q $PLACE2 enabled_functions +cnt=`cat enabled_functions | wc -l` +if [ $cnt -ne 2 ]; then + exit_fail +fi + grep -q myevent1 dynamic_events grep -q myevent2 dynamic_events +grep -q myevent3 dynamic_events test -d events/fprobes/myevent1 test -d events/fprobes/myevent2 @@ -21,6 +47,34 @@ echo "-:myevent2" >> dynamic_events grep -q myevent1 dynamic_events ! grep -q myevent2 dynamic_events +# should still have 2 left +cnt=`cat enabled_functions | wc -l` +if [ $cnt -ne 2 ]; then + exit_fail +fi + echo > dynamic_events +# Should have none left +cnt=`cat enabled_functions | wc -l` +if [ $cnt -ne 0 ]; then + exit_fail +fi + +echo "f:myevent4 $PLACE" >> dynamic_events + +# Should only have one enabled +cnt=`cat enabled_functions | wc -l` +if [ $cnt -ne 1 ]; then + exit_fail +fi + +echo > dynamic_events + +# Should have none left +cnt=`cat enabled_functions | wc -l` +if [ $cnt -ne 0 ]; then + exit_fail +fi + clear_trace -- 2.51.0 From 57b76bedc5c52c66968183b5ef57234894c25ce7 Mon Sep 17 00:00:00 2001 From: Sebastian Andrzej Siewior Date: Thu, 20 Feb 2025 15:07:49 +0100 Subject: [PATCH 02/16] ftrace: Correct preemption accounting for function tracing. The function tracer should record the preemption level at the point when the function is invoked. If the tracing subsystem decrement the preemption counter it needs to correct this before feeding the data into the trace buffer. This was broken in the commit cited below while shifting the preempt-disabled section. Use tracing_gen_ctx_dec() which properly subtracts one from the preemption counter on a preemptible kernel. Cc: stable@vger.kernel.org Cc: Wander Lairson Costa Cc: Masami Hiramatsu Cc: Mathieu Desnoyers Cc: Thomas Gleixner Link: https://lore.kernel.org/20250220140749.pfw8qoNZ@linutronix.de Fixes: ce5e48036c9e7 ("ftrace: disable preemption when recursion locked") Signed-off-by: Sebastian Andrzej Siewior Tested-by: Wander Lairson Costa Signed-off-by: Steven Rostedt (Google) --- kernel/trace/trace_functions.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) diff --git a/kernel/trace/trace_functions.c b/kernel/trace/trace_functions.c index d358c9935164..df56f9b76010 100644 --- a/kernel/trace/trace_functions.c +++ b/kernel/trace/trace_functions.c @@ -216,7 +216,7 @@ function_trace_call(unsigned long ip, unsigned long parent_ip, parent_ip = function_get_true_parent_ip(parent_ip, fregs); - trace_ctx = tracing_gen_ctx(); + trace_ctx = tracing_gen_ctx_dec(); data = this_cpu_ptr(tr->array_buffer.data); if (!atomic_read(&data->disabled)) @@ -321,7 +321,6 @@ function_no_repeats_trace_call(unsigned long ip, unsigned long parent_ip, struct trace_array *tr = op->private; struct trace_array_cpu *data; unsigned int trace_ctx; - unsigned long flags; int bit; if (unlikely(!tr->function_enabled)) @@ -347,8 +346,7 @@ function_no_repeats_trace_call(unsigned long ip, unsigned long parent_ip, if (is_repeat_check(tr, last_info, ip, parent_ip)) goto out; - local_save_flags(flags); - trace_ctx = tracing_gen_ctx_flags(flags); + trace_ctx = tracing_gen_ctx_dec(); process_repeats(tr, ip, parent_ip, last_info, trace_ctx); trace_function(tr, ip, parent_ip, trace_ctx); -- 2.51.0 From 2fa6a01345b538faa7b0fae8f723bb6977312428 Mon Sep 17 00:00:00 2001 From: Adrian Huang Date: Thu, 20 Feb 2025 11:15:28 +0800 Subject: [PATCH 03/16] tracing: Fix memory leak when reading set_event file kmemleak reports the following memory leak after reading set_event file: # cat /sys/kernel/tracing/set_event # cat /sys/kernel/debug/kmemleak unreferenced object 0xff110001234449e0 (size 16): comm "cat", pid 13645, jiffies 4294981880 hex dump (first 16 bytes): 01 00 00 00 00 00 00 00 a8 71 e7 84 ff ff ff ff .........q...... backtrace (crc c43abbc): __kmalloc_cache_noprof+0x3ca/0x4b0 s_start+0x72/0x2d0 seq_read_iter+0x265/0x1080 seq_read+0x2c9/0x420 vfs_read+0x166/0xc30 ksys_read+0xf4/0x1d0 do_syscall_64+0x79/0x150 entry_SYSCALL_64_after_hwframe+0x76/0x7e The issue can be reproduced regardless of whether set_event is empty or not. Here is an example about the valid content of set_event. # cat /sys/kernel/tracing/set_event sched:sched_process_fork sched:sched_switch sched:sched_wakeup *:*:mod:trace_events_sample The root cause is that s_next() returns NULL when nothing is found. This results in s_stop() attempting to free a NULL pointer because its parameter is NULL. Fix the issue by freeing the memory appropriately when s_next() fails to find anything. Cc: Mathieu Desnoyers Link: https://lore.kernel.org/20250220031528.7373-1-ahuang12@lenovo.com Fixes: b355247df104 ("tracing: Cache ":mod:" events for modules not loaded yet") Signed-off-by: Adrian Huang Acked-by: Masami Hiramatsu (Google) Signed-off-by: Steven Rostedt (Google) --- kernel/trace/trace_events.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) diff --git a/kernel/trace/trace_events.c b/kernel/trace/trace_events.c index 4cb275316e51..513de9ceb80e 100644 --- a/kernel/trace/trace_events.c +++ b/kernel/trace/trace_events.c @@ -1591,6 +1591,13 @@ s_next(struct seq_file *m, void *v, loff_t *pos) return iter; #endif + /* + * The iter is allocated in s_start() and passed via the 'v' + * parameter. To stop the iterator, NULL must be returned. But + * the return value is what the 'v' parameter in s_stop() receives + * and frees. Free iter here as it will no longer be used. + */ + kfree(iter); return NULL; } @@ -1667,9 +1674,9 @@ static int s_show(struct seq_file *m, void *v) } #endif -static void s_stop(struct seq_file *m, void *p) +static void s_stop(struct seq_file *m, void *v) { - kfree(p); + kfree(v); t_stop(m, NULL); } -- 2.51.0 From 781813db7909d945c33d3b035822225f3598774d Mon Sep 17 00:00:00 2001 From: Geert Uytterhoeven Date: Thu, 20 Feb 2025 16:12:12 +0100 Subject: [PATCH 04/16] i2c: core: Allocate temporary client dynamically MIME-Version: 1.0 Content-Type: text/plain; charset=utf8 Content-Transfer-Encoding: 8bit drivers/i2c/i2c-core-base.c: In function ‘i2c_detect.isra’: drivers/i2c/i2c-core-base.c:2544:1: warning: the frame size of 1312 bytes is larger than 1024 bytes [-Wframe-larger-than=] 2544 | } | ^ Fix this by allocating the temporary client structure dynamically, as it is a rather large structure (1216 bytes, depending on kernel config). This is basically a revert of the to-be-fixed commit with some checkpatch improvements. Fixes: 735668f8e5c9 ("i2c: core: Allocate temp client on the stack in i2c_detect") Signed-off-by: Geert Uytterhoeven Reviewed-by: Su Hui Reviewed-by: Guenter Roeck [wsa: updated commit message, merged tags from similar patch] Signed-off-by: Wolfram Sang --- drivers/i2c/i2c-core-base.c | 15 ++++++++++----- 1 file changed, 10 insertions(+), 5 deletions(-) diff --git a/drivers/i2c/i2c-core-base.c b/drivers/i2c/i2c-core-base.c index 35a221e2c11c..7ad1ad5c8c3f 100644 --- a/drivers/i2c/i2c-core-base.c +++ b/drivers/i2c/i2c-core-base.c @@ -2506,7 +2506,7 @@ static int i2c_detect_address(struct i2c_client *temp_client, static int i2c_detect(struct i2c_adapter *adapter, struct i2c_driver *driver) { const unsigned short *address_list; - struct i2c_client temp_client; + struct i2c_client *temp_client; int i, err = 0; address_list = driver->address_list; @@ -2527,19 +2527,24 @@ static int i2c_detect(struct i2c_adapter *adapter, struct i2c_driver *driver) return 0; /* Set up a temporary client to help detect callback */ - memset(&temp_client, 0, sizeof(temp_client)); - temp_client.adapter = adapter; + temp_client = kzalloc(sizeof(*temp_client), GFP_KERNEL); + if (!temp_client) + return -ENOMEM; + + temp_client->adapter = adapter; for (i = 0; address_list[i] != I2C_CLIENT_END; i += 1) { dev_dbg(&adapter->dev, "found normal entry for adapter %d, addr 0x%02x\n", i2c_adapter_id(adapter), address_list[i]); - temp_client.addr = address_list[i]; - err = i2c_detect_address(&temp_client, driver); + temp_client->addr = address_list[i]; + err = i2c_detect_address(temp_client, driver); if (unlikely(err)) break; } + kfree(temp_client); + return err; } -- 2.51.0 From d082ecbc71e9e0bf49883ee4afd435a77a5101b6 Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Sun, 23 Feb 2025 12:32:57 -0800 Subject: [PATCH 05/16] Linux 6.14-rc4 --- Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/Makefile b/Makefile index 96407c1d6be1..30dab4c8b012 100644 --- a/Makefile +++ b/Makefile @@ -2,7 +2,7 @@ VERSION = 6 PATCHLEVEL = 14 SUBLEVEL = 0 -EXTRAVERSION = -rc3 +EXTRAVERSION = -rc4 NAME = Baby Opossum Posse # *DOCUMENTATION* -- 2.51.0 From c73f0b69648501978e8b3e8fa7eef7f4197d0481 Mon Sep 17 00:00:00 2001 From: Feng Yang Date: Sun, 23 Feb 2025 15:01:06 +0800 Subject: [PATCH 06/16] ring-buffer: Fix bytes_dropped calculation issue The calculation of bytes-dropped and bytes_dropped_nested is reversed. Although it does not affect the final calculation of total_dropped, it should still be modified. Link: https://lore.kernel.org/20250223070106.6781-1-yangfeng59949@163.com Fixes: 6c43e554a2a5 ("ring-buffer: Add ring buffer startup selftest") Signed-off-by: Feng Yang Signed-off-by: Steven Rostedt (Google) --- kernel/trace/ring_buffer.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index bb6089c2951e..510409f97992 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -7411,9 +7411,9 @@ static __init int rb_write_something(struct rb_test_data *data, bool nested) /* Ignore dropped events before test starts. */ if (started) { if (nested) - data->bytes_dropped += len; - else data->bytes_dropped_nested += len; + else + data->bytes_dropped += len; } return len; } -- 2.51.0 From bcba8d4dbe6880ce9883409df486de35d3946704 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 5 Mar 2025 11:45:40 -0500 Subject: [PATCH 07/16] ring-buffer: Use kaslr address instead of text delta Instead of saving off the text and data pointers and using them to compare with the current boot's text and data pointers, just save off the KASLR offset. Then that can be used to figure out how to read the previous boots buffer. The last_boot_info will now show this offset, but only if it is for a previous boot: ~# cat instances/boot_mapped/last_boot_info 39000000 [kernel] ~# echo function > instances/boot_mapped/current_tracer ~# cat instances/boot_mapped/last_boot_info # Current If the KASLR offset saved is for the current boot, the last_boot_info will show the value of "current". Cc: Mark Rutland Cc: Mathieu Desnoyers Cc: Andrew Morton Link: https://lore.kernel.org/20250305164608.274956504@goodmis.org Reviewed-by: Masami Hiramatsu (Google) Signed-off-by: Steven Rostedt (Google) --- include/linux/ring_buffer.h | 3 +-- kernel/trace/ring_buffer.c | 31 ++++++++++++------------------- kernel/trace/trace.c | 30 +++++++++++++++++++++--------- kernel/trace/trace.h | 9 +++++---- 4 files changed, 39 insertions(+), 34 deletions(-) diff --git a/include/linux/ring_buffer.h b/include/linux/ring_buffer.h index 17fbb7855295..8de035f4f0d9 100644 --- a/include/linux/ring_buffer.h +++ b/include/linux/ring_buffer.h @@ -94,8 +94,7 @@ struct trace_buffer *__ring_buffer_alloc_range(unsigned long size, unsigned flag unsigned long range_size, struct lock_class_key *key); -bool ring_buffer_last_boot_delta(struct trace_buffer *buffer, long *text, - long *data); +bool ring_buffer_last_boot_delta(struct trace_buffer *buffer, unsigned long *kaslr_addr); /* * Because the ring buffer is generic, if other users of the ring buffer get diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index 510409f97992..9aa2e818eced 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -31,6 +31,7 @@ #include #include +#include #include "trace.h" @@ -49,8 +50,7 @@ static void update_pages_handler(struct work_struct *work); struct ring_buffer_meta { int magic; int struct_size; - unsigned long text_addr; - unsigned long data_addr; + unsigned long kaslr_addr; unsigned long first_buffer; unsigned long head_buffer; unsigned long commit_buffer; @@ -550,8 +550,7 @@ struct trace_buffer { unsigned long range_addr_start; unsigned long range_addr_end; - long last_text_delta; - long last_data_delta; + unsigned long kaslr_addr; unsigned int subbuf_size; unsigned int subbuf_order; @@ -1891,16 +1890,13 @@ static void rb_meta_validate_events(struct ring_buffer_per_cpu *cpu_buffer) } } -/* Used to calculate data delta */ -static char rb_data_ptr[] = ""; - -#define THIS_TEXT_PTR ((unsigned long)rb_meta_init_text_addr) -#define THIS_DATA_PTR ((unsigned long)rb_data_ptr) - static void rb_meta_init_text_addr(struct ring_buffer_meta *meta) { - meta->text_addr = THIS_TEXT_PTR; - meta->data_addr = THIS_DATA_PTR; +#ifdef CONFIG_RANDOMIZE_BASE + meta->kaslr_addr = kaslr_offset(); +#else + meta->kaslr_addr = 0; +#endif } static void rb_range_meta_init(struct trace_buffer *buffer, int nr_pages) @@ -1928,8 +1924,7 @@ static void rb_range_meta_init(struct trace_buffer *buffer, int nr_pages) meta->first_buffer += delta; meta->head_buffer += delta; meta->commit_buffer += delta; - buffer->last_text_delta = THIS_TEXT_PTR - meta->text_addr; - buffer->last_data_delta = THIS_DATA_PTR - meta->data_addr; + buffer->kaslr_addr = meta->kaslr_addr; continue; } @@ -2482,17 +2477,15 @@ struct trace_buffer *__ring_buffer_alloc_range(unsigned long size, unsigned flag * * Returns: The true if the delta is non zero */ -bool ring_buffer_last_boot_delta(struct trace_buffer *buffer, long *text, - long *data) +bool ring_buffer_last_boot_delta(struct trace_buffer *buffer, unsigned long *kaslr_addr) { if (!buffer) return false; - if (!buffer->last_text_delta) + if (!buffer->kaslr_addr) return false; - *text = buffer->last_text_delta; - *data = buffer->last_data_delta; + *kaslr_addr = buffer->kaslr_addr; return true; } diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 0e6d517e74e0..934658cda570 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -50,7 +50,7 @@ #include #include -#include /* COMMAND_LINE_SIZE */ +#include /* COMMAND_LINE_SIZE and kaslr_offset() */ #include "trace.h" #include "trace_output.h" @@ -4193,7 +4193,7 @@ static enum print_line_t print_trace_fmt(struct trace_iterator *iter) * safe to use if the array has delta offsets * Force printing via the fields. */ - if ((tr->text_delta || tr->data_delta) && + if ((tr->text_delta) && event->type > __TRACE_LAST_TYPE) return print_event_fields(iter, event); @@ -5990,7 +5990,7 @@ ssize_t tracing_resize_ring_buffer(struct trace_array *tr, static void update_last_data(struct trace_array *tr) { - if (!tr->text_delta && !tr->data_delta) + if (!(tr->flags & TRACE_ARRAY_FL_LAST_BOOT)) return; /* @@ -6003,7 +6003,8 @@ static void update_last_data(struct trace_array *tr) /* Using current data now */ tr->text_delta = 0; - tr->data_delta = 0; + + tr->flags &= ~TRACE_ARRAY_FL_LAST_BOOT; } /** @@ -6821,8 +6822,17 @@ tracing_last_boot_read(struct file *filp, char __user *ubuf, size_t cnt, loff_t seq_buf_init(&seq, buf, 64); - seq_buf_printf(&seq, "text delta:\t%ld\n", tr->text_delta); - seq_buf_printf(&seq, "data delta:\t%ld\n", tr->data_delta); + /* + * Do not leak KASLR address. This only shows the KASLR address of + * the last boot. When the ring buffer is started, the LAST_BOOT + * flag gets cleared, and this should only report "current". + * Otherwise it shows the KASLR address from the previous boot which + * should not be the same as the current boot. + */ + if (tr->flags & TRACE_ARRAY_FL_LAST_BOOT) + seq_buf_printf(&seq, "%lx\t[kernel]\n", tr->kaslr_addr); + else + seq_buf_puts(&seq, "# Current\n"); return simple_read_from_buffer(ubuf, cnt, ppos, buf, seq_buf_used(&seq)); } @@ -9210,8 +9220,10 @@ allocate_trace_buffer(struct trace_array *tr, struct array_buffer *buf, int size tr->range_addr_start, tr->range_addr_size); - ring_buffer_last_boot_delta(buf->buffer, - &tr->text_delta, &tr->data_delta); +#ifdef CONFIG_RANDOMIZE_BASE + if (ring_buffer_last_boot_delta(buf->buffer, &tr->kaslr_addr)) + tr->text_delta = kaslr_offset() - tr->kaslr_addr; +#endif /* * This is basically the same as a mapped buffer, * with the same restrictions. @@ -10459,7 +10471,7 @@ __init static void enable_instances(void) * to it. */ if (start) { - tr->flags |= TRACE_ARRAY_FL_BOOT; + tr->flags |= TRACE_ARRAY_FL_BOOT | TRACE_ARRAY_FL_LAST_BOOT; tr->ref++; } diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 9c21ba45b7af..abe8169c3e87 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -348,8 +348,8 @@ struct trace_array { unsigned int mapped; unsigned long range_addr_start; unsigned long range_addr_size; + unsigned long kaslr_addr; long text_delta; - long data_delta; struct trace_pid_list __rcu *filtered_pids; struct trace_pid_list __rcu *filtered_no_pids; @@ -433,9 +433,10 @@ struct trace_array { }; enum { - TRACE_ARRAY_FL_GLOBAL = BIT(0), - TRACE_ARRAY_FL_BOOT = BIT(1), - TRACE_ARRAY_FL_MOD_INIT = BIT(2), + TRACE_ARRAY_FL_GLOBAL = BIT(0), + TRACE_ARRAY_FL_BOOT = BIT(1), + TRACE_ARRAY_FL_LAST_BOOT = BIT(2), + TRACE_ARRAY_FL_MOD_INIT = BIT(3), }; #ifdef CONFIG_MODULES -- 2.51.0 From 4009cc31e7813ed66a04237ddff76706ff57a771 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 5 Mar 2025 11:45:41 -0500 Subject: [PATCH 08/16] ring-buffer: Add buffer meta data for persistent ring buffer Instead of just having a meta data at the first page of each sub buffer that has duplicate data, add a new meta page to the entire block of memory that holds the duplicate data and remove it from the sub buffer meta data. This will open up the extra memory in this first page to be used by the tracer for its own persistent data. Cc: Mark Rutland Cc: Mathieu Desnoyers Cc: Andrew Morton Link: https://lore.kernel.org/20250305164608.446351513@goodmis.org Reviewed-by: Masami Hiramatsu (Google) Signed-off-by: Steven Rostedt (Google) --- kernel/trace/ring_buffer.c | 171 ++++++++++++++++++++++++++----------- 1 file changed, 119 insertions(+), 52 deletions(-) diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index 9aa2e818eced..0304e431604d 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -49,7 +49,12 @@ static void update_pages_handler(struct work_struct *work); struct ring_buffer_meta { int magic; - int struct_size; + int struct_sizes; + unsigned long total_size; + unsigned long buffers_offset; +}; + +struct ring_buffer_cpu_meta { unsigned long kaslr_addr; unsigned long first_buffer; unsigned long head_buffer; @@ -517,7 +522,7 @@ struct ring_buffer_per_cpu { struct mutex mapping_lock; unsigned long *subbuf_ids; /* ID to subbuf VA */ struct trace_buffer_meta *meta_page; - struct ring_buffer_meta *ring_meta; + struct ring_buffer_cpu_meta *ring_meta; /* ring buffer pages to update, > 0 to add, < 0 to remove */ long nr_pages_to_update; @@ -550,6 +555,8 @@ struct trace_buffer { unsigned long range_addr_start; unsigned long range_addr_end; + struct ring_buffer_meta *meta; + unsigned long kaslr_addr; unsigned int subbuf_size; @@ -1270,7 +1277,7 @@ static void rb_head_page_activate(struct ring_buffer_per_cpu *cpu_buffer) rb_set_list_to_head(head->list.prev); if (cpu_buffer->ring_meta) { - struct ring_buffer_meta *meta = cpu_buffer->ring_meta; + struct ring_buffer_cpu_meta *meta = cpu_buffer->ring_meta; meta->head_buffer = (unsigned long)head->page; } } @@ -1568,7 +1575,7 @@ out_locked: static unsigned long rb_range_align_subbuf(unsigned long addr, int subbuf_size, int nr_subbufs) { - addr += sizeof(struct ring_buffer_meta) + + addr += sizeof(struct ring_buffer_cpu_meta) + sizeof(int) * nr_subbufs; return ALIGN(addr, subbuf_size); } @@ -1579,19 +1586,22 @@ rb_range_align_subbuf(unsigned long addr, int subbuf_size, int nr_subbufs) static void *rb_range_meta(struct trace_buffer *buffer, int nr_pages, int cpu) { int subbuf_size = buffer->subbuf_size + BUF_PAGE_HDR_SIZE; - unsigned long ptr = buffer->range_addr_start; - struct ring_buffer_meta *meta; + struct ring_buffer_cpu_meta *meta; + struct ring_buffer_meta *bmeta; + unsigned long ptr; int nr_subbufs; - if (!ptr) + bmeta = buffer->meta; + if (!bmeta) return NULL; + ptr = (unsigned long)bmeta + bmeta->buffers_offset; + meta = (struct ring_buffer_cpu_meta *)ptr; + /* When nr_pages passed in is zero, the first meta has already been initialized */ if (!nr_pages) { - meta = (struct ring_buffer_meta *)ptr; nr_subbufs = meta->nr_subbufs; } else { - meta = NULL; /* Include the reader page */ nr_subbufs = nr_pages + 1; } @@ -1623,7 +1633,7 @@ static void *rb_range_meta(struct trace_buffer *buffer, int nr_pages, int cpu) } /* Return the start of subbufs given the meta pointer */ -static void *rb_subbufs_from_meta(struct ring_buffer_meta *meta) +static void *rb_subbufs_from_meta(struct ring_buffer_cpu_meta *meta) { int subbuf_size = meta->subbuf_size; unsigned long ptr; @@ -1639,7 +1649,7 @@ static void *rb_subbufs_from_meta(struct ring_buffer_meta *meta) */ static void *rb_range_buffer(struct ring_buffer_per_cpu *cpu_buffer, int idx) { - struct ring_buffer_meta *meta; + struct ring_buffer_cpu_meta *meta; unsigned long ptr; int subbuf_size; @@ -1664,15 +1674,74 @@ static void *rb_range_buffer(struct ring_buffer_per_cpu *cpu_buffer, int idx) return (void *)ptr; } +/* + * See if the existing memory contains a valid meta section. + * if so, use that, otherwise initialize it. + */ +static bool rb_meta_init(struct trace_buffer *buffer) +{ + unsigned long ptr = buffer->range_addr_start; + struct ring_buffer_meta *bmeta; + unsigned long total_size; + int struct_sizes; + + bmeta = (struct ring_buffer_meta *)ptr; + buffer->meta = bmeta; + + total_size = buffer->range_addr_end - buffer->range_addr_start; + + struct_sizes = sizeof(struct ring_buffer_cpu_meta); + struct_sizes |= sizeof(*bmeta) << 16; + + /* The first buffer will start word size after the meta page */ + ptr += sizeof(*bmeta); + ptr = ALIGN(ptr, sizeof(long)); + + if (bmeta->magic != RING_BUFFER_META_MAGIC) { + pr_info("Ring buffer boot meta mismatch of magic\n"); + goto init; + } + + if (bmeta->struct_sizes != struct_sizes) { + pr_info("Ring buffer boot meta mismatch of struct size\n"); + goto init; + } + + if (bmeta->total_size != total_size) { + pr_info("Ring buffer boot meta mismatch of total size\n"); + goto init; + } + + if (bmeta->buffers_offset > bmeta->total_size) { + pr_info("Ring buffer boot meta mismatch of offset outside of total size\n"); + goto init; + } + + if (bmeta->buffers_offset != (void *)ptr - (void *)bmeta) { + pr_info("Ring buffer boot meta mismatch of first buffer offset\n"); + goto init; + } + + return true; + + init: + bmeta->magic = RING_BUFFER_META_MAGIC; + bmeta->struct_sizes = struct_sizes; + bmeta->total_size = total_size; + bmeta->buffers_offset = (void *)ptr - (void *)bmeta; + + return false; +} + /* * See if the existing memory contains valid ring buffer data. * As the previous kernel must be the same as this kernel, all * the calculations (size of buffers and number of buffers) * must be the same. */ -static bool rb_meta_valid(struct ring_buffer_meta *meta, int cpu, - struct trace_buffer *buffer, int nr_pages, - unsigned long *subbuf_mask) +static bool rb_cpu_meta_valid(struct ring_buffer_cpu_meta *meta, int cpu, + struct trace_buffer *buffer, int nr_pages, + unsigned long *subbuf_mask) { int subbuf_size = PAGE_SIZE; struct buffer_data_page *subbuf; @@ -1683,20 +1752,6 @@ static bool rb_meta_valid(struct ring_buffer_meta *meta, int cpu, if (!subbuf_mask) return false; - /* Check the meta magic and meta struct size */ - if (meta->magic != RING_BUFFER_META_MAGIC || - meta->struct_size != sizeof(*meta)) { - pr_info("Ring buffer boot meta[%d] mismatch of magic or struct size\n", cpu); - return false; - } - - /* The subbuffer's size and number of subbuffers must match */ - if (meta->subbuf_size != subbuf_size || - meta->nr_subbufs != nr_pages + 1) { - pr_info("Ring buffer boot meta [%d] mismatch of subbuf_size/nr_pages\n", cpu); - return false; - } - buffers_start = meta->first_buffer; buffers_end = meta->first_buffer + (subbuf_size * meta->nr_subbufs); @@ -1742,7 +1797,7 @@ static bool rb_meta_valid(struct ring_buffer_meta *meta, int cpu, return true; } -static int rb_meta_subbuf_idx(struct ring_buffer_meta *meta, void *subbuf); +static int rb_meta_subbuf_idx(struct ring_buffer_cpu_meta *meta, void *subbuf); static int rb_read_data_buffer(struct buffer_data_page *dpage, int tail, int cpu, unsigned long long *timestamp, u64 *delta_ptr) @@ -1809,7 +1864,7 @@ static int rb_validate_buffer(struct buffer_data_page *dpage, int cpu) /* If the meta data has been validated, now validate the events */ static void rb_meta_validate_events(struct ring_buffer_per_cpu *cpu_buffer) { - struct ring_buffer_meta *meta = cpu_buffer->ring_meta; + struct ring_buffer_cpu_meta *meta = cpu_buffer->ring_meta; struct buffer_page *head_page; unsigned long entry_bytes = 0; unsigned long entries = 0; @@ -1890,7 +1945,7 @@ static void rb_meta_validate_events(struct ring_buffer_per_cpu *cpu_buffer) } } -static void rb_meta_init_text_addr(struct ring_buffer_meta *meta) +static void rb_meta_init_text_addr(struct ring_buffer_cpu_meta *meta) { #ifdef CONFIG_RANDOMIZE_BASE meta->kaslr_addr = kaslr_offset(); @@ -1901,10 +1956,12 @@ static void rb_meta_init_text_addr(struct ring_buffer_meta *meta) static void rb_range_meta_init(struct trace_buffer *buffer, int nr_pages) { - struct ring_buffer_meta *meta; + struct ring_buffer_cpu_meta *meta; + struct ring_buffer_meta *bmeta; unsigned long *subbuf_mask; unsigned long delta; void *subbuf; + bool valid = false; int cpu; int i; @@ -1912,12 +1969,17 @@ static void rb_range_meta_init(struct trace_buffer *buffer, int nr_pages) subbuf_mask = bitmap_alloc(nr_pages + 1, GFP_KERNEL); /* If subbuf_mask fails to allocate, then rb_meta_valid() will return false */ + if (rb_meta_init(buffer)) + valid = true; + + bmeta = buffer->meta; + for (cpu = 0; cpu < nr_cpu_ids; cpu++) { void *next_meta; meta = rb_range_meta(buffer, nr_pages, cpu); - if (rb_meta_valid(meta, cpu, buffer, nr_pages, subbuf_mask)) { + if (valid && rb_cpu_meta_valid(meta, cpu, buffer, nr_pages, subbuf_mask)) { /* Make the mappings match the current address */ subbuf = rb_subbufs_from_meta(meta); delta = (unsigned long)subbuf - meta->first_buffer; @@ -1935,9 +1997,6 @@ static void rb_range_meta_init(struct trace_buffer *buffer, int nr_pages) memset(meta, 0, next_meta - (void *)meta); - meta->magic = RING_BUFFER_META_MAGIC; - meta->struct_size = sizeof(*meta); - meta->nr_subbufs = nr_pages + 1; meta->subbuf_size = PAGE_SIZE; @@ -1966,7 +2025,7 @@ static void rb_range_meta_init(struct trace_buffer *buffer, int nr_pages) static void *rbm_start(struct seq_file *m, loff_t *pos) { struct ring_buffer_per_cpu *cpu_buffer = m->private; - struct ring_buffer_meta *meta = cpu_buffer->ring_meta; + struct ring_buffer_cpu_meta *meta = cpu_buffer->ring_meta; unsigned long val; if (!meta) @@ -1991,7 +2050,7 @@ static void *rbm_next(struct seq_file *m, void *v, loff_t *pos) static int rbm_show(struct seq_file *m, void *v) { struct ring_buffer_per_cpu *cpu_buffer = m->private; - struct ring_buffer_meta *meta = cpu_buffer->ring_meta; + struct ring_buffer_cpu_meta *meta = cpu_buffer->ring_meta; unsigned long val = (unsigned long)v; if (val == 1) { @@ -2040,7 +2099,7 @@ int ring_buffer_meta_seq_init(struct file *file, struct trace_buffer *buffer, in static void rb_meta_buffer_update(struct ring_buffer_per_cpu *cpu_buffer, struct buffer_page *bpage) { - struct ring_buffer_meta *meta = cpu_buffer->ring_meta; + struct ring_buffer_cpu_meta *meta = cpu_buffer->ring_meta; if (meta->head_buffer == (unsigned long)bpage->page) cpu_buffer->head_page = bpage; @@ -2055,7 +2114,7 @@ static int __rb_allocate_pages(struct ring_buffer_per_cpu *cpu_buffer, long nr_pages, struct list_head *pages) { struct trace_buffer *buffer = cpu_buffer->buffer; - struct ring_buffer_meta *meta = NULL; + struct ring_buffer_cpu_meta *meta = NULL; struct buffer_page *bpage, *tmp; bool user_thread = current->mm != NULL; gfp_t mflags; @@ -2179,7 +2238,7 @@ static struct ring_buffer_per_cpu * rb_allocate_cpu_buffer(struct trace_buffer *buffer, long nr_pages, int cpu) { struct ring_buffer_per_cpu *cpu_buffer; - struct ring_buffer_meta *meta; + struct ring_buffer_cpu_meta *meta; struct buffer_page *bpage; struct page *page; int ret; @@ -2350,10 +2409,18 @@ static struct trace_buffer *alloc_buffer(unsigned long size, unsigned flags, /* If start/end are specified, then that overrides size */ if (start && end) { + unsigned long buffers_start; unsigned long ptr; int n; - size = end - start; + /* Make sure that start is word aligned */ + start = ALIGN(start, sizeof(long)); + + /* Subtract the buffer meta data and word aligned */ + buffers_start = start + sizeof(struct ring_buffer_cpu_meta); + buffers_start = ALIGN(buffers_start, sizeof(long)); + + size = end - buffers_start; size = size / nr_cpu_ids; /* @@ -2363,7 +2430,7 @@ static struct trace_buffer *alloc_buffer(unsigned long size, unsigned flags, * needed, plus account for the integer array index that * will be appended to the meta data. */ - nr_pages = (size - sizeof(struct ring_buffer_meta)) / + nr_pages = (size - sizeof(struct ring_buffer_cpu_meta)) / (subbuf_size + sizeof(int)); /* Need at least two pages plus the reader page */ if (nr_pages < 3) @@ -2371,8 +2438,8 @@ static struct trace_buffer *alloc_buffer(unsigned long size, unsigned flags, again: /* Make sure that the size fits aligned */ - for (n = 0, ptr = start; n < nr_cpu_ids; n++) { - ptr += sizeof(struct ring_buffer_meta) + + for (n = 0, ptr = buffers_start; n < nr_cpu_ids; n++) { + ptr += sizeof(struct ring_buffer_cpu_meta) + sizeof(int) * nr_pages; ptr = ALIGN(ptr, subbuf_size); ptr += subbuf_size * nr_pages; @@ -3098,7 +3165,7 @@ static void rb_inc_iter(struct ring_buffer_iter *iter) } /* Return the index into the sub-buffers for a given sub-buffer */ -static int rb_meta_subbuf_idx(struct ring_buffer_meta *meta, void *subbuf) +static int rb_meta_subbuf_idx(struct ring_buffer_cpu_meta *meta, void *subbuf) { void *subbuf_array; @@ -3110,7 +3177,7 @@ static int rb_meta_subbuf_idx(struct ring_buffer_meta *meta, void *subbuf) static void rb_update_meta_head(struct ring_buffer_per_cpu *cpu_buffer, struct buffer_page *next_page) { - struct ring_buffer_meta *meta = cpu_buffer->ring_meta; + struct ring_buffer_cpu_meta *meta = cpu_buffer->ring_meta; unsigned long old_head = (unsigned long)next_page->page; unsigned long new_head; @@ -3127,7 +3194,7 @@ static void rb_update_meta_head(struct ring_buffer_per_cpu *cpu_buffer, static void rb_update_meta_reader(struct ring_buffer_per_cpu *cpu_buffer, struct buffer_page *reader) { - struct ring_buffer_meta *meta = cpu_buffer->ring_meta; + struct ring_buffer_cpu_meta *meta = cpu_buffer->ring_meta; void *old_reader = cpu_buffer->reader_page->page; void *new_reader = reader->page; int id; @@ -3756,7 +3823,7 @@ rb_set_commit_to_write(struct ring_buffer_per_cpu *cpu_buffer) rb_page_write(cpu_buffer->commit_page)); rb_inc_page(&cpu_buffer->commit_page); if (cpu_buffer->ring_meta) { - struct ring_buffer_meta *meta = cpu_buffer->ring_meta; + struct ring_buffer_cpu_meta *meta = cpu_buffer->ring_meta; meta->commit_buffer = (unsigned long)cpu_buffer->commit_page->page; } /* add barrier to keep gcc from optimizing too much */ @@ -6009,7 +6076,7 @@ rb_reset_cpu(struct ring_buffer_per_cpu *cpu_buffer) if (cpu_buffer->mapped) { rb_update_meta_page(cpu_buffer); if (cpu_buffer->ring_meta) { - struct ring_buffer_meta *meta = cpu_buffer->ring_meta; + struct ring_buffer_cpu_meta *meta = cpu_buffer->ring_meta; meta->commit_buffer = meta->head_buffer; } } @@ -6043,7 +6110,7 @@ static void reset_disabled_cpu_buffer(struct ring_buffer_per_cpu *cpu_buffer) void ring_buffer_reset_cpu(struct trace_buffer *buffer, int cpu) { struct ring_buffer_per_cpu *cpu_buffer = buffer->buffers[cpu]; - struct ring_buffer_meta *meta; + struct ring_buffer_cpu_meta *meta; if (!cpumask_test_cpu(cpu, buffer->cpumask)) return; @@ -6081,7 +6148,7 @@ EXPORT_SYMBOL_GPL(ring_buffer_reset_cpu); void ring_buffer_reset_online_cpus(struct trace_buffer *buffer) { struct ring_buffer_per_cpu *cpu_buffer; - struct ring_buffer_meta *meta; + struct ring_buffer_cpu_meta *meta; int cpu; /* prevent another thread from changing buffer sizes */ -- 2.51.0 From 4af0a9c518522892b36cb7ecedf0c6004dc0a581 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 5 Mar 2025 11:45:42 -0500 Subject: [PATCH 09/16] ring-buffer: Add ring_buffer_meta_scratch() Now that there's one meta data at the start of the persistent memory used by the ring buffer, allow the caller to request some memory right after that data that it can use as its own persistent memory. Also fix some white space issues with ring_buffer_alloc(). Cc: Mark Rutland Cc: Mathieu Desnoyers Cc: Andrew Morton Link: https://lore.kernel.org/20250305164608.619631731@goodmis.org Reviewed-by: Masami Hiramatsu (Google) Signed-off-by: Steven Rostedt (Google) --- include/linux/ring_buffer.h | 6 ++++-- kernel/trace/ring_buffer.c | 35 +++++++++++++++++++++++++++++------ kernel/trace/trace.c | 2 +- 3 files changed, 34 insertions(+), 9 deletions(-) diff --git a/include/linux/ring_buffer.h b/include/linux/ring_buffer.h index 8de035f4f0d9..019b59a0bbc9 100644 --- a/include/linux/ring_buffer.h +++ b/include/linux/ring_buffer.h @@ -92,9 +92,11 @@ __ring_buffer_alloc(unsigned long size, unsigned flags, struct lock_class_key *k struct trace_buffer *__ring_buffer_alloc_range(unsigned long size, unsigned flags, int order, unsigned long start, unsigned long range_size, + unsigned long scratch_size, struct lock_class_key *key); bool ring_buffer_last_boot_delta(struct trace_buffer *buffer, unsigned long *kaslr_addr); +void *ring_buffer_meta_scratch(struct trace_buffer *buffer, unsigned int *size); /* * Because the ring buffer is generic, if other users of the ring buffer get @@ -112,11 +114,11 @@ bool ring_buffer_last_boot_delta(struct trace_buffer *buffer, unsigned long *kas * traced by ftrace, it can produce lockdep warnings. We need to keep each * ring buffer's lock class separate. */ -#define ring_buffer_alloc_range(size, flags, order, start, range_size) \ +#define ring_buffer_alloc_range(size, flags, order, start, range_size, s_size) \ ({ \ static struct lock_class_key __key; \ __ring_buffer_alloc_range((size), (flags), (order), (start), \ - (range_size), &__key); \ + (range_size), (s_size), &__key); \ }) typedef bool (*ring_buffer_cond_fn)(void *data); diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index 0304e431604d..c511b3283dc6 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -1678,7 +1678,7 @@ static void *rb_range_buffer(struct ring_buffer_per_cpu *cpu_buffer, int idx) * See if the existing memory contains a valid meta section. * if so, use that, otherwise initialize it. */ -static bool rb_meta_init(struct trace_buffer *buffer) +static bool rb_meta_init(struct trace_buffer *buffer, int scratch_size) { unsigned long ptr = buffer->range_addr_start; struct ring_buffer_meta *bmeta; @@ -1696,6 +1696,7 @@ static bool rb_meta_init(struct trace_buffer *buffer) /* The first buffer will start word size after the meta page */ ptr += sizeof(*bmeta); ptr = ALIGN(ptr, sizeof(long)); + ptr += scratch_size; if (bmeta->magic != RING_BUFFER_META_MAGIC) { pr_info("Ring buffer boot meta mismatch of magic\n"); @@ -1730,6 +1731,9 @@ static bool rb_meta_init(struct trace_buffer *buffer) bmeta->total_size = total_size; bmeta->buffers_offset = (void *)ptr - (void *)bmeta; + /* Zero out the scatch pad */ + memset((void *)bmeta + sizeof(*bmeta), 0, bmeta->buffers_offset - sizeof(*bmeta)); + return false; } @@ -1954,7 +1958,7 @@ static void rb_meta_init_text_addr(struct ring_buffer_cpu_meta *meta) #endif } -static void rb_range_meta_init(struct trace_buffer *buffer, int nr_pages) +static void rb_range_meta_init(struct trace_buffer *buffer, int nr_pages, int scratch_size) { struct ring_buffer_cpu_meta *meta; struct ring_buffer_meta *bmeta; @@ -1969,7 +1973,7 @@ static void rb_range_meta_init(struct trace_buffer *buffer, int nr_pages) subbuf_mask = bitmap_alloc(nr_pages + 1, GFP_KERNEL); /* If subbuf_mask fails to allocate, then rb_meta_valid() will return false */ - if (rb_meta_init(buffer)) + if (rb_meta_init(buffer, scratch_size)) valid = true; bmeta = buffer->meta; @@ -2367,6 +2371,7 @@ static void rb_free_cpu_buffer(struct ring_buffer_per_cpu *cpu_buffer) static struct trace_buffer *alloc_buffer(unsigned long size, unsigned flags, int order, unsigned long start, unsigned long end, + unsigned long scratch_size, struct lock_class_key *key) { struct trace_buffer *buffer; @@ -2416,10 +2421,15 @@ static struct trace_buffer *alloc_buffer(unsigned long size, unsigned flags, /* Make sure that start is word aligned */ start = ALIGN(start, sizeof(long)); + /* scratch_size needs to be aligned too */ + scratch_size = ALIGN(scratch_size, sizeof(long)); + /* Subtract the buffer meta data and word aligned */ buffers_start = start + sizeof(struct ring_buffer_cpu_meta); buffers_start = ALIGN(buffers_start, sizeof(long)); + buffers_start += scratch_size; + /* Calculate the size for the per CPU data */ size = end - buffers_start; size = size / nr_cpu_ids; @@ -2456,7 +2466,7 @@ static struct trace_buffer *alloc_buffer(unsigned long size, unsigned flags, buffer->range_addr_start = start; buffer->range_addr_end = end; - rb_range_meta_init(buffer, nr_pages); + rb_range_meta_init(buffer, nr_pages, scratch_size); } else { /* need at least two pages */ @@ -2509,7 +2519,7 @@ struct trace_buffer *__ring_buffer_alloc(unsigned long size, unsigned flags, struct lock_class_key *key) { /* Default buffer page size - one system page */ - return alloc_buffer(size, flags, 0, 0, 0,key); + return alloc_buffer(size, flags, 0, 0, 0, 0, key); } EXPORT_SYMBOL_GPL(__ring_buffer_alloc); @@ -2521,6 +2531,7 @@ EXPORT_SYMBOL_GPL(__ring_buffer_alloc); * @order: sub-buffer order * @start: start of allocated range * @range_size: size of allocated range + * @scratch_size: size of scratch area (for preallocated memory buffers) * @key: ring buffer reader_lock_key. * * Currently the only flag that is available is the RB_FL_OVERWRITE @@ -2531,9 +2542,11 @@ EXPORT_SYMBOL_GPL(__ring_buffer_alloc); struct trace_buffer *__ring_buffer_alloc_range(unsigned long size, unsigned flags, int order, unsigned long start, unsigned long range_size, + unsigned long scratch_size, struct lock_class_key *key) { - return alloc_buffer(size, flags, order, start, start + range_size, key); + return alloc_buffer(size, flags, order, start, start + range_size, + scratch_size, key); } /** @@ -2557,6 +2570,16 @@ bool ring_buffer_last_boot_delta(struct trace_buffer *buffer, unsigned long *kas return true; } +void *ring_buffer_meta_scratch(struct trace_buffer *buffer, unsigned int *size) +{ + if (!buffer || !buffer->meta) + return NULL; + + *size = PAGE_SIZE - sizeof(*buffer->meta); + + return (void *)buffer->meta + sizeof(*buffer->meta); +} + /** * ring_buffer_free - free a ring buffer. * @buffer: the buffer to free. diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 934658cda570..73c8df12c33c 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -9218,7 +9218,7 @@ allocate_trace_buffer(struct trace_array *tr, struct array_buffer *buf, int size if (tr->range_addr_start && tr->range_addr_size) { buf->buffer = ring_buffer_alloc_range(size, rb_flags, 0, tr->range_addr_start, - tr->range_addr_size); + tr->range_addr_size, 0); #ifdef CONFIG_RANDOMIZE_BASE if (ring_buffer_last_boot_delta(buf->buffer, &tr->kaslr_addr)) -- 2.51.0 From b65334825fb14cae15e93d627c3cfc2986c7eea6 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 5 Mar 2025 11:45:43 -0500 Subject: [PATCH 10/16] tracing: Have persistent trace instances save KASLR offset There's no reason to save the KASLR offset for the ring buffer itself. That is used by the tracer. Now that the tracer has a way to save data in the persistent memory of the ring buffer, have the tracing infrastructure take care of the saving of the KASLR offset. Cc: Mark Rutland Cc: Mathieu Desnoyers Cc: Andrew Morton Link: https://lore.kernel.org/20250305164608.792722274@goodmis.org Reviewed-by: Masami Hiramatsu (Google) Signed-off-by: Steven Rostedt (Google) --- include/linux/ring_buffer.h | 1 - kernel/trace/ring_buffer.c | 59 +++++++------------------------------ kernel/trace/trace.c | 37 +++++++++++++++++++---- kernel/trace/trace.h | 6 ++-- 4 files changed, 46 insertions(+), 57 deletions(-) diff --git a/include/linux/ring_buffer.h b/include/linux/ring_buffer.h index 019b59a0bbc9..56e27263acf8 100644 --- a/include/linux/ring_buffer.h +++ b/include/linux/ring_buffer.h @@ -95,7 +95,6 @@ struct trace_buffer *__ring_buffer_alloc_range(unsigned long size, unsigned flag unsigned long scratch_size, struct lock_class_key *key); -bool ring_buffer_last_boot_delta(struct trace_buffer *buffer, unsigned long *kaslr_addr); void *ring_buffer_meta_scratch(struct trace_buffer *buffer, unsigned int *size); /* diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index c511b3283dc6..b981eb20c206 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -55,7 +55,6 @@ struct ring_buffer_meta { }; struct ring_buffer_cpu_meta { - unsigned long kaslr_addr; unsigned long first_buffer; unsigned long head_buffer; unsigned long commit_buffer; @@ -557,8 +556,6 @@ struct trace_buffer { struct ring_buffer_meta *meta; - unsigned long kaslr_addr; - unsigned int subbuf_size; unsigned int subbuf_order; unsigned int max_data_size; @@ -1949,15 +1946,6 @@ static void rb_meta_validate_events(struct ring_buffer_per_cpu *cpu_buffer) } } -static void rb_meta_init_text_addr(struct ring_buffer_cpu_meta *meta) -{ -#ifdef CONFIG_RANDOMIZE_BASE - meta->kaslr_addr = kaslr_offset(); -#else - meta->kaslr_addr = 0; -#endif -} - static void rb_range_meta_init(struct trace_buffer *buffer, int nr_pages, int scratch_size) { struct ring_buffer_cpu_meta *meta; @@ -1990,7 +1978,6 @@ static void rb_range_meta_init(struct trace_buffer *buffer, int nr_pages, int sc meta->first_buffer += delta; meta->head_buffer += delta; meta->commit_buffer += delta; - buffer->kaslr_addr = meta->kaslr_addr; continue; } @@ -2007,7 +1994,6 @@ static void rb_range_meta_init(struct trace_buffer *buffer, int nr_pages, int sc subbuf = rb_subbufs_from_meta(meta); meta->first_buffer = (unsigned long)subbuf; - rb_meta_init_text_addr(meta); /* * The buffers[] array holds the order of the sub-buffers @@ -2549,35 +2535,22 @@ struct trace_buffer *__ring_buffer_alloc_range(unsigned long size, unsigned flag scratch_size, key); } -/** - * ring_buffer_last_boot_delta - return the delta offset from last boot - * @buffer: The buffer to return the delta from - * @text: Return text delta - * @data: Return data delta - * - * Returns: The true if the delta is non zero - */ -bool ring_buffer_last_boot_delta(struct trace_buffer *buffer, unsigned long *kaslr_addr) -{ - if (!buffer) - return false; - - if (!buffer->kaslr_addr) - return false; - - *kaslr_addr = buffer->kaslr_addr; - - return true; -} - void *ring_buffer_meta_scratch(struct trace_buffer *buffer, unsigned int *size) { + struct ring_buffer_meta *meta; + void *ptr; + if (!buffer || !buffer->meta) return NULL; - *size = PAGE_SIZE - sizeof(*buffer->meta); + meta = buffer->meta; - return (void *)buffer->meta + sizeof(*buffer->meta); + ptr = (void *)ALIGN((unsigned long)meta + sizeof(*meta), sizeof(long)); + + if (size) + *size = (void *)meta + meta->buffers_offset - ptr; + + return ptr; } /** @@ -6133,7 +6106,6 @@ static void reset_disabled_cpu_buffer(struct ring_buffer_per_cpu *cpu_buffer) void ring_buffer_reset_cpu(struct trace_buffer *buffer, int cpu) { struct ring_buffer_per_cpu *cpu_buffer = buffer->buffers[cpu]; - struct ring_buffer_cpu_meta *meta; if (!cpumask_test_cpu(cpu, buffer->cpumask)) return; @@ -6152,11 +6124,6 @@ void ring_buffer_reset_cpu(struct trace_buffer *buffer, int cpu) atomic_dec(&cpu_buffer->record_disabled); atomic_dec(&cpu_buffer->resize_disabled); - /* Make sure persistent meta now uses this buffer's addresses */ - meta = rb_range_meta(buffer, 0, cpu_buffer->cpu); - if (meta) - rb_meta_init_text_addr(meta); - mutex_unlock(&buffer->mutex); } EXPORT_SYMBOL_GPL(ring_buffer_reset_cpu); @@ -6171,7 +6138,6 @@ EXPORT_SYMBOL_GPL(ring_buffer_reset_cpu); void ring_buffer_reset_online_cpus(struct trace_buffer *buffer) { struct ring_buffer_per_cpu *cpu_buffer; - struct ring_buffer_cpu_meta *meta; int cpu; /* prevent another thread from changing buffer sizes */ @@ -6199,11 +6165,6 @@ void ring_buffer_reset_online_cpus(struct trace_buffer *buffer) reset_disabled_cpu_buffer(cpu_buffer); - /* Make sure persistent meta now uses this buffer's addresses */ - meta = rb_range_meta(buffer, 0, cpu_buffer->cpu); - if (meta) - rb_meta_init_text_addr(meta); - atomic_dec(&cpu_buffer->record_disabled); atomic_sub(RESET_BIT, &cpu_buffer->resize_disabled); } diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 73c8df12c33c..d4df56657b9d 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -5988,8 +5988,14 @@ ssize_t tracing_resize_ring_buffer(struct trace_array *tr, return __tracing_resize_ring_buffer(tr, size, cpu_id); } +struct trace_scratch { + unsigned long kaslr_addr; +}; + static void update_last_data(struct trace_array *tr) { + struct trace_scratch *tscratch; + if (!(tr->flags & TRACE_ARRAY_FL_LAST_BOOT)) return; @@ -6004,6 +6010,17 @@ static void update_last_data(struct trace_array *tr) /* Using current data now */ tr->text_delta = 0; + if (!tr->scratch) + return; + + tscratch = tr->scratch; + + /* Set the persistent ring buffer meta data to this address */ +#ifdef CONFIG_RANDOMIZE_BASE + tscratch->kaslr_addr = kaslr_offset(); +#else + tscratch->kaslr_addr = 0; +#endif tr->flags &= ~TRACE_ARRAY_FL_LAST_BOOT; } @@ -6817,6 +6834,7 @@ static ssize_t tracing_last_boot_read(struct file *filp, char __user *ubuf, size_t cnt, loff_t *ppos) { struct trace_array *tr = filp->private_data; + struct trace_scratch *tscratch = tr->scratch; struct seq_buf seq; char buf[64]; @@ -6829,8 +6847,8 @@ tracing_last_boot_read(struct file *filp, char __user *ubuf, size_t cnt, loff_t * Otherwise it shows the KASLR address from the previous boot which * should not be the same as the current boot. */ - if (tr->flags & TRACE_ARRAY_FL_LAST_BOOT) - seq_buf_printf(&seq, "%lx\t[kernel]\n", tr->kaslr_addr); + if (tscratch && (tr->flags & TRACE_ARRAY_FL_LAST_BOOT)) + seq_buf_printf(&seq, "%lx\t[kernel]\n", tscratch->kaslr_addr); else seq_buf_puts(&seq, "# Current\n"); @@ -9210,6 +9228,8 @@ static int allocate_trace_buffer(struct trace_array *tr, struct array_buffer *buf, int size) { enum ring_buffer_flags rb_flags; + struct trace_scratch *tscratch; + unsigned int scratch_size; rb_flags = tr->trace_flags & TRACE_ITER_OVERWRITE ? RB_FL_OVERWRITE : 0; @@ -9218,12 +9238,19 @@ allocate_trace_buffer(struct trace_array *tr, struct array_buffer *buf, int size if (tr->range_addr_start && tr->range_addr_size) { buf->buffer = ring_buffer_alloc_range(size, rb_flags, 0, tr->range_addr_start, - tr->range_addr_size, 0); + tr->range_addr_size, + sizeof(*tscratch)); + + tscratch = ring_buffer_meta_scratch(buf->buffer, &scratch_size); + if (tscratch) { + tr->scratch = tscratch; + tr->scratch_size = scratch_size; #ifdef CONFIG_RANDOMIZE_BASE - if (ring_buffer_last_boot_delta(buf->buffer, &tr->kaslr_addr)) - tr->text_delta = kaslr_offset() - tr->kaslr_addr; + if (tscratch->kaslr_addr) + tr->text_delta = kaslr_offset() - tscratch->kaslr_addr; #endif + } /* * This is basically the same as a mapped buffer, * with the same restrictions. diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index abe8169c3e87..3a020fb82a34 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -348,8 +348,11 @@ struct trace_array { unsigned int mapped; unsigned long range_addr_start; unsigned long range_addr_size; - unsigned long kaslr_addr; long text_delta; + void *scratch; /* pointer in persistent memory */ + int scratch_size; + + int buffer_disabled; struct trace_pid_list __rcu *filtered_pids; struct trace_pid_list __rcu *filtered_no_pids; @@ -367,7 +370,6 @@ struct trace_array { * CONFIG_TRACER_MAX_TRACE. */ arch_spinlock_t max_lock; - int buffer_disabled; #ifdef CONFIG_FTRACE_SYSCALLS int sys_refcount_enter; int sys_refcount_exit; -- 2.51.0 From 966b7d0e524da03a18ba1b111d0fa0d81e840f77 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 5 Mar 2025 11:45:44 -0500 Subject: [PATCH 11/16] module: Add module_for_each_mod() function The tracing system needs a way to save all the currently loaded modules and their addresses into persistent memory so that it can evaluate the addresses on a reboot from a crash. When the persistent memory trace starts, it will load the module addresses and names into the persistent memory. To do so, it will call the module_for_each_mod() function and pass it a function and data structure to get called on each loaded module. Then it can record the memory. This only implements that function. Cc: Luis Chamberlain Cc: Masami Hiramatsu Cc: Mark Rutland Cc: Mathieu Desnoyers Cc: Andrew Morton Cc: Sami Tolvanen Cc: Daniel Gomez Cc: linux-modules@vger.kernel.org Link: https://lore.kernel.org/20250305164608.962615966@goodmis.org Acked-by: Petr Pavlu Signed-off-by: Steven Rostedt (Google) --- include/linux/module.h | 6 ++++++ kernel/module/main.c | 13 +++++++++++++ 2 files changed, 19 insertions(+) diff --git a/include/linux/module.h b/include/linux/module.h index 30e5b19bafa9..9a71dd2cb11f 100644 --- a/include/linux/module.h +++ b/include/linux/module.h @@ -782,6 +782,8 @@ static inline void *module_writable_address(struct module *mod, void *loc) return __module_writable_address(mod, loc); } +void module_for_each_mod(int(*func)(struct module *mod, void *data), void *data); + #else /* !CONFIG_MODULES... */ static inline struct module *__module_address(unsigned long addr) @@ -894,6 +896,10 @@ static inline void *module_writable_address(struct module *mod, void *loc) { return loc; } + +static inline void module_for_each_mod(int(*func)(struct module *mod, void *data), void *data) +{ +} #endif /* CONFIG_MODULES */ #ifdef CONFIG_SYSFS diff --git a/kernel/module/main.c b/kernel/module/main.c index 1fb9ad289a6f..927a2e0ffd5f 100644 --- a/kernel/module/main.c +++ b/kernel/module/main.c @@ -3809,6 +3809,19 @@ bool is_module_text_address(unsigned long addr) return ret; } +void module_for_each_mod(int(*func)(struct module *mod, void *data), void *data) +{ + struct module *mod; + + guard(rcu)(); + list_for_each_entry_rcu(mod, &modules, list) { + if (mod->state == MODULE_STATE_UNFORMED) + continue; + if (func(mod, data)) + break; + } +} + /** * __module_text_address() - get the module whose code contains an address. * @addr: the address. -- 2.51.0 From fd39e48fe8eddfa84a2dbaf468fde8c2b2679211 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 5 Mar 2025 11:45:45 -0500 Subject: [PATCH 12/16] tracing: Have persistent trace instances save module addresses For trace instances that are mapped to persistent memory, have them use the scratch area to save the currently loaded modules. This will allow where the modules have been loaded on the next boot so that their addresses can be deciphered by using where they were loaded previously. Cc: Mark Rutland Cc: Mathieu Desnoyers Cc: Andrew Morton Link: https://lore.kernel.org/20250305164609.129741650@goodmis.org Reviewed-by: Masami Hiramatsu (Google) Signed-off-by: Steven Rostedt (Google) --- kernel/trace/trace.c | 98 ++++++++++++++++++++++++++++++++++++++++---- 1 file changed, 89 insertions(+), 9 deletions(-) diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index d4df56657b9d..a75e03994312 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -5988,14 +5988,60 @@ ssize_t tracing_resize_ring_buffer(struct trace_array *tr, return __tracing_resize_ring_buffer(tr, size, cpu_id); } +struct trace_mod_entry { + unsigned long mod_addr; + char mod_name[MODULE_NAME_LEN]; +}; + struct trace_scratch { unsigned long kaslr_addr; + unsigned long nr_entries; + struct trace_mod_entry entries[]; }; +static int save_mod(struct module *mod, void *data) +{ + struct trace_array *tr = data; + struct trace_scratch *tscratch; + struct trace_mod_entry *entry; + unsigned int size; + + tscratch = tr->scratch; + if (!tscratch) + return -1; + size = tr->scratch_size; + + if (struct_size(tscratch, entries, tscratch->nr_entries + 1) > size) + return -1; + + entry = &tscratch->entries[tscratch->nr_entries]; + + tscratch->nr_entries++; + + entry->mod_addr = (unsigned long)mod->mem[MOD_TEXT].base; + strscpy(entry->mod_name, mod->name); + + return 0; +} + static void update_last_data(struct trace_array *tr) { struct trace_scratch *tscratch; + if (!(tr->flags & TRACE_ARRAY_FL_BOOT)) + return; + + /* Reset the module list and reload them */ + if (tr->scratch) { + struct trace_scratch *tscratch = tr->scratch; + + memset(tscratch->entries, 0, + flex_array_size(tscratch, entries, tscratch->nr_entries)); + tscratch->nr_entries = 0; + + module_for_each_mod(save_mod, tr); + } + if (!(tr->flags & TRACE_ARRAY_FL_LAST_BOOT)) return; @@ -9224,6 +9270,46 @@ static struct dentry *trace_instance_dir; static void init_tracer_tracefs(struct trace_array *tr, struct dentry *d_tracer); +static void setup_trace_scratch(struct trace_array *tr, + struct trace_scratch *tscratch, unsigned int size) +{ + struct trace_mod_entry *entry; + + if (!tscratch) + return; + + tr->scratch = tscratch; + tr->scratch_size = size; + +#ifdef CONFIG_RANDOMIZE_BASE + if (tscratch->kaslr_addr) + tr->text_delta = kaslr_offset() - tscratch->kaslr_addr; +#endif + + if (struct_size(tscratch, entries, tscratch->nr_entries) > size) + goto reset; + + /* Check if each module name is a valid string */ + for (int i = 0; i < tscratch->nr_entries; i++) { + int n; + + entry = &tscratch->entries[i]; + + for (n = 0; n < MODULE_NAME_LEN; n++) { + if (entry->mod_name[n] == '\0') + break; + if (!isprint(entry->mod_name[n])) + goto reset; + } + if (n == MODULE_NAME_LEN) + goto reset; + } + return; + reset: + /* Invalid trace modules */ + memset(tscratch, 0, size); +} + static int allocate_trace_buffer(struct trace_array *tr, struct array_buffer *buf, int size) { @@ -9236,21 +9322,15 @@ allocate_trace_buffer(struct trace_array *tr, struct array_buffer *buf, int size buf->tr = tr; if (tr->range_addr_start && tr->range_addr_size) { + /* Add scratch buffer to handle 128 modules */ buf->buffer = ring_buffer_alloc_range(size, rb_flags, 0, tr->range_addr_start, tr->range_addr_size, - sizeof(*tscratch)); + struct_size(tscratch, entries, 128)); tscratch = ring_buffer_meta_scratch(buf->buffer, &scratch_size); - if (tscratch) { - tr->scratch = tscratch; - tr->scratch_size = scratch_size; + setup_trace_scratch(tr, tscratch, scratch_size); -#ifdef CONFIG_RANDOMIZE_BASE - if (tscratch->kaslr_addr) - tr->text_delta = kaslr_offset() - tscratch->kaslr_addr; -#endif - } /* * This is basically the same as a mapped buffer, * with the same restrictions. -- 2.51.0 From 1bd25a6f7173c005844cd0076718b3c87bccd891 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 5 Mar 2025 11:45:46 -0500 Subject: [PATCH 13/16] tracing: Show module names and addresses of last boot Add the last boot module's names and addresses to the last_boot_info file. This only shows the module information from a previous boot. If the buffer is started and is recording the current boot, this file still will only show "current". ~# cat instances/boot_mapped/last_boot_info 10c00000 [kernel] ffffffffc00ca000 usb_serial_simple ffffffffc00ae000 usbserial ffffffffc008b000 bfq ~# echo function > instances/boot_mapped/current_tracer ~# cat instances/boot_mapped/last_boot_info # Current Cc: Masami Hiramatsu Cc: Mark Rutland Cc: Mathieu Desnoyers Cc: Andrew Morton Link: https://lore.kernel.org/20250305164609.299186021@goodmis.org Signed-off-by: Steven Rostedt (Google) --- kernel/trace/trace.c | 102 +++++++++++++++++++++++++++++++++++++------ 1 file changed, 89 insertions(+), 13 deletions(-) diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index a75e03994312..d22f8d34b18d 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -5999,6 +5999,8 @@ struct trace_scratch { struct trace_mod_entry entries[]; }; +static DEFINE_MUTEX(scratch_mutex); + static int save_mod(struct module *mod, void *data) { struct trace_array *tr = data; @@ -6039,6 +6041,7 @@ static void update_last_data(struct trace_array *tr) flex_array_size(tscratch, entries, tscratch->nr_entries)); tscratch->nr_entries = 0; + guard(mutex)(&scratch_mutex); module_for_each_mod(save_mod, tr); } @@ -6876,15 +6879,47 @@ tracing_total_entries_read(struct file *filp, char __user *ubuf, return simple_read_from_buffer(ubuf, cnt, ppos, buf, r); } -static ssize_t -tracing_last_boot_read(struct file *filp, char __user *ubuf, size_t cnt, loff_t *ppos) +#define LAST_BOOT_HEADER ((void *)1) + +static void *l_next(struct seq_file *m, void *v, loff_t *pos) { - struct trace_array *tr = filp->private_data; + struct trace_array *tr = m->private; struct trace_scratch *tscratch = tr->scratch; - struct seq_buf seq; - char buf[64]; + unsigned int index = *pos; + + (*pos)++; - seq_buf_init(&seq, buf, 64); + if (*pos == 1) + return LAST_BOOT_HEADER; + + /* Only show offsets of the last boot data */ + if (!tscratch || !(tr->flags & TRACE_ARRAY_FL_LAST_BOOT)) + return NULL; + + /* *pos 0 is for the header, 1 is for the first module */ + index--; + + if (index >= tscratch->nr_entries) + return NULL; + + return &tscratch->entries[index]; +} + +static void *l_start(struct seq_file *m, loff_t *pos) +{ + mutex_lock(&scratch_mutex); + + return l_next(m, NULL, pos); +} + +static void l_stop(struct seq_file *m, void *p) +{ + mutex_unlock(&scratch_mutex); +} + +static void show_last_boot_header(struct seq_file *m, struct trace_array *tr) +{ + struct trace_scratch *tscratch = tr->scratch; /* * Do not leak KASLR address. This only shows the KASLR address of @@ -6894,11 +6929,52 @@ tracing_last_boot_read(struct file *filp, char __user *ubuf, size_t cnt, loff_t * should not be the same as the current boot. */ if (tscratch && (tr->flags & TRACE_ARRAY_FL_LAST_BOOT)) - seq_buf_printf(&seq, "%lx\t[kernel]\n", tscratch->kaslr_addr); + seq_printf(m, "%lx\t[kernel]\n", tscratch->kaslr_addr); else - seq_buf_puts(&seq, "# Current\n"); + seq_puts(m, "# Current\n"); +} - return simple_read_from_buffer(ubuf, cnt, ppos, buf, seq_buf_used(&seq)); +static int l_show(struct seq_file *m, void *v) +{ + struct trace_array *tr = m->private; + struct trace_mod_entry *entry = v; + + if (v == LAST_BOOT_HEADER) { + show_last_boot_header(m, tr); + return 0; + } + + seq_printf(m, "%lx\t%s\n", entry->mod_addr, entry->mod_name); + return 0; +} + +static const struct seq_operations last_boot_seq_ops = { + .start = l_start, + .next = l_next, + .stop = l_stop, + .show = l_show, +}; + +static int tracing_last_boot_open(struct inode *inode, struct file *file) +{ + struct trace_array *tr = inode->i_private; + struct seq_file *m; + int ret; + + ret = tracing_check_open_get_tr(tr); + if (ret) + return ret; + + ret = seq_open(file, &last_boot_seq_ops); + if (ret) { + trace_array_put(tr); + return ret; + } + + m = file->private_data; + m->private = tr; + + return 0; } static int tracing_buffer_meta_open(struct inode *inode, struct file *filp) @@ -7527,10 +7603,10 @@ static const struct file_operations trace_time_stamp_mode_fops = { }; static const struct file_operations last_boot_fops = { - .open = tracing_open_generic_tr, - .read = tracing_last_boot_read, - .llseek = generic_file_llseek, - .release = tracing_release_generic_tr, + .open = tracing_last_boot_open, + .read = seq_read, + .llseek = seq_lseek, + .release = tracing_seq_release, }; #ifdef CONFIG_TRACER_SNAPSHOT -- 2.51.0 From 5f3719f697c3fdfae5cd6805f10ac7a04b0f4e43 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 5 Mar 2025 11:45:47 -0500 Subject: [PATCH 14/16] tracing: Update modules to persistent instances when loaded When a module is loaded and a persistent buffer is actively tracing, add it to the list of modules in the persistent memory. Cc: Masami Hiramatsu Cc: Mark Rutland Cc: Mathieu Desnoyers Cc: Andrew Morton Link: https://lore.kernel.org/20250305164609.469844721@goodmis.org Signed-off-by: Steven Rostedt (Google) --- kernel/trace/trace.c | 27 +++++++++++++++++++++++++ kernel/trace/trace.h | 2 ++ kernel/trace/trace_events.c | 40 ++++++++++++++++++++++++++----------- 3 files changed, 57 insertions(+), 12 deletions(-) diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index d22f8d34b18d..66c1683fa1dd 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -10087,6 +10087,32 @@ static void trace_module_remove_evals(struct module *mod) static inline void trace_module_remove_evals(struct module *mod) { } #endif /* CONFIG_TRACE_EVAL_MAP_FILE */ +static bool trace_array_active(struct trace_array *tr) +{ + if (tr->current_trace != &nop_trace) + return true; + + /* 0 is no events, 1 is all disabled */ + return trace_events_enabled(tr, NULL) > 1; +} + +static void trace_module_record(struct module *mod) +{ + struct trace_array *tr; + + list_for_each_entry(tr, &ftrace_trace_arrays, list) { + /* Update any persistent trace array that has already been started */ + if ((tr->flags & (TRACE_ARRAY_FL_BOOT | TRACE_ARRAY_FL_LAST_BOOT)) == + TRACE_ARRAY_FL_BOOT) { + /* Only update if the trace array is active */ + if (trace_array_active(tr)) { + guard(mutex)(&scratch_mutex); + save_mod(mod, tr); + } + } + } +} + static int trace_module_notify(struct notifier_block *self, unsigned long val, void *data) { @@ -10095,6 +10121,7 @@ static int trace_module_notify(struct notifier_block *self, switch (val) { case MODULE_STATE_COMING: trace_module_add_evals(mod); + trace_module_record(mod); break; case MODULE_STATE_GOING: trace_module_remove_evals(mod); diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 3a020fb82a34..90493220c362 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -786,6 +786,8 @@ extern void trace_find_cmdline(int pid, char comm[]); extern int trace_find_tgid(int pid); extern void trace_event_follow_fork(struct trace_array *tr, bool enable); +extern int trace_events_enabled(struct trace_array *tr, const char *system); + #ifdef CONFIG_DYNAMIC_FTRACE extern unsigned long ftrace_update_tot_cnt; extern unsigned long ftrace_number_of_pages; diff --git a/kernel/trace/trace_events.c b/kernel/trace/trace_events.c index 513de9ceb80e..7b3ef1d26167 100644 --- a/kernel/trace/trace_events.c +++ b/kernel/trace/trace_events.c @@ -1818,28 +1818,28 @@ event_enable_write(struct file *filp, const char __user *ubuf, size_t cnt, return cnt; } -static ssize_t -system_enable_read(struct file *filp, char __user *ubuf, size_t cnt, - loff_t *ppos) +/* + * Returns: + * 0 : no events exist? + * 1 : all events are disabled + * 2 : all events are enabled + * 3 : some events are enabled and some are enabled + */ +int trace_events_enabled(struct trace_array *tr, const char *system) { - const char set_to_char[4] = { '?', '0', '1', 'X' }; - struct trace_subsystem_dir *dir = filp->private_data; - struct event_subsystem *system = dir->subsystem; struct trace_event_call *call; struct trace_event_file *file; - struct trace_array *tr = dir->tr; - char buf[2]; int set = 0; - int ret; - mutex_lock(&event_mutex); + guard(mutex)(&event_mutex); + list_for_each_entry(file, &tr->events, list) { call = file->event_call; if ((call->flags & TRACE_EVENT_FL_IGNORE_ENABLE) || !trace_event_name(call) || !call->class || !call->class->reg) continue; - if (system && strcmp(call->class->system, system->name) != 0) + if (system && strcmp(call->class->system, system) != 0) continue; /* @@ -1855,7 +1855,23 @@ system_enable_read(struct file *filp, char __user *ubuf, size_t cnt, if (set == 3) break; } - mutex_unlock(&event_mutex); + + return set; +} + +static ssize_t +system_enable_read(struct file *filp, char __user *ubuf, size_t cnt, + loff_t *ppos) +{ + const char set_to_char[4] = { '?', '0', '1', 'X' }; + struct trace_subsystem_dir *dir = filp->private_data; + struct event_subsystem *system = dir->subsystem; + struct trace_array *tr = dir->tr; + char buf[2]; + int set; + int ret; + + set = trace_events_enabled(tr, system ? system->name : NULL); buf[0] = set_to_char[set]; buf[1] = '\n'; -- 2.51.0 From 74e2498ccf7b303e7fdd881f58a849e884afb486 Mon Sep 17 00:00:00 2001 From: "Masami Hiramatsu (Google)" Date: Wed, 19 Feb 2025 00:08:58 +0900 Subject: [PATCH 15/16] mm/memblock: Add reserved memory release function Add reserve_mem_release_by_name() to release a reserved memory region with a given name. This allows us to release reserved memory which is defined by kernel cmdline, after boot. Signed-off-by: Masami Hiramatsu (Google) Acked-by: Mike Rapoport (Microsoft) Cc: Andrew Morton Cc: Mark Rutland Cc: Mathieu Desnoyers Cc: linux-mm@kvack.org Link: https://lore.kernel.org/173989133862.230693.14094993331347437600.stgit@devnote2 Signed-off-by: Steven Rostedt (Google) --- include/linux/mm.h | 1 + mm/memblock.c | 66 +++++++++++++++++++++++++++++++++++++--------- 2 files changed, 55 insertions(+), 12 deletions(-) diff --git a/include/linux/mm.h b/include/linux/mm.h index 7b1068ddcbb7..1ee9e7447485 100644 --- a/include/linux/mm.h +++ b/include/linux/mm.h @@ -4123,6 +4123,7 @@ void vma_pgtable_walk_begin(struct vm_area_struct *vma); void vma_pgtable_walk_end(struct vm_area_struct *vma); int reserve_mem_find_by_name(const char *name, phys_addr_t *start, phys_addr_t *size); +int reserve_mem_release_by_name(const char *name); #ifdef CONFIG_64BIT int do_mseal(unsigned long start, size_t len_in, unsigned long flags); diff --git a/mm/memblock.c b/mm/memblock.c index 95af35fd1389..8cd95f60015d 100644 --- a/mm/memblock.c +++ b/mm/memblock.c @@ -16,6 +16,7 @@ #include #include #include +#include #include #include @@ -2283,6 +2284,7 @@ struct reserve_mem_table { }; static struct reserve_mem_table reserved_mem_table[RESERVE_MEM_MAX_ENTRIES]; static int reserved_mem_count; +static DEFINE_MUTEX(reserve_mem_lock); /* Add wildcard region with a lookup name */ static void __init reserved_mem_add(phys_addr_t start, phys_addr_t size, @@ -2296,6 +2298,21 @@ static void __init reserved_mem_add(phys_addr_t start, phys_addr_t size, strscpy(map->name, name); } +static struct reserve_mem_table *reserve_mem_find_by_name_nolock(const char *name) +{ + struct reserve_mem_table *map; + int i; + + for (i = 0; i < reserved_mem_count; i++) { + map = &reserved_mem_table[i]; + if (!map->size) + continue; + if (strcmp(name, map->name) == 0) + return map; + } + return NULL; +} + /** * reserve_mem_find_by_name - Find reserved memory region with a given name * @name: The name that is attached to a reserved memory region @@ -2309,22 +2326,47 @@ static void __init reserved_mem_add(phys_addr_t start, phys_addr_t size, int reserve_mem_find_by_name(const char *name, phys_addr_t *start, phys_addr_t *size) { struct reserve_mem_table *map; - int i; - for (i = 0; i < reserved_mem_count; i++) { - map = &reserved_mem_table[i]; - if (!map->size) - continue; - if (strcmp(name, map->name) == 0) { - *start = map->start; - *size = map->size; - return 1; - } - } - return 0; + guard(mutex)(&reserve_mem_lock); + map = reserve_mem_find_by_name_nolock(name); + if (!map) + return 0; + + *start = map->start; + *size = map->size; + return 1; } EXPORT_SYMBOL_GPL(reserve_mem_find_by_name); +/** + * reserve_mem_release_by_name - Release reserved memory region with a given name + * @name: The name that is attatched to a reserved memory region + * + * Forcibly release the pages in the reserved memory region so that those memory + * can be used as free memory. After released the reserved region size becomes 0. + * + * Returns: 1 if released or 0 if not found. + */ +int reserve_mem_release_by_name(const char *name) +{ + char buf[RESERVE_MEM_NAME_SIZE + 12]; + struct reserve_mem_table *map; + void *start, *end; + + guard(mutex)(&reserve_mem_lock); + map = reserve_mem_find_by_name_nolock(name); + if (!map) + return 0; + + start = phys_to_virt(map->start); + end = start + map->size - 1; + snprintf(buf, sizeof(buf), "reserve_mem:%s", name); + free_reserved_area(start, end, 0, buf); + map->size = 0; + + return 1; +} + /* * Parse reserve_mem=nn:align:name */ -- 2.51.0 From fb6d03238e35f96cc1d6a5411ee1d684221d1c39 Mon Sep 17 00:00:00 2001 From: "Masami Hiramatsu (Google)" Date: Wed, 19 Feb 2025 00:09:08 +0900 Subject: [PATCH 16/16] tracing: Freeable reserved ring buffer Make the ring buffer on reserved memory to be freeable. This allows us to free the trace instance on the reserved memory without changing cmdline and rebooting. Even if we can not change the kernel cmdline for security reason, we can release the reserved memory for the ring buffer as free (available) memory. For example, boot kernel with reserved memory; "reserve_mem=20M:2M:trace trace_instance=boot_mapped^traceoff@trace" ~ # free total used free shared buff/cache available Mem: 1995548 50544 1927568 14964 17436 1911480 Swap: 0 0 0 ~ # rmdir /sys/kernel/tracing/instances/boot_mapped/ [ 23.704023] Freeing reserve_mem:trace memory: 20476K ~ # free total used free shared buff/cache available Mem: 2016024 41844 1956740 14968 17440 1940572 Swap: 0 0 0 Cc: Mark Rutland Cc: Mathieu Desnoyers Cc: Andrew Morton Cc: Mike Rapoport Link: https://lore.kernel.org/173989134814.230693.18199312930337815629.stgit@devnote2 Signed-off-by: Masami Hiramatsu (Google) Signed-off-by: Steven Rostedt (Google) --- kernel/trace/trace.c | 13 ++++++++++++- kernel/trace/trace.h | 1 + 2 files changed, 13 insertions(+), 1 deletion(-) diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 66c1683fa1dd..e7a14bb7d7aa 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -9479,6 +9479,9 @@ static void free_trace_buffers(struct trace_array *tr) #ifdef CONFIG_TRACER_MAX_TRACE free_trace_buffer(&tr->max_buffer); #endif + + if (tr->range_addr_start) + vunmap((void *)tr->range_addr_start); } static void init_trace_flags_index(struct trace_array *tr) @@ -9640,6 +9643,7 @@ trace_array_create_systems(const char *name, const char *systems, free_cpumask_var(tr->pipe_cpumask); free_cpumask_var(tr->tracing_cpumask); kfree_const(tr->system_names); + kfree(tr->range_name); kfree(tr->name); kfree(tr); @@ -9766,6 +9770,11 @@ static int __remove_instance(struct trace_array *tr) free_trace_buffers(tr); clear_tracing_err_log(tr); + if (tr->range_name) { + reserve_mem_release_by_name(tr->range_name); + kfree(tr->range_name); + } + for (i = 0; i < tr->nr_topts; i++) { kfree(tr->topts[i].topts); } @@ -10590,6 +10599,7 @@ __init static void enable_instances(void) bool traceoff = false; char *flag_delim; char *addr_delim; + char *rname __free(kfree) = NULL; tok = strsep(&curr_str, ","); @@ -10646,6 +10656,7 @@ __init static void enable_instances(void) pr_warn("Failed to map boot instance %s to %s\n", name, tok); continue; } + rname = kstrdup(tok, GFP_KERNEL); } if (start) { @@ -10682,7 +10693,7 @@ __init static void enable_instances(void) */ if (start) { tr->flags |= TRACE_ARRAY_FL_BOOT | TRACE_ARRAY_FL_LAST_BOOT; - tr->ref++; + tr->range_name = no_free_ptr(rname); } while ((tok = strsep(&curr_str, ","))) { diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 90493220c362..0d6efb8a1179 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -348,6 +348,7 @@ struct trace_array { unsigned int mapped; unsigned long range_addr_start; unsigned long range_addr_size; + char *range_name; long text_delta; void *scratch; /* pointer in persistent memory */ int scratch_size; -- 2.51.0