[ Upstream commit 45d99ea451d0c30bfd4864f0fe485d7dac014902 ]
The 'bytes' info in file 'per_cpu/cpu<X>/stats' means the number of
bytes in cpu buffer that have not been consumed. However, currently
after consuming data by reading file 'trace_pipe', the 'bytes' info
was not changed as expected.
# cat per_cpu/cpu0/stats
entries: 0
overrun: 0
commit overrun: 0
bytes: 568 <--- 'bytes' is problematical !!!
oldest event ts: 8651.371479
now ts: 8653.912224
dropped events: 0
read events: 8
The root cause is incorrect stat on cpu_buffer->read_bytes. To fix it:
1. When stat 'read_bytes', account consumed event in rb_advance_reader();
2. When stat 'entries_bytes', exclude the discarded padding event which
is smaller than minimum size because it is invisible to reader. Then
use rb_page_commit() instead of BUF_PAGE_SIZE at where accounting for
page-based read/remove/overrun.
Also correct the comments of ring_buffer_bytes_cpu() in this patch.
Link: https://lore.kernel.org/linux-trace-kernel/20230921125425.1708423-1-zhengyejian1@huawei.com
Cc: stable@vger.kernel.org
Fixes: c64e148a3b ("trace: Add ring buffer stats to measure rate of events")
Signed-off-by: Zheng Yejian <zhengyejian1@huawei.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit a98151ad53b53f010ee364ec2fd06445b328578b ]
The comment refers to mm/slob.c which is being removed. It comes from
commit ed56829cb3 ("ring_buffer: reset buffer page when freeing") and
according to Steven the borrowed code was a page mapcount and mapping
reset, which was later removed by commit e4c2ce82ca ("ring_buffer:
allocate buffer page pointer"). Thus the comment is not accurate anyway,
remove it.
Link: https://lore.kernel.org/linux-trace-kernel/20230315142446.27040-1-vbabka@suse.cz
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Ingo Molnar <mingo@elte.hu>
Reported-by: Mike Rapoport <mike.rapoport@gmail.com>
Suggested-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Fixes: e4c2ce82ca ("ring_buffer: allocate buffer page pointer")
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Mukesh Ojha <quic_mojha@quicinc.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Stable-dep-of: 45d99ea451d0 ("ring-buffer: Fix bytes info in per_cpu buffer stats")
Signed-off-by: Sasha Levin <sashal@kernel.org>
commit 1e0cb399c7653462d9dadf8ab9425337c355d358 upstream.
It was discovered that the ring buffer polling was incorrectly stating
that read would not block, but that's because polling did not take into
account that reads will block if the "buffer-percent" was set. Instead,
the ring buffer polling would say reads would not block if there was any
data in the ring buffer. This was incorrect behavior from a user space
point of view. This was fixed by commit 42fb0a1e84ff by having the polling
code check if the ring buffer had more data than what the user specified
"buffer percent" had.
The problem now is that the polling code did not register itself to the
writer that it wanted to wait for a specific "full" value of the ring
buffer. The result was that the writer would wake the polling waiter
whenever there was a new event. The polling waiter would then wake up, see
that there's not enough data in the ring buffer to notify user space and
then go back to sleep. The next event would wake it up again.
Before the polling fix was added, the code would wake up around 100 times
for a hackbench 30 benchmark. After the "fix", due to the constant waking
of the writer, it would wake up over 11,0000 times! It would never leave
the kernel, so the user space behavior was still "correct", but this
definitely is not the desired effect.
To fix this, have the polling code add what it's waiting for to the
"shortest_full" variable, to tell the writer not to wake it up if the
buffer is not as full as it expects to be.
Note, after this fix, it appears that the waiter is now woken up around 2x
the times it was before (~200). This is a tremendous improvement from the
11,000 times, but I will need to spend some time to see why polling is
more aggressive in its wakeups than the read blocking code.
Link: https://lore.kernel.org/linux-trace-kernel/20230929180113.01c2cae3@rorschach.local.home
Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Fixes: 42fb0a1e84ff ("tracing/ring-buffer: Have polling block on watermark")
Reported-by: Julia Lawall <julia.lawall@inria.fr>
Tested-by: Julia Lawall <julia.lawall@inria.fr>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit 95a404bd60af6c4d9d8db01ad14fe8957ece31ca ]
When iterating over the ring buffer while the ring buffer is active, the
writer can corrupt the reader. There's barriers to help detect this and
handle it, but that code missed the case where the last event was at the
very end of the page and has only 4 bytes left.
The checks to detect the corruption by the writer to reads needs to see the
length of the event. If the length in the first 4 bytes is zero then the
length is stored in the second 4 bytes. But if the writer is in the process
of updating that code, there's a small window where the length in the first
4 bytes could be zero even though the length is only 4 bytes. That will
cause rb_event_length() to read the next 4 bytes which could happen to be off the
allocated page.
To protect against this, fail immediately if the next event pointer is
less than 8 bytes from the end of the commit (last byte of data), as all
events must be a minimum of 8 bytes anyway.
Link: https://lore.kernel.org/all/20230905141245.26470-1-Tze-nan.Wu@mediatek.com/
Link: https://lore.kernel.org/linux-trace-kernel/20230907122820.0899019c@gandalf.local.home
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Reported-by: Tze-nan Wu <Tze-nan.Wu@mediatek.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit f6bd2c92488c30ef53b5bd80c52f0a7eee9d545a ]
When user resize all trace ring buffer through file 'buffer_size_kb',
then in ring_buffer_resize(), kernel allocates buffer pages for each
cpu in a loop.
If the kernel preemption model is PREEMPT_NONE and there are many cpus
and there are many buffer pages to be allocated, it may not give up cpu
for a long time and finally cause a softlockup.
To avoid it, call cond_resched() after each cpu buffer allocation.
Link: https://lore.kernel.org/linux-trace-kernel/20230906081930.3939106-1-zhengyejian1@huawei.com
Cc: <mhiramat@kernel.org>
Signed-off-by: Zheng Yejian <zhengyejian1@huawei.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 2d093282b0d4357373497f65db6a05eb0c28b7c8 ]
When pages are removed in rb_remove_pages(), 'cpu_buffer->read' is set
to 0 in order to make sure any read iterators reset themselves. However,
this will mess 'entries' stating, see following steps:
# cd /sys/kernel/tracing/
# 1. Enlarge ring buffer prepare for later reducing:
# echo 20 > per_cpu/cpu0/buffer_size_kb
# 2. Write a log into ring buffer of cpu0:
# taskset -c 0 echo "hello1" > trace_marker
# 3. Read the log:
# cat per_cpu/cpu0/trace_pipe
<...>-332 [000] ..... 62.406844: tracing_mark_write: hello1
# 4. Stop reading and see the stats, now 0 entries, and 1 event readed:
# cat per_cpu/cpu0/stats
entries: 0
[...]
read events: 1
# 5. Reduce the ring buffer
# echo 7 > per_cpu/cpu0/buffer_size_kb
# 6. Now entries became unexpected 1 because actually no entries!!!
# cat per_cpu/cpu0/stats
entries: 1
[...]
read events: 0
To fix it, introduce 'page_removed' field to count total removed pages
since last reset, then use it to let read iterators reset themselves
instead of changing the 'read' pointer.
Link: https://lore.kernel.org/linux-trace-kernel/20230724054040.3489499-1-zhengyejian1@huawei.com
Cc: <mhiramat@kernel.org>
Cc: <vnagarnaik@google.com>
Fixes: 83f40318da ("ring-buffer: Make removal of ring buffer pages atomic")
Signed-off-by: Zheng Yejian <zhengyejian1@huawei.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
commit 7e42907f3a7b4ce3a2d1757f6d78336984daf8f5 upstream.
Soft lockup occurs when reading file 'trace_pipe':
watchdog: BUG: soft lockup - CPU#6 stuck for 22s! [cat:4488]
[...]
RIP: 0010:ring_buffer_empty_cpu+0xed/0x170
RSP: 0018:ffff88810dd6fc48 EFLAGS: 00000246
RAX: 0000000000000000 RBX: 0000000000000246 RCX: ffffffff93d1aaeb
RDX: ffff88810a280040 RSI: 0000000000000008 RDI: ffff88811164b218
RBP: ffff88811164b218 R08: 0000000000000000 R09: ffff88815156600f
R10: ffffed102a2acc01 R11: 0000000000000001 R12: 0000000051651901
R13: 0000000000000000 R14: ffff888115e49500 R15: 0000000000000000
[...]
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f8d853c2000 CR3: 000000010dcd8000 CR4: 00000000000006e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
__find_next_entry+0x1a8/0x4b0
? peek_next_entry+0x250/0x250
? down_write+0xa5/0x120
? down_write_killable+0x130/0x130
trace_find_next_entry_inc+0x3b/0x1d0
tracing_read_pipe+0x423/0xae0
? tracing_splice_read_pipe+0xcb0/0xcb0
vfs_read+0x16b/0x490
ksys_read+0x105/0x210
? __ia32_sys_pwrite64+0x200/0x200
? switch_fpu_return+0x108/0x220
do_syscall_64+0x33/0x40
entry_SYSCALL_64_after_hwframe+0x61/0xc6
Through the vmcore, I found it's because in tracing_read_pipe(),
ring_buffer_empty_cpu() found some buffer is not empty but then it
cannot read anything due to "rb_num_of_entries() == 0" always true,
Then it infinitely loop the procedure due to user buffer not been
filled, see following code path:
tracing_read_pipe() {
... ...
waitagain:
tracing_wait_pipe() // 1. find non-empty buffer here
trace_find_next_entry_inc() // 2. loop here try to find an entry
__find_next_entry()
ring_buffer_empty_cpu(); // 3. find non-empty buffer
peek_next_entry() // 4. but peek always return NULL
ring_buffer_peek()
rb_buffer_peek()
rb_get_reader_page()
// 5. because rb_num_of_entries() == 0 always true here
// then return NULL
// 6. user buffer not been filled so goto 'waitgain'
// and eventually leads to an deadloop in kernel!!!
}
By some analyzing, I found that when resetting ringbuffer, the 'entries'
of its pages are not all cleared (see rb_reset_cpu()). Then when reducing
the ringbuffer, and if some reduced pages exist dirty 'entries' data, they
will be added into 'cpu_buffer->overrun' (see rb_remove_pages()), which
cause wrong 'overrun' count and eventually cause the deadloop issue.
To fix it, we need to clear every pages in rb_reset_cpu().
Link: https://lore.kernel.org/linux-trace-kernel/20230708225144.3785600-1-zhengyejian1@huawei.com
Cc: stable@vger.kernel.org
Fixes: a5fb833172 ("ring-buffer: Fix uninitialized read_stamp")
Signed-off-by: Zheng Yejian <zhengyejian1@huawei.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit 7c339fb4d8577792378136c15fde773cfb863cb8 ]
In ring_buffer_reset_online_cpus, the buffer_size_kb write operation
may permanently fail if the cpu_online_mask changes between two
for_each_online_buffer_cpu loops. The number of increases and decreases
on both cpu_buffer->resize_disabled and cpu_buffer->record_disabled may be
inconsistent, causing some CPUs to have non-zero values for these atomic
variables after the function returns.
This issue can be reproduced by "echo 0 > trace" while hotplugging cpu.
After reproducing success, we can find out buffer_size_kb will not be
functional anymore.
To prevent leaving 'resize_disabled' and 'record_disabled' non-zero after
ring_buffer_reset_online_cpus returns, we ensure that each atomic variable
has been set up before atomic_sub() to it.
Link: https://lore.kernel.org/linux-trace-kernel/20230426062027.17451-1-Tze-nan.Wu@mediatek.com
Cc: stable@vger.kernel.org
Cc: <mhiramat@kernel.org>
Cc: npiggin@gmail.com
Fixes: b23d7a5f4a ("ring-buffer: speed up buffer resets by avoiding synchronize_rcu for each CPU")
Reviewed-by: Cheng-Jui Wang <cheng-jui.wang@mediatek.com>
Signed-off-by: Tze-nan Wu <Tze-nan.Wu@mediatek.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
commit 675751bb20634f981498c7d66161584080cc061e upstream.
If something was written to the buffer just before destruction,
it may be possible (maybe not in a real system, but it did
happen in ARCH=um with time-travel) to destroy the ringbuffer
before the IRQ work ran, leading this KASAN report (or a crash
without KASAN):
BUG: KASAN: slab-use-after-free in irq_work_run_list+0x11a/0x13a
Read of size 8 at addr 000000006d640a48 by task swapper/0
CPU: 0 PID: 0 Comm: swapper Tainted: G W O 6.3.0-rc1 #7
Stack:
60c4f20f 0c203d48 41b58ab3 60f224fc
600477fa 60f35687 60c4f20f 601273dd
00000008 6101eb00 6101eab0 615be548
Call Trace:
[<60047a58>] show_stack+0x25e/0x282
[<60c609e0>] dump_stack_lvl+0x96/0xfd
[<60c50d4c>] print_report+0x1a7/0x5a8
[<603078d3>] kasan_report+0xc1/0xe9
[<60308950>] __asan_report_load8_noabort+0x1b/0x1d
[<60232844>] irq_work_run_list+0x11a/0x13a
[<602328b4>] irq_work_tick+0x24/0x34
[<6017f9dc>] update_process_times+0x162/0x196
[<6019f335>] tick_sched_handle+0x1a4/0x1c3
[<6019fd9e>] tick_sched_timer+0x79/0x10c
[<601812b9>] __hrtimer_run_queues.constprop.0+0x425/0x695
[<60182913>] hrtimer_interrupt+0x16c/0x2c4
[<600486a3>] um_timer+0x164/0x183
[...]
Allocated by task 411:
save_stack_trace+0x99/0xb5
stack_trace_save+0x81/0x9b
kasan_save_stack+0x2d/0x54
kasan_set_track+0x34/0x3e
kasan_save_alloc_info+0x25/0x28
____kasan_kmalloc+0x8b/0x97
__kasan_kmalloc+0x10/0x12
__kmalloc+0xb2/0xe8
load_elf_phdrs+0xee/0x182
[...]
The buggy address belongs to the object at 000000006d640800
which belongs to the cache kmalloc-1k of size 1024
The buggy address is located 584 bytes inside of
freed 1024-byte region [000000006d640800, 000000006d640c00)
Add the appropriate irq_work_sync() so the work finishes before
the buffers are destroyed.
Prior to the commit in the Fixes tag below, there was only a
single global IRQ work, so this issue didn't exist.
Link: https://lore.kernel.org/linux-trace-kernel/20230427175920.a76159263122.I8295e405c44362a86c995e9c2c37e3e03810aa56@changeid
Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Fixes: 15693458c4 ("tracing/ring-buffer: Move poll wake ups into ring buffer code")
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 6455b6163d8c680366663cdb8c679514d55fc30c upstream.
When user reads file 'trace_pipe', kernel keeps printing following logs
that warn at "cpu_buffer->reader_page->read > rb_page_size(reader)" in
rb_get_reader_page(). It just looks like there's an infinite loop in
tracing_read_pipe(). This problem occurs several times on arm64 platform
when testing v5.10 and below.
Call trace:
rb_get_reader_page+0x248/0x1300
rb_buffer_peek+0x34/0x160
ring_buffer_peek+0xbc/0x224
peek_next_entry+0x98/0xbc
__find_next_entry+0xc4/0x1c0
trace_find_next_entry_inc+0x30/0x94
tracing_read_pipe+0x198/0x304
vfs_read+0xb4/0x1e0
ksys_read+0x74/0x100
__arm64_sys_read+0x24/0x30
el0_svc_common.constprop.0+0x7c/0x1bc
do_el0_svc+0x2c/0x94
el0_svc+0x20/0x30
el0_sync_handler+0xb0/0xb4
el0_sync+0x160/0x180
Then I dump the vmcore and look into the problematic per_cpu ring_buffer,
I found that tail_page/commit_page/reader_page are on the same page while
reader_page->read is obviously abnormal:
tail_page == commit_page == reader_page == {
.write = 0x100d20,
.read = 0x8f9f4805, // Far greater than 0xd20, obviously abnormal!!!
.entries = 0x10004c,
.real_end = 0x0,
.page = {
.time_stamp = 0x857257416af0,
.commit = 0xd20, // This page hasn't been full filled.
// .data[0...0xd20] seems normal.
}
}
The root cause is most likely the race that reader and writer are on the
same page while reader saw an event that not fully committed by writer.
To fix this, add memory barriers to make sure the reader can see the
content of what is committed. Since commit a0fcaaed0c46 ("ring-buffer: Fix
race between reset page and reading page") has added the read barrier in
rb_get_reader_page(), here we just need to add the write barrier.
Link: https://lore.kernel.org/linux-trace-kernel/20230325021247.2923907-1-zhengyejian1@huawei.com
Cc: stable@vger.kernel.org
Fixes: 77ae365eca ("ring-buffer: make lockless")
Suggested-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Zheng Yejian <zhengyejian1@huawei.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit 3e4272b9954094907f16861199728f14002fcaf6 ]
In a previous commit 7433632c9ff6, buffer, buffer->buffers and
buffer->buffers[cpu] in ring_buffer_wake_waiters() can be NULL,
and thus the related checks are added.
However, in the same call stack, these variables are also used in
ring_buffer_free_read_page():
tracing_buffers_release()
ring_buffer_wake_waiters(iter->array_buffer->buffer)
cpu_buffer = buffer->buffers[cpu] -> Add checks by previous commit
ring_buffer_free_read_page(iter->array_buffer->buffer)
cpu_buffer = buffer->buffers[cpu] -> No check
Thus, to avod possible null-pointer derefernces, the related checks
should be added.
These results are reported by a static tool designed by myself.
Link: https://lkml.kernel.org/r/20230113125501.760324-1-baijiaju1990@gmail.com
Reported-by: TOTE Robot <oslab@tsinghua.edu.cn>
Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
commit 8843e06f67b14f71c044bf6267b2387784c7e198 upstream.
It seems a data race between ring_buffer writing and integrity check.
That is, RB_FLAG of head_page is been updating, while at same time
RB_FLAG was cleared when doing integrity check rb_check_pages():
rb_check_pages() rb_handle_head_page():
-------- --------
rb_head_page_deactivate()
rb_head_page_set_normal()
rb_head_page_activate()
We do intergrity test of the list to check if the list is corrupted and
it is still worth doing it. So, let's refactor rb_check_pages() such that
we no longer clear and set flag during the list sanity checking.
[1] and [2] are the test to reproduce and the crash report respectively.
1:
``` read_trace.sh
while true;
do
# the "trace" file is closed after read
head -1 /sys/kernel/tracing/trace > /dev/null
done
```
``` repro.sh
sysctl -w kernel.panic_on_warn=1
# function tracer will writing enough data into ring_buffer
echo function > /sys/kernel/tracing/current_tracer
./read_trace.sh &
./read_trace.sh &
./read_trace.sh &
./read_trace.sh &
./read_trace.sh &
./read_trace.sh &
./read_trace.sh &
./read_trace.sh &
```
2:
------------[ cut here ]------------
WARNING: CPU: 9 PID: 62 at kernel/trace/ring_buffer.c:2653
rb_move_tail+0x450/0x470
Modules linked in:
CPU: 9 PID: 62 Comm: ksoftirqd/9 Tainted: G W 6.2.0-rc6+
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS
rel-1.15.0-0-g2dd4b9b3f840-prebuilt.qemu.org 04/01/2014
RIP: 0010:rb_move_tail+0x450/0x470
Code: ff ff 4c 89 c8 f0 4d 0f b1 02 48 89 c2 48 83 e2 fc 49 39 d0 75 24
83 e0 03 83 f8 02 0f 84 e1 fb ff ff 48 8b 57 10 f0 ff 42 08 <0f> 0b 83
f8 02 0f 84 ce fb ff ff e9 db
RSP: 0018:ffffb5564089bd00 EFLAGS: 00000203
RAX: 0000000000000000 RBX: ffff9db385a2bf81 RCX: ffffb5564089bd18
RDX: ffff9db281110100 RSI: 0000000000000fe4 RDI: ffff9db380145400
RBP: ffff9db385a2bf80 R08: ffff9db385a2bfc0 R09: ffff9db385a2bfc2
R10: ffff9db385a6c000 R11: ffff9db385a2bf80 R12: 0000000000000000
R13: 00000000000003e8 R14: ffff9db281110100 R15: ffffffffbb006108
FS: 0000000000000000(0000) GS:ffff9db3bdcc0000(0000)
knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00005602323024c8 CR3: 0000000022e0c000 CR4: 00000000000006e0
Call Trace:
<TASK>
ring_buffer_lock_reserve+0x136/0x360
? __do_softirq+0x287/0x2df
? __pfx_rcu_softirq_qs+0x10/0x10
trace_function+0x21/0x110
? __pfx_rcu_softirq_qs+0x10/0x10
? __do_softirq+0x287/0x2df
function_trace_call+0xf6/0x120
0xffffffffc038f097
? rcu_softirq_qs+0x5/0x140
rcu_softirq_qs+0x5/0x140
__do_softirq+0x287/0x2df
run_ksoftirqd+0x2a/0x30
smpboot_thread_fn+0x188/0x220
? __pfx_smpboot_thread_fn+0x10/0x10
kthread+0xe7/0x110
? __pfx_kthread+0x10/0x10
ret_from_fork+0x2c/0x50
</TASK>
---[ end trace 0000000000000000 ]---
[ crash report and test reproducer credit goes to Zheng Yejian]
Link: https://lore.kernel.org/linux-trace-kernel/1676376403-16462-1-git-send-email-quic_mojha@quicinc.com
Cc: <mhiramat@kernel.org>
Cc: stable@vger.kernel.org
Fixes: 1039221cc2 ("ring-buffer: Do not disable recording when there is an iterator")
Reported-by: Zheng Yejian <zhengyejian1@huawei.com>
Signed-off-by: Mukesh Ojha <quic_mojha@quicinc.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit 31029a8b2c7e656a0289194ef16415050ae4c4ac ]
The function ring_buffer_nr_dirty_pages() was created to find out how many
pages are filled in the ring buffer. There's two running counters. One is
incremented whenever a new page is touched (pages_touched) and the other
is whenever a page is read (pages_read). The dirty count is the number
touched minus the number read. This is used to determine if a blocked task
should be woken up if the percentage of the ring buffer it is waiting for
is hit.
The problem is that it does not take into account dropped pages (when the
new writes overwrite pages that were not read). And then the dirty pages
will always be greater than the percentage.
This makes the "buffer_percent" file inaccurate, as the number of dirty
pages end up always being larger than the percentage, event when it's not
and this causes user space to be woken up more than it wants to be.
Add a new counter to keep track of lost pages, and include that in the
accounting of dirty pages so that it is actually accurate.
Link: https://lkml.kernel.org/r/20221021123013.55fb6055@gandalf.local.home
Fixes: 2c2b0a78b3 ("ring-buffer: Add percentage of ring buffer full to wake up reader")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
commit 42fb0a1e84ff525ebe560e2baf9451ab69127e2b upstream.
Currently the way polling works on the ring buffer is broken. It will
return immediately if there's any data in the ring buffer whereas a read
will block until the watermark (defined by the tracefs buffer_percent file)
is hit.
That is, a select() or poll() will return as if there's data available,
but then the following read will block. This is broken for the way
select()s and poll()s are supposed to work.
Have the polling on the ring buffer also block the same way reads and
splice does on the ring buffer.
Link: https://lkml.kernel.org/r/20221020231427.41be3f26@gandalf.local.home
Cc: Linux Trace Kernel <linux-trace-kernel@vger.kernel.org>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Primiano Tucci <primiano@google.com>
Cc: stable@vger.kernel.org
Fixes: 1e0d6714ac ("ring-buffer: Do not wake up a splice waiter when page is not full")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit a0fcaaed0c46cf9399d3a2d6e0c87ddb3df0e044 upstream.
The ring buffer is broken up into sub buffers (currently of page size).
Each sub buffer has a pointer to its "tail" (the last event written to the
sub buffer). When a new event is requested, the tail is locally
incremented to cover the size of the new event. This is done in a way that
there is no need for locking.
If the tail goes past the end of the sub buffer, the process of moving to
the next sub buffer takes place. After setting the current sub buffer to
the next one, the previous one that had the tail go passed the end of the
sub buffer needs to be reset back to the original tail location (before
the new event was requested) and the rest of the sub buffer needs to be
"padded".
The race happens when a reader takes control of the sub buffer. As readers
do a "swap" of sub buffers from the ring buffer to get exclusive access to
the sub buffer, it replaces the "head" sub buffer with an empty sub buffer
that goes back into the writable portion of the ring buffer. This swap can
happen as soon as the writer moves to the next sub buffer and before it
updates the last sub buffer with padding.
Because the sub buffer can be released to the reader while the writer is
still updating the padding, it is possible for the reader to see the event
that goes past the end of the sub buffer. This can cause obvious issues.
To fix this, add a few memory barriers so that the reader definitely sees
the updates to the sub buffer, and also waits until the writer has put
back the "tail" of the sub buffer back to the last event that was written
on it.
To be paranoid, it will only spin for 1 second, otherwise it will
warn and shutdown the ring buffer code. 1 second should be enough as
the writer does have preemption disabled. If the writer doesn't move
within 1 second (with preemption disabled) something is horribly
wrong. No interrupt should last 1 second!
Link: https://lore.kernel.org/all/20220830120854.7545-1-jiazi.li@transsion.com/
Link: https://bugzilla.kernel.org/show_bug.cgi?id=216369
Link: https://lkml.kernel.org/r/20220929104909.0650a36c@gandalf.local.home
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: stable@vger.kernel.org
Fixes: c7b0930857 ("ring-buffer: prevent adding write in discarded area")
Reported-by: Jiazi.Li <jiazi.li@transsion.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit ec0bbc5ec5664dcee344f79373852117dc672c86 upstream.
The wake up waiters only checks the "wakeup_full" variable and not the
"full_waiters_pending". The full_waiters_pending is set when a waiter is
added to the wait queue. The wakeup_full is only set when an event is
triggered, and it clears the full_waiters_pending to avoid multiple calls
to irq_work_queue().
The irq_work callback really needs to check both wakeup_full as well as
full_waiters_pending such that this code can be used to wake up waiters
when a file is closed that represents the ring buffer and the waiters need
to be woken up.
Link: https://lkml.kernel.org/r/20220927231824.209460321@goodmis.org
Cc: stable@vger.kernel.org
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Fixes: 15693458c4 ("tracing/ring-buffer: Move poll wake ups into ring buffer code")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit fa8f4a89736b654125fb254b0db753ac68a5fced upstream.
If a page is partially read, and then the splice system call is run
against the ring buffer, it will always fail to read, no matter how much
is in the ring buffer. That's because the code path for a partial read of
the page does will fail if the "full" flag is set.
The splice system call wants full pages, so if the read of the ring buffer
is not yet full, it should return zero, and the splice will block. But if
a previous read was done, where the beginning has been consumed, it should
still be given to the splice caller if the rest of the page has been
written to.
This caused the splice command to never consume data in this scenario, and
let the ring buffer just fill up and lose events.
Link: https://lkml.kernel.org/r/20220927144317.46be6b80@gandalf.local.home
Cc: stable@vger.kernel.org
Fixes: 8789a9e7df ("ring-buffer: read page interface")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 51d157946666382e779f94c39891e8e9a020da78 upstream.
The resetting of the entire ring buffer use to simply go through and reset
each individual CPU buffer that had its own protection and synchronization.
But this was very slow, due to performing a synchronization for each CPU.
The code was reshuffled to do one disabling of all CPU buffers, followed
by a single RCU synchronization, and then the resetting of each of the CPU
buffers. But unfortunately, the mutex that prevented multiple occurrences
of resetting the buffer was not moved to the upper function, and there is
nothing to protect from it.
Take the ring buffer mutex around the global reset.
Cc: stable@vger.kernel.org
Fixes: b23d7a5f4a ("ring-buffer: speed up buffer resets by avoiding synchronize_rcu for each CPU")
Reported-by: "Tzvetomir Stoyanov (VMware)" <tz.stoyanov@gmail.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 67f0d6d9883c13174669f88adac4f0ee656cc16a upstream.
The "rb_per_cpu_empty()" misinterpret the condition (as not-empty) when
"head_page" and "commit_page" of "struct ring_buffer_per_cpu" points to
the same buffer page, whose "buffer_data_page" is empty and "read" field
is non-zero.
An error scenario could be constructed as followed (kernel perspective):
1. All pages in the buffer has been accessed by reader(s) so that all of
them will have non-zero "read" field.
2. Read and clear all buffer pages so that "rb_num_of_entries()" will
return 0 rendering there's no more data to read. It is also required
that the "read_page", "commit_page" and "tail_page" points to the same
page, while "head_page" is the next page of them.
3. Invoke "ring_buffer_lock_reserve()" with large enough "length"
so that it shot pass the end of current tail buffer page. Now the
"head_page", "commit_page" and "tail_page" points to the same page.
4. Discard current event with "ring_buffer_discard_commit()", so that
"head_page", "commit_page" and "tail_page" points to a page whose buffer
data page is now empty.
When the error scenario has been constructed, "tracing_read_pipe" will
be trapped inside a deadloop: "trace_empty()" returns 0 since
"rb_per_cpu_empty()" returns 0 when it hits the CPU containing such
constructed ring buffer. Then "trace_find_next_entry_inc()" always
return NULL since "rb_num_of_entries()" reports there's no more entry
to read. Finally "trace_seq_to_user()" returns "-EBUSY" spanking
"tracing_read_pipe" back to the start of the "waitagain" loop.
I've also written a proof-of-concept script to construct the scenario
and trigger the bug automatically, you can use it to trace and validate
my reasoning above:
https://github.com/aegistudio/RingBufferDetonator.git
Tests has been carried out on linux kernel 5.14-rc2
(2734d6c1b1a089fb593ef6a23d4b70903526fe0c), my fixed version
of kernel (for testing whether my update fixes the bug) and
some older kernels (for range of affected kernels). Test result is
also attached to the proof-of-concept repository.
Link: https://lore.kernel.org/linux-trace-devel/YPaNxsIlb2yjSi5Y@aegistudio/
Link: https://lore.kernel.org/linux-trace-devel/YPgrN85WL9VyrZ55@aegistudio
Cc: stable@vger.kernel.org
Fixes: bf41a158ca ("ring-buffer: make reentrant")
Suggested-by: Linus Torvalds <torvalds@linuxfoundation.org>
Signed-off-by: Haoran Luo <www@aegistudio.net>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 6f6be606e763f2da9fc21de00538c97fe4ca1492 upstream.
Part of the logic of the new time stamp code depends on the before_stamp and
the write_stamp to be different if the write_stamp does not match the last
event on the buffer, as it will be used to calculate the delta of the next
event written on the buffer.
The discard logic depends on this, as the next event to come in needs to
inject a full timestamp as it can not rely on the last event timestamp in
the buffer because it is unknown due to events after it being discarded. But
by changing the write_stamp back to the time before it, it forces the next
event to use a full time stamp, instead of relying on it.
The issue came when a full time stamp was used for the event, and
rb_time_delta() returns zero in that case. The update to the write_stamp
(which subtracts delta) made it not change. Then when the event is removed
from the buffer, because the before_stamp and write_stamp still match, the
next event written would calculate its delta from the write_stamp, but that
would be wrong as the write_stamp is of the time of the event that was
discarded.
In the case that the delta change being made to write_stamp is zero, set the
before_stamp to zero as well, and this will force the next event to inject a
full timestamp and not use the current write_stamp.
Cc: stable@vger.kernel.org
Fixes: a389d86f7f ("ring-buffer: Have nested events still record running time stamp")
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit adab66b71abfe206a020f11e561f4df41f0b2aba upstream.
It was believed that metag was the only architecture that required the ring
buffer to keep 8 byte words aligned on 8 byte architectures, and with its
removal, it was assumed that the ring buffer code did not need to handle
this case. It appears that sparc64 also requires this.
The following was reported on a sparc64 boot up:
kernel: futex hash table entries: 65536 (order: 9, 4194304 bytes, linear)
kernel: Running postponed tracer tests:
kernel: Testing tracer function:
kernel: Kernel unaligned access at TPC[552a20] trace_function+0x40/0x140
kernel: Kernel unaligned access at TPC[552a24] trace_function+0x44/0x140
kernel: Kernel unaligned access at TPC[552a20] trace_function+0x40/0x140
kernel: Kernel unaligned access at TPC[552a24] trace_function+0x44/0x140
kernel: Kernel unaligned access at TPC[552a20] trace_function+0x40/0x140
kernel: PASSED
Need to put back the 64BIT aligned code for the ring buffer.
Link: https://lore.kernel.org/r/CADxRZqzXQRYgKc=y-KV=S_yHL+Y8Ay2mh5ezeZUnpRvg+syWKw@mail.gmail.com
Cc: stable@vger.kernel.org
Fixes: 86b3de60a0 ("ring-buffer: Remove HAVE_64BIT_ALIGNED_ACCESS")
Reported-by: Anatoly Pugachev <matorola@gmail.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The current ring buffer logic checks to see if the updating of the event
buffer was interrupted, and if it is, it will try to fix up the before stamp
with the write stamp to make them equal again. This logic is flawed, because
if it is not interrupted, the two are guaranteed to be different, as the
current event just updated the before stamp before allocation. This
guarantees that the next event (this one or another interrupting one) will
think it interrupted the time updates of a previous event and inject an
absolute time stamp to compensate.
The correct logic is to always update the timestamps when traversing to a
new sub buffer.
Cc: stable@vger.kernel.org
Fixes: a389d86f7f ("ring-buffer: Have nested events still record running time stamp")
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
The recursion protection of the ring buffer depends on preempt_count() to be
correct. But it is possible that the ring buffer gets called after an
interrupt comes in but before it updates the preempt_count(). This will
trigger a false positive in the recursion code.
Use the same trick from the ftrace function callback recursion code which
uses a "transition" bit that gets set, to allow for a single recursion for
to handle transitions between contexts.
Cc: stable@vger.kernel.org
Fixes: 567cd4da54 ("ring-buffer: User context bit recursion checking")
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
We don't need to check the new buffer size, and the return value
had confused resize_buffer_duplicate_size().
...
ret = ring_buffer_resize(trace_buf->buffer,
per_cpu_ptr(size_buf->data,cpu_id)->entries, cpu_id);
if (ret == 0)
per_cpu_ptr(trace_buf->data, cpu_id)->entries =
per_cpu_ptr(size_buf->data, cpu_id)->entries;
...
Link: https://lkml.kernel.org/r/20201019142242.11560-1-hqjagain@gmail.com
Cc: stable@vger.kernel.org
Fixes: d60da506cb ("tracing: Add a resize function to make one buffer equivalent to another buffer")
Signed-off-by: Qiujun Huang <hqjagain@gmail.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Below race can come, if trace_open and resize of
cpu buffer is running parallely on different cpus
CPUX CPUY
ring_buffer_resize
atomic_read(&buffer->resize_disabled)
tracing_open
tracing_reset_online_cpus
ring_buffer_reset_cpu
rb_reset_cpu
rb_update_pages
remove/insert pages
resetting pointer
This race can cause data abort or some times infinte loop in
rb_remove_pages and rb_insert_pages while checking pages
for sanity.
Take buffer lock to fix this.
Link: https://lkml.kernel.org/r/1601976833-24377-1-git-send-email-gkohli@codeaurora.org
Cc: stable@vger.kernel.org
Fixes: b23d7a5f4a ("ring-buffer: speed up buffer resets by avoiding synchronize_rcu for each CPU")
Signed-off-by: Gaurav Kohli <gkohli@codeaurora.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Pull tracing updates from Steven Rostedt:
- The biggest news in that the tracing ring buffer can now time events
that interrupted other ring buffer events.
Before this change, if an interrupt came in while recording another
event, and that interrupt also had an event, those events would all
have the same time stamp as the event it interrupted.
Now, with the new design, those events will have a unique time stamp
and rightfully display the time for those events that were recorded
while interrupting another event.
- Bootconfig how has an "override" operator that lets the users have a
default config, but then add options to override the default.
- A fix was made to properly filter function graph tracing to the
ftrace PIDs. This came in at the end of the -rc cycle, and needs to
be backported.
- Several clean ups, performance updates, and minor fixes as well.
* tag 'trace-v5.9' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace: (39 commits)
tracing: Add trace_array_init_printk() to initialize instance trace_printk() buffers
kprobes: Fix compiler warning for !CONFIG_KPROBES_ON_FTRACE
tracing: Use trace_sched_process_free() instead of exit() for pid tracing
bootconfig: Fix to find the initargs correctly
Documentation: bootconfig: Add bootconfig override operator
tools/bootconfig: Add testcases for value override operator
lib/bootconfig: Add override operator support
kprobes: Remove show_registers() function prototype
tracing/uprobe: Remove dead code in trace_uprobe_register()
kprobes: Fix NULL pointer dereference at kprobe_ftrace_handler
ftrace: Fix ftrace_trace_task return value
tracepoint: Use __used attribute definitions from compiler_attributes.h
tracepoint: Mark __tracepoint_string's __used
trace : Have tracing buffer info use kvzalloc instead of kzalloc
tracing: Remove outdated comment in stack handling
ftrace: Do not let direct or IPMODIFY ftrace_ops be added to module and set trampolines
ftrace: Setup correct FTRACE_FL_REGS flags for module
tracing/hwlat: Honor the tracing_cpumask
tracing/hwlat: Drop the duplicate assignment in start_kthread()
tracing: Save one trace_event->type by using __TRACE_LAST_TYPE
...
After tweaking the ring buffer to be a bit faster, a warning is triggering
on one of my machines, and causing my tests to fail. This warning is caused
when the delta (current time stamp minus previous time stamp), is larger
than the max time held by the ring buffer (59 bits).
If the clock were to go backwards slightly, this would then easily trigger
this warning. The machine that it triggered on, the clock did go backwards
by around 450 nanoseconds, and this happened after a recalibration of the
TSC clock. Now that the ring buffer is faster, it detects this, and the
delta that is used larger than the max, the warning is triggered and my test
fails.
To handle the clock going backwards, look at the saved before and after time
stamps. If they are the same, it means that the current event did not
interrupt another event, and that those timestamp are of a previous event
that was recorded. If the max delta is triggered, look at those time stamps,
make sure they are the same, then use them to compare with the current
timestamp. If the current timestamp is less than the before/after time
stamps, then that means the clock being used went backward.
Print out a message that this has happened, but do not warn about it (and
only print the message once).
Still do the warning if the delta is indeed larger than what can be used.
Also remove the unneeded KERN_WARNING from the WARN_ONCE() print.
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
After doing some benchmarks and examining the code, I found that the ring
buffer clock calls were quite expensive, and noticed that it uses
retpolines. This is because the ring buffer clock is programmable, and can
be set. But in most cases it simply uses the fastest ns unit clock which is
the trace_clock_local(). For RETPOLINE builds, checking if the ring buffer
clock is set to trace_clock_local() and then calling it directly has brought
the time of an event on my i7 box from an average of 93 nanoseconds an event
down to 83 nanoseconds an event, and the minimum time from 81 nanoseconds to
68 nanoseconds!
Suggested-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Make a helper function rb_add_timestamp() that moves the adding of the
extended time stamps into its own function. Also, remove the noinline and
inline for the functions it calls, as recent benchmarks appear they do not
make a difference (just let gcc decide).
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Reorganize a little the logic to handle adding the absolute time stamp,
extended and forced time stamps, in such a way to remove a branch or two.
This is just a micro optimization.
Also add before and after time stamps to the rb_event_info structure to
display those values in the rb_check_timestamps() code, if something were to
go wrong.
Suggested-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
It is the uncommon case where an event crosses a sub buffer boundary (page)
mark that check at the end of reserving an event as unlikely.
Suggested-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
On a 144 thread system, `perf ftrace` takes about 20 seconds to start
up, due to calling synchronize_rcu() for each CPU.
cat /proc/108560/stack
0xc0003e7eb336f470
__switch_to+0x2e0/0x480
__wait_rcu_gp+0x20c/0x220
synchronize_rcu+0x9c/0xc0
ring_buffer_reset_cpu+0x88/0x2e0
tracing_reset_online_cpus+0x84/0xe0
tracing_open+0x1d4/0x1f0
On a system with 10x more threads, it starts to become an annoyance.
Batch these up so we disable all the per-cpu buffers first, then
synchronize_rcu() once, then reset each of the buffers. This brings
the time down to about 0.5s.
Link: https://lkml.kernel.org/r/20200625053403.2386972-1-npiggin@gmail.com
Tested-by: Anton Blanchard <anton@ozlabs.org>
Acked-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
After a discussion with the new time algorithm to have nested events still
have proper time keeping but required using local64_t atomic operations.
Mathieu was concerned about the performance this would have on 32 bit
machines, as in most cases, atomic 64 bit operations on them can be
expensive.
As the ring buffer's timing needs do not require full features of local64_t,
a wrapper is made to implement a new rb_time_t operation that uses two longs
on 32 bit machines but still uses the local64_t operations on 64 bit
machines. There's a switch that can be made in the file to force 64 bit to
use the 32 bit version just for testing purposes.
All reads do not need to succeed if a read happened while the stamp being
read is in the process of being updated. The requirement is that all reads
must succed that were done by an interrupting event (where this event was
interrupted by another event that did the write). Or if the event itself did
the write first. That is: rb_time_set(t, x) followed by rb_time_read(t) will
always succeed (even if it gets interrupted by another event that writes to
t. The result of the read will be either the previous set, or a set
performed by an interrupting event.
If the read is done by an event that interrupted another event that was in
the process of setting the time stamp, and no other event came along to
write to that time stamp, it will fail and the rb_time_read() will return
that it failed (the value to read will be undefined).
A set will always write to the time stamp and return with a valid time
stamp, such that any read after it will be valid.
A cmpxchg may fail if it interrupted an event that was in the process of
updating the time stamp just like the reads do. Other than that, it will act
like a normal cmpxchg.
The way this works is that the rb_time_t is made of of three fields. A cnt,
that gets updated atomically everyting a modification is made. A top that
represents the most significant 30 bits of the time, and a bottom to
represent the least significant 30 bits of the time. Notice, that the time
values is only 60 bits long (where the ring buffer only uses 59 bits, which
gives us 18 years of nanoseconds!).
The top two bits of both the top and bottom is a 2 bit counter that gets set
by the value of the least two significant bits of the cnt. A read of the top
and the bottom where both the top and bottom have the same most significant
top 2 bits, are considered a match and a valid 60 bit number can be created
from it. If they do not match, then the number is considered invalid, and
this must only happen if an event interrupted another event in the midst of
updating the time stamp.
This is only used for 32 bits machines as 64 bit machines can get better
performance out of the local64_t. This has been tested heavily by forcing 64
bit to use this logic.
Link: https://lore.kernel.org/r/20200625225345.18cf5881@oasis.local.home
Link: http://lkml.kernel.org/r/20200629025259.309232719@goodmis.org
Inspired-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Instead of calling out the absolute test for each time to check if the
ring buffer wants absolute time stamps for all its recording, incorporate it
with the add_timestamp field and turn it into flags for faster processing
between wanting a absolute tag and needing to force one.
Link: http://lkml.kernel.org/r/20200629025259.154892368@goodmis.org
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Up until now, if an event is interrupted while it is recorded by an
interrupt, and that interrupt records events, the time of those events will
all be the same. This is because events only record the delta of the time
since the previous event (or beginning of a page), and to handle updating
the time keeping for that of nested events is extremely racy. After years of
thinking about this and several failed attempts, I finally have a solution
to solve this puzzle.
The problem is that you need to atomically calculate the delta and then
update the time stamp you made the delta from, as well as then record it
into the buffer, all this while at any time an interrupt can come in and
do the same thing. This is easy to solve with heavy weight atomics, but that
would be detrimental to the performance of the ring buffer. The current
state of affairs sacrificed the time deltas for nested events for
performance.
The reason for previous failed attempts at solving this puzzle was because I
was trying to completely avoid slow atomic operations like cmpxchg. I final
came to the conclusion to always avoid cmpxchg is not possible, which is why
those previous attempts always failed. But it is possible to pick one path
(the most common case) and avoid cmpxchg in that path, which is the "fast
path". The most common case is that an event will not be interrupted and
have other events added into it. An event can detect if it has
interrupted another event, and for these cases we can make it the slow
path and use the heavy operations like cmpxchg.
One more player was added to the game that made this possible, and that is
the "absolute timestamp" (by Tom Zanussi) that allows us to inject a full 59
bit time stamp. (Of course this breaks if a machine is running for more than
18 years without a reboot!).
There's barrier() placements around for being paranoid, even when they
are not needed because of other atomic functions near by. But those
should not hurt, as if they are not needed, they basically become a nop.
Note, this also makes the race window much smaller, which means there
are less slow paths to slow down the performance.
The basic idea is that there's two main paths taken.
1) Not being interrupted between time stamps and reserving buffer space.
In this case, the time stamps taken are true to the location in the
buffer.
2) Was interrupted by another path between taking time stamps and reserving
buffer space.
The objective is to know what the delta is from the last reserved location
in the buffer.
As it is possible to detect if an event is interrupting another event before
reserving data, space is added to the length to be reserved to inject a full
time stamp along with the event being reserved.
When an event is not interrupted, the write stamp is always the time of the
last event written to the buffer.
In path 1, there's two sub paths we care about:
a) The event did not interrupt another event.
b) The event interrupted another event.
In case a, as the write stamp was read and known to be correct, the delta
between the current time stamp and the write stamp is the delta between the
current event and the previously recorded event.
In case b, extra space was reserved to just put the full time stamp into the
buffer. Which is done, as stated, in this path the time stamp taken is known
to match the location in the buffer.
In path 2, there's also two sub paths we care about:
a) The event was not interrupted by another event since it reserved space
on the buffer and re-reading the write stamp.
b) The event was interrupted by another event.
In case a, the write stamp is that of the last event that interrupted this
event between taking the time stamps and reserving. As no event came in
after re-reading the write stamp, that event is known to be the time of the
event directly before this event and the delta can be the new time stamp and
the write stamp.
In case b, one or more events came in between reserving the event and
re-reading he write stamp. Since this event's buffer reservation is between
other events at this path, there's no way to know what the delta is. But
because an event interrupted this event after it started, its fine to just
give a zero delta, and take the same time stamp as the events that happened
within the event being recorded.
Here's the implementation of the design of this solution:
All this is per cpu, and only needs to worry about nested events (not
parallel events).
The players:
write_tail: The index in the buffer where new events can be written to.
It is incremented via local_add() to reserve space for a new event.
before_stamp: A time stamp set by all events before reserving space.
write_stamp: A time stamp updated by events after it has successfully
reserved space.
/* Save the current position of write */
[A] w = local_read(write_tail);
barrier();
/* Read both before and write stamps before touching anything */
before = local_read(before_stamp);
after = local_read(write_stamp);
barrier();
/*
* If before and after are the same, then this event is not
* interrupting a time update. If it is, then reserve space for adding
* a full time stamp (this can turn into a time extend which is
* just an extended time delta but fill up the extra space).
*/
if (after != before)
abs = true;
ts = clock();
/* Now update the before_stamp (everyone does this!) */
[B] local_set(before_stamp, ts);
/* Now reserve space on the buffer */
[C] write = local_add_return(len, write_tail);
/* Set tail to be were this event's data is */
tail = write - len;
if (w == tail) {
/* Nothing interrupted this between A and C */
[D] local_set(write_stamp, ts);
barrier();
[E] save_before = local_read(before_stamp);
if (!abs) {
/* This did not interrupt a time update */
delta = ts - after;
} else {
delta = ts; /* The full time stamp will be in use */
}
if (ts != save_before) {
/* slow path - Was interrupted between C and E */
/* The update to write_stamp could have overwritten the update to
* it by the interrupting event, but before and after should be
* the same for all completed top events */
after = local_read(write_stamp);
if (save_before > after)
local_cmpxchg(write_stamp, after, save_before);
}
} else {
/* slow path - Interrupted between A and C */
after = local_read(write_stamp);
temp_ts = clock();
barrier();
[F] if (write == local_read(write_tail) && after < temp_ts) {
/* This was not interrupted since C and F
* The last write_stamp is still valid for the previous event
* in the buffer. */
delta = temp_ts - after;
/* OK to keep this new time stamp */
ts = temp_ts;
} else {
/* Interrupted between C and F
* Well, there's no use to try to know what the time stamp
* is for the previous event. Just set delta to zero and
* be the same time as that event that interrupted us before
* the reservation of the buffer. */
delta = 0;
}
/* No need to use full timestamps here */
abs = 0;
}
Link: https://lkml.kernel.org/r/20200625094454.732790f7@oasis.local.home
Link: https://lore.kernel.org/r/20200627010041.517736087@goodmis.org
Link: http://lkml.kernel.org/r/20200629025258.957440797@goodmis.org
Reviewed-by: Masami Hiramatsu <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Currently the ring buffer makes events that happen in interrupts that preempt
another event have a delta of zero. (Hopefully we can change this soon). But
this is to deal with the races of updating a global counter with lockless
and nesting functions updating deltas.
With the addition of absolute time stamps, the time extend didn't follow
this rule. A time extend can happen if two events happen longer than 2^27
nanoseconds appart, as the delta time field in each event is only 27 bits.
If that happens, then a time extend is injected with 2^59 bits of
nanoseconds to use (18 years). But if the 2^27 nanoseconds happen between
two events, and as it is writing the event, an interrupt triggers, it will
see the 2^27 difference as well and inject a time extend of its own. But a
recent change made the time extend logic not take into account the nesting,
and this can cause two time extend deltas to happen moving the time stamp
much further ahead than the current time. This gets all reset when the ring
buffer moves to the next page, but that can cause time to appear to go
backwards.
This was observed in a trace-cmd recording, and since the data is saved in a
file, with trace-cmd report --debug, it was possible to see that this indeed
did happen!
bash-52501 110d... 81778.908247: sched_switch: bash:52501 [120] S ==> swapper/110:0 [120] [12770284:0x2e8:64]
<idle>-0 110d... 81778.908757: sched_switch: swapper/110:0 [120] R ==> bash:52501 [120] [509947:0x32c:64]
TIME EXTEND: delta:306454770 length:0
bash-52501 110.... 81779.215212: sched_swap_numa: src_pid=52501 src_tgid=52388 src_ngid=52501 src_cpu=110 src_nid=2 dst_pid=52509 dst_tgid=52388 dst_ngid=52501 dst_cpu=49 dst_nid=1 [0:0x378:48]
TIME EXTEND: delta:306458165 length:0
bash-52501 110dNh. 81779.521670: sched_wakeup: migration/110:565 [0] success=1 CPU:110 [0:0x3b4:40]
and at the next page, caused the time to go backwards:
bash-52504 110d... 81779.685411: sched_switch: bash:52504 [120] S ==> swapper/110:0 [120] [8347057:0xfb4:64]
CPU:110 [SUBBUFFER START] [81779379165886:0x1320000]
<idle>-0 110dN.. 81779.379166: sched_wakeup: bash:52504 [120] success=1 CPU:110 [0:0x10:40]
<idle>-0 110d... 81779.379167: sched_switch: swapper/110:0 [120] R ==> bash:52504 [120] [1168:0x3c:64]
Link: https://lkml.kernel.org/r/20200622151815.345d1bf5@oasis.local.home
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Tom Zanussi <zanussi@kernel.org>
Cc: stable@vger.kernel.org
Fixes: dc4e2801d4 ("ring-buffer: Redefine the unimplemented RINGBUF_TYPE_TIME_STAMP")
Reported-by: Julia Lawall <julia.lawall@inria.fr>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
There's a lot of checks to make sure the ring buffer is working, and if an
anomaly is detected, it safely shuts itself down. But there's a few cases
that it will call BUG(), which defeats the point of being safe (it crashes
the kernel when an anomaly is found!). There's no reason for them. Switch
them all to either WARN_ON_ONCE() (when no ring buffer descriptor is present),
or to RB_WARN_ON() (when a ring buffer descriptor is present).
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
If the function tracer is running and the trace file is read (which uses the
ring buffer iterator), the iterator can get in sync with the writes, and
caues it to fail to find a page with content it can read three times. This
causes a warning and deactivation of the ring buffer code.
Looking at the other cases of failure to get an event, it appears that
there's a chance that the writer could cause them too. Since the iterator is
a "best effort" to read the ring buffer if there's an active writer (the
consumer reader is made for this case "see trace_pipe"), if it fails to get
an event after three tries, simply give up and return NULL. Don't warn, nor
disable the ring buffer on this failure.
Link: https://lore.kernel.org/r/20200429090508.GG5770@shao2-debian
Reported-by: kernel test robot <lkp@intel.com>
Fixes: ff84c50cfb ("ring-buffer: Do not die if rb_iter_peek() fails more than thrice")
Tested-by: Sven Schnelle <svens@linux.ibm.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Have the ring_buffer_iterator set a flag if events were dropped as it were
to go and peek at the next event. Have the trace file display this fact if
it happened with a "LOST EVENTS" message.
Link: http://lkml.kernel.org/r/20200317213417.045858900@goodmis.org
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
When the ring buffer becomes writable for even when the trace file is read,
it must still not be resized. But since tracers can be activated while the
trace file is being read, the irqsoff tracer can modify the per CPU buffers,
and this can cause the reader of the trace file to update the wrong buffer's
resize disable bit, as the irqsoff tracer swaps out cpu buffers.
By making the resize disable per cpu_buffer, it makes the update follow the
per cpu_buffer even if it's swapped out with the snapshot buffer and keeps
the release of the trace file modifying the same data as the open did.
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
As it is fine to perform several "peeks" of event data in the ring buffer
via the iterator before moving it forward, do not re-read the event, just
return what was read before. Otherwise, it can cause inconsistent results,
especially when testing multiple CPU buffers to interleave them.
Link: http://lkml.kernel.org/r/20200317213416.592032170@goodmis.org
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
