commit 097350d1c6e1f5808cae142006f18a0bbc57018d upstream.
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: dc4e2801d400b ("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>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 6784beada631800f2c5afd567e5628c843362cee upstream.
Fix the event trigger to accept redundant spaces in
the trigger input.
For example, these return -EINVAL
echo " traceon" > events/ftrace/print/trigger
echo "traceon if common_pid == 0" > events/ftrace/print/trigger
echo "disable_event:kmem:kmalloc " > events/ftrace/print/trigger
But these are hard to find what is wrong.
To fix this issue, use skip_spaces() to remove spaces
in front of actual tokens, and set NULL if there is no
token.
Link: http://lkml.kernel.org/r/159262476352.185015.5261566783045364186.stgit@devnote2
Cc: Tom Zanussi <zanussi@kernel.org>
Cc: stable@vger.kernel.org
Fixes: 85f2b08268c0 ("tracing: Add basic event trigger framework")
Reviewed-by: Tom Zanussi <zanussi@kernel.org>
Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit 1b0b283648163dae2a214ca28ed5a99f62a77319 ]
We use one blktrace per request_queue, that means one per the entire
disk. So we cannot run one blktrace on say /dev/vda and then /dev/vda1,
or just two calls on /dev/vda.
We check for concurrent setup only at the very end of the blktrace setup though.
If we try to run two concurrent blktraces on the same block device the
second one will fail, and the first one seems to go on. However when
one tries to kill the first one one will see things like this:
The kernel will show these:
```
debugfs: File 'dropped' in directory 'nvme1n1' already present!
debugfs: File 'msg' in directory 'nvme1n1' already present!
debugfs: File 'trace0' in directory 'nvme1n1' already present!
``
And userspace just sees this error message for the second call:
```
blktrace /dev/nvme1n1
BLKTRACESETUP(2) /dev/nvme1n1 failed: 5/Input/output error
```
The first userspace process #1 will also claim that the files
were taken underneath their nose as well. The files are taken
away form the first process given that when the second blktrace
fails, it will follow up with a BLKTRACESTOP and BLKTRACETEARDOWN.
This means that even if go-happy process #1 is waiting for blktrace
data, we *have* been asked to take teardown the blktrace.
This can easily be reproduced with break-blktrace [0] run_0005.sh test.
Just break out early if we know we're already going to fail, this will
prevent trying to create the files all over again, which we know still
exist.
[0] https://github.com/mcgrof/break-blktrace
Signed-off-by: Luis Chamberlain <mcgrof@kernel.org>
Signed-off-by: Jan Kara <jack@suse.cz>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 22d5bd6867364b41576a712755271a7d6161abd6 ]
Commit 60d53e2c3b75 ("tracing/probe: Split trace_event related data from
trace_probe") removed the trace_[ku]probe structure from the
trace_event_call->data pointer. As bpf_get_[ku]probe_info() were
forgotten in that change, fix them now. These functions are currently
only used by the bpf_task_fd_query() syscall handler to collect
information about a perf event.
Fixes: 60d53e2c3b75 ("tracing/probe: Split trace_event related data from trace_probe")
Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Yonghong Song <yhs@fb.com>
Acked-by: Masami Hiramatsu <mhiramat@kernel.org>
Link: https://lore.kernel.org/bpf/20200608124531.819838-1-jean-philippe@linaro.org
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 5aec598c456fe3c1b71a1202cbb42bdc2a643277 ]
The function blk_log_remap() can be simplified by removing the
call to get_pdu_remap() that copies the values into extra variable to
print the data, which also fixes the endiannness warning reported by
sparse.
Signed-off-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 48bc3cd3e07a1486f45d9971c75d6090976c3b1b ]
In blk_add_trace_spliti() blk_add_trace_bio_remap() use
blk_status_to_errno() to pass the error instead of pasing the bi_status.
This fixes the sparse warning.
Signed-off-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>
commit 78a5255ffb6a1af189a83e493d916ba1c54d8c75 upstream.
We have some rather random rules about when we accept the
"maybe-initialized" warnings, and when we don't.
For example, we consider it unreliable for gcc versions < 4.9, but also
if -O3 is enabled, or if optimizing for size. And then various kernel
config options disabled it, because they know that they trigger that
warning by confusing gcc sufficiently (ie PROFILE_ALL_BRANCHES).
And now gcc-10 seems to be introducing a lot of those warnings too, so
it falls under the same heading as 4.9 did.
At the same time, we have a very straightforward way to _enable_ that
warning when wanted: use "W=2" to enable more warnings.
So stop playing these ad-hoc games, and just disable that warning by
default, with the known and straight-forward "if you want to work on the
extra compiler warnings, use W=123".
Would it be great to have code that is always so obvious that it never
confuses the compiler whether a variable is used initialized or not?
Yes, it would. In a perfect world, the compilers would be smarter, and
our source code would be simpler.
That's currently not the world we live in, though.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 11f5efc3ab66284f7aaacc926e9351d658e2577b upstream.
x86_64 lazily maps in the vmalloc pages, and the way this works with per_cpu
areas can be complex, to say the least. Mappings may happen at boot up, and
if nothing synchronizes the page tables, those page mappings may not be
synced till they are used. This causes issues for anything that might touch
one of those mappings in the path of the page fault handler. When one of
those unmapped mappings is touched in the page fault handler, it will cause
another page fault, which in turn will cause a page fault, and leave us in
a loop of page faults.
Commit 763802b53a42 ("x86/mm: split vmalloc_sync_all()") split
vmalloc_sync_all() into vmalloc_sync_unmappings() and
vmalloc_sync_mappings(), as on system exit, it did not need to do a full
sync on x86_64 (although it still needed to be done on x86_32). By chance,
the vmalloc_sync_all() would synchronize the page mappings done at boot up
and prevent the per cpu area from being a problem for tracing in the page
fault handler. But when that synchronization in the exit of a task became a
nop, it caused the problem to appear.
Link: https://lore.kernel.org/r/20200429054857.66e8e333@oasis.local.home
Cc: stable@vger.kernel.org
Fixes: 737223fbca3b1 ("tracing: Consolidate buffer allocation code")
Reported-by: "Tzvetomir Stoyanov (VMware)" <tz.stoyanov@gmail.com>
Suggested-by: Joerg Roedel <jroedel@suse.de>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 6a13a0d7b4d1171ef9b80ad69abc37e1daa941b3 upstream.
Show maxactive parameter on kprobe_events.
This allows user to save the current configuration and
restore it without losing maxactive parameter.
Link: http://lkml.kernel.org/r/4762764a-6df7-bc93-ed60-e336146dce1f@gmail.com
Link: http://lkml.kernel.org/r/158503528846.22706.5549974121212526020.stgit@devnote2
Cc: stable@vger.kernel.org
Fixes: 696ced4fb1d76 ("tracing/kprobes: expose maxactive for kretprobe in kprobe_events")
Reported-by: Taeung Song <treeze.taeung@gmail.com>
Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit 1bc7896e9ef44fd77858b3ef0b8a6840be3a4494 ]
When experimenting with bpf_send_signal() helper in our production
environment (5.2 based), we experienced a deadlock in NMI mode:
#5 [ffffc9002219f770] queued_spin_lock_slowpath at ffffffff8110be24
#6 [ffffc9002219f770] _raw_spin_lock_irqsave at ffffffff81a43012
#7 [ffffc9002219f780] try_to_wake_up at ffffffff810e7ecd
#8 [ffffc9002219f7e0] signal_wake_up_state at ffffffff810c7b55
#9 [ffffc9002219f7f0] __send_signal at ffffffff810c8602
#10 [ffffc9002219f830] do_send_sig_info at ffffffff810ca31a
#11 [ffffc9002219f868] bpf_send_signal at ffffffff8119d227
#12 [ffffc9002219f988] bpf_overflow_handler at ffffffff811d4140
#13 [ffffc9002219f9e0] __perf_event_overflow at ffffffff811d68cf
#14 [ffffc9002219fa10] perf_swevent_overflow at ffffffff811d6a09
#15 [ffffc9002219fa38] ___perf_sw_event at ffffffff811e0f47
#16 [ffffc9002219fc30] __schedule at ffffffff81a3e04d
#17 [ffffc9002219fc90] schedule at ffffffff81a3e219
#18 [ffffc9002219fca0] futex_wait_queue_me at ffffffff8113d1b9
#19 [ffffc9002219fcd8] futex_wait at ffffffff8113e529
#20 [ffffc9002219fdf0] do_futex at ffffffff8113ffbc
#21 [ffffc9002219fec0] __x64_sys_futex at ffffffff81140d1c
#22 [ffffc9002219ff38] do_syscall_64 at ffffffff81002602
#23 [ffffc9002219ff50] entry_SYSCALL_64_after_hwframe at ffffffff81c00068
The above call stack is actually very similar to an issue
reported by Commit eac9153f2b58 ("bpf/stackmap: Fix deadlock with
rq_lock in bpf_get_stack()") by Song Liu. The only difference is
bpf_send_signal() helper instead of bpf_get_stack() helper.
The above deadlock is triggered with a perf_sw_event.
Similar to Commit eac9153f2b58, the below almost identical reproducer
used tracepoint point sched/sched_switch so the issue can be easily caught.
/* stress_test.c */
#include <stdio.h>
#include <stdlib.h>
#include <sys/mman.h>
#include <pthread.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#define THREAD_COUNT 1000
char *filename;
void *worker(void *p)
{
void *ptr;
int fd;
char *pptr;
fd = open(filename, O_RDONLY);
if (fd < 0)
return NULL;
while (1) {
struct timespec ts = {0, 1000 + rand() % 2000};
ptr = mmap(NULL, 4096 * 64, PROT_READ, MAP_PRIVATE, fd, 0);
usleep(1);
if (ptr == MAP_FAILED) {
printf("failed to mmap\n");
break;
}
munmap(ptr, 4096 * 64);
usleep(1);
pptr = malloc(1);
usleep(1);
pptr[0] = 1;
usleep(1);
free(pptr);
usleep(1);
nanosleep(&ts, NULL);
}
close(fd);
return NULL;
}
int main(int argc, char *argv[])
{
void *ptr;
int i;
pthread_t threads[THREAD_COUNT];
if (argc < 2)
return 0;
filename = argv[1];
for (i = 0; i < THREAD_COUNT; i++) {
if (pthread_create(threads + i, NULL, worker, NULL)) {
fprintf(stderr, "Error creating thread\n");
return 0;
}
}
for (i = 0; i < THREAD_COUNT; i++)
pthread_join(threads[i], NULL);
return 0;
}
and the following command:
1. run `stress_test /bin/ls` in one windown
2. hack bcc trace.py with the following change:
# --- a/tools/trace.py
# +++ b/tools/trace.py
@@ -513,6 +513,7 @@ BPF_PERF_OUTPUT(%s);
__data.tgid = __tgid;
__data.pid = __pid;
bpf_get_current_comm(&__data.comm, sizeof(__data.comm));
+ bpf_send_signal(10);
%s
%s
%s.perf_submit(%s, &__data, sizeof(__data));
3. in a different window run
./trace.py -p $(pidof stress_test) t:sched:sched_switch
The deadlock can be reproduced in our production system.
Similar to Song's fix, the fix is to delay sending signal if
irqs is disabled to avoid deadlocks involving with rq_lock.
With this change, my above stress-test in our production system
won't cause deadlock any more.
I also implemented a scale-down version of reproducer in the
selftest (a subsequent commit). With latest bpf-next,
it complains for the following potential deadlock.
[ 32.832450] -> #1 (&p->pi_lock){-.-.}:
[ 32.833100] _raw_spin_lock_irqsave+0x44/0x80
[ 32.833696] task_rq_lock+0x2c/0xa0
[ 32.834182] task_sched_runtime+0x59/0xd0
[ 32.834721] thread_group_cputime+0x250/0x270
[ 32.835304] thread_group_cputime_adjusted+0x2e/0x70
[ 32.835959] do_task_stat+0x8a7/0xb80
[ 32.836461] proc_single_show+0x51/0xb0
...
[ 32.839512] -> #0 (&(&sighand->siglock)->rlock){....}:
[ 32.840275] __lock_acquire+0x1358/0x1a20
[ 32.840826] lock_acquire+0xc7/0x1d0
[ 32.841309] _raw_spin_lock_irqsave+0x44/0x80
[ 32.841916] __lock_task_sighand+0x79/0x160
[ 32.842465] do_send_sig_info+0x35/0x90
[ 32.842977] bpf_send_signal+0xa/0x10
[ 32.843464] bpf_prog_bc13ed9e4d3163e3_send_signal_tp_sched+0x465/0x1000
[ 32.844301] trace_call_bpf+0x115/0x270
[ 32.844809] perf_trace_run_bpf_submit+0x4a/0xc0
[ 32.845411] perf_trace_sched_switch+0x10f/0x180
[ 32.846014] __schedule+0x45d/0x880
[ 32.846483] schedule+0x5f/0xd0
...
[ 32.853148] Chain exists of:
[ 32.853148] &(&sighand->siglock)->rlock --> &p->pi_lock --> &rq->lock
[ 32.853148]
[ 32.854451] Possible unsafe locking scenario:
[ 32.854451]
[ 32.855173] CPU0 CPU1
[ 32.855745] ---- ----
[ 32.856278] lock(&rq->lock);
[ 32.856671] lock(&p->pi_lock);
[ 32.857332] lock(&rq->lock);
[ 32.857999] lock(&(&sighand->siglock)->rlock);
Deadlock happens on CPU0 when it tries to acquire &sighand->siglock
but it has been held by CPU1 and CPU1 tries to grab &rq->lock
and cannot get it.
This is not exactly the callstack in our production environment,
but sympotom is similar and both locks are using spin_lock_irqsave()
to acquire the lock, and both involves rq_lock. The fix to delay
sending signal when irq is disabled also fixed this issue.
Signed-off-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Cc: Song Liu <songliubraving@fb.com>
Link: https://lore.kernel.org/bpf/20200304191104.2796501-1-yhs@fb.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Steven Rostedt (VMware)