linux.git/kernel/trace, branch v6.6.56

tracing/timerlat: Fix duplicated kthread creation due to CPU online/offline

2024-10-10T09:57:59+00:00

commit 0bb0a5c12ecf36ad561542bbb95f96355e036a02 upstream.

osnoise_hotplug_workfn() is the asynchronous online callback for
"trace/osnoise:online". It may be congested when a CPU goes online and
offline repeatedly and is invoked for multiple times after a certain
online.

This will lead to kthread leak and timer corruption. Add a check
in start_kthread() to prevent this situation.

Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu 
Cc: Mathieu Desnoyers 
Link: https://lore.kernel.org/20240924094515.3561410-2-liwei391@huawei.com
Fixes: c8895e271f79 ("trace/osnoise: Support hotplug operations")
Signed-off-by: Wei Li 
Signed-off-by: Steven Rostedt (Google) 
Signed-off-by: Greg Kroah-Hartman

tracing/timerlat: Fix a race during cpuhp processing

2024-10-10T09:57:59+00:00

commit 829e0c9f0855f26b3ae830d17b24aec103f7e915 upstream.

There is another found exception that the "timerlat/1" thread was
scheduled on CPU0, and lead to timer corruption finally:

```
ODEBUG: init active (active state 0) object: ffff888237c2e108 object type: hrtimer hint: timerlat_irq+0x0/0x220
WARNING: CPU: 0 PID: 426 at lib/debugobjects.c:518 debug_print_object+0x7d/0xb0
Modules linked in:
CPU: 0 UID: 0 PID: 426 Comm: timerlat/1 Not tainted 6.11.0-rc7+ #45
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.13.0-1ubuntu1.1 04/01/2014
RIP: 0010:debug_print_object+0x7d/0xb0
...
Call Trace:
 
 ? __warn+0x7c/0x110
 ? debug_print_object+0x7d/0xb0
 ? report_bug+0xf1/0x1d0
 ? prb_read_valid+0x17/0x20
 ? handle_bug+0x3f/0x70
 ? exc_invalid_op+0x13/0x60
 ? asm_exc_invalid_op+0x16/0x20
 ? debug_print_object+0x7d/0xb0
 ? debug_print_object+0x7d/0xb0
 ? __pfx_timerlat_irq+0x10/0x10
 __debug_object_init+0x110/0x150
 hrtimer_init+0x1d/0x60
 timerlat_main+0xab/0x2d0
 ? __pfx_timerlat_main+0x10/0x10
 kthread+0xb7/0xe0
 ? __pfx_kthread+0x10/0x10
 ret_from_fork+0x2d/0x40
 ? __pfx_kthread+0x10/0x10
 ret_from_fork_asm+0x1a/0x30
 
```

After tracing the scheduling event, it was discovered that the migration
of the "timerlat/1" thread was performed during thread creation. Further
analysis confirmed that it is because the CPU online processing for
osnoise is implemented through workers, which is asynchronous with the
offline processing. When the worker was scheduled to create a thread, the
CPU may has already been removed from the cpu_online_mask during the offline
process, resulting in the inability to select the right CPU:

T1                       | T2
[CPUHP_ONLINE]           | cpu_device_down()
osnoise_hotplug_workfn() |
                         |     cpus_write_lock()
                         |     takedown_cpu(1)
                         |     cpus_write_unlock()
[CPUHP_OFFLINE]          |
    cpus_read_lock()     |
    start_kthread(1)     |
    cpus_read_unlock()   |

To fix this, skip online processing if the CPU is already offline.

Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu 
Cc: Mathieu Desnoyers 
Link: https://lore.kernel.org/20240924094515.3561410-4-liwei391@huawei.com
Fixes: c8895e271f79 ("trace/osnoise: Support hotplug operations")
Signed-off-by: Wei Li 
Signed-off-by: Steven Rostedt (Google) 
Signed-off-by: Greg Kroah-Hartman

tracing/timerlat: Drop interface_lock in stop_kthread()

2024-10-10T09:57:59+00:00

commit b484a02c9cedf8703eff8f0756f94618004bd165 upstream.

stop_kthread() is the offline callback for "trace/osnoise:online", since
commit 5bfbcd1ee57b ("tracing/timerlat: Add interface_lock around clearing
of kthread in stop_kthread()"), the following ABBA deadlock scenario is
introduced:

T1                            | T2 [BP]               | T3 [AP]
osnoise_hotplug_workfn()      | work_for_cpu_fn()     | cpuhp_thread_fun()
                              |   _cpu_down()         |   osnoise_cpu_die()
  mutex_lock(&interface_lock) |                       |     stop_kthread()
                              |     cpus_write_lock() |       mutex_lock(&interface_lock)
  cpus_read_lock()            |     cpuhp_kick_ap()   |

As the interface_lock here in just for protecting the "kthread" field of
the osn_var, use xchg() instead to fix this issue. Also use
for_each_online_cpu() back in stop_per_cpu_kthreads() as it can take
cpu_read_lock() again.

Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu 
Cc: Mathieu Desnoyers 
Link: https://lore.kernel.org/20240924094515.3561410-3-liwei391@huawei.com
Fixes: 5bfbcd1ee57b ("tracing/timerlat: Add interface_lock around clearing of kthread in stop_kthread()")
Signed-off-by: Wei Li 
Signed-off-by: Steven Rostedt (Google) 
Signed-off-by: Greg Kroah-Hartman

tracing/hwlat: Fix a race during cpuhp processing

2024-10-10T09:57:59+00:00

commit 2a13ca2e8abb12ee43ada8a107dadca83f140937 upstream.

The cpuhp online/offline processing race also exists in percpu-mode hwlat
tracer in theory, apply the fix too. That is:

    T1                       | T2
    [CPUHP_ONLINE]           | cpu_device_down()
     hwlat_hotplug_workfn()  |
                             |     cpus_write_lock()
                             |     takedown_cpu(1)
                             |     cpus_write_unlock()
    [CPUHP_OFFLINE]          |
        cpus_read_lock()     |
        start_kthread(1)     |
        cpus_read_unlock()   |

Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu 
Cc: Mathieu Desnoyers 
Link: https://lore.kernel.org/20240924094515.3561410-5-liwei391@huawei.com
Fixes: ba998f7d9531 ("trace/hwlat: Support hotplug operations")
Signed-off-by: Wei Li 
Signed-off-by: Steven Rostedt (Google) 
Signed-off-by: Greg Kroah-Hartman

bpf: Fix use-after-free in bpf_uprobe_multi_link_attach()

2024-10-04T14:30:03+00:00

commit 5fe6e308abaea082c20fbf2aa5df8e14495622cf upstream.

If bpf_link_prime() fails, bpf_uprobe_multi_link_attach() goes to the
error_free label and frees the array of bpf_uprobe's without calling
bpf_uprobe_unregister().

This leaks bpf_uprobe->uprobe and worse, this frees bpf_uprobe->consumer
without removing it from the uprobe->consumers list.

Fixes: 89ae89f53d20 ("bpf: Add multi uprobe link")
Closes: https://lore.kernel.org/all/000000000000382d39061f59f2dd@google.com/
Reported-by: syzbot+f7a1c2c2711e4a780f19@syzkaller.appspotmail.com
Signed-off-by: Oleg Nesterov 
Signed-off-by: Peter Zijlstra (Intel) 
Acked-by: Andrii Nakryiko 
Acked-by: Jiri Olsa 
Tested-by: syzbot+f7a1c2c2711e4a780f19@syzkaller.appspotmail.com
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20240813152524.GA7292@redhat.com
Signed-off-by: Greg Kroah-Hartman

bpf: Fix helper writes to read-only maps

2024-10-04T14:29:23+00:00

[ Upstream commit 32556ce93bc45c730829083cb60f95a2728ea48b ]

Lonial found an issue that despite user- and BPF-side frozen BPF map
(like in case of .rodata), it was still possible to write into it from
a BPF program side through specific helpers having ARG_PTR_TO_{LONG,INT}
as arguments.

In check_func_arg() when the argument is as mentioned, the meta->raw_mode
is never set. Later, check_helper_mem_access(), under the case of
PTR_TO_MAP_VALUE as register base type, it assumes BPF_READ for the
subsequent call to check_map_access_type() and given the BPF map is
read-only it succeeds.

The helpers really need to be annotated as ARG_PTR_TO_{LONG,INT} | MEM_UNINIT
when results are written into them as opposed to read out of them. The
latter indicates that it's okay to pass a pointer to uninitialized memory
as the memory is written to anyway.

However, ARG_PTR_TO_{LONG,INT} is a special case of ARG_PTR_TO_FIXED_SIZE_MEM
just with additional alignment requirement. So it is better to just get
rid of the ARG_PTR_TO_{LONG,INT} special cases altogether and reuse the
fixed size memory types. For this, add MEM_ALIGNED to additionally ensure
alignment given these helpers write directly into the args via * = val.
The .arg*_size has been initialized reflecting the actual sizeof(*).

MEM_ALIGNED can only be used in combination with MEM_FIXED_SIZE annotated
argument types, since in !MEM_FIXED_SIZE cases the verifier does not know
the buffer size a priori and therefore cannot blindly write * = val.

Fixes: 57c3bb725a3d ("bpf: Introduce ARG_PTR_TO_{INT,LONG} arg types")
Reported-by: Lonial Con 
Signed-off-by: Daniel Borkmann 
Acked-by: Andrii Nakryiko 
Acked-by: Shung-Hsi Yu 
Link: https://lore.kernel.org/r/20240913191754.13290-3-daniel@iogearbox.net
Signed-off-by: Alexei Starovoitov 
Signed-off-by: Sasha Levin

tracing/osnoise: Fix build when timerlat is not enabled

2024-09-18T17:24:09+00:00

commit af178143343028fdec9d5960a22d17f5587fd3f5 upstream.

To fix some critical section races, the interface_lock was added to a few
locations. One of those locations was above where the interface_lock was
declared, so the declaration was moved up before that usage.
Unfortunately, where it was placed was inside a CONFIG_TIMERLAT_TRACER
ifdef block. As the interface_lock is used outside that config, this broke
the build when CONFIG_OSNOISE_TRACER was enabled but
CONFIG_TIMERLAT_TRACER was not.

Cc: Masami Hiramatsu 
Cc: Mathieu Desnoyers 
Cc: "Helena Anna" 
Cc: "Luis Claudio R. Goncalves" 
Cc: Tomas Glozar 
Link: https://lore.kernel.org/20240909103231.23a289e2@gandalf.local.home
Fixes: e6a53481da29 ("tracing/timerlat: Only clear timer if a kthread exists")
Reported-by: "Bityutskiy, Artem" 
Signed-off-by: Steven Rostedt (Google) 
Signed-off-by: Greg Kroah-Hartman

tracing/timerlat: Add interface_lock around clearing of kthread in stop_kthread()

2024-09-12T09:11:27+00:00

commit 5bfbcd1ee57b607fd29e4645c7f350dd385dd9ad upstream.

The timerlat interface will get and put the task that is part of the
"kthread" field of the osn_var to keep it around until all references are
released. But here's a race in the "stop_kthread()" code that will call
put_task_struct() on the kthread if it is not a kernel thread. This can
race with the releasing of the references to that task struct and the
put_task_struct() can be called twice when it should have been called just
once.

Take the interface_lock() in stop_kthread() to synchronize this change.
But to do so, the function stop_per_cpu_kthreads() needs to change the
loop from for_each_online_cpu() to for_each_possible_cpu() and remove the
cpu_read_lock(), as the interface_lock can not be taken while the cpu
locks are held. The only side effect of this change is that it may do some
extra work, as the per_cpu variables of the offline CPUs would not be set
anyway, and would simply be skipped in the loop.

Remove unneeded "return;" in stop_kthread().

Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu 
Cc: Mathieu Desnoyers 
Cc: Tomas Glozar 
Cc: John Kacur 
Cc: "Luis Claudio R. Goncalves" 
Link: https://lore.kernel.org/20240905113359.2b934242@gandalf.local.home
Fixes: e88ed227f639e ("tracing/timerlat: Add user-space interface")
Signed-off-by: Steven Rostedt (Google) 
Signed-off-by: Greg Kroah-Hartman

tracing: Avoid possible softlockup in tracing_iter_reset()

2024-09-12T09:11:27+00:00

commit 49aa8a1f4d6800721c7971ed383078257f12e8f9 upstream.

In __tracing_open(), when max latency tracers took place on the cpu,
the time start of its buffer would be updated, then event entries with
timestamps being earlier than start of the buffer would be skipped
(see tracing_iter_reset()).

Softlockup will occur if the kernel is non-preemptible and too many
entries were skipped in the loop that reset every cpu buffer, so add
cond_resched() to avoid it.

Cc: stable@vger.kernel.org
Fixes: 2f26ebd549b9a ("tracing: use timestamp to determine start of latency traces")
Link: https://lore.kernel.org/20240827124654.3817443-1-zhengyejian@huaweicloud.com
Suggested-by: Steven Rostedt 
Signed-off-by: Zheng Yejian 
Signed-off-by: Steven Rostedt (Google) 
Signed-off-by: Greg Kroah-Hartman

tracing/timerlat: Only clear timer if a kthread exists

2024-09-12T09:11:27+00:00

commit e6a53481da292d970d1edf0d8831121d1c5e2f0d upstream.

The timerlat tracer can use user space threads to check for osnoise and
timer latency. If the program using this is killed via a SIGTERM, the
threads are shutdown one at a time and another tracing instance can start
up resetting the threads before they are fully closed. That causes the
hrtimer assigned to the kthread to be shutdown and freed twice when the
dying thread finally closes the file descriptors, causing a use-after-free
bug.

Only cancel the hrtimer if the associated thread is still around. Also add
the interface_lock around the resetting of the tlat_var->kthread.

Note, this is just a quick fix that can be backported to stable. A real
fix is to have a better synchronization between the shutdown of old
threads and the starting of new ones.

Link: https://lore.kernel.org/all/20240820130001.124768-1-tglozar@redhat.com/

Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu 
Cc: Mathieu Desnoyers 
Cc: "Luis Claudio R. Goncalves" 
Link: https://lore.kernel.org/20240905085330.45985730@gandalf.local.home
Fixes: e88ed227f639e ("tracing/timerlat: Add user-space interface")
Reported-by: Tomas Glozar 
Signed-off-by: Steven Rostedt (Google) 
Signed-off-by: Greg Kroah-Hartman