exit_itimers() loops through every timer in the process to delete it. This
requires taking the system-wide hash_lock for each of these timers, and
contends with other processes trying to create or delete timers.
When a process creates hundreds of thousands of timers, and then exits
while other processes contend with it, this can trigger softlockups on
CONFIG_PREEMPT=n.
Add a cond_resched() invocation into the loop to allow the system to make
progress.
Signed-off-by: Ben Segall <bsegall@google.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/all/xm2634gg2n23.fsf@google.com
Clang and GCC complain about overlapped initialisers in the
hrtimer_clock_to_base_table definition. With `make W=1` and CONFIG_WERROR=y
(which is default nowadays) this breaks the build:
CC kernel/time/hrtimer.o
kernel/time/hrtimer.c:124:21: error: initializer overrides prior initialization of this subobject [-Werror,-Winitializer-overrides]
124 | [CLOCK_REALTIME] = HRTIMER_BASE_REALTIME,
kernel/time/hrtimer.c:122:27: note: previous initialization is here
122 | [0 ... MAX_CLOCKS - 1] = HRTIMER_MAX_CLOCK_BASES,
(and similar for CLOCK_MONOTONIC, CLOCK_BOOTTIME, and CLOCK_TAI).
hrtimer_clockid_to_base(), which uses the table, is only used in
__hrtimer_init(), which is not a hotpath.
Therefore replace the table lookup with a switch case in
hrtimer_clockid_to_base() to avoid this warning.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/all/20250214134424.3367619-1-andriy.shevchenko@linux.intel.com
When a process reduces its number of threads or clears bits in its CPU
affinity mask, the mm_cid allocation should eventually converge towards
smaller values.
However, the change introduced by:
commit 7e019dcc47 ("sched: Improve cache locality of RSEQ concurrency
IDs for intermittent workloads")
adds a per-mm/CPU recent_cid which is never unset unless a thread
migrates.
This is a tradeoff between:
A) Preserving cache locality after a transition from many threads to few
threads, or after reducing the hamming weight of the allowed CPU mask.
B) Making the mm_cid upper bounds wrt nr threads and allowed CPU mask
easy to document and understand.
C) Allowing applications to eventually react to mm_cid compaction after
reduction of the nr threads or allowed CPU mask, making the tracking
of mm_cid compaction easier by shrinking it back towards 0 or not.
D) Making sure applications that periodically reduce and then increase
again the nr threads or allowed CPU mask still benefit from good
cache locality with mm_cid.
Introduce the following changes:
* After shrinking the number of threads or reducing the number of
allowed CPUs, reduce the value of max_nr_cid so expansion of CID
allocation will preserve cache locality if the number of threads or
allowed CPUs increase again.
* Only re-use a recent_cid if it is within the max_nr_cid upper bound,
else find the first available CID.
Fixes: 7e019dcc47 ("sched: Improve cache locality of RSEQ concurrency IDs for intermittent workloads")
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Gabriele Monaco <gmonaco@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Gabriele Monaco <gmonaco@redhat.com>
Link: https://lkml.kernel.org/r/20250210153253.460471-2-gmonaco@redhat.com
Allow a struct_ops program to return a referenced kptr if the struct_ops
operator's return type is a struct pointer. To make sure the returned
pointer continues to be valid in the kernel, several constraints are
required:
1) The type of the pointer must matches the return type
2) The pointer originally comes from the kernel (not locally allocated)
3) The pointer is in its unmodified form
Implementation wise, a referenced kptr first needs to be allowed to _leak_
in check_reference_leak() if it is in the return register. Then, in
check_return_code(), constraints 1-3 are checked. During struct_ops
registration, a check is also added to warn about operators with
non-struct pointer return.
In addition, since the first user, Qdisc_ops::dequeue, allows a NULL
pointer to be returned when there is no skb to be dequeued, we will allow
a scalar value with value equals to NULL to be returned.
In the future when there is a struct_ops user that always expects a valid
pointer to be returned from an operator, we may extend tagging to the
return value. We can tell the verifier to only allow NULL pointer return
if the return value is tagged with MAY_BE_NULL.
Signed-off-by: Amery Hung <amery.hung@bytedance.com>
Acked-by: Eduard Zingerman <eddyz87@gmail.com>
Acked-by: Martin KaFai Lau <martin.lau@kernel.org>
Link: https://lore.kernel.org/r/20250217190640.1748177-5-ameryhung@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Allows struct_ops programs to acqurie referenced kptrs from arguments
by directly reading the argument.
The verifier will acquire a reference for struct_ops a argument tagged
with "__ref" in the stub function in the beginning of the main program.
The user will be able to access the referenced kptr directly by reading
the context as long as it has not been released by the program.
This new mechanism to acquire referenced kptr (compared to the existing
"kfunc with KF_ACQUIRE") is introduced for ergonomic and semantic reasons.
In the first use case, Qdisc_ops, an skb is passed to .enqueue in the
first argument. This mechanism provides a natural way for users to get a
referenced kptr in the .enqueue struct_ops programs and makes sure that a
qdisc will always enqueue or drop the skb.
Signed-off-by: Amery Hung <amery.hung@bytedance.com>
Acked-by: Eduard Zingerman <eddyz87@gmail.com>
Acked-by: Martin KaFai Lau <martin.lau@kernel.org>
Link: https://lore.kernel.org/r/20250217190640.1748177-3-ameryhung@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Currently, ctx_arg_info is read-only in the view of the verifier since
it is shared among programs of the same attach type. Make each program
have their own copy of ctx_arg_info so that we can use it to store
program specific information.
In the next patch where we support acquiring a referenced kptr through a
struct_ops argument tagged with "__ref", ctx_arg_info->ref_obj_id will
be used to store the unique reference object id of the argument. This
avoids creating a requirement in the verifier that "__ref" tagged
arguments must be the first set of references acquired [0].
[0] https://lore.kernel.org/bpf/20241220195619.2022866-2-amery.hung@gmail.com/
Signed-off-by: Amery Hung <ameryhung@gmail.com>
Acked-by: Eduard Zingerman <eddyz87@gmail.com>
Acked-by: Martin KaFai Lau <martin.lau@kernel.org>
Link: https://lore.kernel.org/r/20250217190640.1748177-2-ameryhung@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Pull vfs fixes from Christian Brauner:
"It was reported that the acct(2) system call can be used to trigger a
NULL deref in cases where it is set to write to a file that triggers
an internal lookup.
This can e.g., happen when pointing acct(2) to /sys/power/resume. At
the point the where the write to this file happens the calling task
has already exited and called exit_fs() but an internal lookup might
be triggered through lookup_bdev(). This may trigger a NULL-deref when
accessing current->fs.
Reorganize the code so that the the final write happens from the
workqueue but with the caller's credentials. This preserves the
(strange) permission model and has almost no regression risk.
Also block access to kernel internal filesystems as well as procfs and
sysfs in the first place.
Various fixes for netfslib:
- Fix a number of read-retry hangs, including:
- Incorrect getting/putting of references on subreqs as we retry
them
- Failure to track whether a last old subrequest in a retried set
is superfluous
- Inconsistency in the usage of wait queues used for subrequests
(ie. using clear_and_wake_up_bit() whilst waiting on a private
waitqueue)
- Add stats counters for retries and publish in /proc/fs/netfs/stats.
This is not a fix per se, but is useful in debugging and shouldn't
otherwise change the operation of the code
- Fix the ordering of queuing subrequests with respect to setting the
request flag that says we've now queued them all"
* tag 'vfs-6.14-rc4.fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs:
netfs: Fix setting NETFS_RREQ_ALL_QUEUED to be after all subreqs queued
netfs: Add retry stat counters
netfs: Fix a number of read-retry hangs
acct: block access to kernel internal filesystems
acct: perform last write from workqueue
Pull scheduler fix from Borislav Petkov:
- Clarify what happens when a task is woken up from the wake queue and
make clear its removal from that queue is atomic
* tag 'sched_urgent_for_v6.14_rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
sched: Clarify wake_up_q()'s write to task->wake_q.next
Using a single global idle mask can lead to inefficiencies and a lot of
stress on the cache coherency protocol on large systems with multiple
NUMA nodes, since all the CPUs can create a really intense read/write
activity on the single global cpumask.
Therefore, split the global cpumask into multiple per-NUMA node cpumasks
to improve scalability and performance on large systems.
The concept is that each cpumask will track only the idle CPUs within
its corresponding NUMA node, treating CPUs in other NUMA nodes as busy.
In this way concurrent access to the idle cpumask will be restricted
within each NUMA node.
The split of multiple per-node idle cpumasks can be controlled using the
SCX_OPS_BUILTIN_IDLE_PER_NODE flag.
By default SCX_OPS_BUILTIN_IDLE_PER_NODE is not enabled and a global
host-wide idle cpumask is used, maintaining the previous behavior.
NOTE: if a scheduler explicitly enables the per-node idle cpumasks (via
SCX_OPS_BUILTIN_IDLE_PER_NODE), scx_bpf_get_idle_cpu/smtmask() will
trigger an scx error, since there are no system-wide cpumasks.
= Test =
Hardware:
- System: DGX B200
- CPUs: 224 SMT threads (112 physical cores)
- Processor: INTEL(R) XEON(R) PLATINUM 8570
- 2 NUMA nodes
Scheduler:
- scx_simple [1] (so that we can focus at the built-in idle selection
policy and not at the scheduling policy itself)
Test:
- Run a parallel kernel build `make -j $(nproc)` and measure the average
elapsed time over 10 runs:
avg time | stdev
---------+------
before: 52.431s | 2.895
after: 50.342s | 2.895
= Conclusion =
Splitting the global cpumask into multiple per-NUMA cpumasks helped to
achieve a speedup of approximately +4% with this particular architecture
and test case.
The same test on a DGX-1 (40 physical cores, Intel Xeon E5-2698 v4 @
2.20GHz, 2 NUMA nodes) shows a speedup of around 1.5-3%.
On smaller systems, I haven't noticed any measurable regressions or
improvements with the same test (parallel kernel build) and scheduler
(scx_simple).
Moreover, with a modified scx_bpfland that uses the new NUMA-aware APIs
I observed an additional +2-2.5% performance improvement with the same
test.
[1] https://github.com/sched-ext/scx/blob/main/scheds/c/scx_simple.bpf.c
Cc: Yury Norov [NVIDIA] <yury.norov@gmail.com>
Signed-off-by: Andrea Righi <arighi@nvidia.com>
Reviewed-by: Yury Norov [NVIDIA] <yury.norov@gmail.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Add the new scheduler flag SCX_OPS_BUILTIN_IDLE_PER_NODE, which allows
BPF schedulers to select between using a global flat idle cpumask or
multiple per-node cpumasks.
This only introduces the flag and the mechanism to enable/disable this
feature without affecting any scheduling behavior.
Cc: Yury Norov [NVIDIA] <yury.norov@gmail.com>
Signed-off-by: Andrea Righi <arighi@nvidia.com>
Reviewed-by: Yury Norov [NVIDIA] <yury.norov@gmail.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Make all the static keys used by the idle CPU selection policy private
to ext_idle.c. This avoids unnecessary exposure in headers and improves
code encapsulation.
Cc: Yury Norov <yury.norov@gmail.com>
Signed-off-by: Andrea Righi <arighi@nvidia.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Pull trace ring buffer fixes from Steven Rostedt:
- Enable resize on mmap() error
When a process mmaps a ring buffer, its size is locked and resizing
is disabled. But if the user passes in a wrong parameter, the mmap()
can fail after the resize was disabled and the mmap() exits with
error without reenabling the ring buffer resize. This prevents the
ring buffer from ever being resized after that. Reenable resizing of
the ring buffer on mmap() error.
- Have resizing return proper error and not always -ENOMEM
If the ring buffer is mmapped by one task and another task tries to
resize the buffer it will error with -ENOMEM. This is confusing to
the user as there may be plenty of memory available. Have it return
the error that actually happens (in this case -EBUSY) where the user
can understand why the resize failed.
- Test the sub-buffer array to validate persistent memory buffer
On boot up, the initialization of the persistent memory buffer will
do a validation check to see if the content of the data is valid, and
if so, it will use the memory as is, otherwise it re-initializes it.
There's meta data in this persistent memory that keeps track of which
sub-buffer is the reader page and an array that states the order of
the sub-buffers. The values in this array are indexes into the
sub-buffers. The validator checks to make sure that all the entries
in the array are within the sub-buffer list index, but it does not
check for duplications.
While working on this code, the array got corrupted and had
duplicates, where not all the sub-buffers were accounted for. This
passed the validator as all entries were valid, but the link list was
incorrect and could have caused a crash. The corruption only produced
incorrect data, but it could have been more severe. To fix this,
create a bitmask that covers all the sub-buffer indexes and set it to
all zeros. While iterating the array checking the values of the array
content, have it set a bit corresponding to the index in the array.
If the bit was already set, then it is a duplicate and mark the
buffer as invalid and reset it.
- Prevent mmap()ing persistent ring buffer
The persistent ring buffer uses vmap() to map the persistent memory.
Currently, the mmap() logic only uses virt_to_page() to get the page
from the ring buffer memory and use that to map to user space. This
works because a normal ring buffer uses alloc_page() to allocate its
memory. But because the persistent ring buffer use vmap() it causes a
kernel crash.
Fixing this to work with vmap() is not hard, but since mmap() on
persistent memory buffers never worked, just have the mmap() return
-ENODEV (what was returned before mmap() for persistent memory ring
buffers, as they never supported mmap. Normal buffers will still
allow mmap(). Implementing mmap() for persistent memory ring buffers
can wait till the next merge window.
- Fix polling on persistent ring buffers
There's a "buffer_percent" option (default set to 50), that is used
to have reads of the ring buffer binary data block until the buffer
fills to that percentage. The field "pages_touched" is incremented
every time a new sub-buffer has content added to it. This field is
used in the calculations to determine the amount of content is in the
buffer and if it exceeds the "buffer_percent" then it will wake the
task polling on the buffer.
As persistent ring buffers can be created by the content from a
previous boot, the "pages_touched" field was not updated. This means
that if a task were to poll on the persistent buffer, it would block
even if the buffer was completely full. It would block even if the
"buffer_percent" was zero, because with "pages_touched" as zero, it
would be calculated as the buffer having no content. Update
pages_touched when initializing the persistent ring buffer from a
previous boot.
* tag 'trace-ring-buffer-v6.14-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
ring-buffer: Update pages_touched to reflect persistent buffer content
tracing: Do not allow mmap() of persistent ring buffer
ring-buffer: Validate the persistent meta data subbuf array
tracing: Have the error of __tracing_resize_ring_buffer() passed to user
ring-buffer: Unlock resize on mmap error
The pages_touched field represents the number of subbuffers in the ring
buffer that have content that can be read. This is used in accounting of
"dirty_pages" and "buffer_percent" to allow the user to wait for the
buffer to be filled to a certain amount before it reads the buffer in
blocking mode.
The persistent buffer never updated this value so it was set to zero, and
this accounting would take it as it had no content. This would cause user
space to wait for content even though there's enough content in the ring
buffer that satisfies the buffer_percent.
Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Vincent Donnefort <vdonnefort@google.com>
Link: https://lore.kernel.org/20250214123512.0631436e@gandalf.local.home
Fixes: 5f3b6e839f ("ring-buffer: Validate boot range memory events")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
When trying to mmap a trace instance buffer that is attached to
reserve_mem, it would crash:
BUG: unable to handle page fault for address: ffffe97bd00025c8
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
PGD 2862f3067 P4D 2862f3067 PUD 0
Oops: Oops: 0000 [#1] PREEMPT_RT SMP PTI
CPU: 4 UID: 0 PID: 981 Comm: mmap-rb Not tainted 6.14.0-rc2-test-00003-g7f1a5e3fbf9e-dirty #233
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.3-debian-1.16.3-2 04/01/2014
RIP: 0010:validate_page_before_insert+0x5/0xb0
Code: e2 01 89 d0 c3 cc cc cc cc 66 66 2e 0f 1f 84 00 00 00 00 00 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 0f 1f 44 00 00 <48> 8b 46 08 a8 01 75 67 66 90 48 89 f0 8b 50 34 85 d2 74 76 48 89
RSP: 0018:ffffb148c2f3f968 EFLAGS: 00010246
RAX: ffff9fa5d3322000 RBX: ffff9fa5ccff9c08 RCX: 00000000b879ed29
RDX: ffffe97bd00025c0 RSI: ffffe97bd00025c0 RDI: ffff9fa5ccff9c08
RBP: ffffb148c2f3f9f0 R08: 0000000000000004 R09: 0000000000000004
R10: 0000000000000000 R11: 0000000000000200 R12: 0000000000000000
R13: 00007f16a18d5000 R14: ffff9fa5c48db6a8 R15: 0000000000000000
FS: 00007f16a1b54740(0000) GS:ffff9fa73df00000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: ffffe97bd00025c8 CR3: 00000001048c6006 CR4: 0000000000172ef0
Call Trace:
<TASK>
? __die_body.cold+0x19/0x1f
? __die+0x2e/0x40
? page_fault_oops+0x157/0x2b0
? search_module_extables+0x53/0x80
? validate_page_before_insert+0x5/0xb0
? kernelmode_fixup_or_oops.isra.0+0x5f/0x70
? __bad_area_nosemaphore+0x16e/0x1b0
? bad_area_nosemaphore+0x16/0x20
? do_kern_addr_fault+0x77/0x90
? exc_page_fault+0x22b/0x230
? asm_exc_page_fault+0x2b/0x30
? validate_page_before_insert+0x5/0xb0
? vm_insert_pages+0x151/0x400
__rb_map_vma+0x21f/0x3f0
ring_buffer_map+0x21b/0x2f0
tracing_buffers_mmap+0x70/0xd0
__mmap_region+0x6f0/0xbd0
mmap_region+0x7f/0x130
do_mmap+0x475/0x610
vm_mmap_pgoff+0xf2/0x1d0
ksys_mmap_pgoff+0x166/0x200
__x64_sys_mmap+0x37/0x50
x64_sys_call+0x1670/0x1d70
do_syscall_64+0xbb/0x1d0
entry_SYSCALL_64_after_hwframe+0x77/0x7f
The reason was that the code that maps the ring buffer pages to user space
has:
page = virt_to_page((void *)cpu_buffer->subbuf_ids[s]);
And uses that in:
vm_insert_pages(vma, vma->vm_start, pages, &nr_pages);
But virt_to_page() does not work with vmap()'d memory which is what the
persistent ring buffer has. It is rather trivial to allow this, but for
now just disable mmap() of instances that have their ring buffer from the
reserve_mem option.
If an mmap() is performed on a persistent buffer it will return -ENODEV
just like it would if the .mmap field wasn't defined in the
file_operations structure.
Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Vincent Donnefort <vdonnefort@google.com>
Link: https://lore.kernel.org/20250214115547.0d7287d3@gandalf.local.home
Fixes: 9b7bdf6f6e ("tracing: Have trace_printk not use binary prints if boot buffer")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
kernfs_rename_lock is used to obtain stable kernfs_node::{name|parent}
pointer. This is a preparation to access kernfs_node::parent under RCU
and ensure that the pointer remains stable under the RCU lifetime
guarantees.
For a complete path, as it is done in kernfs_path_from_node(), the
kernfs_rename_lock is still required in order to obtain a stable parent
relationship while computing the relevant node depth. This must not
change while the nodes are inspected in order to build the path.
If the kernfs user never moves the nodes (changes the parent) then the
kernfs_rename_lock is not required and the RCU guarantees are
sufficient. This "restriction" can be set with
KERNFS_ROOT_INVARIANT_PARENT. Otherwise the lock is required.
Rename kernfs_node::parent to kernfs_node::__parent to denote the RCU
access and use RCU accessor while accessing the node.
Make cgroup use KERNFS_ROOT_INVARIANT_PARENT since the parent here can
not change.
Acked-by: Tejun Heo <tj@kernel.org>
Cc: Yonghong Song <yonghong.song@linux.dev>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Link: https://lore.kernel.org/r/20250213145023.2820193-6-bigeasy@linutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
may_goto uses an additional 8 bytes on the stack, which causes the
interpreters[] array to go out of bounds when calculating index by
stack_size.
1. If a BPF program is rewritten, re-evaluate the stack size. For non-JIT
cases, reject loading directly.
2. For non-JIT cases, calculating interpreters[idx] may still cause
out-of-bounds array access, and just warn about it.
3. For jit_requested cases, the execution of bpf_func also needs to be
warned. So move the definition of function __bpf_prog_ret0_warn out of
the macro definition CONFIG_BPF_JIT_ALWAYS_ON.
Reported-by: syzbot+d2a2c639d03ac200a4f1@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/bpf/0000000000000f823606139faa5d@google.com/
Fixes: 011832b97b ("bpf: Introduce may_goto instruction")
Signed-off-by: Jiayuan Chen <mrpre@163.com>
Link: https://lore.kernel.org/r/20250214091823.46042-2-mrpre@163.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Pull sched_ext fixes from Tejun Heo:
- Fix lock imbalance in a corner case of dispatch_to_local_dsq()
- Migration disabled tasks were confusing some BPF schedulers and its
handling had a bug. Fix it and simplify the default behavior by
dispatching them automatically
- ops.tick(), ops.disable() and ops.exit_task() were incorrectly
disallowing kfuncs that require the task argument to be the rq
operation is currently operating on and thus is rq-locked.
Allow them.
- Fix autogroup migration handling bug which was occasionally
triggering a warning in the cgroup migration path
- tools/sched_ext, selftest and other misc updates
* tag 'sched_ext-for-6.14-rc2-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/sched_ext:
sched_ext: Use SCX_CALL_OP_TASK in task_tick_scx
sched_ext: Fix the incorrect bpf_list kfunc API in common.bpf.h.
sched_ext: selftests: Fix grammar in tests description
sched_ext: Fix incorrect assumption about migration disabled tasks in task_can_run_on_remote_rq()
sched_ext: Fix migration disabled handling in targeted dispatches
sched_ext: Implement auto local dispatching of migration disabled tasks
sched_ext: Fix incorrect time delta calculation in time_delta()
sched_ext: Fix lock imbalance in dispatch_to_local_dsq()
sched_ext: selftests/dsp_local_on: Fix selftest on UP systems
tools/sched_ext: Add helper to check task migration state
sched_ext: Fix incorrect autogroup migration detection
sched_ext: selftests/dsp_local_on: Fix sporadic failures
selftests/sched_ext: Fix enum resolution
sched_ext: Include task weight in the error state dump
sched_ext: Fixes typos in comments
Pull cgroup fixes from Tejun Heo:
- Fix a race window where a newly forked task could escape cgroup.kill
- Remove incorrectly included steal time from cpu.stat::usage_usec
- Minor update in selftest
* tag 'cgroup-for-6.14-rc2-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup:
cgroup: Remove steal time from usage_usec
selftests/cgroup: use bash in test_cpuset_v1_hp.sh
cgroup: fix race between fork and cgroup.kill
Pull workqueue fix from Tejun Heo:
- Fix a regression where a worker pool can be freed before rescuer
workers are done with it leading to user-after-free
* tag 'wq-for-6.14-rc2-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq:
workqueue: Put the pwq after detaching the rescuer from the pool
The meta data for a mapped ring buffer contains an array of indexes of all
the subbuffers. The first entry is the reader page, and the rest of the
entries lay out the order of the subbuffers in how the ring buffer link
list is to be created.
The validator currently makes sure that all the entries are within the
range of 0 and nr_subbufs. But it does not check if there are any
duplicates.
While working on the ring buffer, I corrupted this array, where I added
duplicates. The validator did not catch it and created the ring buffer
link list on top of it. Luckily, the corruption was only that the reader
page was also in the writer path and only presented corrupted data but did
not crash the kernel. But if there were duplicates in the writer side,
then it could corrupt the ring buffer link list and cause a crash.
Create a bitmask array with the size of the number of subbuffers. Then
clear it. When walking through the subbuf array checking to see if the
entries are within the range, test if its bit is already set in the
subbuf_mask. If it is, then there is duplicates and fail the validation.
If not, set the corresponding bit and continue.
Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Vincent Donnefort <vdonnefort@google.com>
Link: https://lore.kernel.org/20250214102820.7509ddea@gandalf.local.home
Fixes: c76883f18e ("ring-buffer: Add test if range of boot buffer is valid")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Currently if __tracing_resize_ring_buffer() returns an error, the
tracing_resize_ringbuffer() returns -ENOMEM. But it may not be a memory
issue that caused the function to fail. If the ring buffer is memory
mapped, then the resizing of the ring buffer will be disabled. But if the
user tries to resize the buffer, it will get an -ENOMEM returned, which is
confusing because there is plenty of memory. The actual error returned was
-EBUSY, which would make much more sense to the user.
Cc: stable@vger.kernel.org
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Vincent Donnefort <vdonnefort@google.com>
Link: https://lore.kernel.org/20250213134132.7e4505d7@gandalf.local.home
Fixes: 117c39200d ("ring-buffer: Introducing ring-buffer mapping functions")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Reviewed-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Syzbot regularly runs into the following warning on arm64:
| WARNING: CPU: 1 PID: 6023 at kernel/workqueue.c:2257 current_wq_worker kernel/workqueue_internal.h:69 [inline]
| WARNING: CPU: 1 PID: 6023 at kernel/workqueue.c:2257 is_chained_work kernel/workqueue.c:2199 [inline]
| WARNING: CPU: 1 PID: 6023 at kernel/workqueue.c:2257 __queue_work+0xe50/0x1308 kernel/workqueue.c:2256
| Modules linked in:
| CPU: 1 UID: 0 PID: 6023 Comm: klogd Not tainted 6.13.0-rc2-syzkaller-g2e7aff49b5da #0
| Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 09/13/2024
| pstate: 404000c5 (nZcv daIF +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
| pc : __queue_work+0xe50/0x1308 kernel/workqueue_internal.h:69
| lr : current_wq_worker kernel/workqueue_internal.h:69 [inline]
| lr : is_chained_work kernel/workqueue.c:2199 [inline]
| lr : __queue_work+0xe50/0x1308 kernel/workqueue.c:2256
[...]
| __queue_work+0xe50/0x1308 kernel/workqueue.c:2256 (L)
| delayed_work_timer_fn+0x74/0x90 kernel/workqueue.c:2485
| call_timer_fn+0x1b4/0x8b8 kernel/time/timer.c:1793
| expire_timers kernel/time/timer.c:1839 [inline]
| __run_timers kernel/time/timer.c:2418 [inline]
| __run_timer_base+0x59c/0x7b4 kernel/time/timer.c:2430
| run_timer_base kernel/time/timer.c:2439 [inline]
| run_timer_softirq+0xcc/0x194 kernel/time/timer.c:2449
The warning is probably because we are trying to queue work into a
destroyed workqueue, but the softirq context makes it hard to pinpoint
the problematic caller.
Extend the warning diagnostics to print both the function we are trying
to queue as well as the name of the workqueue.
Cc: Tejun Heo <tj@kernel.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Marc Zyngier <maz@kernel.org>
Cc: Lai Jiangshan <jiangshanlai@gmail.com>
Link: https://syzkaller.appspot.com/bug?extid=e13e654d315d4da1277c
Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Tejun Heo <tj@kernel.org>
Add a sysfs entry at /sys/kernel/sched_ext/root/events to expose core
event counters through the files system interface. Each line of the file
shows the event name and its counter value.
In addition, the format of scx_dump_event() is adjusted as the event name
gets longer.
Signed-off-by: Changwoo Min <changwoo@igalia.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Depends on the simplifications from commit 1d7e707af4 ("Revert "x86/module: prepare module loading for ROX allocations of text"")
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
The function "can_migrate_task()" utilize "for_each_cpu_and" with a
"if" statement inside to find the destination cpu. It's the same logic
to find the first set bit of the result of the bitwise-AND of
"env->dst_grpmask", "env->cpus" and "p->cpus_ptr".
Refactor it by using "cpumask_first_and_and()" to perform bitwise-AND
for "env->dst_grpmask", "env->cpus" and "p->cpus_ptr" and pick the
first cpu within the intersection as the destination cpu, so we can
elimate the need of looping and multiple times of branch.
After the refactoring this part of the code can speed up from ~115ns
to ~54ns, according to the test below.
Ran the test for 5 times and the result is showned in the following
table, and the test script is paste in next section.
-------------------------------------------------------
|Old method| 130| 118| 115| 109| 106| avg ~115ns|
-------------------------------------------------------
|New method| 58| 55| 54| 48| 55| avg ~54ns|
-------------------------------------------------------
Signed-off-by: I Hsin Cheng <richard120310@gmail.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20250210103019.283824-1-richard120310@gmail.com
When a task is enqueued and its parent cgroup se is already on_rq, this
parent cgroup se will not be enqueued again, and hence the root->min_slice
leaves unchanged. The same issue happens when a task is dequeued and its
parent cgroup se has other runnable entities, and the parent cgroup se
will not be dequeued.
Force propagating min_slice when se doesn't need to be enqueued or
dequeued. Ensure the se hierarchy always get the latest min_slice.
Fixes: aef6987d89 ("sched/eevdf: Propagate min_slice up the cgroup hierarchy")
Signed-off-by: Tianchen Ding <dtcccc@linux.alibaba.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20250211063659.7180-1-dtcccc@linux.alibaba.com
The sched_clock_irqtime was defined as a static key in commit 8722903cbb
('sched: Define sched_clock_irqtime as static key'). However, this change
introduces a 'sleeping in atomic context' warning, as shown below:
arch/x86/kernel/tsc.c:1214 mark_tsc_unstable()
warn: sleeping in atomic context
As analyzed by Dan, the affected code path is as follows:
vcpu_load() <- disables preempt
-> kvm_arch_vcpu_load()
-> mark_tsc_unstable() <- sleeps
virt/kvm/kvm_main.c
166 void vcpu_load(struct kvm_vcpu *vcpu)
167 {
168 int cpu = get_cpu();
^^^^^^^^^^
This get_cpu() disables preemption.
169
170 __this_cpu_write(kvm_running_vcpu, vcpu);
171 preempt_notifier_register(&vcpu->preempt_notifier);
172 kvm_arch_vcpu_load(vcpu, cpu);
173 put_cpu();
174 }
arch/x86/kvm/x86.c
4979 if (unlikely(vcpu->cpu != cpu) || kvm_check_tsc_unstable()) {
4980 s64 tsc_delta = !vcpu->arch.last_host_tsc ? 0 :
4981 rdtsc() - vcpu->arch.last_host_tsc;
4982 if (tsc_delta < 0)
4983 mark_tsc_unstable("KVM discovered backwards TSC");
arch/x86/kernel/tsc.c
1206 void mark_tsc_unstable(char *reason)
1207 {
1208 if (tsc_unstable)
1209 return;
1210
1211 tsc_unstable = 1;
1212 if (using_native_sched_clock())
1213 clear_sched_clock_stable();
--> 1214 disable_sched_clock_irqtime();
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
kernel/jump_label.c
245 void static_key_disable(struct static_key *key)
246 {
247 cpus_read_lock();
^^^^^^^^^^^^^^^^
This lock has a might_sleep() in it which triggers the static checker
warning.
248 static_key_disable_cpuslocked(key);
249 cpus_read_unlock();
250 }
Let revert this change for now as {disable,enable}_sched_clock_irqtime
are used in many places, as pointed out by Sean, including the following:
The code path in clocksource_watchdog():
clocksource_watchdog()
|
-> spin_lock(&watchdog_lock);
|
-> __clocksource_unstable()
|
-> clocksource.mark_unstable() == tsc_cs_mark_unstable()
|
-> disable_sched_clock_irqtime()
And the code path in sched_clock_register():
/* Cannot register a sched_clock with interrupts on */
local_irq_save(flags);
...
/* Enable IRQ time accounting if we have a fast enough sched_clock() */
if (irqtime > 0 || (irqtime == -1 && rate >= 1000000))
enable_sched_clock_irqtime();
local_irq_restore(flags);
[lkp@intel.com: reported a build error in the prev version]
Closes: https://lore.kernel.org/kvm/37a79ba3-9ce0-479c-a5b0-2bd75d573ed3@stanley.mountain/
Fixes: 8722903cbb ("sched: Define sched_clock_irqtime as static key")
Reported-by: Dan Carpenter <dan.carpenter@linaro.org>
Debugged-by: Dan Carpenter <dan.carpenter@linaro.org>
Debugged-by: Sean Christopherson <seanjc@google.com>
Debugged-by: Michal Koutný <mkoutny@suse.com>
Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Vincent Guittot <vincent.guittot@linaro.org>
Link: https://lkml.kernel.org/r/20250205032438.14668-1-laoar.shao@gmail.com
The old default value for slice is 0.75 msec * (1 + ilog(ncpus)) which
means that we have a default slice of:
0.75 for 1 cpu
1.50 up to 3 cpus
2.25 up to 7 cpus
3.00 for 8 cpus and above.
For HZ=250 and HZ=100, because of the tick accuracy, the runtime of
tasks is far higher than their slice.
For HZ=1000 with 8 cpus or more, the accuracy of tick is already
satisfactory, but there is still an issue that tasks will get an extra
tick because the tick often arrives a little faster than expected. In
this case, the task can only wait until the next tick to consider that it
has reached its deadline, and will run 1ms longer.
vruntime + sysctl_sched_base_slice = deadline
|-----------|-----------|-----------|-----------|
1ms 1ms 1ms 1ms
^ ^ ^ ^
tick1 tick2 tick3 tick4(nearly 4ms)
There are two reasons for tick error: clockevent precision and the
CONFIG_IRQ_TIME_ACCOUNTING/CONFIG_PARAVIRT_TIME_ACCOUNTING. with
CONFIG_IRQ_TIME_ACCOUNTING every tick will be less than 1ms, but even
without it, because of clockevent precision, tick still often less than
1ms.
In order to make scheduling more precise, we changed 0.75 to 0.70,
Using 0.70 instead of 0.75 should not change much for other configs
and would fix this issue:
0.70 for 1 cpu
1.40 up to 3 cpus
2.10 up to 7 cpus
2.8 for 8 cpus and above.
This does not guarantee that tasks can run the slice time accurately
every time, but occasionally running an extra tick has little impact.
Signed-off-by: zihan zhou <15645113830zzh@gmail.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Vincent Guittot <vincent.guittot@linaro.org>
Link: https://lkml.kernel.org/r/20250208075322.13139-1-15645113830zzh@gmail.com
A wakeup non-idle entity should preempt idle entity at any time,
but because of the slice protection of the idle entity, the non-idle
entity has to wait, so just cancel it.
This patch is aimed at minimizing the impact of SCHED_IDLE on
SCHED_NORMAL. For example, a task with SCHED_IDLE policy that sleeps for
1s and then runs for 3 ms, running cyclictest on the same cpu, has a
maximum latency of 3 ms, which is caused by the slice protection of the
idle entity. It is unreasonable. With this patch, the cyclictest latency
under the same conditions is basically the same on the cpu with idle
processes and on empty cpu.
[peterz: add helpers]
Fixes: 63304558ba ("sched/eevdf: Curb wakeup-preemption")
Signed-off-by: zihan zhou <15645113830zzh@gmail.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Vincent Guittot <vincent.guittot@linaro.org>
Tested-by: Vincent Guittot <vincent.guittot@linaro.org>
Link: https://lkml.kernel.org/r/20250208080850.16300-1-15645113830zzh@gmail.com
Add the following kfuncs to set and remove xattrs from BPF programs:
bpf_set_dentry_xattr
bpf_remove_dentry_xattr
bpf_set_dentry_xattr_locked
bpf_remove_dentry_xattr_locked
The _locked version of these kfuncs are called from hooks where
dentry->d_inode is already locked. Instead of requiring the user
to know which version of the kfuncs to use, the verifier will pick
the proper kfunc based on the calling hook.
Signed-off-by: Song Liu <song@kernel.org>
Acked-by: Christian Brauner <brauner@kernel.org>
Reviewed-by: Matt Bobrowski <mattbobrowski@google.com>
Link: https://lore.kernel.org/r/20250130213549.3353349-5-song@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
A task wakeup can be either processed on the waker's CPU or bounced to the
wakee's previous CPU using an IPI (ttwu_queue). Bouncing to the wakee's CPU
avoids the waker's CPU locking and accessing the wakee's rq which can be
expensive across cache and node boundaries.
When ttwu_queue path is taken, select_task_rq() and thus ops.select_cpu()
may be skipped in some cases (racing against the wakee switching out). As
this confused some BPF schedulers, there wasn't a good way for a BPF
scheduler to tell whether idle CPU selection has been skipped, ops.enqueue()
couldn't insert tasks into foreign local DSQs, and the performance
difference on machines with simple toplogies were minimal, sched_ext
disabled ttwu_queue.
However, this optimization makes noticeable difference on more complex
topologies and a BPF scheduler now has an easy way tell whether
ops.select_cpu() was skipped since 9b671793c7 ("sched_ext, scx_qmap: Add
and use SCX_ENQ_CPU_SELECTED") and can insert tasks into foreign local DSQs
since 5b26f7b920 ("sched_ext: Allow SCX_DSQ_LOCAL_ON for direct
dispatches").
Implement SCX_OPS_ALLOW_QUEUED_WAKEUP which allows BPF schedulers to choose
to enable ttwu_queue optimization.
v2: Update the patch description and comment re. ops.select_cpu() being
skipped in some cases as opposed to always as per Neel.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Neel Natu <neelnatu@google.com>
Reported-by: Barret Rhoden <brho@google.com>
Cc: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Andrea Righi <arighi@nvidia.com>
Now when we use scx_bpf_task_cgroup() in ops.tick() to get the cgroup of
the current task, the following error will occur:
scx_foo[3795244] triggered exit kind 1024:
runtime error (called on a task not being operated on)
The reason is that we are using SCX_CALL_OP() instead of SCX_CALL_OP_TASK()
when calling ops.tick(), which triggers the error during the subsequent
scx_kf_allowed_on_arg_tasks() check.
SCX_CALL_OP_TASK() was first introduced in commit 36454023f5 ("sched_ext:
Track tasks that are subjects of the in-flight SCX operation") to ensure
task's rq lock is held when accessing task's sched_group. Since ops.tick()
is marked as SCX_KF_TERMINAL and task_tick_scx() is protected by the rq
lock, we can use SCX_CALL_OP_TASK() to avoid the above issue. Similarly,
the same changes should be made for ops.disable() and ops.exit_task(), as
they are also protected by task_rq_lock() and it's safe to access the
task's task_group.
Fixes: 36454023f5 ("sched_ext: Track tasks that are subjects of the in-flight SCX operation")
Signed-off-by: Chuyi Zhou <zhouchuyi@bytedance.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
