| Age | Commit message (Collapse) | Author |
|---|
|
commit 1af6239d1d3e61d33fd2f0ba53d3d1a67cc50574 upstream.
With the advent of CFI it is no longer acceptible to cast function
pointers.
The robot complains thusly:
kernel-events-core.c:warning:cast-from-int-(-)(struct-perf_cpu_pmu_context-)-to-remote_function_f-(aka-int-(-)(void-)-)-converts-to-incompatible-function-type
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Cixi Geng <cixi.geng1@unisoc.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
[ Upstream commit 15def34e2635ab7e0e96f1bc32e1b69609f14942 ]
commit e050e3f0a71bf ("perf: Fix broken interrupt rate throttling")
introduces a change in throttling threshold judgment. Before this,
compare hwc->interrupts and max_samples_per_tick, then increase
hwc->interrupts by 1, but this commit reverses order of these two
behaviors, causing the semantics of max_samples_per_tick to change.
In literal sense of "max_samples_per_tick", if hwc->interrupts ==
max_samples_per_tick, it should not be throttled, therefore, the judgment
condition should be changed to "hwc->interrupts > max_samples_per_tick".
In fact, this may cause the hardlockup to fail, The minimum value of
max_samples_per_tick may be 1, in this case, the return value of
__perf_event_account_interrupt function is 1.
As a result, nmi_watchdog gets throttled, which would stop PMU (Use x86
architecture as an example, see x86_pmu_handle_irq).
Fixes: e050e3f0a71b ("perf: Fix broken interrupt rate throttling")
Signed-off-by: Yang Jihong <yangjihong1@huawei.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20230227023508.102230-1-yangjihong1@huawei.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
|
|
[ Upstream commit 24d3ae2f37d8bc3c14b31d353c5d27baf582b6a6 ]
The same task check in perf_event_set_output has some potential issues
for some usages.
For the current perf code, there is a problem if using of
perf_event_open() to have multiple samples getting into the same mmap’d
memory when they are both attached to the same process.
https://lore.kernel.org/all/92645262-D319-4068-9C44-2409EF44888E@gmail.com/
Because the event->ctx is not ready when the perf_event_set_output() is
invoked in the perf_event_open().
Besides the above issue, before the commit bd2756811766 ("perf: Rewrite
core context handling"), perf record can errors out when sampling with
a hardware event and a software event as below.
$ perf record -e cycles,dummy --per-thread ls
failed to mmap with 22 (Invalid argument)
That's because that prior to the commit a hardware event and a software
event are from different task context.
The problem should be a long time issue since commit c3f00c70276d
("perk: Separate find_get_context() from event initialization").
The task struct is stored in the event->hw.target for each per-thread
event. It is a more reliable way to determine whether two events are
attached to the same task.
The event->hw.target was also introduced several years ago by the
commit 50f16a8bf9d7 ("perf: Remove type specific target pointers"). It
can not only be used to fix the issue with the current code, but also
back port to fix the issues with an older kernel.
Note: The event->hw.target was introduced later than commit
c3f00c70276d. The patch may cannot be applied between the commit
c3f00c70276d and commit 50f16a8bf9d7. Anybody that wants to back-port
this at that period may have to find other solutions.
Fixes: c3f00c70276d ("perf: Separate find_get_context() from event initialization")
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Zhengjun Xing <zhengjun.xing@linux.intel.com>
Link: https://lkml.kernel.org/r/20230322202449.512091-1-kan.liang@linux.intel.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
|
|
[ Upstream commit baf1b12a67f5b24f395baca03e442ce27cab0c18 ]
Time readers rely on perf_event_context->[time|timestamp|timeoffset] to get
accurate time_enabled and time_running for an event. The difference between
ctx->timestamp and ctx->time is the among of time when the context is not
enabled. __update_context_time(ctx, false) is used to increase timestamp,
but not time. Therefore, it should only be called in ctx_sched_in() when
EVENT_TIME was not enabled.
Fixes: 09f5e7dc7ad7 ("perf: Fix perf_event_read_local() time")
Signed-off-by: Song Liu <song@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/r/20230313171608.298734-1-song@kernel.org
Signed-off-by: Sasha Levin <sashal@kernel.org>
|
|
perf_event_bpf_output
[ Upstream commit eb81a2ed4f52be831c9fb879752d89645a312c13 ]
syzkaller reportes a KASAN issue with stack-out-of-bounds.
The call trace is as follows:
dump_stack+0x9c/0xd3
print_address_description.constprop.0+0x19/0x170
__kasan_report.cold+0x6c/0x84
kasan_report+0x3a/0x50
__perf_event_header__init_id+0x34/0x290
perf_event_header__init_id+0x48/0x60
perf_output_begin+0x4a4/0x560
perf_event_bpf_output+0x161/0x1e0
perf_iterate_sb_cpu+0x29e/0x340
perf_iterate_sb+0x4c/0xc0
perf_event_bpf_event+0x194/0x2c0
__bpf_prog_put.constprop.0+0x55/0xf0
__cls_bpf_delete_prog+0xea/0x120 [cls_bpf]
cls_bpf_delete_prog_work+0x1c/0x30 [cls_bpf]
process_one_work+0x3c2/0x730
worker_thread+0x93/0x650
kthread+0x1b8/0x210
ret_from_fork+0x1f/0x30
commit 267fb27352b6 ("perf: Reduce stack usage of perf_output_begin()")
use on-stack struct perf_sample_data of the caller function.
However, perf_event_bpf_output uses incorrect parameter to convert
small-sized data (struct perf_bpf_event) into large-sized data
(struct perf_sample_data), which causes memory overwriting occurs in
__perf_event_header__init_id.
Fixes: 267fb27352b6 ("perf: Reduce stack usage of perf_output_begin()")
Signed-off-by: Yang Jihong <yangjihong1@huawei.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20230314044735.56551-1-yangjihong1@huawei.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
|
|
commit fd0815f632c24878e325821943edccc7fde947a2 upstream.
Events should only be added to a groups rb tree if they have not been
removed from their context by list_del_event(). Since remove_on_exec
made it possible to call list_del_event() on individual events before
they are detached from their group, perf_group_detach() should check each
sibling's attach_state before calling add_event_to_groups() on it.
Fixes: 2e498d0a74e5 ("perf: Add support for event removal on exec")
Signed-off-by: Budimir Markovic <markovicbudimir@gmail.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/ZBFzvQV9tEqoHEtH@gentoo
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
commit 0a041ebca4956292cadfb14a63ace3a9c1dcb0a3 upstream.
It passes the attr struct to the security_perf_event_open() but it's
not initialized yet.
Fixes: da97e18458fb ("perf_event: Add support for LSM and SELinux checks")
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: stable@vger.kernel.org
Link: https://lkml.kernel.org/r/20221220223140.4020470-1-namhyung@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
[ Upstream commit e8d7a90c08ce963c592fb49845f2ccc606a2ac21 ]
In pmu_dev_alloc(), when dev_set_name() failed, it will goto free_dev
and call put_device(pmu->dev) to release it.
However pmu->dev->release is assigned after this, which makes warning
and memleak.
Call dev_set_name() after pmu->dev->release = pmu_dev_release to fix it.
Device '(null)' does not have a release() function...
WARNING: CPU: 2 PID: 441 at drivers/base/core.c:2332 device_release+0x1b9/0x240
...
Call Trace:
<TASK>
kobject_put+0x17f/0x460
put_device+0x20/0x30
pmu_dev_alloc+0x152/0x400
perf_pmu_register+0x96b/0xee0
...
kmemleak: 1 new suspected memory leaks (see /sys/kernel/debug/kmemleak)
unreferenced object 0xffff888014759000 (size 2048):
comm "modprobe", pid 441, jiffies 4294931444 (age 38.332s)
backtrace:
[<0000000005aed3b4>] kmalloc_trace+0x27/0x110
[<000000006b38f9b8>] pmu_dev_alloc+0x50/0x400
[<00000000735f17be>] perf_pmu_register+0x96b/0xee0
[<00000000e38477f1>] 0xffffffffc0ad8603
[<000000004e162216>] do_one_initcall+0xd0/0x4e0
...
Fixes: abe43400579d ("perf: Sysfs enumeration")
Signed-off-by: Chen Zhongjin <chenzhongjin@huawei.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20221111103653.91058-1-chenzhongjin@huawei.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull perf fix from Borislav Petkov:
- Fix a use-after-free case where the perf pending task callback would
see an already freed event
* tag 'perf_urgent_for_v6.1_rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
perf: Fix perf_pending_task() UaF
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Pull networking fixes from Jakub Kicinski:
"Including fixes from bpf, can and wifi.
Current release - new code bugs:
- eth: mlx5e:
- use kvfree() in mlx5e_accel_fs_tcp_create()
- MACsec, fix RX data path 16 RX security channel limit
- MACsec, fix memory leak when MACsec device is deleted
- MACsec, fix update Rx secure channel active field
- MACsec, fix add Rx security association (SA) rule memory leak
Previous releases - regressions:
- wifi: cfg80211: don't allow multi-BSSID in S1G
- stmmac: set MAC's flow control register to reflect current settings
- eth: mlx5:
- E-switch, fix duplicate lag creation
- fix use-after-free when reverting termination table
Previous releases - always broken:
- ipv4: fix route deletion when nexthop info is not specified
- bpf: fix a local storage BPF map bug where the value's spin lock
field can get initialized incorrectly
- tipc: re-fetch skb cb after tipc_msg_validate
- wifi: wilc1000: fix Information Element parsing
- packet: do not set TP_STATUS_CSUM_VALID on CHECKSUM_COMPLETE
- sctp: fix memory leak in sctp_stream_outq_migrate()
- can: can327: fix potential skb leak when netdev is down
- can: add number of missing netdev freeing on error paths
- aquantia: do not purge addresses when setting the number of rings
- wwan: iosm:
- fix incorrect skb length leading to truncated packet
- fix crash in peek throughput test due to skb UAF"
* tag 'net-6.1-rc8-2' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (79 commits)
net: ethernet: renesas: ravb: Fix promiscuous mode after system resumed
MAINTAINERS: Update maintainer list for chelsio drivers
ionic: update MAINTAINERS entry
sctp: fix memory leak in sctp_stream_outq_migrate()
packet: do not set TP_STATUS_CSUM_VALID on CHECKSUM_COMPLETE
net/mlx5: Lag, Fix for loop when checking lag
Revert "net/mlx5e: MACsec, remove replay window size limitation in offload path"
net: marvell: prestera: Fix a NULL vs IS_ERR() check in some functions
net: tun: Fix use-after-free in tun_detach()
net: mdiobus: fix unbalanced node reference count
net: hsr: Fix potential use-after-free
tipc: re-fetch skb cb after tipc_msg_validate
mptcp: fix sleep in atomic at close time
mptcp: don't orphan ssk in mptcp_close()
dsa: lan9303: Correct stat name
ipv4: Fix route deletion when nexthop info is not specified
net: wwan: iosm: fix incorrect skb length
net: wwan: iosm: fix crash in peek throughput test
net: wwan: iosm: fix dma_alloc_coherent incompatible pointer type
net: wwan: iosm: fix kernel test robot reported error
...
|
|
Per syzbot it is possible for perf_pending_task() to run after the
event is free()'d. There are two related but distinct cases:
- the task_work was already queued before destroying the event;
- destroying the event itself queues the task_work.
The first cannot be solved using task_work_cancel() since
perf_release() itself might be called from a task_work (____fput),
which means the current->task_works list is already empty and
task_work_cancel() won't be able to find the perf_pending_task()
entry.
The simplest alternative is extending the perf_event lifetime to cover
the task_work.
The second is just silly, queueing a task_work while you know the
event is going away makes no sense and is easily avoided by
re-arranging how the event is marked STATE_DEAD and ensuring it goes
through STATE_OFF on the way down.
Reported-by: syzbot+9228d6098455bb209ec8@syzkaller.appspotmail.com
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Marco Elver <elver@google.com>
|
|
Some PMUs (notably the traditional hardware kind) have boundary issues
with the OS filter. Specifically, it is possible for
perf_event_attr::exclude_kernel=1 events to trigger in-kernel due to
SKID or errata.
This can upset the sigtrap logic some and trigger the WARN.
However, if this invalid sample is the first we must not loose the
SIGTRAP, OTOH if it is the second, it must not override the
pending_addr with a (possibly) invalid one.
Fixes: ca6c21327c6a ("perf: Fix missing SIGTRAPs")
Reported-by: Pengfei Xu <pengfei.xu@intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Marco Elver <elver@google.com>
Tested-by: Pengfei Xu <pengfei.xu@intel.com>
Link: https://lkml.kernel.org/r/Y3hDYiXwRnJr8RYG@xpf.sh.intel.com
|
|
The perf_event_attr::sigtrap functionality relies on data->addr being
set. However commit 7b0846301531 ("perf: Use sample_flags for addr")
changed this to only initialize data->addr when not 0.
Fixes: 7b0846301531 ("perf: Use sample_flags for addr")
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/Y3426b4OimE%2FI5po%40hirez.programming.kicks-ass.net
|
|
Since commit bfea9a8574f3 ("bpf: Add name to struct bpf_ksym"), when
reporting subprog ksymbol to perf, prog name instead of subprog name is
used. The backtrace of bpf program with subprogs will be incorrect as
shown below:
ffffffffc02deace bpf_prog_e44a3057dcb151f8_overwrite+0x66
ffffffffc02de9f7 bpf_prog_e44a3057dcb151f8_overwrite+0x9f
ffffffffa71d8d4e trace_call_bpf+0xce
ffffffffa71c2938 perf_call_bpf_enter.isra.0+0x48
overwrite is the entry program and it invokes the overwrite_htab subprog
through bpf_loop, but in above backtrace, overwrite program just jumps
inside itself.
Fixing it by using subprog name when reporting subprog ksymbol. After
the fix, the output of perf script will be correct as shown below:
ffffffffc031aad2 bpf_prog_37c0bec7d7c764a4_overwrite_htab+0x66
ffffffffc031a9e7 bpf_prog_c7eb827ef4f23e71_overwrite+0x9f
ffffffffa3dd8d4e trace_call_bpf+0xce
ffffffffa3dc2938 perf_call_bpf_enter.isra.0+0x48
Fixes: bfea9a8574f3 ("bpf: Add name to struct bpf_ksym")
Signed-off-by: Hou Tao <houtao1@huawei.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Link: https://lore.kernel.org/bpf/20221114095733.158588-1-houtao@huaweicloud.com
|
|
To catch missing SIGTRAP we employ a WARN in __perf_event_overflow(),
which fires if pending_sigtrap was already set: returning to user space
without consuming pending_sigtrap, and then having the event fire again
would re-enter the kernel and trigger the WARN.
This, however, seemed to miss the case where some events not associated
with progress in the user space task can fire and the interrupt handler
runs before the IRQ work meant to consume pending_sigtrap (and generate
the SIGTRAP).
syzbot gifted us this stack trace:
| WARNING: CPU: 0 PID: 3607 at kernel/events/core.c:9313 __perf_event_overflow
| Modules linked in:
| CPU: 0 PID: 3607 Comm: syz-executor100 Not tainted 6.1.0-rc2-syzkaller-00073-g88619e77b33d #0
| Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 10/11/2022
| RIP: 0010:__perf_event_overflow+0x498/0x540 kernel/events/core.c:9313
| <...>
| Call Trace:
| <TASK>
| perf_swevent_hrtimer+0x34f/0x3c0 kernel/events/core.c:10729
| __run_hrtimer kernel/time/hrtimer.c:1685 [inline]
| __hrtimer_run_queues+0x1c6/0xfb0 kernel/time/hrtimer.c:1749
| hrtimer_interrupt+0x31c/0x790 kernel/time/hrtimer.c:1811
| local_apic_timer_interrupt arch/x86/kernel/apic/apic.c:1096 [inline]
| __sysvec_apic_timer_interrupt+0x17c/0x640 arch/x86/kernel/apic/apic.c:1113
| sysvec_apic_timer_interrupt+0x40/0xc0 arch/x86/kernel/apic/apic.c:1107
| asm_sysvec_apic_timer_interrupt+0x16/0x20 arch/x86/include/asm/idtentry.h:649
| <...>
| </TASK>
In this case, syzbot produced a program with event type
PERF_TYPE_SOFTWARE and config PERF_COUNT_SW_CPU_CLOCK. The hrtimer
manages to fire again before the IRQ work got a chance to run, all while
never having returned to user space.
Improve the WARN to check for real progress in user space: approximate
this by storing a 32-bit hash of the current IP into pending_sigtrap,
and if an event fires while pending_sigtrap still matches the previous
IP, we assume no progress (false negatives are possible given we could
return to user space and trigger again on the same IP).
Fixes: ca6c21327c6a ("perf: Fix missing SIGTRAPs")
Reported-by: syzbot+b8ded3e2e2c6adde4990@syzkaller.appspotmail.com
Signed-off-by: Marco Elver <elver@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20221031093513.3032814-1-elver@google.com
|
|
Running the test currently fails on non-SMP systems, despite being
enabled by default. This means that running the test with:
./tools/testing/kunit/kunit.py run --arch x86_64 hw_breakpoint
results in every hw_breakpoint test failing with:
# test_one_cpu: failed to initialize: -22
not ok 1 - test_one_cpu
Instead, use kunit_skip(), which will mark the test as skipped, and give
a more comprehensible message:
ok 1 - test_one_cpu # SKIP not enough cpus
This makes it more obvious that the test is not suited to the test
environment, and so wasn't run, rather than having run and failed.
Signed-off-by: David Gow <davidgow@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Daniel Latypov <dlatypov@google.com>
Acked-by: Marco Elver <elver@google.com>
Link: https://lore.kernel.org/r/20221026141040.1609203-1-davidgow@google.com
|
|
Since commit 838d9bb62d13 ("perf: Use sample_flags for raw_data")
raw data is not being output on tracepoints due to the PERF_SAMPLE_RAW
field not being set. Fix this by setting it for tracepoint events.
This fixes the following test failure:
perf test "sched_switch" -vvv
35: Track with sched_switch
--- start ---
test child forked, pid 1828
...
Using CPUID 0x00000000410fd400
sched_switch: cpu: 2 prev_tid -14687 next_tid 0
sched_switch: cpu: 2 prev_tid -14687 next_tid 0
Missing sched_switch events
4613 events recorded
test child finished with -1
---- end ----
Track with sched_switch: FAILED!
Fixes: 838d9bb62d13 ("perf: Use sample_flags for raw_data")
Signed-off-by: James Clark <james.clark@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Tested-by: SeongJae Park <sj@kernel.org>
Tested-by: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Link: https://lore.kernel.org/r/20221012143857.48198-1-james.clark@arm.com
|
|
Marco reported:
Due to the implementation of how SIGTRAP are delivered if
perf_event_attr::sigtrap is set, we've noticed 3 issues:
1. Missing SIGTRAP due to a race with event_sched_out() (more
details below).
2. Hardware PMU events being disabled due to returning 1 from
perf_event_overflow(). The only way to re-enable the event is
for user space to first "properly" disable the event and then
re-enable it.
3. The inability to automatically disable an event after a
specified number of overflows via PERF_EVENT_IOC_REFRESH.
The worst of the 3 issues is problem (1), which occurs when a
pending_disable is "consumed" by a racing event_sched_out(), observed
as follows:
CPU0 | CPU1
--------------------------------+---------------------------
__perf_event_overflow() |
perf_event_disable_inatomic() |
pending_disable = CPU0 | ...
| _perf_event_enable()
| event_function_call()
| task_function_call()
| /* sends IPI to CPU0 */
<IPI> | ...
__perf_event_enable() +---------------------------
ctx_resched()
task_ctx_sched_out()
ctx_sched_out()
group_sched_out()
event_sched_out()
pending_disable = -1
</IPI>
<IRQ-work>
perf_pending_event()
perf_pending_event_disable()
/* Fails to send SIGTRAP because no pending_disable! */
</IRQ-work>
In the above case, not only is that particular SIGTRAP missed, but also
all future SIGTRAPs because 'event_limit' is not reset back to 1.
To fix, rework pending delivery of SIGTRAP via IRQ-work by introduction
of a separate 'pending_sigtrap', no longer using 'event_limit' and
'pending_disable' for its delivery.
Additionally; and different to Marco's proposed patch:
- recognise that pending_disable effectively duplicates oncpu for
the case where it is set. As such, change the irq_work handler to
use ->oncpu to target the event and use pending_* as boolean toggles.
- observe that SIGTRAP targets the ctx->task, so the context switch
optimization that carries contexts between tasks is invalid. If
the irq_work were delayed enough to hit after a context switch the
SIGTRAP would be delivered to the wrong task.
- observe that if the event gets scheduled out
(rotation/migration/context-switch/...) the irq-work would be
insufficient to deliver the SIGTRAP when the event gets scheduled
back in (the irq-work might still be pending on the old CPU).
Therefore have event_sched_out() convert the pending sigtrap into a
task_work which will deliver the signal at return_to_user.
Fixes: 97ba62b27867 ("perf: Add support for SIGTRAP on perf events")
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Debugged-by: Dmitry Vyukov <dvyukov@google.com>
Reported-by: Marco Elver <elver@google.com>
Debugged-by: Marco Elver <elver@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Marco Elver <elver@google.com>
Tested-by: Marco Elver <elver@google.com>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
Pull MM updates from Andrew Morton:
- Yu Zhao's Multi-Gen LRU patches are here. They've been under test in
linux-next for a couple of months without, to my knowledge, any
negative reports (or any positive ones, come to that).
- Also the Maple Tree from Liam Howlett. An overlapping range-based
tree for vmas. It it apparently slightly more efficient in its own
right, but is mainly targeted at enabling work to reduce mmap_lock
contention.
Liam has identified a number of other tree users in the kernel which
could be beneficially onverted to mapletrees.
Yu Zhao has identified a hard-to-hit but "easy to fix" lockdep splat
at [1]. This has yet to be addressed due to Liam's unfortunately
timed vacation. He is now back and we'll get this fixed up.
- Dmitry Vyukov introduces KMSAN: the Kernel Memory Sanitizer. It uses
clang-generated instrumentation to detect used-unintialized bugs down
to the single bit level.
KMSAN keeps finding bugs. New ones, as well as the legacy ones.
- Yang Shi adds a userspace mechanism (madvise) to induce a collapse of
memory into THPs.
- Zach O'Keefe has expanded Yang Shi's madvise(MADV_COLLAPSE) to
support file/shmem-backed pages.
- userfaultfd updates from Axel Rasmussen
- zsmalloc cleanups from Alexey Romanov
- cleanups from Miaohe Lin: vmscan, hugetlb_cgroup, hugetlb and
memory-failure
- Huang Ying adds enhancements to NUMA balancing memory tiering mode's
page promotion, with a new way of detecting hot pages.
- memcg updates from Shakeel Butt: charging optimizations and reduced
memory consumption.
- memcg cleanups from Kairui Song.
- memcg fixes and cleanups from Johannes Weiner.
- Vishal Moola provides more folio conversions
- Zhang Yi removed ll_rw_block() :(
- migration enhancements from Peter Xu
- migration error-path bugfixes from Huang Ying
- Aneesh Kumar added ability for a device driver to alter the memory
tiering promotion paths. For optimizations by PMEM drivers, DRM
drivers, etc.
- vma merging improvements from Jakub Matěn.
- NUMA hinting cleanups from David Hildenbrand.
- xu xin added aditional userspace visibility into KSM merging
activity.
- THP & KSM code consolidation from Qi Zheng.
- more folio work from Matthew Wilcox.
- KASAN updates from Andrey Konovalov.
- DAMON cleanups from Kaixu Xia.
- DAMON work from SeongJae Park: fixes, cleanups.
- hugetlb sysfs cleanups from Muchun Song.
- Mike Kravetz fixes locking issues in hugetlbfs and in hugetlb core.
Link: https://lkml.kernel.org/r/CAOUHufZabH85CeUN-MEMgL8gJGzJEWUrkiM58JkTbBhh-jew0Q@mail.gmail.com [1]
* tag 'mm-stable-2022-10-08' of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm: (555 commits)
hugetlb: allocate vma lock for all sharable vmas
hugetlb: take hugetlb vma_lock when clearing vma_lock->vma pointer
hugetlb: fix vma lock handling during split vma and range unmapping
mglru: mm/vmscan.c: fix imprecise comments
mm/mglru: don't sync disk for each aging cycle
mm: memcontrol: drop dead CONFIG_MEMCG_SWAP config symbol
mm: memcontrol: use do_memsw_account() in a few more places
mm: memcontrol: deprecate swapaccounting=0 mode
mm: memcontrol: don't allocate cgroup swap arrays when memcg is disabled
mm/secretmem: remove reduntant return value
mm/hugetlb: add available_huge_pages() func
mm: remove unused inline functions from include/linux/mm_inline.h
selftests/vm: add selftest for MADV_COLLAPSE of uffd-minor memory
selftests/vm: add file/shmem MADV_COLLAPSE selftest for cleared pmd
selftests/vm: add thp collapse shmem testing
selftests/vm: add thp collapse file and tmpfs testing
selftests/vm: modularize thp collapse memory operations
selftests/vm: dedup THP helpers
mm/khugepaged: add tracepoint to hpage_collapse_scan_file()
mm/madvise: add file and shmem support to MADV_COLLAPSE
...
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull perf events updates from Ingo Molnar:
"PMU driver updates:
- Add AMD Last Branch Record Extension Version 2 (LbrExtV2) feature
support for Zen 4 processors.
- Extend the perf ABI to provide branch speculation information, if
available, and use this on CPUs that have it (eg. LbrExtV2).
- Improve Intel PEBS TSC timestamp handling & integration.
- Add Intel Raptor Lake S CPU support.
- Add 'perf mem' and 'perf c2c' memory profiling support on AMD CPUs
by utilizing IBS tagged load/store samples.
- Clean up & optimize various x86 PMU details.
HW breakpoints:
- Big rework to optimize the code for systems with hundreds of CPUs
and thousands of breakpoints:
- Replace the nr_bp_mutex global mutex with the bp_cpuinfo_sem
per-CPU rwsem that is read-locked during most of the key
operations.
- Improve the O(#cpus * #tasks) logic in toggle_bp_slot() and
fetch_bp_busy_slots().
- Apply micro-optimizations & cleanups.
- Misc cleanups & enhancements"
* tag 'perf-core-2022-10-07' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (75 commits)
perf/hw_breakpoint: Annotate tsk->perf_event_mutex vs ctx->mutex
perf: Fix pmu_filter_match()
perf: Fix lockdep_assert_event_ctx()
perf/x86/amd/lbr: Adjust LBR regardless of filtering
perf/x86/utils: Fix uninitialized var in get_branch_type()
perf/uapi: Define PERF_MEM_SNOOPX_PEER in kernel header file
perf/x86/amd: Support PERF_SAMPLE_PHY_ADDR
perf/x86/amd: Support PERF_SAMPLE_ADDR
perf/x86/amd: Support PERF_SAMPLE_{WEIGHT|WEIGHT_STRUCT}
perf/x86/amd: Support PERF_SAMPLE_DATA_SRC
perf/x86/amd: Add IBS OP_DATA2 DataSrc bit definitions
perf/mem: Introduce PERF_MEM_LVLNUM_{EXTN_MEM|IO}
perf/x86/uncore: Add new Raptor Lake S support
perf/x86/cstate: Add new Raptor Lake S support
perf/x86/msr: Add new Raptor Lake S support
perf/x86: Add new Raptor Lake S support
bpf: Check flags for branch stack in bpf_read_branch_records helper
perf, hw_breakpoint: Fix use-after-free if perf_event_open() fails
perf: Use sample_flags for raw_data
perf: Use sample_flags for addr
...
|
|
Perf fuzzer gifted a lockdep splat:
perf_event_init_context()
mutex_lock(parent_ctx->mutex); (B)
inherit_task_group()
inherit_group()
inherit_event()
perf_event_alloc()
perf_try_init_event() := hw_breakpoint_event_init()
register_perf_hw_breakpoint()
mutex_lock(child->perf_event_mutex); (A)
Which is against the normal (documented) order. Now, this is a false
positive in that child is not published yet, but also inherited events
never end up on ->perf_event_list.
Annotate this one away.
Fixes: 0912037fec11 ("perf/hw_breakpoint: Reduce contention with large number of tasks")
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
|
|
Mark reported that the new for_each_sibling_event() assertion triggers
in pmu_filter_match() -- which isn't always called with IRQs disabled
or ctx->mutex held.
Fixes: f3c0eba28704 ("perf: Add a few assertions")
Reported-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/YvvJq2f/7eFVcnNy@FVFF77S0Q05N
|
|
Add support for MADV_COLLAPSE to collapse shmem-backed and file-backed
memory into THPs (requires CONFIG_READ_ONLY_THP_FOR_FS=y).
On success, the backing memory will be a hugepage. For the memory range
and process provided, the page tables will synchronously have a huge pmd
installed, mapping the THP. Other mappings of the file extent mapped by
the memory range may be added to a set of entries that khugepaged will
later process and attempt update their page tables to map the THP by a
pmd.
This functionality unlocks two important uses:
(1) Immediately back executable text by THPs. Current support provided
by CONFIG_READ_ONLY_THP_FOR_FS may take a long time on a large
system which might impair services from serving at their full rated
load after (re)starting. Tricks like mremap(2)'ing text onto
anonymous memory to immediately realize iTLB performance prevents
page sharing and demand paging, both of which increase steady state
memory footprint. Now, we can have the best of both worlds: Peak
upfront performance and lower RAM footprints.
(2) userfaultfd-based live migration of virtual machines satisfy UFFD
faults by fetching native-sized pages over the network (to avoid
latency of transferring an entire hugepage). However, after guest
memory has been fully copied to the new host, MADV_COLLAPSE can
be used to immediately increase guest performance.
Since khugepaged is single threaded, this change now introduces
possibility of collapse contexts racing in file collapse path. There a
important few places to consider:
(1) hpage_collapse_scan_file(), when we xas_pause() and drop RCU.
We could have the memory collapsed out from under us, but
the next xas_for_each() iteration will correctly pick up the
hugepage. The hugepage might not be up to date (insofar as
copying of small page contents might not have completed - the
page still may be locked), but regardless what small page index
we were iterating over, we'll find the hugepage and identify it
as a suitably aligned compound page of order HPAGE_PMD_ORDER.
In khugepaged path, we locklessly check the value of the pmd,
and only add it to deferred collapse array if we find pmd
mapping pte table. This is fine, since other values that could
have raced in right afterwards denote failure, or that the
memory was successfully collapsed, so we don't need further
processing.
In madvise path, we'll take mmap_lock() in write to serialize
against page table updates and will know what to do based on the
true value of the pmd: recheck all ptes if we point to a pte table,
directly install the pmd, if the pmd has been cleared, but
memory not yet faulted, or nothing at all if we find a huge pmd.
It's worth putting emphasis here on how we treat the none pmd
here. If khugepaged has processed this mm's page tables
already, it will have left the pmd cleared (ready for refault by
the process). Depending on the VMA flags and sysfs settings,
amount of RAM on the machine, and the current load, could be a
relatively common occurrence - and as such is one we'd like to
handle successfully in MADV_COLLAPSE. When we see the none pmd
in collapse_pte_mapped_thp(), we've locked mmap_lock in write
and checked (a) huepaged_vma_check() to see if the backing
memory is appropriate still, along with VMA sizing and
appropriate hugepage alignment within the file, and (b) we've
found a hugepage head of order HPAGE_PMD_ORDER at the offset
in the file mapped by our hugepage-aligned virtual address.
Even though the common-case is likely race with khugepaged,
given these checks (regardless how we got here - we could be
operating on a completely different file than originally checked
in hpage_collapse_scan_file() for all we know) it should be safe
to directly make the pmd a huge pmd pointing to this hugepage.
(2) collapse_file() is mostly serialized on the same file extent by
lock sequence:
| lock hupepage
| lock mapping->i_pages
| lock 1st page
| unlock mapping->i_pages
| <page checks>
| lock mapping->i_pages
| page_ref_freeze(3)
| xas_store(hugepage)
| unlock mapping->i_pages
| page_ref_unfreeze(1)
| unlock 1st page
V unlock hugepage
Once a context (who already has their fresh hugepage locked)
locks mapping->i_pages exclusively, it will hold said lock
until it locks the first page, and it will hold that lock until
the after the hugepage has been added to the page cache (and
will unlock the hugepage after page table update, though that
isn't important here).
A racing context that loses the race for mapping->i_pages will
then lose the race to locking the first page. Here - depending
on how far the other racing context has gotten - we might find
the new hugepage (in which case we'll exit cleanly when we
check PageTransCompound()), or we'll find the "old" 1st small
page (in which we'll exit cleanly when we discover unexpected
refcount of 2 after isolate_lru_page()). This is assuming we
are able to successfully lock the page we find - in shmem path,
we could just fail the trylock and exit cleanly anyways.
Failure path in collapse_file() is similar: once we hold lock
on 1st small page, we are serialized against other collapse
contexts. Before the 1st small page is unlocked, we add it
back to the pagecache and unfreeze the refcount appropriately.
Contexts who lost the race to the 1st small page will then find
the same 1st small page with the correct refcount and will be
able to proceed.
[zokeefe@google.com: don't check pmd value twice in collapse_pte_mapped_thp()]
Link: https://lkml.kernel.org/r/20220927033854.477018-1-zokeefe@google.com
[shy828301@gmail.com: Delete hugepage_vma_revalidate_anon(), remove
check for multi-add in khugepaged_add_pte_mapped_thp()]
Link: https://lore.kernel.org/linux-mm/CAHbLzkrtpM=ic7cYAHcqkubah5VTR8N5=k5RT8MTvv5rN1Y91w@mail.gmail.com/
Link: https://lkml.kernel.org/r/20220907144521.3115321-4-zokeefe@google.com
Link: https://lkml.kernel.org/r/20220922224046.1143204-4-zokeefe@google.com
Signed-off-by: Zach O'Keefe <zokeefe@google.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: Chris Kennelly <ckennelly@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: James Houghton <jthoughton@google.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Pasha Tatashin <pasha.tatashin@soleen.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Rongwei Wang <rongwei.wang@linux.alibaba.com>
Cc: SeongJae Park <sj@kernel.org>
Cc: Song Liu <songliubraving@fb.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yang Shi <shy828301@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
Saves several calls to compound_head().
Link: https://lkml.kernel.org/r/20220902194653.1739778-57-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
Remove a few hidden calls to compound_head().
Link: https://lkml.kernel.org/r/20220902194653.1739778-45-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
IBS_DC_PHYSADDR provides the physical data address for the tagged load/
store operation. Populate perf sample physical address using it.
Signed-off-by: Ravi Bangoria <ravi.bangoria@amd.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20220928095805.596-7-ravi.bangoria@amd.com
|
|
Local testing revealed that we can trigger a use-after-free during
rhashtable lookup as follows:
| BUG: KASAN: use-after-free in memcmp lib/string.c:757
| Read of size 8 at addr ffff888107544dc0 by task perf-rhltable-n/1293
|
| CPU: 0 PID: 1293 Comm: perf-rhltable-n Not tainted 6.0.0-rc3-00014-g85260862789c #46
| Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.0-debian-1.16.0-4 04/01/2014
| Call Trace:
| <TASK>
| memcmp lib/string.c:757
| rhashtable_compare include/linux/rhashtable.h:577 [inline]
| __rhashtable_lookup include/linux/rhashtable.h:602 [inline]
| rhltable_lookup include/linux/rhashtable.h:688 [inline]
| task_bp_pinned kernel/events/hw_breakpoint.c:324
| toggle_bp_slot kernel/events/hw_breakpoint.c:462
| __release_bp_slot kernel/events/hw_breakpoint.c:631 [inline]
| release_bp_slot kernel/events/hw_breakpoint.c:639
| register_perf_hw_breakpoint kernel/events/hw_breakpoint.c:742
| hw_breakpoint_event_init kernel/events/hw_breakpoint.c:976
| perf_try_init_event kernel/events/core.c:11261
| perf_init_event kernel/events/core.c:11325 [inline]
| perf_event_alloc kernel/events/core.c:11619
| __do_sys_perf_event_open kernel/events/core.c:12157
| do_syscall_x64 arch/x86/entry/common.c:50 [inline]
| do_syscall_64 arch/x86/entry/common.c:80
| entry_SYSCALL_64_after_hwframe
| </TASK>
|
| Allocated by task 1292:
| perf_event_alloc kernel/events/core.c:11505
| __do_sys_perf_event_open kernel/events/core.c:12157
| do_syscall_x64 arch/x86/entry/common.c:50 [inline]
| do_syscall_64 arch/x86/entry/common.c:80
| entry_SYSCALL_64_after_hwframe
|
| Freed by task 1292:
| perf_event_alloc kernel/events/core.c:11716
| __do_sys_perf_event_open kernel/events/core.c:12157
| do_syscall_x64 arch/x86/entry/common.c:50 [inline]
| do_syscall_64 arch/x86/entry/common.c:80
| entry_SYSCALL_64_after_hwframe
|
| The buggy address belongs to the object at ffff888107544c00
| which belongs to the cache perf_event of size 1352
| The buggy address is located 448 bytes inside of
| 1352-byte region [ffff888107544c00, ffff888107545148)
This happens because the first perf_event_open() managed to reserve a HW
breakpoint slot, however, later fails for other reasons and returns. The
second perf_event_open() runs concurrently, and during rhltable_lookup()
looks up an entry which is being freed: since rhltable_lookup() may run
concurrently (under the RCU read lock) with rhltable_remove(), we may
end up with a stale entry, for which memory may also have already been
freed when being accessed.
To fix, only free the failed perf_event after an RCU grace period. This
allows subsystems that store references to an event to always access it
concurrently under the RCU read lock, even if initialization will fail.
Given failure is unlikely and a slow-path, turning the immediate free
into a call_rcu()-wrapped free does not affect performance elsewhere.
Fixes: 0370dc314df3 ("perf/hw_breakpoint: Optimize list of per-task breakpoints")
Reported-by: syzkaller <syzkaller@googlegroups.com>
Signed-off-by: Marco Elver <elver@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20220927172025.1636995-1-elver@google.com
|
|
Use the new sample_flags to indicate whether the raw data field is
filled by the PMU driver. Although it could check with the NULL,
follow the same rule with other fields.
Remove the raw field from the perf_sample_data_init() to minimize
the number of cache lines touched.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20220921220032.2858517-2-namhyung@kernel.org
|
|
Use the new sample_flags to indicate whether the addr field is filled by
the PMU driver. As most PMU drivers pass 0, it can set the flag only if
it has a non-zero value. And use 0 in perf_sample_output() if it's not
filled already.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20220921220032.2858517-1-namhyung@kernel.org
|
|
The VMA iterator is faster than the linked list and removing the linked
list will shrink the vm_area_struct.
Link: https://lkml.kernel.org/r/20220906194824.2110408-48-Liam.Howlett@oracle.com
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Liam R. Howlett <Liam.Howlett@Oracle.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Davidlohr Bueso <dave@stgolabs.net>
Tested-by: Yu Zhao <yuzhao@google.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: David Howells <dhowells@redhat.com>
Cc: SeongJae Park <sj@kernel.org>
Cc: Sven Schnelle <svens@linux.ibm.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
Coccinelle reports a warning:
WARNING: use scnprintf or sprintf
This LWN article explains the rationale for this change:
https: //lwn.net/Articles/69419/
Ie. snprintf() returns what *would* be the resulting length,
while scnprintf() returns the actual length.
Adding to that, there has also been some slow migration from snprintf to scnprintf,
here's the shift in usage in the past 3.5 years, in all fs/ files:
v5.0 v6.0-rc6
--------------------------------------
snprintf() uses: 63 213
scnprintf() uses: 374 186
No intended change in behavior.
[ mingo: Improved the changelog & reviewed the usage sites. ]
Signed-off-by: Jules Irenge <jbi.octave@gmail.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
|
|
If the perf_event has PERF_SAMPLE_CALLCHAIN, BPF can use it for stack trace.
The problematic cases like PEBS and IBS already handled in the PMU driver and
they filled the callchain info in the sample data. For others, we can call
perf_callchain() before the BPF handler.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220908214104.3851807-2-namhyung@kernel.org
|
|
So that it can call perf_callchain() only if needed. Historically it used
__PERF_SAMPLE_CALLCHAIN_EARLY but we can do that with sample_flags in the
struct perf_sample_data.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220908214104.3851807-1-namhyung@kernel.org
|
|
While auditing 6b959ba22d34 ("perf/core: Fix reentry problem in
perf_output_read_group()") a few spots were found that wanted
assertions.
Notable for_each_sibling_event() relies on exclusion from
modification. This would normally be holding either ctx->lock or
ctx->mutex, however due to how things are constructed disabling IRQs
is a valid and sufficient substitute for ctx->lock.
Another possible site to add assertions would be the various
pmu::{add,del,read,..}() methods, but that's not trivially expressable
in C -- the best option is wrappers, but those are easy enough to
forget.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
|
|
Besides the branch type filtering requests, 'event.attr.branch_sample_type'
also contains various flags indicating which additional information should
be captured, along with the base branch record. These flags help configure
the underlying hardware, and capture the branch records appropriately when
required e.g after PMU interrupt. But first, this moves an existing helper
perf_sample_save_hw_index() into the header before adding some more helpers
for other branch sample filter flags.
Signed-off-by: Anshuman Khandual <anshuman.khandual@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20220906084414.396220-1-anshuman.khandual@arm.com
|
|
Use the new sample_flags to indicate whether the txn field is filled by
the PMU driver.
Remove the txn field from the perf_sample_data_init() to minimize the
number of cache lines touched.
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220901130959.1285717-7-kan.liang@linux.intel.com
|
|
Use the new sample_flags to indicate whether the data_src field is
filled by the PMU driver.
Remove the data_src field from the perf_sample_data_init() to minimize
the number of cache lines touched.
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220901130959.1285717-6-kan.liang@linux.intel.com
|
|
Use the new sample_flags to indicate whether the weight field is filled
by the PMU driver.
Remove the weight field from the perf_sample_data_init() to minimize the
number of cache lines touched.
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220901130959.1285717-5-kan.liang@linux.intel.com
|
|
Use the new sample_flags to indicate whether the branch stack is filled
by the PMU driver.
Remove the br_stack from the perf_sample_data_init() to minimize the number
of cache lines touched.
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220901130959.1285717-4-kan.liang@linux.intel.com
|
|
On some platforms, some data e.g., timestamps, can be retrieved from
the PMU driver. Usually, the data from the PMU driver is more accurate.
The current perf kernel should output the PMU-filled sample data if
it's available.
To check the availability of the PMU-filled sample data, the current
perf kernel initializes the related fields in the
perf_sample_data_init(). When outputting a sample, the perf checks
whether the field is updated by the PMU driver. If yes, the updated
value will be output. If not, the perf uses an SW way to calculate the
value or just outputs the initialized value if an SW way is unavailable
either.
With more and more data being provided by the PMU driver, more fields
has to be initialized in the perf_sample_data_init(). That will
increase the number of cache lines touched in perf_sample_data_init()
and be harmful to the performance.
Add new "sample_flags" to indicate the PMU-filled sample data. The PMU
driver should set the corresponding PERF_SAMPLE_ flag when the field is
updated. The initialization of the corresponding field is not required
anymore. The following patches will make use of it and remove the
corresponding fields from the perf_sample_data_init(), which will
further minimize the number of cache lines touched.
Only clear the sample flags that have already been done by the PMU
driver in the perf_prepare_sample() for the PERF_RECORD_SAMPLE. For the
other PERF_RECORD_ event type, the sample data is not available.
Suggested-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220901130959.1285717-2-kan.liang@linux.intel.com
|
|
perf_output_read_group may respond to IPI request of other cores and invoke
__perf_install_in_context function. As a result, hwc configuration is modified.
causing inconsistency and unexpected consequences.
Interrupts are not disabled when perf_output_read_group reads PMU counter.
In this case, IPI request may be received from other cores.
As a result, PMU configuration is modified and an error occurs when
reading PMU counter:
CPU0 CPU1
__se_sys_perf_event_open
perf_install_in_context
perf_output_read_group smp_call_function_single
for_each_sibling_event(sub, leader) { generic_exec_single
if ((sub != event) && remote_function
(sub->state == PERF_EVENT_STATE_ACTIVE)) |
<enter IPI handler: __perf_install_in_context> <----RAISE IPI-----+
__perf_install_in_context
ctx_resched
event_sched_out
armpmu_del
...
hwc->idx = -1; // event->hwc.idx is set to -1
...
<exit IPI>
sub->pmu->read(sub);
armpmu_read
armv8pmu_read_counter
armv8pmu_read_hw_counter
int idx = event->hw.idx; // idx = -1
u64 val = armv8pmu_read_evcntr(idx);
u32 counter = ARMV8_IDX_TO_COUNTER(idx); // invalid counter = 30
read_pmevcntrn(counter) // undefined instruction
Signed-off-by: Yang Jihong <yangjihong1@huawei.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20220902082918.179248-1-yangjihong1@huawei.com
|
|
We can still see that a majority of the time is spent hashing task pointers:
...
16.98% [kernel] [k] rhashtable_jhash2
...
Doing the bookkeeping in toggle_bp_slots() is currently O(#cpus),
calling task_bp_pinned() for each CPU, even if task_bp_pinned() is
CPU-independent. The reason for this is to update the per-CPU
'tsk_pinned' histogram.
To optimize the CPU-independent case to O(1), keep a separate
CPU-independent 'tsk_pinned_all' histogram.
The major source of complexity are transitions between "all
CPU-independent task breakpoints" and "mixed CPU-independent and
CPU-dependent task breakpoints". The code comments list all cases that
require handling.
After this optimization:
| $> perf bench -r 100 breakpoint thread -b 4 -p 128 -t 512
| # Running 'breakpoint/thread' benchmark:
| # Created/joined 100 threads with 4 breakpoints and 128 parallelism
| Total time: 1.758 [sec]
|
| 34.336621 usecs/op
| 4395.087500 usecs/op/cpu
38.08% [kernel] [k] queued_spin_lock_slowpath
10.81% [kernel] [k] smp_cfm_core_cond
3.01% [kernel] [k] update_sg_lb_stats
2.58% [kernel] [k] osq_lock
2.57% [kernel] [k] llist_reverse_order
1.45% [kernel] [k] find_next_bit
1.21% [kernel] [k] flush_tlb_func_common
1.01% [kernel] [k] arch_install_hw_breakpoint
Showing that the time spent hashing keys has become insignificant.
With the given benchmark parameters, that's an improvement of 12%
compared with the old O(#cpus) version.
And finally, using the less aggressive parameters from the preceding
changes, we now observe:
| $> perf bench -r 30 breakpoint thread -b 4 -p 64 -t 64
| # Running 'breakpoint/thread' benchmark:
| # Created/joined 30 threads with 4 breakpoints and 64 parallelism
| Total time: 0.067 [sec]
|
| 35.292187 usecs/op
| 2258.700000 usecs/op/cpu
Which is an improvement of 12% compared to without the histogram
optimizations (baseline is 40 usecs/op). This is now on par with the
theoretical ideal (constraints disabled), and only 12% slower than no
breakpoints at all.
Signed-off-by: Marco Elver <elver@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
Acked-by: Ian Rogers <irogers@google.com>
Link: https://lore.kernel.org/r/20220829124719.675715-15-elver@google.com
|
|
targets
Running the perf benchmark with (note: more aggressive parameters vs.
preceding changes, but same 256 CPUs host):
| $> perf bench -r 100 breakpoint thread -b 4 -p 128 -t 512
| # Running 'breakpoint/thread' benchmark:
| # Created/joined 100 threads with 4 breakpoints and 128 parallelism
| Total time: 1.989 [sec]
|
| 38.854160 usecs/op
| 4973.332500 usecs/op/cpu
20.43% [kernel] [k] queued_spin_lock_slowpath
18.75% [kernel] [k] osq_lock
16.98% [kernel] [k] rhashtable_jhash2
8.34% [kernel] [k] task_bp_pinned
4.23% [kernel] [k] smp_cfm_core_cond
3.65% [kernel] [k] bcmp
2.83% [kernel] [k] toggle_bp_slot
1.87% [kernel] [k] find_next_bit
1.49% [kernel] [k] __reserve_bp_slot
We can see that a majority of the time is now spent hashing task
pointers to index into task_bps_ht in task_bp_pinned().
Obtaining the max_bp_pinned_slots() for CPU-independent task targets
currently is O(#cpus), and calls task_bp_pinned() for each CPU, even if
the result of task_bp_pinned() is CPU-independent.
The loop in max_bp_pinned_slots() wants to compute the maximum slots
across all CPUs. If task_bp_pinned() is CPU-independent, we can do so by
obtaining the max slots across all CPUs and adding task_bp_pinned().
To do so in O(1), use a bp_slots_histogram for CPU-pinned slots.
After this optimization:
| $> perf bench -r 100 breakpoint thread -b 4 -p 128 -t 512
| # Running 'breakpoint/thread' benchmark:
| # Created/joined 100 threads with 4 breakpoints and 128 parallelism
| Total time: 1.930 [sec]
|
| 37.697832 usecs/op
| 4825.322500 usecs/op/cpu
19.13% [kernel] [k] queued_spin_lock_slowpath
18.21% [kernel] [k] rhashtable_jhash2
15.46% [kernel] [k] osq_lock
6.27% [kernel] [k] toggle_bp_slot
5.91% [kernel] [k] task_bp_pinned
5.05% [kernel] [k] smp_cfm_core_cond
1.78% [kernel] [k] update_sg_lb_stats
1.36% [kernel] [k] llist_reverse_order
1.34% [kernel] [k] find_next_bit
1.19% [kernel] [k] bcmp
Suggesting that time spent in task_bp_pinned() has been reduced.
However, we're still hashing too much, which will be addressed in the
subsequent change.
Signed-off-by: Marco Elver <elver@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
Acked-by: Ian Rogers <irogers@google.com>
Link: https://lore.kernel.org/r/20220829124719.675715-14-elver@google.com
|
|
Factor out the existing `atomic_t count[N]` into its own struct called
'bp_slots_histogram', to generalize and make its intent clearer in
preparation of reusing elsewhere. The basic idea of bucketing "total
uses of N slots" resembles a histogram, so calling it such seems most
intuitive.
No functional change.
Signed-off-by: Marco Elver <elver@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
Acked-by: Ian Rogers <irogers@google.com>
Link: https://lore.kernel.org/r/20220829124719.675715-13-elver@google.com
|
|
While optimizing task_bp_pinned()'s runtime complexity to O(1) on
average helps reduce time spent in the critical section, we still suffer
due to serializing everything via 'nr_bp_mutex'. Indeed, a profile shows
that now contention is the biggest issue:
95.93% [kernel] [k] osq_lock
0.70% [kernel] [k] mutex_spin_on_owner
0.22% [kernel] [k] smp_cfm_core_cond
0.18% [kernel] [k] task_bp_pinned
0.18% [kernel] [k] rhashtable_jhash2
0.15% [kernel] [k] queued_spin_lock_slowpath
when running the breakpoint benchmark with (system with 256 CPUs):
| $> perf bench -r 30 breakpoint thread -b 4 -p 64 -t 64
| # Running 'breakpoint/thread' benchmark:
| # Created/joined 30 threads with 4 breakpoints and 64 parallelism
| Total time: 0.207 [sec]
|
| 108.267188 usecs/op
| 6929.100000 usecs/op/cpu
The main concern for synchronizing the breakpoint constraints data is
that a consistent snapshot of the per-CPU and per-task data is observed.
The access pattern is as follows:
1. If the target is a task: the task's pinned breakpoints are counted,
checked for space, and then appended to; only bp_cpuinfo::cpu_pinned
is used to check for conflicts with CPU-only breakpoints;
bp_cpuinfo::tsk_pinned are incremented/decremented, but otherwise
unused.
2. If the target is a CPU: bp_cpuinfo::cpu_pinned are counted, along
with bp_cpuinfo::tsk_pinned; after a successful check, cpu_pinned is
incremented. No per-task breakpoints are checked.
Since rhltable safely synchronizes insertions/deletions, we can allow
concurrency as follows:
1. If the target is a task: independent tasks may update and check the
constraints concurrently, but same-task target calls need to be
serialized; since bp_cpuinfo::tsk_pinned is only updated, but not
checked, these modifications can happen concurrently by switching
tsk_pinned to atomic_t.
2. If the target is a CPU: access to the per-CPU constraints needs to
be serialized with other CPU-target and task-target callers (to
stabilize the bp_cpuinfo::tsk_pinned snapshot).
We can allow the above concurrency by introducing a per-CPU constraints
data reader-writer lock (bp_cpuinfo_sem), and per-task mutexes (reuses
task_struct::perf_event_mutex):
1. If the target is a task: acquires perf_event_mutex, and acquires
bp_cpuinfo_sem as a reader. The choice of percpu-rwsem minimizes
contention in the presence of many read-lock but few write-lock
acquisitions: we assume many orders of magnitude more task target
breakpoints creations/destructions than CPU target breakpoints.
2. If the target is a CPU: acquires bp_cpuinfo_sem as a writer.
With these changes, contention with thousands of tasks is reduced to the
point where waiting on locking no longer dominates the profile:
| $> perf bench -r 30 breakpoint thread -b 4 -p 64 -t 64
| # Running 'breakpoint/thread' benchmark:
| # Created/joined 30 threads with 4 breakpoints and 64 parallelism
| Total time: 0.077 [sec]
|
| 40.201563 usecs/op
| 2572.900000 usecs/op/cpu
21.54% [kernel] [k] task_bp_pinned
20.18% [kernel] [k] rhashtable_jhash2
6.81% [kernel] [k] toggle_bp_slot
5.47% [kernel] [k] queued_spin_lock_slowpath
3.75% [kernel] [k] smp_cfm_core_cond
3.48% [kernel] [k] bcmp
On this particular setup that's a speedup of 2.7x.
We're also getting closer to the theoretical ideal performance through
optimizations in hw_breakpoint.c -- constraints accounting disabled:
| perf bench -r 30 breakpoint thread -b 4 -p 64 -t 64
| # Running 'breakpoint/thread' benchmark:
| # Created/joined 30 threads with 4 breakpoints and 64 parallelism
| Total time: 0.067 [sec]
|
| 35.286458 usecs/op
| 2258.333333 usecs/op/cpu
Which means the current implementation is ~12% slower than the
theoretical ideal.
For reference, performance without any breakpoints:
| $> bench -r 30 breakpoint thread -b 0 -p 64 -t 64
| # Running 'breakpoint/thread' benchmark:
| # Created/joined 30 threads with 0 breakpoints and 64 parallelism
| Total time: 0.060 [sec]
|
| 31.365625 usecs/op
| 2007.400000 usecs/op/cpu
On a system with 256 CPUs, the theoretical ideal is only ~12% slower
than no breakpoints at all; the current implementation is ~28% slower.
Signed-off-by: Marco Elver <elver@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
Acked-by: Ian Rogers <irogers@google.com>
Link: https://lore.kernel.org/r/20220829124719.675715-12-elver@google.com
|
|
Flexible breakpoints have never been implemented, with
bp_cpuinfo::flexible always being 0. Unfortunately, they still occupy 4
bytes in each bp_cpuinfo and bp_busy_slots, as well as computing the max
flexible count in fetch_bp_busy_slots().
This again causes suboptimal code generation, when we always know that
`!!slots.flexible` will be 0.
Just get rid of the flexible "placeholder" and remove all real code
related to it. Make a note in the comment related to the constraints
algorithm but don't remove them from the algorithm, so that if in future
flexible breakpoints need supporting, it should be trivial to revive
them (along with reverting this change).
Signed-off-by: Marco Elver <elver@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
Acked-by: Ian Rogers <irogers@google.com>
Link: https://lore.kernel.org/r/20220829124719.675715-9-elver@google.com
|
|
Due to being a __weak function, hw_breakpoint_weight() will cause the
compiler to always emit a call to it. This generates unnecessarily bad
code (register spills etc.) for no good reason; in fact it appears in
profiles of `perf bench -r 100 breakpoint thread -b 4 -p 128 -t 512`:
...
0.70% [kernel] [k] hw_breakpoint_weight
...
While a small percentage, no architecture defines its own
hw_breakpoint_weight() nor are there users outside hw_breakpoint.c,
which makes the fact it is currently __weak a poor choice.
Change hw_breakpoint_weight()'s definition to follow a similar protocol
to hw_breakpoint_slots(), such that if <asm/hw_breakpoint.h> defines
hw_breakpoint_weight(), we'll use it instead.
The result is that it is inlined and no longer shows up in profiles.
Signed-off-by: Marco Elver <elver@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
Acked-by: Ian Rogers <irogers@google.com>
Link: https://lore.kernel.org/r/20220829124719.675715-8-elver@google.com
|
|
Optimize internal hw_breakpoint state if the architecture's number of
breakpoint slots is constant. This avoids several kmalloc() calls and
potentially unnecessary failures if the allocations fail, as well as
subtly improves code generation and cache locality.
The protocol is that if an architecture defines hw_breakpoint_slots via
the preprocessor, it must be constant and the same for all types.
Signed-off-by: Marco Elver <elver@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Dmitry Vyukov <dvyukov@google.com>
Acked-by: Ian Rogers <irogers@google.com>
Link: https://lore.kernel.org/r/20220829124719.675715-7-elver@google.com
|
|
Mark read-only data after initialization as __ro_after_init.
While we are here, turn 'constraints_initialized' into a bool.
Signed-off-by: Marco Elver <elver@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
Acked-by: Ian Rogers <irogers@google.com>
Link: https://lore.kernel.org/r/20220829124719.675715-6-elver@google.com
|
|
On a machine with 256 CPUs, running the recently added perf breakpoint
benchmark results in:
| $> perf bench -r 30 breakpoint thread -b 4 -p 64 -t 64
| # Running 'breakpoint/thread' benchmark:
| # Created/joined 30 threads with 4 breakpoints and 64 parallelism
| Total time: 236.418 [sec]
|
| 123134.794271 usecs/op
| 7880626.833333 usecs/op/cpu
The benchmark tests inherited breakpoint perf events across many
threads.
Looking at a perf profile, we can see that the majority of the time is
spent in various hw_breakpoint.c functions, which execute within the
'nr_bp_mutex' critical sections which then results in contention on that
mutex as well:
37.27% [kernel] [k] osq_lock
34.92% [kernel] [k] mutex_spin_on_owner
12.15% [kernel] [k] toggle_bp_slot
11.90% [kernel] [k] __reserve_bp_slot
The culprit here is task_bp_pinned(), which has a runtime complexity of
O(#tasks) due to storing all task breakpoints in the same list and
iterating through that list looking for a matching task. Clearly, this
does not scale to thousands of tasks.
Instead, make use of the "rhashtable" variant "rhltable" which stores
multiple items with the same key in a list. This results in average
runtime complexity of O(1) for task_bp_pinned().
With the optimization, the benchmark shows:
| $> perf bench -r 30 breakpoint thread -b 4 -p 64 -t 64
| # Running 'breakpoint/thread' benchmark:
| # Created/joined 30 threads with 4 breakpoints and 64 parallelism
| Total time: 0.208 [sec]
|
| 108.422396 usecs/op
| 6939.033333 usecs/op/cpu
On this particular setup that's a speedup of ~1135x.
While one option would be to make task_struct a breakpoint list node,
this would only further bloat task_struct for infrequently used data.
Furthermore, after all optimizations in this series, there's no evidence
it would result in better performance: later optimizations make the time
spent looking up entries in the hash table negligible (we'll reach the
theoretical ideal performance i.e. no constraints).
Signed-off-by: Marco Elver <elver@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
Acked-by: Ian Rogers <irogers@google.com>
Link: https://lore.kernel.org/r/20220829124719.675715-5-elver@google.com
|