user/sven/linux.git/kernel/events, branch v5.15.81

perf: Improve missing SIGTRAP checking

2022-11-26T08:24:49Z

[ Upstream commit bb88f9695460bec25aa30ba9072595025cf6c8af ] 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: | | 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 | <...> | 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 Signed-off-by: Peter Zijlstra (Intel) Link: https://lkml.kernel.org/r/20221031093513.3032814-1-elver@google.com Signed-off-by: Sasha Levin

perf: Fix missing SIGTRAPs

2022-11-03T14:59:15Z

[ Upstream commit ca6c21327c6af02b7eec31ce4b9a740a18c6c13f ] 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 */ | ... __perf_event_enable() +--------------------------- ctx_resched() task_ctx_sched_out() ctx_sched_out() group_sched_out() event_sched_out() pending_disable = -1 perf_pending_event() perf_pending_event_disable() /* Fails to send SIGTRAP because no pending_disable! */ 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 Debugged-by: Dmitry Vyukov Reported-by: Marco Elver Debugged-by: Marco Elver Signed-off-by: Peter Zijlstra (Intel) Reviewed-by: Marco Elver Tested-by: Marco Elver Signed-off-by: Sasha Levin

perf/core: Fix data race between perf_event_set_output() and perf_mmap_close()

2022-07-29T15:25:11Z

[ Upstream commit 68e3c69803dada336893640110cb87221bb01dcf ] Yang Jihing reported a race between perf_event_set_output() and perf_mmap_close(): CPU1 CPU2 perf_mmap_close(e2) if (atomic_dec_and_test(&e2->rb->mmap_count)) // 1 - > 0 detach_rest = true ioctl(e1, IOC_SET_OUTPUT, e2) perf_event_set_output(e1, e2) ... list_for_each_entry_rcu(e, &e2->rb->event_list, rb_entry) ring_buffer_attach(e, NULL); // e1 isn't yet added and // therefore not detached ring_buffer_attach(e1, e2->rb) list_add_rcu(&e1->rb_entry, &e2->rb->event_list) After this; e1 is attached to an unmapped rb and a subsequent perf_mmap() will loop forever more: again: mutex_lock(&e->mmap_mutex); if (event->rb) { ... if (!atomic_inc_not_zero(&e->rb->mmap_count)) { ... mutex_unlock(&e->mmap_mutex); goto again; } } The loop in perf_mmap_close() holds e2->mmap_mutex, while the attach in perf_event_set_output() holds e1->mmap_mutex. As such there is no serialization to avoid this race. Change perf_event_set_output() to take both e1->mmap_mutex and e2->mmap_mutex to alleviate that problem. Additionally, have the loop in perf_mmap() detach the rb directly, this avoids having to wait for the concurrent perf_mmap_close() to get around to doing it to make progress. Fixes: 9bb5d40cd93c ("perf: Fix mmap() accounting hole") Reported-by: Yang Jihong Signed-off-by: Peter Zijlstra (Intel) Tested-by: Yang Jihong Link: https://lkml.kernel.org/r/YsQ3jm2GR38SW7uD@worktop.programming.kicks-ass.net Signed-off-by: Sasha Levin

signal: Deliver SIGTRAP on perf event asynchronously if blocked

2022-06-09T08:22:48Z

[ Upstream commit 78ed93d72ded679e3caf0758357209887bda885f ] With SIGTRAP on perf events, we have encountered termination of processes due to user space attempting to block delivery of SIGTRAP. Consider this case: ... sigset_t s; sigemptyset(&s); sigaddset(&s, SIGTRAP | ); sigprocmask(SIG_BLOCK, &s, ...); ... When the perf event triggers, while SIGTRAP is blocked, force_sig_perf() will force the signal, but revert back to the default handler, thus terminating the task. This makes sense for error conditions, but not so much for explicitly requested monitoring. However, the expectation is still that signals generated by perf events are synchronous, which will no longer be the case if the signal is blocked and delivered later. To give user space the ability to clearly distinguish synchronous from asynchronous signals, introduce siginfo_t::si_perf_flags and TRAP_PERF_FLAG_ASYNC (opted for flags in case more binary information is required in future). The resolution to the problem is then to (a) no longer force the signal (avoiding the terminations), but (b) tell user space via si_perf_flags if the signal was synchronous or not, so that such signals can be handled differently (e.g. let user space decide to ignore or consider the data imprecise). The alternative of making the kernel ignore SIGTRAP on perf events if the signal is blocked may work for some usecases, but likely causes issues in others that then have to revert back to interception of sigprocmask() (which we want to avoid). [ A concrete example: when using breakpoint perf events to track data-flow, in a region of code where signals are blocked, data-flow can no longer be tracked accurately. When a relevant asynchronous signal is received after unblocking the signal, the data-flow tracking logic needs to know its state is imprecise. ] Fixes: 97ba62b27867 ("perf: Add support for SIGTRAP on perf events") Reported-by: Dmitry Vyukov Signed-off-by: Marco Elver Signed-off-by: Peter Zijlstra (Intel) Acked-by: Geert Uytterhoeven Tested-by: Dmitry Vyukov Link: https://lore.kernel.org/r/20220404111204.935357-1-elver@google.com Signed-off-by: Sasha Levin

perf: Fix sys_perf_event_open() race against self

2022-05-25T07:57:27Z

commit 3ac6487e584a1eb54071dbe1212e05b884136704 upstream. Norbert reported that it's possible to race sys_perf_event_open() such that the looser ends up in another context from the group leader, triggering many WARNs. The move_group case checks for races against itself, but the !move_group case doesn't, seemingly relying on the previous group_leader->ctx == ctx check. However, that check is racy due to not holding any locks at that time. Therefore, re-check the result after acquiring locks and bailing if they no longer match. Additionally, clarify the not_move_group case from the move_group-vs-move_group race. Fixes: f63a8daa5812 ("perf: Fix event->ctx locking") Reported-by: Norbert Slusarek Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Linus Torvalds Signed-off-by: Greg Kroah-Hartman

perf/core: Fix perf_mmap fail when CONFIG_PERF_USE_VMALLOC enabled

2022-04-27T12:38:58Z

[ Upstream commit 60490e7966659b26d74bf1fa4aa8693d9a94ca88 ] This problem can be reproduced with CONFIG_PERF_USE_VMALLOC enabled on both x86_64 and aarch64 arch when using sysdig -B(using ebpf)[1]. sysdig -B works fine after rebuilding the kernel with CONFIG_PERF_USE_VMALLOC disabled. I tracked it down to the if condition event->rb->nr_pages != nr_pages in perf_mmap is true when CONFIG_PERF_USE_VMALLOC is enabled where event->rb->nr_pages = 1 and nr_pages = 2048 resulting perf_mmap to return -EINVAL. This is because when CONFIG_PERF_USE_VMALLOC is enabled, rb->nr_pages is always equal to 1. Arch with CONFIG_PERF_USE_VMALLOC enabled by default: arc/arm/csky/mips/sh/sparc/xtensa Arch with CONFIG_PERF_USE_VMALLOC disabled by default: x86_64/aarch64/... Fix this problem by using data_page_nr() [1] https://github.com/draios/sysdig Fixes: 906010b2134e ("perf_event: Provide vmalloc() based mmap() backing") Signed-off-by: Zhipeng Xie Signed-off-by: Peter Zijlstra (Intel) Link: https://lkml.kernel.org/r/20220209145417.6495-1-xiezhipeng1@huawei.com Signed-off-by: Sasha Levin

perf/core: Inherit event_caps

2022-04-13T18:59:24Z

commit e3265a4386428d3d157d9565bb520aabff8b4bf0 upstream. It was reported that some perf event setup can make fork failed on ARM64. It was the case of a group of mixed hw and sw events and it failed in perf_event_init_task() due to armpmu_event_init(). The ARM PMU code checks if all the events in a group belong to the same PMU except for software events. But it didn't set the event_caps of inherited events and no longer identify them as software events. Therefore the test failed in a child process. A simple reproducer is: $ perf stat -e '{cycles,cs,instructions}' perf bench sched messaging # Running 'sched/messaging' benchmark: perf: fork(): Invalid argument The perf stat was fine but the perf bench failed in fork(). Let's inherit the event caps from the parent. Signed-off-by: Namhyung Kim Signed-off-by: Peter Zijlstra (Intel) Cc: Link: https://lkml.kernel.org/r/20220328200112.457740-1-namhyung@kernel.org Signed-off-by: Greg Kroah-Hartman

perf/core: Fix address filter parser for multiple filters

2022-04-08T12:23:10Z

[ Upstream commit d680ff24e9e14444c63945b43a37ede7cd6958f9 ] Reset appropriate variables in the parser loop between parsing separate filters, so that they do not interfere with parsing the next filter. Fixes: 375637bc524952 ("perf/core: Introduce address range filtering") Signed-off-by: Adrian Hunter Signed-off-by: Peter Zijlstra (Intel) Link: https://lore.kernel.org/r/20220131072453.2839535-4-adrian.hunter@intel.com Signed-off-by: Sasha Levin

perf: Fix list corruption in perf_cgroup_switch()

2022-02-16T11:56:41Z

commit 5f4e5ce638e6a490b976ade4a40017b40abb2da0 upstream. There's list corruption on cgrp_cpuctx_list. This happens on the following path: perf_cgroup_switch: list_for_each_entry(cgrp_cpuctx_list) cpu_ctx_sched_in ctx_sched_in ctx_pinned_sched_in merge_sched_in perf_cgroup_event_disable: remove the event from the list Use list_for_each_entry_safe() to allow removing an entry during iteration. Fixes: 058fe1c0440e ("perf/core: Make cgroup switch visit only cpuctxs with cgroup events") Signed-off-by: Song Liu Reviewed-by: Rik van Riel Signed-off-by: Peter Zijlstra (Intel) Link: https://lkml.kernel.org/r/20220204004057.2961252-1-song@kernel.org Signed-off-by: Greg Kroah-Hartman

perf: Always wake the parent event

2022-02-16T11:56:15Z

[ Upstream commit 961c39121759ad09a89598ec4ccdd34ae0468a19 ] When using per-process mode and event inheritance is set to true, forked processes will create a new perf events via inherit_event() -> perf_event_alloc(). But these events will not have ring buffers assigned to them. Any call to wakeup will be dropped if it's called on an event with no ring buffer assigned because that's the object that holds the wakeup list. If the child event is disabled due to a call to perf_aux_output_begin() or perf_aux_output_end(), the wakeup is dropped leaving userspace hanging forever on the poll. Normally the event is explicitly re-enabled by userspace after it wakes up to read the aux data, but in this case it does not get woken up so the event remains disabled. This can be reproduced when using Arm SPE and 'stress' which forks once before running the workload. By looking at the list of aux buffers read, it's apparent that they stop after the fork: perf record -e arm_spe// -vvv -- stress -c 1 With this patch applied they continue to be printed. This behaviour doesn't happen when using systemwide or per-cpu mode. Reported-by: Ruben Ayrapetyan Signed-off-by: James Clark Signed-off-by: Peter Zijlstra (Intel) Link: https://lkml.kernel.org/r/20211206113840.130802-2-james.clark@arm.com Signed-off-by: Sasha Levin