HomeDefectsLIN1025-1935
Acknowledged

LIN1025-1935 : Security Advisory - linux - CVE-2022-50118

Created: Jun 18, 2025    Updated: Sep 4, 2026
Resolved Date: Jun 19, 2025
Found In Version: 10.25.33.1
Severity: Standard
Applicable for: Wind River Linux LTS 25
Component/s: Kernel

Description

In the Linux kernel, the following vulnerability has been resolved:EOL][EOL]powerpc/perf: Optimize clearing the pending PMI and remove WARN_ON for PMI check in power_pmu_disable[EOL][EOL]commit 2c9ac51b850d ("powerpc/perf: Fix PMU callbacks to clear[EOL]pending PMI before resetting an overflown PMC") added a new[EOL]function "pmi_irq_pending" in hw_irq.h. This function is to check[EOL]if there is a PMI marked as pending in Paca (PACA_IRQ_PMI).This is[EOL]used in power_pmu_disable in a WARN_ON. The intention here is to[EOL]provide a warning if there is PMI pending, but no counter is found[EOL]overflown.[EOL][EOL]During some of the perf runs, below warning is hit:[EOL][EOL]WARNING: CPU: 36 PID: 0 at arch/powerpc/perf/core-book3s.c:1332 power_pmu_disable+0x25c/0x2c0[EOL] Modules linked in:[EOL] -----[EOL][EOL] NIP [c000000000141c3c] power_pmu_disable+0x25c/0x2c0[EOL] LR [c000000000141c8c] power_pmu_disable+0x2ac/0x2c0[EOL] Call Trace:[EOL] [c000000baffcfb90] [c000000000141c8c] power_pmu_disable+0x2ac/0x2c0 (unreliable)[EOL] [c000000baffcfc10] [c0000000003e2f8c] perf_pmu_disable+0x4c/0x60[EOL] [c000000baffcfc30] [c0000000003e3344] group_sched_out.part.124+0x44/0x100[EOL] [c000000baffcfc80] [c0000000003e353c] __perf_event_disable+0x13c/0x240[EOL] [c000000baffcfcd0] [c0000000003dd334] event_function+0xc4/0x140[EOL] [c000000baffcfd20] [c0000000003d855c] remote_function+0x7c/0xa0[EOL] [c000000baffcfd50] [c00000000026c394] flush_smp_call_function_queue+0xd4/0x300[EOL] [c000000baffcfde0] [c000000000065b24] smp_ipi_demux_relaxed+0xa4/0x100[EOL] [c000000baffcfe20] [c0000000000cb2b0] xive_muxed_ipi_action+0x20/0x40[EOL] [c000000baffcfe40] [c000000000207c3c] __handle_irq_event_percpu+0x8c/0x250[EOL] [c000000baffcfee0] [c000000000207e2c] handle_irq_event_percpu+0x2c/0xa0[EOL] [c000000baffcff10] [c000000000210a04] handle_percpu_irq+0x84/0xc0[EOL] [c000000baffcff40] [c000000000205f14] generic_handle_irq+0x54/0x80[EOL] [c000000baffcff60] [c000000000015740] __do_irq+0x90/0x1d0[EOL] [c000000baffcff90] [c000000000016990] __do_IRQ+0xc0/0x140[EOL] [c0000009732f3940] [c000000bafceaca8] 0xc000000bafceaca8[EOL] [c0000009732f39d0] [c000000000016b78] do_IRQ+0x168/0x1c0[EOL] [c0000009732f3a00] [c0000000000090c8] hardware_interrupt_common_virt+0x218/0x220[EOL][EOL]This means that there is no PMC overflown among the active events[EOL]in the PMU, but there is a PMU pending in Paca. The function[EOL]"any_pmc_overflown" checks the PMCs on active events in[EOL]cpuhw->n_events. Code snippet:[EOL][EOL]<<>>[EOL]if (any_pmc_overflown(cpuhw))[EOL] \tclear_pmi_irq_pending();[EOL] else[EOL] \tWARN_ON(pmi_irq_pending());[EOL]<<>>[EOL][EOL]Here the PMC overflown is not from active event. Example: When we do[EOL]perf record, default cycles and instructions will be running on PMC6[EOL]and PMC5 respectively. It could happen that overflowed event is currently[EOL]not active and pending PMI is for the inactive event. Debug logs from[EOL]trace_printk:[EOL][EOL]<<>>[EOL]any_pmc_overflown: idx is 5: pmc value is 0xd9a[EOL]power_pmu_disable: PMC1: 0x0, PMC2: 0x0, PMC3: 0x0, PMC4: 0x0, PMC5: 0xd9a, PMC6: 0x80002011[EOL]<<>>[EOL][EOL]Here active PMC (from idx) is PMC5 , but overflown PMC is PMC6(0x80002011).[EOL]When we handle PMI interrupt for such cases, if the PMC overflown is[EOL]from inactive event, it will be ignored. Reference commit:[EOL]commit bc09c219b2e6 ("powerpc/perf: Fix finding overflowed PMC in interrupt")[EOL][EOL]Patch addresses two changes:[EOL]1) Fix 1 : Removal of warning ( WARN_ON(pmi_irq_pending()); )[EOL]   We were printing warning if no PMC is found overflown among active PMU[EOL]   events, but PMI pending in PACA. But this could happen in cases where[EOL]   PMC overflown is not in active PMC. An inactive event could have caused[EOL]   the overflow. Hence the warning is not needed. To know pending PMI is[EOL]   from an inactive event, we need to loop through all PMC's which will[EOL]   cause more SPR reads via mfspr and increase in context switch. Also in[EOL]   existing function: perf_event_interrupt, already we ignore PMI's[EOL]   overflown when it is from an inactive PMC.[EOL][EOL]2) Fix 2: optimization in clearing pending PMI.[EOL]   Currently we check for any active PMC overflown before clearing PMI[EOL]   pending in Paca. This is causing additional SP[EOL]---truncated---

CREATE(Triage):(User=lchen-cn) [CVE-2022-50118 (https://nvd.nist.gov/vuln/detail/CVE-2022-50118)