From: Michael Jeanson Date: Tue, 12 Nov 2024 16:19:23 +0000 (-0500) Subject: fix: mm/page_alloc: fix tracepoint mm_page_alloc_zone_locked() (v5.15.171) X-Git-Url: https://git.lttng.org./?a=commitdiff_plain;ds=sidebyside;p=lttng-modules.git fix: mm/page_alloc: fix tracepoint mm_page_alloc_zone_locked() (v5.15.171) See upstream backported commit: commit 28e7a507196fefd119e7ca2286840f1a9aad5e8a Author: Wonhyuk Yang Date: Thu May 19 14:08:54 2022 -0700 mm/page_alloc: fix tracepoint mm_page_alloc_zone_locked() [ Upstream commit 10e0f7530205799e7e971aba699a7cb3a47456de ] Currently, trace point mm_page_alloc_zone_locked() doesn't show correct information. First, when alloc_flag has ALLOC_HARDER/ALLOC_CMA, page can be allocated from MIGRATE_HIGHATOMIC/MIGRATE_CMA. Nevertheless, tracepoint use requested migration type not MIGRATE_HIGHATOMIC and MIGRATE_CMA. Second, after commit 44042b4498728 ("mm/page_alloc: allow high-order pages to be stored on the per-cpu lists") percpu-list can store high order pages. But trace point determine whether it is a refiil of percpu-list by comparing requested order and 0. To handle these problems, make mm_page_alloc_zone_locked() only be called by __rmqueue_smallest with correct migration type. With a new argument called percpu_refill, it can show roughly whether it is a refill of percpu-list. Link: https://lkml.kernel.org/r/20220512025307.57924-1-vvghjk1234@gmail.com Change-Id: Ib76feb79d95e9f93c84c3aa1b946e57ac2e2666a Signed-off-by: Michael Jeanson Signed-off-by: Mathieu Desnoyers --- diff --git a/include/instrumentation/events/kmem.h b/include/instrumentation/events/kmem.h index 3c616cb2..b761a67b 100644 --- a/include/instrumentation/events/kmem.h +++ b/include/instrumentation/events/kmem.h @@ -367,6 +367,7 @@ LTTNG_TRACEPOINT_EVENT_MAP(mm_page_alloc, kmem_mm_page_alloc, ) #if (LTTNG_LINUX_VERSION_CODE >= LTTNG_KERNEL_VERSION(5,19,0) || \ + LTTNG_KERNEL_RANGE(5,15,171, 5,16,0) || \ LTTNG_RHEL_KERNEL_RANGE(5,14,0,163,0,0, 5,15,0,0,0,0)) LTTNG_TRACEPOINT_EVENT_CLASS(kmem_mm_page,