| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Fri, 21 Jan 2022 12:08:14 +0530
From: Anshuman Khandual <anshuman.khandual@....com>
To: Naoya Horiguchi <naoya.horiguchi@...ux.dev>
Cc: linux-mm@...ck.org, Steven Rostedt <rostedt@...dmis.org>,
Ingo Molnar <mingo@...hat.com>,
Andrew Morton <akpm@...ux-foundation.org>,
Zi Yan <ziy@...dia.com>,
Naoya Horiguchi <naoya.horiguchi@....com>,
John Hubbard <jhubbard@...dia.com>,
Matthew Wilcox <willy@...radead.org>,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH] mm/migration: Add trace events for THP migrations
On 1/11/22 12:27 PM, Naoya Horiguchi wrote:
> On Tue, Jan 11, 2022 at 10:31:21AM +0530, Anshuman Khandual wrote:
>>
>>
>> On 1/11/22 7:28 AM, Naoya Horiguchi wrote:
>>> Hi Anshuman,
>>>
>>> On Fri, Jan 07, 2022 at 10:29:35AM +0530, Anshuman Khandual wrote:
>>>> This adds two trace events for PMD based THP migration without split. These
>>>> events closely follow the implementation details like setting and removing
>>>> of PMD migration entries, which are essential operations for THP migration.
>>>
>>> I often want to check which individual pages are migrated to which places
>>> (or not migrated) for testing, so these new tracepoints could help me.
>>> Maybe these can be much greater if they can handle other types of page
>>> migration for raw pages and hugetlb pages. Is it hard to cover all such
>>> page migration events?
>>
>> Are you suggesting to cover all migration entry transitions for normal
>> and HugeTLB pages as well ?
>
> Yes if you like the idea. I think that some events listed below can be grouped
> into one tracepoint event with showing args like pgsize or read/write flags
> (or implementation detail is up to you).
In its simplest form, something like this will work ? Although the THP migration
patch still remains (almost) unchanged.
include/trace/events/migrate.h | 38 ++++++++++++++++++++++++++++++++++
mm/migrate.c | 10 +++++++--
mm/rmap.c | 3 +++
3 files changed, 49 insertions(+), 2 deletions(-)
diff --git a/include/trace/events/migrate.h b/include/trace/events/migrate.h
index 779f3fad9ecd..b66652fcc8af 100644
--- a/include/trace/events/migrate.h
+++ b/include/trace/events/migrate.h
@@ -105,6 +105,44 @@ TRACE_EVENT(mm_migrate_pages_start,
__print_symbolic(__entry->reason, MIGRATE_REASON))
);
+TRACE_EVENT(set_migration_pte,
+
+ TP_PROTO(unsigned long addr, unsigned long pte),
+
+ TP_ARGS(addr, pte),
+
+ TP_STRUCT__entry(
+ __field(unsigned long, addr)
+ __field(unsigned long, pte)
+ ),
+
+ TP_fast_assign(
+ __entry->addr = addr;
+ __entry->pte = pte;
+ ),
+
+ TP_printk("create pte migration entry addr=%lx, pte=%lx", __entry->addr, __entry->pte)
+);
+
+TRACE_EVENT(remove_migration_ptes,
+
+ TP_PROTO(struct page *old_page, struct page *new_page),
+
+ TP_ARGS(old_page, new_page),
+
+ TP_STRUCT__entry(
+ __field(struct page *, old_page)
+ __field(struct page *, new_page)
+ ),
+
+ TP_fast_assign(
+ __entry->old_page = old_page;
+ __entry->new_page = new_page;
+ ),
+
+ TP_printk("remove pte migration entry old_page=%lx new_page=%lx", __entry->old_page, __entry->new_page);
+);
+
#endif /* _TRACE_MIGRATE_H */
/* This part must be outside protection */
diff --git a/mm/migrate.c b/mm/migrate.c
index 18ce840914f0..271b1d565642 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -1053,9 +1053,12 @@ static int __unmap_and_move(struct page *page, struct page *newpage,
if (!page_mapped(page))
rc = move_to_new_page(newpage, page, mode);
- if (page_was_mapped)
+ if (page_was_mapped) {
remove_migration_ptes(page,
rc == MIGRATEPAGE_SUCCESS ? newpage : page, false);
+ trace_remove_migration_ptes(page,
+ rc == MIGRATEPAGE_SUCCESS ? newpage : page);
+ }
out_unlock_both:
unlock_page(newpage);
@@ -1275,9 +1278,12 @@ static int unmap_and_move_huge_page(new_page_t get_new_page,
if (!page_mapped(hpage))
rc = move_to_new_page(new_hpage, hpage, mode);
- if (page_was_mapped)
+ if (page_was_mapped) {
remove_migration_ptes(hpage,
rc == MIGRATEPAGE_SUCCESS ? new_hpage : hpage, false);
+ trace_remove_migration_ptes(hpage,
+ rc == MIGRATEPAGE_SUCCESS ? new_hpage : hpage);
+ }
unlock_put_anon:
unlock_page(new_hpage);
diff --git a/mm/rmap.c b/mm/rmap.c
index 6a1e8c7f6213..cd373e378694 100644
--- a/mm/rmap.c
+++ b/mm/rmap.c
@@ -77,6 +77,7 @@
#include <asm/tlbflush.h>
#include <trace/events/tlb.h>
+#include <trace/events/migrate.h>
#include "internal.h"
@@ -1861,6 +1862,7 @@ static bool try_to_migrate_one(struct page *page, struct vm_area_struct *vma,
if (pte_swp_uffd_wp(pteval))
swp_pte = pte_swp_mkuffd_wp(swp_pte);
set_pte_at(mm, pvmw.address, pvmw.pte, swp_pte);
+ trace_set_migration_pte(pvmw.address, pte_val(swp_pte));
/*
* No need to invalidate here it will synchronize on
* against the special swap migration pte.
@@ -1929,6 +1931,7 @@ static bool try_to_migrate_one(struct page *page, struct vm_area_struct *vma,
if (pte_uffd_wp(pteval))
swp_pte = pte_swp_mkuffd_wp(swp_pte);
set_pte_at(mm, address, pvmw.pte, swp_pte);
+ trace_set_migration_pte(address, pte_val(swp_pte));
/*
* No need to invalidate here it will synchronize on
* against the special swap migration pte.
--
2.20.1
>
>>
>> migrate_pages()
>> unmap_and_move_huge_page()
>> try_to_migrate()
>> make_writable_migration_entry() <---
>> make_readable_migration_entry() <---
>> remove_migration_ptes() <---
>> unmap_and_move()
>> __unmap_and_move()
>> try_to_migrate()
>> make_writable_migration_entry() <---
>> make_readable_migration_entry() <---
>> remove_migration_ptes() <---
>
> Thanks,
> Naoya Horiguchi
>
Powered by blists - more mailing lists