From: "David Hildenbrand (Arm)" <david@kernel.org>
To: Alexander Gordeev <agordeev@linux.ibm.com>,
Andrew Morton <akpm@linux-foundation.org>,
Lorenzo Stoakes <ljs@kernel.org>
Cc: Gerald Schaefer <gerald.schaefer@linux.ibm.com>,
Heiko Carstens <hca@linux.ibm.com>,
Vasily Gorbik <gor@linux.ibm.com>,
linux-s390@vger.kernel.org, linux-mm@kvack.org,
linux-kernel@vger.kernel.org, Ryan Roberts <Ryan.Roberts@arm.com>,
"Christophe Leroy (CS GROUP)" <chleroy@kernel.org>
Subject: Re: [PATCH] mm/page_vma_mapped_walk: add missing pgtable entry accessors
Date: Mon, 27 Apr 2026 11:24:31 +0200 [thread overview]
Message-ID: <08ecabe9-0664-4aea-82fb-f9cb1739f762@kernel.org> (raw)
In-Reply-To: <69c4c12c-7cdc-4a17-9eb7-6fb1a61f9834@kernel.org>
On 4/27/26 11:02, David Hildenbrand (Arm) wrote:
> On 4/27/26 07:20, Alexander Gordeev wrote:
>> Convert pgtable direct entry dereferences to the corresponding
>> pXdp_get() accessors. Use ptep_get_lockless() variant for PTE
>> reads when no lock is taken.
>>
>> Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>
>> ---
>> mm/page_vma_mapped.c | 12 ++++++------
>> 1 file changed, 6 insertions(+), 6 deletions(-)
>>
>> diff --git a/mm/page_vma_mapped.c b/mm/page_vma_mapped.c
>> index b38a1d00c971..a4520bb10d2a 100644
>> --- a/mm/page_vma_mapped.c
>> +++ b/mm/page_vma_mapped.c
>> @@ -41,7 +41,7 @@ static bool map_pte(struct page_vma_mapped_walk *pvmw, pmd_t *pmdvalp,
>> if (!pvmw->pte)
>> return false;
>>
>> - ptent = ptep_get(pvmw->pte);
>> + ptent = ptep_get_lockless(pvmw->pte);
>>
>> if (pte_none(ptent)) {
>> return false;
>> @@ -219,17 +219,17 @@ bool page_vma_mapped_walk(struct page_vma_mapped_walk *pvmw)
>> restart:
>> do {
>> pgd = pgd_offset(mm, pvmw->address);
>> - if (!pgd_present(*pgd)) {
>> + if (!pgd_present(pgdp_get(pgd))) {
>> step_forward(pvmw, PGDIR_SIZE);
>> continue;
>> }
>> p4d = p4d_offset(pgd, pvmw->address);
>> - if (!p4d_present(*p4d)) {
>> + if (!p4d_present(p4dp_get(p4d))) {
>> step_forward(pvmw, P4D_SIZE);
>> continue;
>> }
>> pud = pud_offset(p4d, pvmw->address);
>> - if (!pud_present(*pud)) {
>> + if (!pud_present(pudp_get(pud))) {
>> step_forward(pvmw, PUD_SIZE);
>> continue;
>
> Wasn't there a problem with folded page tables, where we would no longer be able
> to optimize out the folded page table accesses?
>
> I thought we discussed ways to resolve that. Let me dig ...
Here is what I have after the discussion following
https://lore.kernel.org/all/0019d675-ce3d-4a5c-89ed-f126c45145c9@kernel.org/.
Completely untested of course:
From b4715d49feb588032c6bb44bd97cf06739b5c99f Mon Sep 17 00:00:00 2001
From: "David Hildenbrand (Arm)" <david@kernel.org>
Date: Mon, 27 Apr 2026 11:16:21 +0200
Subject: [PATCH] mm: optimize pmdp_get() and friends for folded pagetable
levels
Using pmdp_get() and friends in common code on a kernel config with
folded page tables is suboptimal: pdmp_get() and friends defaults to a
READ_ONCE(), forcing the compiler to actually read that value even though
it will not actually be used afterwards.
This was recently reported by Christophe Leroy [1].
Once we realize that the output of pdmp_get() on these kernel configs is
entirely ignored, as pmd_present()==1 and pmd_leaf()==0 are just
hard-coded, we can just make it return some dummy value.
pmd_offset() is expected to be called with the pudp afterwards, simply
performing a typecast of the pudp pointer to a pmdp pointer.
Let's introduce a pmd_offset_lockless() that does exactly the same:
perform a typecast.
[1] https://lore.kernel.org/all/0019d675-ce3d-4a5c-89ed-f126c45145c9@kernel.org/
Signed-off-by: David Hildenbrand (Arm) <david@kernel.org>
---
include/asm-generic/pgtable-nop4d.h | 8 ++++++++
include/asm-generic/pgtable-nopmd.h | 19 +++++++++++++++++++
include/asm-generic/pgtable-nopud.h | 8 ++++++++
3 files changed, 35 insertions(+)
diff --git a/include/asm-generic/pgtable-nop4d.h b/include/asm-generic/pgtable-nop4d.h
index 03b7dae47dd4..cfe87036d61c 100644
--- a/include/asm-generic/pgtable-nop4d.h
+++ b/include/asm-generic/pgtable-nop4d.h
@@ -32,6 +32,14 @@ static inline void pgd_clear(pgd_t *pgd) { }
*/
#define set_pgd(pgdptr, pgdval) set_p4d((p4d_t *)(pgdptr), (p4d_t) { pgdval })
+static inline pgd_t pgdp_get(pgd_t *p4dp)
+{
+ pgd_t dummy = { 0 };
+
+ return dummy;
+}
+#define pgdp_get pgdp_get
+
static inline p4d_t *p4d_offset(pgd_t *pgd, unsigned long address)
{
return (p4d_t *)pgd;
diff --git a/include/asm-generic/pgtable-nopmd.h b/include/asm-generic/pgtable-nopmd.h
index 8ffd64e7a24c..e87a42407563 100644
--- a/include/asm-generic/pgtable-nopmd.h
+++ b/include/asm-generic/pgtable-nopmd.h
@@ -43,12 +43,31 @@ static inline void pud_clear(pud_t *pud) { }
*/
#define set_pud(pudptr, pudval) set_pmd((pmd_t *)(pudptr), (pmd_t) { pudval })
+static inline pud_t pudp_get(pud_t *pudp)
+{
+ pud_t dummy = { 0 };
+
+ /*
+ * Given that pud_present()==1 and pud_leaf==0, page table walking code
+ * treats this like a page table and calls pmd_offset() /
+ * pmd_offset_lockless() with pudp, ignoring the returned value.
+ */
+ return dummy;
+}
+#define pudp_get pudp_get
+
static inline pmd_t * pmd_offset(pud_t * pud, unsigned long address)
{
return (pmd_t *)pud;
}
#define pmd_offset pmd_offset
+static inline pmd_t * pmd_offset_lockless(pud_t *pud, puf_t pud, unsigned long address)
+{
+ return (pmd_t *)pud;
+}
+#define pmd_offset_lockless pmd_offset_lockless
+
#define pmd_val(x) (pud_val((x).pud))
#define __pmd(x) ((pmd_t) { __pud(x) } )
diff --git a/include/asm-generic/pgtable-nopud.h b/include/asm-generic/pgtable-nopud.h
index eb70c6d7ceff..2cacd9571b2f 100644
--- a/include/asm-generic/pgtable-nopud.h
+++ b/include/asm-generic/pgtable-nopud.h
@@ -39,6 +39,14 @@ static inline void p4d_clear(p4d_t *p4d) { }
*/
#define set_p4d(p4dptr, p4dval) set_pud((pud_t *)(p4dptr), (pud_t) { p4dval })
+static inline p4d_t p4dp_get(p4d_t *p4dp)
+{
+ p4d_t dummy = { 0 };
+
+ return dummy;
+}
+#define p4dp_get p4dp_get
+
static inline pud_t *pud_offset(p4d_t *p4d, unsigned long address)
{
return (pud_t *)p4d;
--
2.43.0
--
Cheers,
David
prev parent reply other threads:[~2026-04-27 9:24 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-04-27 5:20 [PATCH] mm/page_vma_mapped_walk: add missing pgtable entry accessors Alexander Gordeev
2026-04-27 5:47 ` Anshuman Khandual
2026-04-27 8:40 ` Oscar Salvador
2026-04-27 8:49 ` Anshuman Khandual
2026-04-27 9:01 ` Alexander Gordeev
2026-04-27 9:02 ` David Hildenbrand (Arm)
2026-04-27 9:24 ` David Hildenbrand (Arm) [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=08ecabe9-0664-4aea-82fb-f9cb1739f762@kernel.org \
--to=david@kernel.org \
--cc=Ryan.Roberts@arm.com \
--cc=agordeev@linux.ibm.com \
--cc=akpm@linux-foundation.org \
--cc=chleroy@kernel.org \
--cc=gerald.schaefer@linux.ibm.com \
--cc=gor@linux.ibm.com \
--cc=hca@linux.ibm.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=linux-s390@vger.kernel.org \
--cc=ljs@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox