The patch titled
Subject: mm/mm_init.c: not always search next deferred_init_pfn from very beginning
has been added to the -mm mm-unstable branch. Its filename is
mm-mm_initc-not-always-search-next-deferred_init_pfn-from-very-beginning.patch
This patch will shortly appear at
https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patches/mm-mm_initc-not-always-search-next-deferred_init_pfn-from-very-beginning.patch
This patch will later appear in the mm-unstable branch at
git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
Before you just go and hit "reply", please:
a) Consider who else should be cc'ed
b) Prefer to cc a suitable mailing list as well
c) Ideally: find the original patch on the mailing list and do a
reply-to-all to that, adding suitable additional cc's
*** Remember to use Documentation/process/submit-checklist.rst when testing your code ***
The -mm tree is included into linux-next via the mm-everything
branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
and is updated there every 2-3 working days
------------------------------------------------------
From: Wei Yang <richard.weiyang@xxxxxxxxx>
Subject: mm/mm_init.c: not always search next deferred_init_pfn from very beginning
Date: Fri, 31 May 2024 00:26:13 +0000
In deferred_init_memmap(), we call deferred_init_mem_pfn_range_in_zone()
to get the next deferred_init_pfn. But we always search it from the very
beginning.
Since we save the index in i, we can leverage this to search from i next
time.
Link: https://lkml.kernel.org/r/20240531002613.5231-3-richard.weiyang@xxxxxxxxx
Signed-off-by: Wei Yang <richard.weiyang@xxxxxxxxx>
Cc: Mike Rapoport (IBM) <rppt@xxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
include/linux/memblock.h | 19 -------------------
mm/mm_init.c | 21 ++++++++++++---------
2 files changed, 12 insertions(+), 28 deletions(-)
--- a/include/linux/memblock.h~mm-mm_initc-not-always-search-next-deferred_init_pfn-from-very-beginning
+++ a/include/linux/memblock.h
@@ -299,25 +299,6 @@ void __next_mem_pfn_range(int *idx, int
void __next_mem_pfn_range_in_zone(u64 *idx, struct zone *zone,
unsigned long *out_spfn,
unsigned long *out_epfn);
-/**
- * for_each_free_mem_pfn_range_in_zone - iterate through zone specific free
- * memblock areas
- * @i: u64 used as loop variable
- * @zone: zone in which all of the memory blocks reside
- * @p_start: ptr to phys_addr_t for start address of the range, can be %NULL
- * @p_end: ptr to phys_addr_t for end address of the range, can be %NULL
- *
- * Walks over free (memory && !reserved) areas of memblock in a specific
- * zone. Available once memblock and an empty zone is initialized. The main
- * assumption is that the zone start, end, and pgdat have been associated.
- * This way we can use the zone to determine NUMA node, and if a given part
- * of the memblock is valid for the zone.
- */
-#define for_each_free_mem_pfn_range_in_zone(i, zone, p_start, p_end) \
- for (i = 0, \
- __next_mem_pfn_range_in_zone(&i, zone, p_start, p_end); \
- i != U64_MAX; \
- __next_mem_pfn_range_in_zone(&i, zone, p_start, p_end))
/**
* for_each_free_mem_pfn_range_in_zone_from - iterate through zone specific
--- a/mm/mm_init.c~mm-mm_initc-not-always-search-next-deferred_init_pfn-from-very-beginning
+++ a/mm/mm_init.c
@@ -2025,18 +2025,21 @@ static unsigned long __init deferred_in
* return false indicating there are no valid ranges left.
*/
static bool __init
-deferred_init_mem_pfn_range_in_zone(u64 *i, struct zone *zone,
+deferred_init_mem_pfn_range_in_zone_from(u64 *i, struct zone *zone,
unsigned long *spfn, unsigned long *epfn,
unsigned long first_init_pfn)
{
- u64 j;
+ u64 j = *i;
+
+ if (j == 0)
+ __next_mem_pfn_range_in_zone(&j, zone, spfn, epfn);
/*
* Start out by walking through the ranges in this zone that have
* already been initialized. We don't need to do anything with them
* so we just need to flush them out of the system.
*/
- for_each_free_mem_pfn_range_in_zone(j, zone, spfn, epfn) {
+ for_each_free_mem_pfn_range_in_zone_from(j, zone, spfn, epfn) {
if (*epfn <= first_init_pfn)
continue;
if (*spfn < first_init_pfn)
@@ -2108,9 +2111,9 @@ deferred_init_memmap_chunk(unsigned long
{
unsigned long spfn, epfn;
struct zone *zone = arg;
- u64 i;
+ u64 i = 0;
- deferred_init_mem_pfn_range_in_zone(&i, zone, &spfn, &epfn, start_pfn);
+ deferred_init_mem_pfn_range_in_zone_from(&i, zone, &spfn, &epfn, start_pfn);
/*
* Initialize and free pages in MAX_PAGE_ORDER sized increments so that
@@ -2139,7 +2142,7 @@ static int __init deferred_init_memmap(v
unsigned long start = jiffies;
struct zone *zone;
int max_threads;
- u64 i;
+ u64 i = 0;
/* Bind memory initialisation thread to a local node if possible */
if (!cpumask_empty(cpumask))
@@ -2170,7 +2173,7 @@ static int __init deferred_init_memmap(v
max_threads = deferred_page_init_max_threads(cpumask);
- while (deferred_init_mem_pfn_range_in_zone(&i, zone, &spfn, &epfn, first_init_pfn)) {
+ while (deferred_init_mem_pfn_range_in_zone_from(&i, zone, &spfn, &epfn, first_init_pfn)) {
first_init_pfn = ALIGN(epfn, PAGES_PER_SECTION);
struct padata_mt_job job = {
.thread_fn = deferred_init_memmap_chunk,
@@ -2214,7 +2217,7 @@ bool __init deferred_grow_zone(struct zo
unsigned long first_deferred_pfn = pgdat->first_deferred_pfn;
unsigned long spfn, epfn, flags;
unsigned long nr_pages = 0;
- u64 i;
+ u64 i = 0;
/* Only the last zone may have deferred pages */
if (zone_end_pfn(zone) != pgdat_end_pfn(pgdat))
@@ -2232,7 +2235,7 @@ bool __init deferred_grow_zone(struct zo
}
/* If the zone is empty somebody else may have cleared out the zone */
- if (!deferred_init_mem_pfn_range_in_zone(&i, zone, &spfn, &epfn,
+ if (!deferred_init_mem_pfn_range_in_zone_from(&i, zone, &spfn, &epfn,
first_deferred_pfn)) {
pgdat->first_deferred_pfn = ULONG_MAX;
pgdat_resize_unlock(pgdat, &flags);
_
Patches currently in -mm which might be from richard.weiyang@xxxxxxxxx are
mm-mm_initc-get-the-highest-zone-directly.patch
mm-mm_initc-use-deferred_init_mem_pfn_range_in_zone-to-decide-loop-condition.patch
mm-mm_initc-not-always-search-next-deferred_init_pfn-from-very-beginning.patch
