[PATCH v2 1/2] memblock: drop for_each_memblock_type() and open code its users

[PATCH v2 1/2] memblock: drop for_each_memblock_type() and open code its users

[Tarun Sahu]

for_each_memblock_type() always starts the iteration at index 0 and its
body is a trivial 'for' loop, so it hides very little. It also gets in
the way of iterating from an arbitrary index, which the next patch
needs.

Remove the macro and open code its three users: memblock_add_range(),
memblock_isolate_range() and memblock_dump(). While at it, move the
region pointer into the loop body scope.

No functional change.

Signed-off-by: Tarun Sahu <tarunsahu@google.com>
---
 mm/memblock.c | 14 ++++++--------
 1 file changed, 6 insertions(+), 8 deletions(-)

diff --git a/mm/memblock.c b/mm/memblock.c
index 9ce86349a29f..5aae75d34dec 100644
--- a/mm/memblock.c
+++ b/mm/memblock.c
@@ -155,11 +155,6 @@ struct memblock_type physmem = {
  */
 static __refdata struct memblock_type *memblock_memory = &memblock.memory;
 
-#define for_each_memblock_type(i, memblock_type, rgn)			\
-	for (i = 0, rgn = &memblock_type->regions[0];			\
-	     i < memblock_type->cnt;					\
-	     i++, rgn = &memblock_type->regions[i])
-
 #define memblock_dbg(fmt, ...)						\
 	do {								\
 		if (memblock_debug)					\
@@ -651,7 +646,8 @@ static int __init_memblock memblock_add_range(struct memblock_type *type,
 	base = obase;
 	nr_new = 0;
 
-	for_each_memblock_type(idx, type, rgn) {
+	for (idx = 0; idx < type->cnt; idx++) {
+		rgn = &type->regions[idx];
 		phys_addr_t rbase = rgn->base;
 		phys_addr_t rend = rbase + rgn->size;
 
@@ -827,7 +823,8 @@ static int __init_memblock memblock_isolate_range(struct memblock_type *type,
 		if (memblock_double_array(type, base, size) < 0)
 			return -ENOMEM;
 
-	for_each_memblock_type(idx, type, rgn) {
+	for (idx = 0; idx < type->cnt; idx++) {
+		rgn = &type->regions[idx];
 		phys_addr_t rbase = rgn->base;
 		phys_addr_t rend = rbase + rgn->size;
 
@@ -2198,7 +2195,8 @@ static void __init_memblock memblock_dump(struct memblock_type *type)
 
 	pr_info(" %s.cnt  = 0x%lx\n", type->name, type->cnt);
 
-	for_each_memblock_type(idx, type, rgn) {
+	for (idx = 0; idx < type->cnt; idx++) {
+		rgn = &type->regions[idx];
 		char nid_buf[32] = "";
 
 		base = rgn->base;
-- 
2.55.0.1082.g2b9226bbc0-goog

[PATCH v2 2/2] memblock: use binary search to locate candidate regions

[Tarun Sahu]

Use binary search (memblock_bsearch_start) in memblock_add_range() and
memblock_isolate_range() to locate candidate regions instead of linearly
scanning from index 0.

Under heavy memory fragmentation (such as KHO page preservation registering
hundreds of thousands of disjoint folios), scanning from index 0 on every
insertion and isolation results in O(N^2) complexity, causing boot-time
memory retrieval to take several minutes (~268s for 393k pages).

Using binary search reduces the worst-case complexity to O(N log N)
(and O(N) for sequential appends), cutting KHO memory retrieval time
from ~268s to ~50ms.

memblock_search() open codes the same binary search, so reimplement it on
top of the new helper.

Signed-off-by: Tarun Sahu <tarunsahu@google.com>
---
 mm/memblock.c | 53 +++++++++++++++++++++++++++++++++++----------------
 1 file changed, 37 insertions(+), 16 deletions(-)

diff --git a/mm/memblock.c b/mm/memblock.c
index 5aae75d34dec..00a8768de988 100644
--- a/mm/memblock.c
+++ b/mm/memblock.c
@@ -586,6 +586,33 @@ static void __init_memblock memblock_insert_region(struct memblock_type *type,
 	type->total_size += size;
 }
 
+/**
+ * memblock_bsearch_start - Find the first region index where rend > base
+ * @type: memblock type to search
+ * @base: base physical address of the candidate range
+ *
+ * Returns the first region index that could potentially overlap @base.
+ */
+static int __init_memblock memblock_bsearch_start(struct memblock_type *type,
+						  phys_addr_t base)
+{
+	int mid, low = 0;
+	int high = type->cnt;
+
+	if (type->cnt && base >= type->regions[type->cnt - 1].base +
+				 type->regions[type->cnt - 1].size)
+		return type->cnt;
+
+	while (low < high) {
+		mid = (low + high) / 2;
+		if (type->regions[mid].base + type->regions[mid].size <= base)
+			low = mid + 1;
+		else
+			high = mid;
+	}
+	return low;
+}
+
 /**
  * memblock_add_range - add new memblock region
  * @type: memblock type to add new region into
@@ -609,7 +636,7 @@ static int __init_memblock memblock_add_range(struct memblock_type *type,
 	bool insert = false;
 	phys_addr_t obase = base;
 	phys_addr_t end = base + memblock_cap_size(base, &size);
-	int idx, nr_new, start_rgn = -1, end_rgn;
+	int idx, start_idx, nr_new, start_rgn = -1, end_rgn;
 	struct memblock_region *rgn;
 
 	if (!size)
@@ -645,8 +672,9 @@ static int __init_memblock memblock_add_range(struct memblock_type *type,
 	 */
 	base = obase;
 	nr_new = 0;
+	start_idx = memblock_bsearch_start(type, base);
 
-	for (idx = 0; idx < type->cnt; idx++) {
+	for (idx = start_idx; idx < type->cnt; idx++) {
 		rgn = &type->regions[idx];
 		phys_addr_t rbase = rgn->base;
 		phys_addr_t rend = rbase + rgn->size;
@@ -810,7 +838,7 @@ static int __init_memblock memblock_isolate_range(struct memblock_type *type,
 					int *start_rgn, int *end_rgn)
 {
 	phys_addr_t end = base + memblock_cap_size(base, &size);
-	int idx;
+	int idx, start_idx;
 	struct memblock_region *rgn;
 
 	*start_rgn = *end_rgn = 0;
@@ -823,7 +851,9 @@ static int __init_memblock memblock_isolate_range(struct memblock_type *type,
 		if (memblock_double_array(type, base, size) < 0)
 			return -ENOMEM;
 
-	for (idx = 0; idx < type->cnt; idx++) {
+	start_idx = memblock_bsearch_start(type, base);
+
+	for (idx = start_idx; idx < type->cnt; idx++) {
 		rgn = &type->regions[idx];
 		phys_addr_t rbase = rgn->base;
 		phys_addr_t rend = rbase + rgn->size;
@@ -2064,19 +2094,10 @@ void __init memblock_mem_limit_remove_map(phys_addr_t limit)
 
 static int __init_memblock memblock_search(struct memblock_type *type, phys_addr_t addr)
 {
-	unsigned int left = 0, right = type->cnt;
+	int idx = memblock_bsearch_start(type, addr);
 
-	do {
-		unsigned int mid = (right + left) / 2;
-
-		if (addr < type->regions[mid].base)
-			right = mid;
-		else if (addr >= (type->regions[mid].base +
-				  type->regions[mid].size))
-			left = mid + 1;
-		else
-			return mid;
-	} while (left < right);
+	if (idx < type->cnt && addr >= type->regions[idx].base)
+		return idx;
 	return -1;
 }
 
-- 
2.55.0.1082.g2b9226bbc0-goog

Re: [PATCH v2 1/2] memblock: drop for_each_memblock_type() and open code its users

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] mm: commit message and implementation mismatch regarding variable scope
--

commit c0dca5a3d7239e89d9dccb549819f0c9f7139656
Author: Tarun Sahu <tarunsahu@google.com>

memblock: drop for_each_memblock_type() and open code its users
    
Remove the for_each_memblock_type() macro and open code its users to
allow iterating from an arbitrary index. The three users modified are
memblock_add_range(), memblock_isolate_range(), and memblock_dump().

> Remove the macro and open code its three users: memblock_add_range(),
> memblock_isolate_range() and memblock_dump(). While at it, move the
> region pointer into the loop body scope.

[Severity: Low]
This isn't a bug, but the commit message mentions moving the region pointer
into the loop body scope.

Looking at memblock_add_range(), memblock_isolate_range(), and
memblock_dump() in mm/memblock.c, the declaration for 
struct memblock_region *rgn remains at the top-level function scope rather
than being moved inside the newly open-coded for loops.

Did this patch miss moving the declaration of the rgn pointer into the
loop blocks as intended?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260918002239.4024231-1-tarunsahu@google.com?part=1

Re: [PATCH v2 2/2] memblock: use binary search to locate candidate regions

[sashiko-bot]

> Use binary search (memblock_bsearch_start) in memblock_add_range() and
> memblock_isolate_range() to locate candidate regions instead of linearly
> scanning from index 0.
> 
> Under heavy memory fragmentation (such as KHO page preservation registering
> hundreds of thousands of disjoint folios), scanning from index 0 on every
> insertion and isolation results in O(N^2) complexity, causing boot-time
> memory retrieval to take several minutes (~268s for 393k pages).
> 
> Using binary search reduces the worst-case complexity to O(N log N)
> (and O(N) for sequential appends), cutting KHO memory retrieval time
> from ~268s to ~50ms.
> 
> memblock_search() open codes the same binary search, so reimplement it on
> top of the new helper.
> 
> Signed-off-by: Tarun Sahu <tarunsahu@google.com>

Sashiko has reviewed this patch and found no issues. It looks great!

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260918002239.4024231-1-tarunsahu@google.com?part=2

Re: [PATCH v2 1/2] memblock: drop for_each_memblock_type() and open code its users

[Pratyush Yadav]

On Fri, Sep 18 2026, Tarun Sahu wrote:

> for_each_memblock_type() always starts the iteration at index 0 and its
> body is a trivial 'for' loop, so it hides very little. It also gets in
> the way of iterating from an arbitrary index, which the next patch
> needs.
>
> Remove the macro and open code its three users: memblock_add_range(),
> memblock_isolate_range() and memblock_dump(). While at it, move the
> region pointer into the loop body scope.
>
> No functional change.
>
> Signed-off-by: Tarun Sahu <tarunsahu@google.com>
> ---
>  mm/memblock.c | 14 ++++++--------
>  1 file changed, 6 insertions(+), 8 deletions(-)
>
> diff --git a/mm/memblock.c b/mm/memblock.c
> index 9ce86349a29f..5aae75d34dec 100644
> --- a/mm/memblock.c
> +++ b/mm/memblock.c
> @@ -155,11 +155,6 @@ struct memblock_type physmem = {
>   */
>  static __refdata struct memblock_type *memblock_memory = &memblock.memory;
>  
> -#define for_each_memblock_type(i, memblock_type, rgn)			\
> -	for (i = 0, rgn = &memblock_type->regions[0];			\
> -	     i < memblock_type->cnt;					\
> -	     i++, rgn = &memblock_type->regions[i])
> -
>  #define memblock_dbg(fmt, ...)						\
>  	do {								\
>  		if (memblock_debug)					\
> @@ -651,7 +646,8 @@ static int __init_memblock memblock_add_range(struct memblock_type *type,
>  	base = obase;
>  	nr_new = 0;
>  
> -	for_each_memblock_type(idx, type, rgn) {
> +	for (idx = 0; idx < type->cnt; idx++) {
> +		rgn = &type->regions[idx];

Nit: this is really odd to read. You assign rgn and then declare rbase
and rend. Doesn't checkpatch complain?

Regardless, I reckon it would be a lot nicer to read of you move the
declaration of rgn into the loop. Same for the others below.

LGTM otherwise, so after fixing this up, feel free to add:

Reviewed-by: Pratyush Yadav <pratyush@kernel.org>

>  		phys_addr_t rbase = rgn->base;
>  		phys_addr_t rend = rbase + rgn->size;
>  
> @@ -827,7 +823,8 @@ static int __init_memblock memblock_isolate_range(struct memblock_type *type,
>  		if (memblock_double_array(type, base, size) < 0)
>  			return -ENOMEM;
>  
> -	for_each_memblock_type(idx, type, rgn) {
> +	for (idx = 0; idx < type->cnt; idx++) {
> +		rgn = &type->regions[idx];
>  		phys_addr_t rbase = rgn->base;
>  		phys_addr_t rend = rbase + rgn->size;
>  
> @@ -2198,7 +2195,8 @@ static void __init_memblock memblock_dump(struct memblock_type *type)
>  
>  	pr_info(" %s.cnt  = 0x%lx\n", type->name, type->cnt);
>  
> -	for_each_memblock_type(idx, type, rgn) {
> +	for (idx = 0; idx < type->cnt; idx++) {
> +		rgn = &type->regions[idx];
>  		char nid_buf[32] = "";
>  
>  		base = rgn->base;

-- 
Regards,
Pratyush Yadav