From: Mike Rapoport <rppt@kernel.org>
To: Tarun Sahu <tarunsahu@google.com>
Cc: dmatlack@google.com, Pasha Tatashin <pasha.tatashin@soleen.com>,
Andrew Morton <akpm@linux-foundation.org>,
Pratyush Yadav <pratyush@kernel.org>,
kexec@lists.infradead.org, linux-mm@kvack.org,
linux-kernel@vger.kernel.org
Subject: Re: [PATCH v4 2/2] memblock: use binary search to locate candidate regions
Date: Fri, 9 Oct 2026 19:00:26 +0200 [thread overview]
Message-ID: <askdqgzm2Cbeo3SB@kernel.org> (raw)
In-Reply-To: <20261008190828.3221718-2-tarunsahu@google.com>
On Thu, Oct 08, 2026 at 07:08:28PM +0000, Tarun Sahu wrote:
> Use binary search (memblock_bsearch_start) in memblock_add_range() and
The function name is __memblock_search now and I think you can drop it
altogether :)
> memblock_isolate_range() to locate candidate regions instead of linearly
> scanning from index 0.
>
> Under heavy memory fragmentation (such as KHO page preservation registering
Comma instead of open parenthesis feels better to me.
> 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 candidate search complexity to O(log N)
> (from O(N)), 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>
> Reviewed-by: Pratyush Yadav <pratyush@kernel.org>
> ---
> mm/memblock.c | 45 +++++++++++++++++++++++++++++++--------------
> 1 file changed, 31 insertions(+), 14 deletions(-)
>
> diff --git a/mm/memblock.c b/mm/memblock.c
> index 59dda7d085f3..614d0a55dd70 100644
> --- a/mm/memblock.c
> +++ b/mm/memblock.c
> @@ -586,6 +586,29 @@ static void __init_memblock memblock_insert_region(struct memblock_type *type,
> type->total_size += size;
> }
>
> +/**
> + * __memblock_search - 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.
Return: for proper kernel-doc format AFAIR
> + */
> +static int __init_memblock __memblock_search(struct memblock_type *type,
> + phys_addr_t base)
> +{
> + int mid, low = 0;
> + int high = 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
> @@ -644,8 +667,9 @@ static int __init_memblock memblock_add_range(struct memblock_type *type,
> */
> base = obase;
> nr_new = 0;
> + idx = __memblock_search(type, base);
>
> - for (idx = 0; idx < type->cnt; idx++) {
> + for (; idx < type->cnt; idx++) {
for (idx = __memblock_search(type, base); ...
> struct memblock_region *rgn = &type->regions[idx];
> phys_addr_t rbase = rgn->base;
> phys_addr_t rend = rbase + rgn->size;
> @@ -821,7 +845,9 @@ 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++) {
> + idx = __memblock_search(type, base);
> +
> + for (; idx < type->cnt; idx++) {
Same here or even
for (int idx = __memblock_search(type, base); ...
> struct memblock_region *rgn = &type->regions[idx];
> phys_addr_t rbase = rgn->base;
> phys_addr_t rend = rbase + rgn->size;
> @@ -2062,19 +2088,10 @@ 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;
> -
> - do {
> - unsigned int mid = (right + left) / 2;
> + int idx = __memblock_search(type, addr);
>
> - 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;
> }
>
> --
> 2.56.0.385.gd3acb90ef8-goog
>
--
Sincerely yours,
Mike.
next prev parent reply other threads:[~2026-10-09 17:00 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-10-08 19:08 [PATCH v4 1/2] memblock: drop for_each_memblock_type() and open code its users Tarun Sahu
2026-10-08 19:08 ` [PATCH v4 2/2] memblock: use binary search to locate candidate regions Tarun Sahu
2026-10-08 19:14 ` sashiko-bot
2026-10-09 17:00 ` Mike Rapoport [this message]
2026-10-08 19:12 ` [PATCH v4 1/2] memblock: drop for_each_memblock_type() and open code its users sashiko-bot
2026-10-13 10:06 ` Mike Rapoport
2026-10-13 10:14 ` Mike Rapoport
2026-10-13 15:18 ` tarunsahu
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=askdqgzm2Cbeo3SB@kernel.org \
--to=rppt@kernel.org \
--cc=akpm@linux-foundation.org \
--cc=dmatlack@google.com \
--cc=kexec@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=pasha.tatashin@soleen.com \
--cc=pratyush@kernel.org \
--cc=tarunsahu@google.com \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.