linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Oscar Salvador <osalvador@suse.de>
To: Mike Rapoport <rppt@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>,
	Alexandre Ghiti <alexghiti@rivosinc.com>,
	David Hildenbrand <david@redhat.com>,
	Pratyush Yadav <ptyadav@amazon.de>,
	linux-kernel@vger.kernel.org, linux-mm@kvack.org
Subject: Re: [PATCH 3/3] cma: move allocation from HIGHMEM to a helper function
Date: Thu, 3 Jul 2025 11:53:06 +0200	[thread overview]
Message-ID: <aGZTApK8WxFrTxI0@localhost.localdomain> (raw)
In-Reply-To: <20250702173605.2198924-4-rppt@kernel.org>

On Wed, Jul 02, 2025 at 08:36:05PM +0300, Mike Rapoport wrote:
> From: "Mike Rapoport (Microsoft)" <rppt@kernel.org>
> 
> When CONFIG_HIGMEM is enabled, __cma_declare_contiguous_nid() first
> tries to allocate the area from HIGHMEM and if that fails it falls back
> to allocation from low memory.
> 
> Split allocation from HIGMEM into a helper function to further decouple
> logic related to CONFIG_HIGHMEM.
> 
> Signed-off-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
> ---
>  mm/cma.c | 52 +++++++++++++++++++++++++++++-----------------------
>  1 file changed, 29 insertions(+), 23 deletions(-)
> 
> diff --git a/mm/cma.c b/mm/cma.c
> index 1df8ff312d99..0a24c46f3296 100644
> --- a/mm/cma.c
> +++ b/mm/cma.c
> @@ -376,6 +376,30 @@ static int __init cma_fixed_reserve(phys_addr_t base, phys_addr_t size)
>  	return 0;
>  }
>  
> +static phys_addr_t __init cma_alloc_highmem(phys_addr_t base, phys_addr_t size,
> +			phys_addr_t align, phys_addr_t *limit, int nid)
> +{
> +	phys_addr_t addr = 0;
> +
> +	if (IS_ENABLED(CONFIG_HIGHMEM)) {
> +		phys_addr_t highmem = __pa(high_memory - 1) + 1;
> +
> +		/*
> +		 * All pages in the reserved area must come from the same zone.
> +		 * If the requested region crosses the low/high memory boundary,
> +		 * try allocating from high memory first and fall back to low
> +		 * memory in case of failure.
> +		 */
> +		if (base < highmem && *limit > highmem) {
> +			addr = memblock_alloc_range_nid(size, align, highmem,
> +							*limit, nid, true);
> +			*limit = highmem;
> +		}
> +	}

Not a big deal, but maybe better to do it in one function? Maybe even move
the CONFIG_PHYS_ADDR_T_64BIT block in there as well? So memblock_alloc_range_nid()
calls would be contained in one place and the X86_64/HIGHMEM comments as
well.
Just a thought.

 diff --git a/mm/cma.c b/mm/cma.c
 index dd7643fc01db..532b56e6971a 100644
 --- a/mm/cma.c
 +++ b/mm/cma.c
 @@ -377,11 +377,12 @@ static int __init cma_fixed_reserve(phys_addr_t base, phys_addr_t size)
  	return 0;
  }
 
 -static phys_addr_t __init cma_alloc_highmem(phys_addr_t base, phys_addr_t size,
 -			phys_addr_t align, phys_addr_t *limit, int nid)
 +static phys_addr_t __init cma_alloc_mem(phys_addr_t base, phys_addr_t size,
 +			phys_addr_t align, phys_addr_t limit, int nid)
  {
  	phys_addr_t addr = 0;
 
 +	/* On systems with HIGHMEM try allocating from there first */
  	if (IS_ENABLED(CONFIG_HIGHMEM)) {
  		phys_addr_t highmem = __pa(high_memory - 1) + 1;
 
 @@ -393,11 +394,15 @@ static phys_addr_t __init cma_alloc_highmem(phys_addr_t base, phys_addr_t size,
  		 */
  		if (base < highmem && *limit > highmem) {
  			addr = memblock_alloc_range_nid(size, align, highmem,
 -							*limit, nid, true);
 +							limit, nid, true);
  			*limit = highmem;
  		}
  	}
 
 +	if (!addr)
 +		addr = memblock_alloc_range_nid(size, alignment, base,
 +						limit, nid, true);
 +
  	return addr;
  }
 
 @@ -487,16 +492,8 @@ static int __init __cma_declare_contiguous_nid(phys_addr_t *basep,
  		}
  #endif
 
 -		/* On systems with HIGHMEM try allocating from there first */
  		if (!addr)
 -			addr = cma_alloc_highmem(base, size, alignment, &limit,
 -						 nid);
 -		if (!addr) {
 -			addr = memblock_alloc_range_nid(size, alignment, base,
 -					limit, nid, true);
 -			if (!addr)
 -				return -ENOMEM;
 -		}
 +			addr = cma_alloc_mem(base, size, alignment, limit, nid);
 
  		/*
  		 * kmemleak scans/reads tracked objects for pointers to other



-- 
Oscar Salvador
SUSE Labs


  reply	other threads:[~2025-07-03  9:53 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-07-02 17:36 [PATCH 0/3] cma: factor out HIGMEM logic from __cma_declare_contiguous_nid Mike Rapoport
2025-07-02 17:36 ` [PATCH 1/3] cma: move __cma_declare_contiguous_nid() before its usage Mike Rapoport
2025-07-03  9:21   ` Oscar Salvador
2025-07-03 11:11   ` David Hildenbrand
2025-07-02 17:36 ` [PATCH 2/3] cma: split resrvation of fixed area into a helper function Mike Rapoport
2025-07-03  9:34   ` Oscar Salvador
2025-07-03 11:12   ` David Hildenbrand
2025-07-02 17:36 ` [PATCH 3/3] cma: move allocation from HIGHMEM to " Mike Rapoport
2025-07-03  9:53   ` Oscar Salvador [this message]
2025-07-03 11:14     ` David Hildenbrand
2025-07-03 17:27     ` Mike Rapoport

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=aGZTApK8WxFrTxI0@localhost.localdomain \
    --to=osalvador@suse.de \
    --cc=akpm@linux-foundation.org \
    --cc=alexghiti@rivosinc.com \
    --cc=david@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=ptyadav@amazon.de \
    --cc=rppt@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;
as well as URLs for NNTP newsgroup(s).