Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20230615142016.419570-1-zhiguangni01@gmail.com> <20230615170011.GK52412@kernel.org>
In-Reply-To: <20230615170011.GK52412@kernel.org>
From:   Liam Ni <zhiguangni01@gmail.com>
Date:   Sat, 17 Jun 2023 18:47:11 +0800
Message-ID: <CACZJ9cVxqeDooeAMi8HBFmeo_85E+NuMYbW9PCvp7Rm-wunOGg@mail.gmail.com>
Subject: Re: [PATCH] x86,NUMA:Get the number of ram pages directly in numa_meminfo_cover_memory()
To:     Mike Rapoport <rppt@kernel.org>
Cc:     dave.hansen@linux.intel.com, luto@kernel.org, peterz@infradead.org,
        tglx@linutronix.de, mingo@redhat.com, bp@alien8.de, hpa@zytor.com,
        akpm@linux-foundation.org, x86@kernel.org,
        linux-kernel@vger.kernel.org, linux-mm@kvack.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

Hi,

On Fri, 16 Jun 2023 at 01:00, Mike Rapoport <rppt@kernel.org> wrote:
>
> Hi,
>
> On Fri, Jun 16, 2023 at 12:20:16AM +1000, Liam Ni wrote:
> > In a previous implementation,The ram page is calculated
> > by counting the number of holes,
> > however,the number of ram pages is calculated during hole calculation.
> > Therefore,we can directly calculate the amount of ram pages.
> >
> > Signed-off-by: Liam Ni <zhiguangni01@gmail.com>
> > ---
> >  arch/x86/mm/numa.c |  4 ++--
> >  include/linux/mm.h |  4 ++++
> >  mm/mm_init.c       | 33 +++++++++++++++++++++++++++++++++
> >  3 files changed, 39 insertions(+), 2 deletions(-)
> >
> > diff --git a/arch/x86/mm/numa.c b/arch/x86/mm/numa.c
> > index 2aadb2019b4f..8ea0e956e3d7 100644
> > --- a/arch/x86/mm/numa.c
> > +++ b/arch/x86/mm/numa.c
> > @@ -461,12 +461,12 @@ static bool __init numa_meminfo_cover_memory(const struct numa_meminfo *mi)
> >               u64 s = mi->blk[i].start >> PAGE_SHIFT;
> >               u64 e = mi->blk[i].end >> PAGE_SHIFT;
> >               numaram += e - s;
> > -             numaram -= __absent_pages_in_range(mi->blk[i].nid, s, e);
> > +             numaram += __available_pages_in_range(mi->blk[i].nid, s, e);
>
> This is wrong. You add number of pages in range core MM sees to the number
> of pages covered by the numa_meminfo.
>
> More generally, rather than traverse all the numa_meminfo's and for each of
> them traverse all the regions in memblock.memory it's enough to count
> memory in memblock that doesn't have the node assigned and compare it to
> memblock_phys_mem_size().

Logic like below?
static bool __init numa_meminfo_cover_memory(const struct numa_meminfo *mi)
{
   u64 error_value;
   error_value = pages_without_nid_in_range(0,max_pfn);
    if ((error_value ) >= (1 << (20 - PAGE_SHIFT))) {
        //print error information
        return false;
     }
     return true;
}

I can't figure out why compare it with memblock_phys_mem_size().
The number of pages in memblock that doesn't have the node
assigned,which also means that these pages are not in numa_info.
So these pages can represent the number of lose pages.

Thanks
Liam Ni