Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <20180726193509.3326-1-pasha.tatashin@oracle.com>
 <20180726193509.3326-3-pasha.tatashin@oracle.com> <20180727115645.GA13637@techadventures.net>
In-Reply-To: <20180727115645.GA13637@techadventures.net>
From:   Pavel Tatashin <pasha.tatashin@oracle.com>
Date:   Fri, 27 Jul 2018 10:53:24 -0400
Message-ID: <CAGM2reZnrwy1Y8MFRgyDLG8VZ6Hf+v-PAmZvUG4H65zunmjWZw@mail.gmail.com>
Subject: Re: [PATCH v2 2/3] mm: calculate deferred pages after skipping
 mirrored memory
To:     osalvador@techadventures.net
Cc:     Steven Sistare <steven.sistare@oracle.com>,
        Daniel Jordan <daniel.m.jordan@oracle.com>,
        LKML <linux-kernel@vger.kernel.org>,
        Andrew Morton <akpm@linux-foundation.org>,
        kirill.shutemov@linux.intel.com, Michal Hocko <mhocko@suse.com>,
        Linux Memory Management List <linux-mm@kvack.org>,
        dan.j.williams@intel.com, jack@suse.cz, jglisse@redhat.com,
        Souptick Joarder <jrdr.linux@gmail.com>, bhe@redhat.com,
        gregkh@linuxfoundation.org, Vlastimil Babka <vbabka@suse.cz>,
        Wei Yang <richard.weiyang@gmail.com>, dave.hansen@intel.com,
        rientjes@google.com, mingo@kernel.org, abdhalee@linux.vnet.ibm.com,
        mpe@ellerman.id.au
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

                         unsigned long *nr_initialised)
> > +static bool __meminit
> > +defer_init(int nid, unsigned long pfn, unsigned long end_pfn)
>
> Hi Pavel,
>
> maybe I do not understand properly the __init/__meminit macros, but should not
> "defer_init" be __init instead of __meminit?
> I think that functions marked as __meminit are not freed up, right?

Not exactly. As I understand: __meminit is the same as __init when
CONFIG_MEMORY_HOTPLUG=n. But, when memory hotplug is configured,
__meminit is not freed, because code that adds memory is shared
between boot and hotplug. In this case defer_init() is called only
during boot, and could be __init, but it is called from
memmap_init_zone() which is __meminit and thus section mismatch would
happen.

We could split memmap_init_zone() into two functions: boot and hotplug
variants, or we could use __ref, but I do not think any of that is
really needed. Keeping defer_init() in __meminit is OK, it does not
take that much memory.

>
> Reviewed-by: Oscar Salvador <osalvador@suse.de>

Thank you,
Pavel