Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20201010151235.20585-1-nsaenzjulienne@suse.de>
 <20201010151235.20585-2-nsaenzjulienne@suse.de> <20201012113748.GE9844@gaia>
In-Reply-To: <20201012113748.GE9844@gaia>
From:   Ard Biesheuvel <ardb@kernel.org>
Date:   Mon, 12 Oct 2020 16:23:47 +0200
Message-ID: <CAMj1kXFUH7gv-dPVNM8xa5SBHhSndCQu4QFy5os2vAkaO_yecg@mail.gmail.com>
Subject: Re: [PATCH v2 1/5] arm64: mm: Move zone_dma_bits initialization into zone_sizes_init()
To:     Catalin Marinas <catalin.marinas@arm.com>
Cc:     Nicolas Saenz Julienne <nsaenzjulienne@suse.de>,
        Rob Herring <robh+dt@kernel.org>,
        Christoph Hellwig <hch@lst.de>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        Robin Murphy <robin.murphy@arm.com>,
        Linux ARM <linux-arm-kernel@lists.infradead.org>,
        linux-rpi-kernel@lists.infradead.org,
        Jeremy Linton <jeremy.linton@arm.com>,
        iommu@lists.linux-foundation.org,
        "open list:OPEN FIRMWARE AND FLATTENED DEVICE TREE BINDINGS" 
        <devicetree@vger.kernel.org>, Will Deacon <will@kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Mon, 12 Oct 2020 at 13:37, Catalin Marinas <catalin.marinas@arm.com> wrote:
>
> On Sat, Oct 10, 2020 at 05:12:31PM +0200, Nicolas Saenz Julienne wrote:
> > diff --git a/arch/arm64/mm/init.c b/arch/arm64/mm/init.c
> > index f6902a2b4ea6..0eca5865dcb1 100644
> > --- a/arch/arm64/mm/init.c
> > +++ b/arch/arm64/mm/init.c
> > @@ -196,14 +196,16 @@ static void __init zone_sizes_init(unsigned long min, unsigned long max)
> >       unsigned long max_zone_pfns[MAX_NR_ZONES]  = {0};
> >
> >  #ifdef CONFIG_ZONE_DMA
> > +     zone_dma_bits = ARM64_ZONE_DMA_BITS;
> > +
> >       if (IS_ENABLED(CONFIG_ACPI)) {
> >               extern unsigned int acpi_iort_get_zone_dma_size(void);
> >
> >               zone_dma_bits = min(zone_dma_bits,
> >                                   acpi_iort_get_zone_dma_size());
> > -             arm64_dma_phys_limit = max_zone_phys(zone_dma_bits);
> >       }
> >
> > +     arm64_dma_phys_limit = max_zone_phys(zone_dma_bits);
> >       max_zone_pfns[ZONE_DMA] = PFN_DOWN(arm64_dma_phys_limit);
> >  #endif
> >  #ifdef CONFIG_ZONE_DMA32
> > @@ -394,11 +396,6 @@ void __init arm64_memblock_init(void)
> >
> >       early_init_fdt_scan_reserved_mem();
> >
> > -     if (IS_ENABLED(CONFIG_ZONE_DMA)) {
> > -             zone_dma_bits = ARM64_ZONE_DMA_BITS;
> > -             arm64_dma_phys_limit = max_zone_phys(ARM64_ZONE_DMA_BITS);
> > -     }
>
> arm64_dma_phys_limit is used by memblock_alloc_low() (via
> ARCH_LOW_ADDRESS_LIMIT). I think it's too late to leave its
> initialisation to zone_sizes_init().
>

The only generic caller of memblock_alloc_low() is swiotlb_init(),
which is called much later. So at that point, we definitely need
ARCH_LOW_ADDRESS_LIMIT to be set correctly, but that means doing it in
zone_sizes_init() is early enough.

So the only problematic reference seems to be crashkernel_reserve() afaict.