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;
Date:   Fri, 03 Apr 2020 08:17:02 -0700 (PDT)
Subject:     Re: [RFC PATCH 1/7] riscv: Get rid of compile time logic with MAX_EARLY_MAPPING_SIZE
In-Reply-To: <20200322110028.18279-2-alex@ghiti.fr>
CC:     Paul Walmsley <paul.walmsley@sifive.com>, zong.li@sifive.com,
        anup@brainfault.org, Christoph Hellwig <hch@lst.de>,
        linux-riscv@lists.infradead.org, linux-kernel@vger.kernel.org,
        alex@ghiti.fr
From:   Palmer Dabbelt <palmer@dabbelt.com>
To:     alex@ghiti.fr
Message-ID: <mhng-ee39eb13-2567-479e-b0f4-5d23dff8496c@palmerdabbelt-glaptop1>
Mime-Version: 1.0 (MHng)
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 8bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Sun, 22 Mar 2020 04:00:22 PDT (-0700), alex@ghiti.fr wrote:
> There is no need to compare at compile time MAX_EARLY_MAPPING_SIZE value
> with PGDIR_SIZE since MAX_EARLY_MAPPING_SIZE is set to 128MB which is less
> than PGDIR_SIZE that is equal to 1GB: that allows to simplify early_pmd
> definition.
>
> Signed-off-by: Alexandre Ghiti <alex@ghiti.fr>
> ---
>  arch/riscv/mm/init.c | 16 ++++------------
>  1 file changed, 4 insertions(+), 12 deletions(-)
>
> diff --git a/arch/riscv/mm/init.c b/arch/riscv/mm/init.c
> index 238bd0033c3f..18bbb426848e 100644
> --- a/arch/riscv/mm/init.c
> +++ b/arch/riscv/mm/init.c
> @@ -247,13 +247,7 @@ static void __init create_pte_mapping(pte_t *ptep,
>
>  pmd_t trampoline_pmd[PTRS_PER_PMD] __page_aligned_bss;
>  pmd_t fixmap_pmd[PTRS_PER_PMD] __page_aligned_bss;
> -
> -#if MAX_EARLY_MAPPING_SIZE < PGDIR_SIZE
> -#define NUM_EARLY_PMDS		1UL
> -#else
> -#define NUM_EARLY_PMDS		(1UL + MAX_EARLY_MAPPING_SIZE / PGDIR_SIZE)
> -#endif
> -pmd_t early_pmd[PTRS_PER_PMD * NUM_EARLY_PMDS] __initdata __aligned(PAGE_SIZE);
> +pmd_t early_pmd[PTRS_PER_PMD] __initdata __aligned(PAGE_SIZE);
>
>  static pmd_t *__init get_pmd_virt(phys_addr_t pa)
>  {
> @@ -267,14 +261,12 @@ static pmd_t *__init get_pmd_virt(phys_addr_t pa)
>
>  static phys_addr_t __init alloc_pmd(uintptr_t va)
>  {
> -	uintptr_t pmd_num;
> -
>  	if (mmu_enabled)
>  		return memblock_phys_alloc(PAGE_SIZE, PAGE_SIZE);
>
> -	pmd_num = (va - PAGE_OFFSET) >> PGDIR_SHIFT;
> -	BUG_ON(pmd_num >= NUM_EARLY_PMDS);
> -	return (uintptr_t)&early_pmd[pmd_num * PTRS_PER_PMD];
> +	BUG_ON((va - PAGE_OFFSET) >> PGDIR_SHIFT);
> +
> +	return (uintptr_t)early_pmd;
>  }
>
>  static void __init create_pmd_mapping(pmd_t *pmdp,

My specific worry here was that allyesconfig kernels are quite large, and that
dropping the code to handle large kernels would make it even harder to boot
them.  That said, I can't actually get one to boot so I'm happy to just push
that off until later and drop the code we can't practically use.

Reviewed-by: Palmer Dabbelt <palmerdabbelt@google.com>

Thanks!