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;
Date:   Mon, 24 Jul 2023 08:49:27 +0100
From:   Conor Dooley <conor.dooley@microchip.com>
To:     Jisheng Zhang <jszhang@kernel.org>
CC:     Paul Walmsley <paul.walmsley@sifive.com>,
        Palmer Dabbelt <palmer@dabbelt.com>,
        Albert Ou <aou@eecs.berkeley.edu>,
        <linux-riscv@lists.infradead.org>, <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH v3 1/2] riscv: allow kmalloc() caches aligned to the
 smallest value
Message-ID: <20230724-next-shifter-3ec32114b34e@wendy>
References: <20230718152214.2907-1-jszhang@kernel.org>
 <20230718152214.2907-2-jszhang@kernel.org>
MIME-Version: 1.0
Content-Type: multipart/signed; micalg=pgp-sha256;
        protocol="application/pgp-signature"; boundary="lbLuWTJ0u5bINZ3+"
Content-Disposition: inline
In-Reply-To: <20230718152214.2907-2-jszhang@kernel.org>
Precedence: bulk

--lbLuWTJ0u5bINZ3+
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Tue, Jul 18, 2023 at 11:22:13PM +0800, Jisheng Zhang wrote:
> Currently, riscv defines ARCH_DMA_MINALIGN as L1_CACHE_BYTES, I.E
> 64Bytes, if CONFIG_RISCV_DMA_NONCOHERENT=3Dy. To support unified kernel
> Image, usually we have to enable CONFIG_RISCV_DMA_NONCOHERENT, thus
> it brings some bad effects to coherent platforms:
>=20
> Firstly, it wastes memory, kmalloc-96, kmalloc-32, kmalloc-16 and
> kmalloc-8 slab caches don't exist any more, they are replaced with
> either kmalloc-128 or kmalloc-64.
>=20
> Secondly, larger than necessary kmalloc aligned allocations results
> in unnecessary cache/TLB pressure.
>=20
> This issue also exists on arm64 platforms. From last year, Catalin
> tried to solve this issue by decoupling ARCH_KMALLOC_MINALIGN from
> ARCH_DMA_MINALIGN, limiting kmalloc() minimum alignment to
> dma_get_cache_alignment() and replacing ARCH_KMALLOC_MINALIGN usage
> in various drivers with ARCH_DMA_MINALIGN etc.[1]
>=20
> One fact we can make use of for riscv: if the CPU doesn't support
> ZICBOM or T-HEAD CMO, we know the platform is coherent. Based on
> Catalin's work and above fact, we can easily solve the kmalloc align
> issue for riscv: we can override dma_get_cache_alignment(), then let
> it return ARCH_DMA_MINALIGN at the beginning and return 1 once we know
> the underlying HW neither supports ZICBOM nor supports T-HEAD CMO.
>=20
> So what about if the CPU supports ZICBOM or T-HEAD CMO, but all the
> devices are dma coherent? Well, we use ARCH_DMA_MINALIGN as the
> kmalloc minimum alignment, nothing changed in this case. This case
> can be improved in the future.
>=20
> After this patch, a simple test of booting to a small buildroot rootfs
> on qemu shows:
>=20
> kmalloc-96           5041    5041     96  ...
> kmalloc-64           9606    9606     64  ...
> kmalloc-32           5128    5128     32  ...
> kmalloc-16           7682    7682     16  ...
> kmalloc-8           10246   10246      8  ...
>=20
> So we save about 1268KB memory. The saving will be much larger in normal
> OS env on real HW platforms.
>=20
> Link: https://lore.kernel.org/linux-arm-kernel/20230524171904.3967031-1-c=
atalin.marinas@arm.com/ [1]
>=20
> Signed-off-by: Jisheng Zhang <jszhang@kernel.org>

Reviewed-by: Conor Dooley <conor.dooley@microchip.com>

Thanks,
Conor.

--lbLuWTJ0u5bINZ3+
Content-Type: application/pgp-signature; name="signature.asc"

-----BEGIN PGP SIGNATURE-----

iHUEABYIAB0WIQRh246EGq/8RLhDjO14tDGHoIJi0gUCZL4tBwAKCRB4tDGHoIJi
0jIWAQDK3i9Q+yWDW+zT3HKMI/9vFe2vBFuaDRF2zhYd1DnQmAD/eQucZDuJ1yvE
Dnzs8qBMYSXxWuN7amY5rN5pjrjroAo=
=PbNw
-----END PGP SIGNATURE-----

--lbLuWTJ0u5bINZ3+--