From: Carlo Caione <ccaione@baylibre.com>
To: Steffen Dirkwinkel <lists@steffen.cc>
Cc: GSS_MTK_Uboot_upstream <GSS_MTK_Uboot_upstream@mediatek.com>,
u-boot@lists.u-boot-project.org, Tom Rini <trini@konsulko.com>,
Ilias Apalodimas <ilias.apalodimas@linaro.org>,
Ryder Lee <ryder.lee@mediatek.com>,
Weijie Gao <weijie.gao@mediatek.com>,
Chunfeng Yun <chunfeng.yun@mediatek.com>,
Igor Belwon <igor.belwon@mentallysanemainliners.org>,
David Lechner <dlechner@baylibre.com>,
Julien Stephan <jstephan@baylibre.com>,
Steffen Dirkwinkel <s.dirkwinkel@beckhoff.com>,
Chris Chen <chris-qj.chen@mediatek.com>,
Julien Masson <jmasson@baylibre.com>,
Johan Jonker <jbx6244@gmail.com>
Subject: Re: [PATCH v3 1/2] arm: mediatek: enable the caches in SPL
Date: Sat, 21 Nov 2026 17:29:21 +0100 [thread overview]
Message-ID: <awHEzAIcwMkHvMIt@macbook> (raw)
In-Reply-To: <20261120-std-upstream-mt8188-spl-caches-v3-1-b5ab70d56006@beckhoff.com>
On Fri, Nov 20, 2026 at 17:29:18 +0100, Steffen Dirkwinkel wrote:
> From: Steffen Dirkwinkel <s.dirkwinkel@beckhoff.com>
>
> Running through spl without the dcache is very slow, especially when
> hashing or checking signatures.
[...]
> --- a/arch/arm/include/asm/arch-mediatek/misc.h
> +++ b/arch/arm/include/asm/arch-mediatek/misc.h
> @@ -14,4 +14,6 @@
> #define APHW_VER 0x08
> #define APSW_VER 0x0c
>
> +void mtk_early_enable_caches(void);
nit: this should really be mtk_early_enable_dcache()
> +
> #endif /* __MEDIATEK_MISC_H_ */
> diff --git a/arch/arm/mach-mediatek/Kconfig b/arch/arm/mach-mediatek/Kconfig
> index cdbcb6b3bc18..9e36b2a2c82d 100644
> --- a/arch/arm/mach-mediatek/Kconfig
> +++ b/arch/arm/mach-mediatek/Kconfig
> @@ -221,6 +221,23 @@ config MTK_BROM_HEADER_INFO
> default "media=emmc" if TARGET_MT8516 || TARGET_MT8365 || TARGET_MT8366 || TARGET_MT8183 || TARGET_MT8189 || TARGET_MT8195
> default "lk=1" if TARGET_MT7623
>
> +config SPL_MTK_EARLY_CACHES
> + bool "Enable caches in SPL"
> + depends on SPL && ARM64 && !SPL_SYS_DCACHE_OFF
> + default y if TARGET_MT8188
> + help
> + Enable the MMU and caches in SPL and disable them before
> + jumping to the next stage.
nit: same here SPL_MTK_EARLY_DCACHE
> +
> +config MTK_EARLY_TLB_ADDR
> + hex "Address of the early page tables"
> + depends on SPL_MTK_EARLY_CACHES
> + default 0x401f8000 if TARGET_MT8188
> + help
> + Address of the 32 KiB of page tables used while the early
> + caches are on. Nothing else may use this memory during that
> + time.
> +
> config SPL_MTK_FIRMWARE_AB
> bool "MediaTek firmware A/B boot in SPL"
> depends on TARGET_MT8188 && SPL_MMC && SPL_LOAD_FIT && SPL_BLK
> diff --git a/arch/arm/mach-mediatek/Makefile b/arch/arm/mach-mediatek/Makefile
> index 4b1524e5e2f1..55bf8790cc92 100644
> --- a/arch/arm/mach-mediatek/Makefile
> +++ b/arch/arm/mach-mediatek/Makefile
> @@ -6,6 +6,7 @@ obj-$(CONFIG_MTK_TZ_MOVABLE) += tzcfg.o
> obj-$(CONFIG_XPL_BUILD) += spl.o
> obj-$(CONFIG_$(PHASE_)MTK_FIRMWARE_AB) += spl_ab.o mtk_ab.o
> obj-$(CONFIG_$(PHASE_)MTK_BROM_DOWNLOAD_AGENT_FASTBOOT) += download_agent.o
> +obj-$(CONFIG_$(PHASE_)MTK_EARLY_CACHES) += early_caches.o
>
> obj-$(CONFIG_TARGET_MT7622) += mt7622/
> obj-$(CONFIG_TARGET_MT7623) += mt7623/
> diff --git a/arch/arm/mach-mediatek/early_caches.c b/arch/arm/mach-mediatek/early_caches.c
> new file mode 100644
> index 000000000000..003393d2f681
> --- /dev/null
> +++ b/arch/arm/mach-mediatek/early_caches.c
> @@ -0,0 +1,21 @@
> +// SPDX-License-Identifier: GPL-2.0-only
> +/*
> + * MMU and caches before U-Boot sets up its final page tables
> + */
> +
> +#include <cpu_func.h>
> +#include <asm/global_data.h>
> +#include <asm/arch/misc.h>
> +#include <linux/sizes.h>
> +
> +DECLARE_GLOBAL_DATA_PTR;
> +
> +/* get_page_table_size() for the largest MediaTek mem_map */
> +#define EARLY_TLB_SIZE SZ_32K
nit: MTK_EARLY_TLB_SIZE for naming consistency
besides the nits:
Reviewed-by: Carlo Caione <ccaione@baylibre.com>
--
Carlo Caione
next prev parent reply other threads:[~2026-11-21 16:29 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-11-20 16:29 [PATCH v3 0/2] arm: mediatek: run SPL and pre-relocation U-Boot with caches Steffen Dirkwinkel
2026-11-20 16:29 ` [PATCH v3 1/2] arm: mediatek: enable the caches in SPL Steffen Dirkwinkel
2026-11-21 16:29 ` Carlo Caione [this message]
2026-11-20 16:29 ` [PATCH v3 2/2] arm: mediatek: run board_init_f() with the caches on Steffen Dirkwinkel
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=awHEzAIcwMkHvMIt@macbook \
--to=ccaione@baylibre.com \
--cc=GSS_MTK_Uboot_upstream@mediatek.com \
--cc=chris-qj.chen@mediatek.com \
--cc=chunfeng.yun@mediatek.com \
--cc=dlechner@baylibre.com \
--cc=igor.belwon@mentallysanemainliners.org \
--cc=ilias.apalodimas@linaro.org \
--cc=jbx6244@gmail.com \
--cc=jmasson@baylibre.com \
--cc=jstephan@baylibre.com \
--cc=lists@steffen.cc \
--cc=ryder.lee@mediatek.com \
--cc=s.dirkwinkel@beckhoff.com \
--cc=trini@konsulko.com \
--cc=u-boot@lists.u-boot-project.org \
--cc=weijie.gao@mediatek.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.