All of lore.kernel.org
 help / color / mirror / Atom feed
From: Trevor Woerner <twoerner@gmail.com>
To: Andrea Ricchi <andrea.ricchi@amarulasolutions.com>
Cc: yocto-patches@lists.yoctoproject.org
Subject: Re: [meta-rockchip][PATCH v2] kernel: restore options the 32-bit arch removal drops
Date: Wed, 18 Nov 2026 07:11:13 -0500	[thread overview]
Message-ID: <av2V4TqDeMg67bZV@localhost.localdomain> (raw)
In-Reply-To: <20261104130230.236614-1-andrea.ricchi@amarulasolutions.com>

On Wed 2026-11-04 @ 02:02:29 PM, Andrea Ricchi wrote:
> multi_v7_defconfig does not list several options the Rockchip boards
> rely on; it gets them through "select" from platforms that
> remove-non-rockchip-arch-arm.cfg switches off (ARCH_OMAP2PLUS,
> ARCH_VEXPRESS, ARCH_VIRT, ARCH_U8500, ...). Removing those platforms
> therefore also removes:
> 
>   REGULATOR_FIXED_VOLTAGE   parent supply of the PMIC regulators
>   VFP, VFPv3, NEON, KERNEL_MODE_NEON
>   ARM_PSCI                  HAVE_ARM_SMCCC_DISCOVERY, which
>                             ROCKCHIP_PM_DOMAINS depends on
>   ROCKCHIP_PM_DOMAINS
>   HIGHMEM
>   REALTEK_PHY
>   USB_EHCI_HCD_PLATFORM, USB_OHCI_HCD_PLATFORM
> 
> On tinker-board-s with linux-yocto 6.18 the kernel stops at "Waiting
> for root device": the RK808 regulators have no parent supply, so the
> SD host never gets vmmc. With the root mounted, every hard-float
> binary would fail and the VOPs would never probe for lack of power
> domains.
> 
> Add the options back in a fragment pulled in by the same scc, so the
> removal no longer takes them along. Found by diffing the expanded
> multi_v7_defconfig against the built .config.
> 
> Signed-off-by: Andrea Ricchi <andrea.ricchi@amarulasolutions.com>
> ---
> v2: add the layer tag to the subject, the patch itself is unchanged.
> 
>  .../rockchip/remove-non-rockchip-arch-arm.scc |  1 +
>  .../rockchip/restore-arm-selected-options.cfg | 26 +++++++++++++++++++
>  2 files changed, 27 insertions(+)
>  create mode 100644 recipes-kernel/linux/linux-rockchip/rockchip-kmeta/bsp/rockchip/restore-arm-selected-options.cfg

Added to meta-rockchip, master branch. Thanks!

The kernel config mechanism has a funny way of working.

ARCH_ROCKCHIP, and several others, do not select VFP, NEON, and
REGULATOR_FIXED_VOLTAGE, even though they rely on these settings.
ARCH_OMAP2PLUS_TYPICAL does, and turning off ARCH_OMAP2PLUS takes it,
and those settings, with it.

But there are more. The same removal also drops eight CPU errata
workarounds, six of which apply to cores this layer ships: five
Cortex-A9 ones for rk3066 and rk3188, and ARM_ERRATA_814220 for the
Cortex-A7 parts. I recently sent this second patch to the list for
review. Both are needed; the errata patch applies on top of this one,
and will be applied at the same time.

> diff --git a/recipes-kernel/linux/linux-rockchip/rockchip-kmeta/bsp/rockchip/remove-non-rockchip-arch-arm.scc b/recipes-kernel/linux/linux-rockchip/rockchip-kmeta/bsp/rockchip/remove-non-rockchip-arch-arm.scc
> index ce77704..da5c12e 100644
> --- a/recipes-kernel/linux/linux-rockchip/rockchip-kmeta/bsp/rockchip/remove-non-rockchip-arch-arm.scc
> +++ b/recipes-kernel/linux/linux-rockchip/rockchip-kmeta/bsp/rockchip/remove-non-rockchip-arch-arm.scc
> @@ -1 +1,2 @@
>  kconf hardware remove-non-rockchip-arch-arm.cfg
> +kconf hardware restore-arm-selected-options.cfg
> diff --git a/recipes-kernel/linux/linux-rockchip/rockchip-kmeta/bsp/rockchip/restore-arm-selected-options.cfg b/recipes-kernel/linux/linux-rockchip/rockchip-kmeta/bsp/rockchip/restore-arm-selected-options.cfg
> new file mode 100644
> index 0000000..ababaf6
> --- /dev/null
> +++ b/recipes-kernel/linux/linux-rockchip/rockchip-kmeta/bsp/rockchip/restore-arm-selected-options.cfg
> @@ -0,0 +1,26 @@
> +# multi_v7_defconfig does not list these options; it gets them through
> +# "select" from platforms that remove-non-rockchip-arch-arm.cfg switches
> +# off (ARCH_OMAP2PLUS, ARCH_VEXPRESS, ARCH_VIRT, ARCH_U8500, ...). Set them
> +# explicitly so removing those platforms does not remove them too.
> +
> +# regulator-fixed: parent supply of the PMIC regulators on most boards
> +CONFIG_REGULATOR_FIXED_VOLTAGE=y
> +
> +# hard-float userspace
> +CONFIG_VFP=y
> +CONFIG_VFPv3=y
> +CONFIG_NEON=y
> +CONFIG_KERNEL_MODE_NEON=y
> +
> +# ROCKCHIP_PM_DOMAINS depends on HAVE_ARM_SMCCC_DISCOVERY, provided by
> +# ARM_PSCI; without the power domains the VOPs, HDMI and IOMMUs never probe
> +CONFIG_ARM_PSCI=y
> +CONFIG_ROCKCHIP_PM_DOMAINS=y
> +
> +# boards with more than 768 MB
> +CONFIG_HIGHMEM=y
> +
> +# Realtek PHYs on most boards, generic-ehci/generic-ohci in rk3288.dtsi
> +CONFIG_REALTEK_PHY=y
> +CONFIG_USB_EHCI_HCD_PLATFORM=y
> +CONFIG_USB_OHCI_HCD_PLATFORM=y
> -- 
> 2.53.0
> 

      parent reply	other threads:[~2026-11-18 12:11 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-04 13:02 [meta-rockchip][PATCH v2] kernel: restore options the 32-bit arch removal drops Andrea Ricchi
2026-11-04 22:24 ` Trevor Woerner
2026-11-18 12:11 ` Trevor Woerner [this message]

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=av2V4TqDeMg67bZV@localhost.localdomain \
    --to=twoerner@gmail.com \
    --cc=andrea.ricchi@amarulasolutions.com \
    --cc=yocto-patches@lists.yoctoproject.org \
    /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.