[PATCH] u-boot: always use GCC to build

[PATCH] u-boot: always use GCC to build

[Ross Burton]

Previously, u-boot was almost always built with GCC because the
kernel-arch inherit set TOOLCHAIN to KERNEL_TOOLCHAIN, which defaulted
to 'gcc'.

However, since kernel-arch inherit was removed, u-boot in theory now
respects the default toolchain. In practise, the recipe hard-codes CC
to be gcc, and there are known caveats building u-boot with clang[2].

In the long term we should be able to make this recipe respect TOOLCHAIN,
as the upstream u-boot CI does build and test with clang, but for now
fix the build by restoring the explicit use of gcc.

[1] oe-core 8fb5470841b ("u-boot: Separate out from kernel-arch.bbclass")
[2] https://docs.u-boot-project.org/en/latest/build/clang.html

Signed-off-by: Ross Burton <ross.burton@arm.com>
---
 meta/recipes-bsp/u-boot/u-boot.inc | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/meta/recipes-bsp/u-boot/u-boot.inc b/meta/recipes-bsp/u-boot/u-boot.inc
index 3e2f5aa1a5f..00a12f1bb91 100644
--- a/meta/recipes-bsp/u-boot/u-boot.inc
+++ b/meta/recipes-bsp/u-boot/u-boot.inc
@@ -3,6 +3,10 @@ PROVIDES = "virtual/bootloader"
 
 PACKAGE_ARCH = "${MACHINE_ARCH}"
 
+# u-boot is best built with gcc
+# https://docs.u-boot-project.org/en/latest/build/clang.html
+TOOLCHAIN = "gcc"
+
 DEPENDS += "${@bb.utils.contains('UBOOT_ENV_SUFFIX', 'scr', 'u-boot-mkimage-native', '', d)}"
 DEPENDS += "${@ 'u-boot-mkenvimage-native' if bb.utils.to_boolean(d.getVar('UBOOT_INITIAL_ENV_BINARY')) else ''}"
 
-- 
2.43.0

Re: [OE-core] [PATCH] u-boot: always use GCC to build

[Khem Raj]

[-- Attachment #1: Type: text/plain, Size: 2360 bytes --]

On Tue, Aug 4, 2026 at 4:41 AM Ross Burton via lists.openembedded.org
<ross.burton=arm.com@lists.openembedded.org> wrote:

> Previously, u-boot was almost always built with GCC because the
> kernel-arch inherit set TOOLCHAIN to KERNEL_TOOLCHAIN, which defaulted
> to 'gcc'.
>
> However, since kernel-arch inherit was removed, u-boot in theory now
> respects the default toolchain. In practise, the recipe hard-codes CC
> to be gcc, and there are known caveats building u-boot with clang[2].
>

None of these caveats are relevant to u-boot and clang versions we use
isn't it?
so it seems to be workaround for some issue that is now brought forth which
was
latent before ? if so, why not fix that


>
> In the long term we should be able to make this recipe respect TOOLCHAIN,
> as the upstream u-boot CI does build and test with clang, but for now
> fix the build by restoring the explicit use of gcc.
>
> [1] oe-core 8fb5470841b ("u-boot: Separate out from kernel-arch.bbclass")
> [2] https://docs.u-boot-project.org/en/latest/build/clang.html
>
> Signed-off-by: Ross Burton <ross.burton@arm.com>
> ---
>  meta/recipes-bsp/u-boot/u-boot.inc | 4 ++++
>  1 file changed, 4 insertions(+)
>
> diff --git a/meta/recipes-bsp/u-boot/u-boot.inc
> b/meta/recipes-bsp/u-boot/u-boot.inc
> index 3e2f5aa1a5f..00a12f1bb91 100644
> --- a/meta/recipes-bsp/u-boot/u-boot.inc
> +++ b/meta/recipes-bsp/u-boot/u-boot.inc
> @@ -3,6 +3,10 @@ PROVIDES = "virtual/bootloader"
>
>  PACKAGE_ARCH = "${MACHINE_ARCH}"
>
> +# u-boot is best built with gcc
> +# https://docs.u-boot-project.org/en/latest/build/clang.html
> +TOOLCHAIN = "gcc"
> +
>  DEPENDS += "${@bb.utils.contains('UBOOT_ENV_SUFFIX', 'scr',
> 'u-boot-mkimage-native', '', d)}"
>  DEPENDS += "${@ 'u-boot-mkenvimage-native' if
> bb.utils.to_boolean(d.getVar('UBOOT_INITIAL_ENV_BINARY')) else ''}"
>
> --
> 2.43.0
>
>
> -=-=-=-=-=-=-=-=-=-=-=-
> Links: You receive all messages sent to this group.
> View/Reply Online (#242718):
> https://lists.openembedded.org/g/openembedded-core/message/242718
> Mute This Topic: https://lists.openembedded.org/mt/120591917/1997914
> Group Owner: openembedded-core+owner@lists.openembedded.org
> Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [
> raj.khem@gmail.com]
> -=-=-=-=-=-=-=-=-=-=-=-
>
>

[-- Attachment #2: Type: text/html, Size: 3873 bytes --]

Re: [OE-core] [PATCH] u-boot: always use GCC to build

[Ross Burton]

On 5 Aug 2026, at 02:08, Khem Raj <raj.khem@gmail.com> wrote:
> 
> None of these caveats are relevant to u-boot and clang versions we use isn't it?
> so it seems to be workaround for some issue that is now brought forth which was
> latent before ? if so, why not fix that

Almost definitely not, but the immediate fix of restoring the original behaviour (always builds with gcc, even when TOOLCHAIN=clang) was provably a quick fix that could be merged.  I filed a bug for myself to do the work so that the uboot recipe actually respects TOOLCHAIN.

Ross 

Re: [OE-core] [PATCH] u-boot: always use GCC to build

[Quentin Schulz]

Hi Ross,

On 8/4/26 1:41 PM, Ross Burton via lists.openembedded.org wrote:
> Previously, u-boot was almost always built with GCC because the
> kernel-arch inherit set TOOLCHAIN to KERNEL_TOOLCHAIN, which defaulted
> to 'gcc'.
> 
> However, since kernel-arch inherit was removed, u-boot in theory now
> respects the default toolchain. In practise, the recipe hard-codes CC
> to be gcc, and there are known caveats building u-boot with clang[2].
> 
> In the long term we should be able to make this recipe respect TOOLCHAIN,
> as the upstream u-boot CI does build and test with clang, but for now
> fix the build by restoring the explicit use of gcc.
> 

You don't say what is the issue we encounter when building this recipe 
with clang so this can be summed up to "fix issue" which isn't 
particularly helpful :) Could you consider providing a bit more info 
about build breakage in the commit log? I've built semi-recently (a few 
months ago) Rockchip (ARM and ARM64) devices with clang and couldn't 
build any of the ARM (thus 32b) SoCs at that time. I am traveling right 
now so cannot give you a link to the U-Boot ML archive but look for 
"Couple of clang warnings for Rockchip boards" from April 22nd, 2026.

Cheers,
Quentin

Re: [OE-core] [PATCH] u-boot: always use GCC to build

[Ross Burton]

On 6 Aug 2026, at 09:20, Quentin Schulz <quentin.schulz@cherry.de> wrote:
> You don't say what is the issue we encounter when building this recipe with clang so this can be summed up to "fix issue" which isn't particularly helpful :) 

Sorry.  At the time the message was sent the failure was “the makefile explicitly calls gcc, but clang is in the sysroot”.

Ross