[Buildroot] [PATCH 1/1] package/gcc-bare-metal: bump to version 15.2.0 to fix build

[Buildroot] [PATCH 1/1] package/gcc-bare-metal: bump to version 15.2.0 to fix build

[Julien Olivain via buildroot]

Since Buildroot commit [1] "package/gcc: bump 15.x series to 15.2.0",
gcc-bare-metal fails to download its sources. For example using the
commands:

	make zynqmp_kria_kv260_defconfig
	make source

fails with error:

	ERROR: No hash found for gcc-15.1.0.tar.xz

This is because the gcc-bare-metal package reuses the hash file and
the 15.x package patch directory from the gcc package (which was
bumped to 15.2.0), but still defines its _VERSION to 15.1.0.

This commit fixes this issue by bumping gcc-bare-metal to 15.2.0.

Fixes:
https://autobuild.buildroot.org/results/eda879fb8e1ccd696f01927087713b88a4b1a2c5
https://gitlab.com/buildroot.org/buildroot/-/jobs/11026775160

[1] https://gitlab.com/buildroot.org/buildroot/-/commit/b1dd235ad0dd91e1919a167a17cc9cd20a3695b2

Cc: Ibai Erkiaga <ibai.erkiaga-elorza@amd.com>
Cc: Neal Frager <neal.frager@amd.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
---
 package/gcc-bare-metal/15.1.0            | 1 -
 package/gcc-bare-metal/15.2.0            | 1 +
 package/gcc-bare-metal/gcc-bare-metal.mk | 2 +-
 3 files changed, 2 insertions(+), 2 deletions(-)
 delete mode 120000 package/gcc-bare-metal/15.1.0
 create mode 120000 package/gcc-bare-metal/15.2.0

diff --git a/package/gcc-bare-metal/15.1.0 b/package/gcc-bare-metal/15.1.0
deleted file mode 120000
index 669da9e956..0000000000
--- a/package/gcc-bare-metal/15.1.0
+++ /dev/null
@@ -1 +0,0 @@
-../gcc/15.1.0/
\ No newline at end of file
diff --git a/package/gcc-bare-metal/15.2.0 b/package/gcc-bare-metal/15.2.0
new file mode 120000
index 0000000000..d3d1bbdefa
--- /dev/null
+++ b/package/gcc-bare-metal/15.2.0
@@ -0,0 +1 @@
+../gcc/15.2.0
\ No newline at end of file
diff --git a/package/gcc-bare-metal/gcc-bare-metal.mk b/package/gcc-bare-metal/gcc-bare-metal.mk
index 17232a8ef1..88fc56ac09 100644
--- a/package/gcc-bare-metal/gcc-bare-metal.mk
+++ b/package/gcc-bare-metal/gcc-bare-metal.mk
@@ -4,7 +4,7 @@
 #
 ################################################################################
 
-GCC_BARE_METAL_VERSION = 15.1.0
+GCC_BARE_METAL_VERSION = 15.2.0
 GCC_BARE_METAL_SITE = $(BR2_GNU_MIRROR)/gcc/gcc-$(GCC_BARE_METAL_VERSION)
 GCC_BARE_METAL_SOURCE = gcc-$(GCC_BARE_METAL_VERSION).tar.xz
 
-- 
2.50.1

_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH 1/1] package/gcc-bare-metal: bump to version 15.2.0 to fix build

[Thomas Petazzoni via buildroot]

On Fri, 15 Aug 2025 17:27:37 +0200
Julien Olivain <ju.o@free.fr> wrote:

> Since Buildroot commit [1] "package/gcc: bump 15.x series to 15.2.0",
> gcc-bare-metal fails to download its sources. For example using the
> commands:
> 
> 	make zynqmp_kria_kv260_defconfig
> 	make source
> 
> fails with error:
> 
> 	ERROR: No hash found for gcc-15.1.0.tar.xz
> 
> This is because the gcc-bare-metal package reuses the hash file and
> the 15.x package patch directory from the gcc package (which was
> bumped to 15.2.0), but still defines its _VERSION to 15.1.0.
> 
> This commit fixes this issue by bumping gcc-bare-metal to 15.2.0.
> 
> Fixes:
> https://autobuild.buildroot.org/results/eda879fb8e1ccd696f01927087713b88a4b1a2c5
> https://gitlab.com/buildroot.org/buildroot/-/jobs/11026775160
> 
> [1] https://gitlab.com/buildroot.org/buildroot/-/commit/b1dd235ad0dd91e1919a167a17cc9cd20a3695b2
> 
> Cc: Ibai Erkiaga <ibai.erkiaga-elorza@amd.com>
> Cc: Neal Frager <neal.frager@amd.com>
> Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
> Signed-off-by: Julien Olivain <ju.o@free.fr>

Dammit, I always forget about gcc-bare-metal. Maybe we need to add a
comment in package/gcc/Config.in.host saying "if you change since, also
change gcc-bare-metal".

Otherwise:

Acked-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

Thomas
-- 
Thomas Petazzoni, co-owner and CEO, Bootlin
Embedded Linux and Kernel engineering and training
https://bootlin.com
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH 1/1] package/gcc-bare-metal: bump to version 15.2.0 to fix build

[Frager, Neal via buildroot]

[AMD Official Use Only - AMD Internal Distribution Only]

Hi Thomas,

> Since Buildroot commit [1] "package/gcc: bump 15.x series to 15.2.0",
> gcc-bare-metal fails to download its sources. For example using the
> commands:
>
>       make zynqmp_kria_kv260_defconfig
>       make source
>
> fails with error:
>
>       ERROR: No hash found for gcc-15.1.0.tar.xz
>
> This is because the gcc-bare-metal package reuses the hash file and
> the 15.x package patch directory from the gcc package (which was
> bumped to 15.2.0), but still defines its _VERSION to 15.1.0.
>
> This commit fixes this issue by bumping gcc-bare-metal to 15.2.0.
>
> Fixes:
> https://autobuild.buildroot.org/results/eda879fb8e1ccd696f01927087713b88a4b1a2c5
> https://gitlab.com/buildroot.org/buildroot/-/jobs/11026775160
>
> [1] https://gitlab.com/buildroot.org/buildroot/-/commit/b1dd235ad0dd91e1919a167a17cc9cd20a3695b2
>
> Cc: Ibai Erkiaga <ibai.erkiaga-elorza@amd.com>
> Cc: Neal Frager <neal.frager@amd.com>
> Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
> Signed-off-by: Julien Olivain <ju.o@free.fr>

> Dammit, I always forget about gcc-bare-metal. Maybe we need to add a
> comment in package/gcc/Config.in.host saying "if you change since, also
> change gcc-bare-metal".

Don't be so hard on yourself!  This type of error just shows you're human... :)

By the way, ever since gcc 15.2 and binutils 2.45 were released, I was watching
for Buildroot bump patches each day to make sure that gcc-bare-metal and
binutils-bare-metal get bumped as well.  You just happened to pick the day I
was leaving for a long August 15th weekend to bump gcc.

I do like your idea of adding a comment to the gcc package.  But one way to
fix the issue better would be to not immediately drop the old hash from the
hash file when you bump gcc.  This way, we can then bump gcc-bare-metal
later without build errors happening because of the missing hash.

Since the binutils hashes are not immediately dropped in the same way, we
do not have this issue with binutils-bare-metal bumping after binutils.

> Otherwise:

> Acked-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

@Julien, thanks for taking care of this!

Acked-by: Neal Frager <neal.frager@amd.com>

Best regards,
Neal Frager
AMD
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH 1/1] package/gcc-bare-metal: bump to version 15.2.0 to fix build

[Frager, Neal via buildroot]

[AMD Official Use Only - AMD Internal Distribution Only]

Hi Thomas,

--- SNIP ---

> I do like your idea of adding a comment to the gcc package.  But one way to
> fix the issue better would be to not immediately drop the old hash from the
> hash file when you bump gcc.  This way, we can then bump gcc-bare-metal
> later without build errors happening because of the missing hash.

I just thought of a better idea.  How about we create an actual hash file for
gcc-bare-metal instead of soft linking to gcc.hash?  This would make the
gcc-bare-metal package 'Thomas-forgetting-proof' for future gcc version bumps.

--- SNIP ---

Best regards,
Neal Frager
AMD
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH 1/1] package/gcc-bare-metal: bump to version 15.2.0 to fix build

[Thomas Petazzoni via buildroot]

On Fri, 15 Aug 2025 17:11:53 +0000
"Frager, Neal" <neal.frager@amd.com> wrote:

> I just thought of a better idea.  How about we create an actual hash file for
> gcc-bare-metal instead of soft linking to gcc.hash?  This would make the
> gcc-bare-metal package 'Thomas-forgetting-proof' for future gcc version bumps.

What about the patches symlink? Perhaps we can decide that the stack of
patches doesn't need to be shared?

Thomas
-- 
Thomas Petazzoni, co-owner and CEO, Bootlin
Embedded Linux and Kernel engineering and training
https://bootlin.com
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH 1/1] package/gcc-bare-metal: bump to version 15.2.0 to fix build

[Frager, Neal via buildroot]

[-- Attachment #1.1: Type: text/plain, Size: 751 bytes --]

[AMD Official Use Only - AMD Internal Distribution Only]

Hi Thomas,

> I just thought of a better idea.  How about we create an actual hash file for
> gcc-bare-metal instead of soft linking to gcc.hash?  This would make the
> gcc-bare-metal package 'Thomas-forgetting-proof' for future gcc version bumps.

> What about the patches symlink? Perhaps we can decide that the stack of
> patches doesn't need to be shared?

At the moment, gcc-bare-metal does not need any of the 3 gcc 15 patches.

We could just delete the patches symlink and cherry pick
any gcc patches we might actually need in the future.

Then we would truly disconnect any dependencies between the
gcc and gcc-bare-metal packages.

Best regards,
Neal Frager
AMD

[-- Attachment #1.2: Type: text/html, Size: 2569 bytes --]

[-- Attachment #2: Type: text/plain, Size: 150 bytes --]

_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH 1/1] package/gcc-bare-metal: bump to version 15.2.0 to fix build

[Thomas Petazzoni via buildroot]

On Fri, 15 Aug 2025 18:38:59 +0000
"Frager, Neal" <neal.frager@amd.com> wrote:

> At the moment, gcc-bare-metal does not need any of the 3 gcc 15 patches.
> 
> We could just delete the patches symlink and cherry pick
> any gcc patches we might actually need in the future.
> 
> Then we would truly disconnect any dependencies between the
> gcc and gcc-bare-metal packages.

That would be fine by me, especially as they don't _need_ to be
connected. In other cases, we really need to have the same version,
like package/glibc and package/localedef, so it makes sense to share
hash file and stuff. But between gcc-{final,initial} and
gcc-bare-metal, there's really no relationship whatsoever.

Thomas
-- 
Thomas Petazzoni, co-owner and CEO, Bootlin
Embedded Linux and Kernel engineering and training
https://bootlin.com
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH 1/1] package/gcc-bare-metal: bump to version 15.2.0 to fix build

[Frager, Neal via buildroot]

[-- Attachment #1.1: Type: text/plain, Size: 1211 bytes --]

[AMD Official Use Only - AMD Internal Distribution Only]

Hi Thomas,

> At the moment, gcc-bare-metal does not need any of the 3 gcc 15 patches.
>
> We could just delete the patches symlink and cherry pick
> any gcc patches we might actually need in the future.
>
> Then we would truly disconnect any dependencies between the
> gcc and gcc-bare-metal packages.

> That would be fine by me, especially as they don't _need_ to be
> connected. In other cases, we really need to have the same version,
> like package/glibc and package/localedef, so it makes sense to share
> hash file and stuff. But between gcc-{final,initial} and
> gcc-bare-metal, there's really no relationship whatsoever.

A good example of this are the zynqmp and versal defconfigs.

The aarch64 gcc version is 14.3 and the microblaze gcc-bare-metal
version is 15.1 (15.2 when Julien's patch is applied).

And having two gcc toolchains of completely different version
causes no problems whatsoever for zynqmp or versal.

The symlinks were only there for convenience. Why create a
second copy of files we already have?

Now we know the unintended consequence of that decision...

Best regards,
Neal Frager
AMD

[-- Attachment #1.2: Type: text/html, Size: 3214 bytes --]

[-- Attachment #2: Type: text/plain, Size: 150 bytes --]

_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH 1/1] package/gcc-bare-metal: bump to version 15.2.0 to fix build

[Thomas Petazzoni via buildroot]

On Fri, 15 Aug 2025 17:27:37 +0200
Julien Olivain via buildroot <buildroot@buildroot.org> wrote:

> Since Buildroot commit [1] "package/gcc: bump 15.x series to 15.2.0",
> gcc-bare-metal fails to download its sources. For example using the
> commands:
> 
> 	make zynqmp_kria_kv260_defconfig
> 	make source
> 
> fails with error:
> 
> 	ERROR: No hash found for gcc-15.1.0.tar.xz
> 
> This is because the gcc-bare-metal package reuses the hash file and
> the 15.x package patch directory from the gcc package (which was
> bumped to 15.2.0), but still defines its _VERSION to 15.1.0.
> 
> This commit fixes this issue by bumping gcc-bare-metal to 15.2.0.
> 
> Fixes:
> https://autobuild.buildroot.org/results/eda879fb8e1ccd696f01927087713b88a4b1a2c5
> https://gitlab.com/buildroot.org/buildroot/-/jobs/11026775160
> 
> [1] https://gitlab.com/buildroot.org/buildroot/-/commit/b1dd235ad0dd91e1919a167a17cc9cd20a3695b2
> 
> Cc: Ibai Erkiaga <ibai.erkiaga-elorza@amd.com>
> Cc: Neal Frager <neal.frager@amd.com>
> Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
> Signed-off-by: Julien Olivain <ju.o@free.fr>
> ---
>  package/gcc-bare-metal/15.1.0            | 1 -
>  package/gcc-bare-metal/15.2.0            | 1 +
>  package/gcc-bare-metal/gcc-bare-metal.mk | 2 +-
>  3 files changed, 2 insertions(+), 2 deletions(-)
>  delete mode 120000 package/gcc-bare-metal/15.1.0
>  create mode 120000 package/gcc-bare-metal/15.2.0

Applied to master, thanks.

Thomas
-- 
Thomas Petazzoni, CTO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot