[Buildroot] [PATCH 1/2] package/pkg-hare: refactor build command

[Buildroot] [PATCH 1/2] package/pkg-hare: refactor build command

[Thomas Petazzoni via buildroot]

In order to avoid duplicating the build command, use an intermediate
"target" shell variable to store the make target to be invoked for the
build. This is in preparation to extending the build command.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
---
 package/pkg-hare.mk | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/package/pkg-hare.mk b/package/pkg-hare.mk
index 14e0cfce5a..15dc769311 100644
--- a/package/pkg-hare.mk
+++ b/package/pkg-hare.mk
@@ -49,10 +49,9 @@ ifndef $(2)_BUILD_CMDS
 # but in real world, the target "all" is not always the default one
 define $(2)_BUILD_CMDS
 	if grep "^all:" $$($(2)_SRCDIR)Makefile > /dev/null; then \
-		$(MAKE) -C $$($(2)_SRCDIR) HARE="$$(HARE_RUN_CMD)" HAREFLAGS="$$(HARE_ARCH_OPT) $$(HARE_LIBDIR_OPT)" all; \
-	else \
-		$(MAKE) -C $$($(2)_SRCDIR) HARE="$$(HARE_RUN_CMD)" HAREFLAGS="$$(HARE_ARCH_OPT) $$(HARE_LIBDIR_OPT)"; \
-	fi
+		target=all ; \
+	fi ; \
+	$(MAKE) -C $$($(2)_SRCDIR) HARE="$$(HARE_RUN_CMD)" HAREFLAGS="$$(HARE_ARCH_OPT) $$(HARE_LIBDIR_OPT)" $${target}
 endef
 endif
 
-- 
2.56.0

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

[Buildroot] [PATCH 2/2] package/pkg-hare: move cache directory outside of $(HOME)

[Thomas Petazzoni via buildroot]

By default, hare will store its cache directory in $(HOME)/.cache,
which is a problem as we don't want a Buildroot build to write outside
of its output directory (to the exception of DL_DIR, of course). The
build of hare-package packages currently fails with:

Error creating /home/br-user/.cache/hare: Permission denied
make[2]: *** [Makefile:15: dbus-introspector] Error 255
Error creating make[2]: *** Waiting for unfinished jobs....
/home/br-user/.cache/hare: Permission denied
make[2]: *** [Makefile:19: dbus-scanner] Error 255

due to this issue.

We fix this by passing XDG_CACHE_HOME when building hare-package
packages.

Fixes: 44d38432343e65ec7a6d898cf982985cc2bc1e88 ("package/pkg-hare.mk: new infrastructure")
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
---
 package/pkg-hare.mk | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/package/pkg-hare.mk b/package/pkg-hare.mk
index 15dc769311..6494d2651c 100644
--- a/package/pkg-hare.mk
+++ b/package/pkg-hare.mk
@@ -51,7 +51,11 @@ define $(2)_BUILD_CMDS
 	if grep "^all:" $$($(2)_SRCDIR)Makefile > /dev/null; then \
 		target=all ; \
 	fi ; \
-	$(MAKE) -C $$($(2)_SRCDIR) HARE="$$(HARE_RUN_CMD)" HAREFLAGS="$$(HARE_ARCH_OPT) $$(HARE_LIBDIR_OPT)" $${target}
+	$(MAKE) -C $$($(2)_SRCDIR) \
+		HARE="$$(HARE_RUN_CMD)" \
+		HAREFLAGS="$$(HARE_ARCH_OPT) $$(HARE_LIBDIR_OPT)" \
+		XDG_CACHE_HOME="$$($(2)_SRCDIR)/.cache" \
+		$${target}
 endef
 endif
 
-- 
2.56.0

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

Re: [Buildroot] [PATCH 2/2] package/pkg-hare: move cache directory outside of $(HOME)

[Fiona Klute via buildroot]

Am 20.10.26 um 14:06 schrieb Thomas Petazzoni via buildroot:
> By default, hare will store its cache directory in $(HOME)/.cache,
> which is a problem as we don't want a Buildroot build to write outside
> of its output directory (to the exception of DL_DIR, of course). The

And CCACHE_DIR, if enabled. Not really critical, but might give someone 
who's new to Buildroot a wrong idea. ;-)

> build of hare-package packages currently fails with:
> 
> Error creating /home/br-user/.cache/hare: Permission denied
> make[2]: *** [Makefile:15: dbus-introspector] Error 255
> Error creating make[2]: *** Waiting for unfinished jobs....
> /home/br-user/.cache/hare: Permission denied
> make[2]: *** [Makefile:19: dbus-scanner] Error 255
> 
> due to this issue.
> 
> We fix this by passing XDG_CACHE_HOME when building hare-package
> packages.

I'm not familiar with Hare, what does it store in its cache dir? The 
default path suggests it's intended as a global cache, so I'm wondering 
if something like $(BUILD_DIR)/.hare-cache might be better than a 
per-package cache. But of course that depends on what goes in there.

Best regards,
Fiona

> Fixes: 44d38432343e65ec7a6d898cf982985cc2bc1e88 ("package/pkg-hare.mk: new infrastructure")
> Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
> ---
>   package/pkg-hare.mk | 6 +++++-
>   1 file changed, 5 insertions(+), 1 deletion(-)
> 
> diff --git a/package/pkg-hare.mk b/package/pkg-hare.mk
> index 15dc769311..6494d2651c 100644
> --- a/package/pkg-hare.mk
> +++ b/package/pkg-hare.mk
> @@ -51,7 +51,11 @@ define $(2)_BUILD_CMDS
>   	if grep "^all:" $$($(2)_SRCDIR)Makefile > /dev/null; then \
>   		target=all ; \
>   	fi ; \
> -	$(MAKE) -C $$($(2)_SRCDIR) HARE="$$(HARE_RUN_CMD)" HAREFLAGS="$$(HARE_ARCH_OPT) $$(HARE_LIBDIR_OPT)" $${target}
> +	$(MAKE) -C $$($(2)_SRCDIR) \
> +		HARE="$$(HARE_RUN_CMD)" \
> +		HAREFLAGS="$$(HARE_ARCH_OPT) $$(HARE_LIBDIR_OPT)" \
> +		XDG_CACHE_HOME="$$($(2)_SRCDIR)/.cache" \
> +		$${target}
>   endef
>   endif
>   
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH 2/2] package/pkg-hare: move cache directory outside of $(HOME)

[Thomas Petazzoni via buildroot]

On Tue, Oct 20, 2026 at 02:55:04PM +0200, Fiona Klute wrote:
> Am 20.10.26 um 14:06 schrieb Thomas Petazzoni via buildroot:
> > By default, hare will store its cache directory in $(HOME)/.cache,
> > which is a problem as we don't want a Buildroot build to write outside
> > of its output directory (to the exception of DL_DIR, of course). The
> 
> And CCACHE_DIR, if enabled. Not really critical, but might give someone
> who's new to Buildroot a wrong idea. ;-)

True.

> I'm not familiar with Hare, what does it store in its cache dir? The default
> path suggests it's intended as a global cache, so I'm wondering if something
> like $(BUILD_DIR)/.hare-cache might be better than a per-package cache. But
> of course that depends on what goes in there.

I must say I'm not sure. On the package/hare patch, I asked the
feedback from François Perrad precisely for this reason. For what I
can see, it contains "object files" for each source file that has been
compiled, along with a bunch of other things.

See:

  https://paste.ack.tf/3839b9@raw

(from this I can see the path within
output/build/hare-dbus-e9a89f8bc5ce783fa25f0eb67d7d8932db90bcc2/.cache
seems a bit crazy though.

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 2/2] package/pkg-hare: move cache directory outside of $(HOME)

[Peter Korsgaard]

>>>>> "Thomas" == Thomas Petazzoni via buildroot <buildroot@buildroot.org> writes:

 > By default, hare will store its cache directory in $(HOME)/.cache,
 > which is a problem as we don't want a Buildroot build to write outside
 > of its output directory (to the exception of DL_DIR, of course). The
 > build of hare-package packages currently fails with:

 > Error creating /home/br-user/.cache/hare: Permission denied
 > make[2]: *** [Makefile:15: dbus-introspector] Error 255
 > Error creating make[2]: *** Waiting for unfinished jobs....
 > /home/br-user/.cache/hare: Permission denied
 > make[2]: *** [Makefile:19: dbus-scanner] Error 255

 > due to this issue.

 > We fix this by passing XDG_CACHE_HOME when building hare-package
 > packages.

Do we have reasons to think that this environment variable will only
ever be used by hare packages? If not, then we should perhaps rather do
it globally / in pkg-generic.mk, similar to E.G. how we set GIT_DIR=. ?

-- 
Bye, Peter Korsgaard
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH 2/2] package/pkg-hare: move cache directory outside of $(HOME)

[François PERRAD]

[-- Attachment #1.1: Type: text/plain, Size: 1403 bytes --]

Le mar. 20 oct. 2026 à 18:43, Peter Korsgaard <peter@korsgaard.com> a
écrit :

> >>>>> "Thomas" == Thomas Petazzoni via buildroot <buildroot@buildroot.org>
> writes:
>
>  > By default, hare will store its cache directory in $(HOME)/.cache,
>  > which is a problem as we don't want a Buildroot build to write outside
>  > of its output directory (to the exception of DL_DIR, of course). The
>  > build of hare-package packages currently fails with:
>
>  > Error creating /home/br-user/.cache/hare: Permission denied
>  > make[2]: *** [Makefile:15: dbus-introspector] Error 255
>  > Error creating make[2]: *** Waiting for unfinished jobs....
>  > /home/br-user/.cache/hare: Permission denied
>  > make[2]: *** [Makefile:19: dbus-scanner] Error 255
>
>  > due to this issue.
>
>  > We fix this by passing XDG_CACHE_HOME when building hare-package
>  > packages.
>
> Do we have reasons to think that this environment variable will only
> ever be used by hare packages? If not, then we should perhaps rather do
> it globally / in pkg-generic.mk, similar to E.G. how we set GIT_DIR=. ?
>
>
in fact, `hare` has its own variable HARECACHE, see
https://manpages.ubuntu.com/manpages/stonking/man1/hare-build.1.html#environment

HARECACHE="$$($(2)_SRCDIR).cache" works fine.

no needs to set a desktop XDG environment variable.

François


> --
> Bye, Peter Korsgaard
>

[-- Attachment #1.2: Type: text/html, Size: 2329 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 2/2] package/pkg-hare: move cache directory outside of $(HOME)

[Thomas Petazzoni via buildroot]

On Tue, Oct 20, 2026 at 06:43:49PM +0200, Peter Korsgaard wrote:

> Do we have reasons to think that this environment variable will only
> ever be used by hare packages? If not, then we should perhaps rather do
> it globally / in pkg-generic.mk, similar to E.G. how we set GIT_DIR=. ?

That is certainly an option. Initially I only saw the problem on
host-hare, but then I saw that the hare packages themselves can be
affected as well. The question is of course whether there will be an
impact on other packages, but if we decide that $HOME should be
read-only, then it makes sense to divert XDG_CACHE_HOME to
BUILD_DIR/.cache, and see whether it causes issues for some packages.

Should I respin a patch in that direction?

Independently from that I believe
https://patchwork.buildroot.org/project/buildroot/patch/20261020120646.1306735-1-thomas.petazzoni@bootlin.com/
can still be a good small refactoring to merge.

Thanks!

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 1/2] package/pkg-hare: refactor build command

[François PERRAD]

[-- Attachment #1.1: Type: text/plain, Size: 1413 bytes --]

Acked-by: Francois Perrad <francois.perrad.86@gmail.com>

Le mar. 20 oct. 2026 à 14:07, Thomas Petazzoni <thomas.petazzoni@bootlin.com>
a écrit :

> In order to avoid duplicating the build command, use an intermediate
> "target" shell variable to store the make target to be invoked for the
> build. This is in preparation to extending the build command.
>
> Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
> ---
>  package/pkg-hare.mk | 7 +++----
>  1 file changed, 3 insertions(+), 4 deletions(-)
>
> diff --git a/package/pkg-hare.mk b/package/pkg-hare.mk
> index 14e0cfce5a..15dc769311 100644
> --- a/package/pkg-hare.mk
> +++ b/package/pkg-hare.mk
> @@ -49,10 +49,9 @@ ifndef $(2)_BUILD_CMDS
>  # but in real world, the target "all" is not always the default one
>  define $(2)_BUILD_CMDS
>         if grep "^all:" $$($(2)_SRCDIR)Makefile > /dev/null; then \
> -               $(MAKE) -C $$($(2)_SRCDIR) HARE="$$(HARE_RUN_CMD)"
> HAREFLAGS="$$(HARE_ARCH_OPT) $$(HARE_LIBDIR_OPT)" all; \
> -       else \
> -               $(MAKE) -C $$($(2)_SRCDIR) HARE="$$(HARE_RUN_CMD)"
> HAREFLAGS="$$(HARE_ARCH_OPT) $$(HARE_LIBDIR_OPT)"; \
> -       fi
> +               target=all ; \
> +       fi ; \
> +       $(MAKE) -C $$($(2)_SRCDIR) HARE="$$(HARE_RUN_CMD)"
> HAREFLAGS="$$(HARE_ARCH_OPT) $$(HARE_LIBDIR_OPT)" $${target}
>  endef
>  endif
>
> --
> 2.56.0
>
>

[-- Attachment #1.2: Type: text/html, Size: 2481 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 2/2] package/pkg-hare: move cache directory outside of $(HOME)

[Peter Korsgaard]

>>>>> "Thomas" == Thomas Petazzoni <thomas.petazzoni@bootlin.com> writes:

 > On Tue, Oct 20, 2026 at 06:43:49PM +0200, Peter Korsgaard wrote:
 >> Do we have reasons to think that this environment variable will only
 >> ever be used by hare packages? If not, then we should perhaps rather do
 >> it globally / in pkg-generic.mk, similar to E.G. how we set GIT_DIR=. ?

 > That is certainly an option. Initially I only saw the problem on
 > host-hare, but then I saw that the hare packages themselves can be
 > affected as well. The question is of course whether there will be an
 > impact on other packages, but if we decide that $HOME should be
 > read-only, then it makes sense to divert XDG_CACHE_HOME to
 > BUILD_DIR/.cache, and see whether it causes issues for some packages.

 > Should I respin a patch in that direction?

I think that would be a good idea, yes.

 > Independently from that I believe
 > https://patchwork.buildroot.org/project/buildroot/patch/20261020120646.1306735-1-thomas.petazzoni@bootlin.com/
 > can still be a good small refactoring to merge.

Sure, I will apply it now.

-- 
Bye, Peter Korsgaard
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH 1/2] package/pkg-hare: refactor build command

[Peter Korsgaard]

>>>>> "Thomas" == Thomas Petazzoni via buildroot <buildroot@buildroot.org> writes:

 > In order to avoid duplicating the build command, use an intermediate
 > "target" shell variable to store the make target to be invoked for the
 > build. This is in preparation to extending the build command.

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

Committed, thanks.

-- 
Bye, Peter Korsgaard
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH 1/2] package/pkg-hare: refactor build command

[Peter Korsgaard]

>>>>> "Thomas" == Thomas Petazzoni via buildroot <buildroot@buildroot.org> writes:

 > In order to avoid duplicating the build command, use an intermediate
 > "target" shell variable to store the make target to be invoked for the
 > build. This is in preparation to extending the build command.

 > Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
 > ---
 >  package/pkg-hare.mk | 7 +++----
 >  1 file changed, 3 insertions(+), 4 deletions(-)

 > diff --git a/package/pkg-hare.mk b/package/pkg-hare.mk
 > index 14e0cfce5a..15dc769311 100644
 > --- a/package/pkg-hare.mk
 > +++ b/package/pkg-hare.mk
 > @@ -49,10 +49,9 @@ ifndef $(2)_BUILD_CMDS
 >  # but in real world, the target "all" is not always the default one
 >  define $(2)_BUILD_CMDS
 >  	if grep "^all:" $$($(2)_SRCDIR)Makefile > /dev/null; then \
 > -		$(MAKE) -C $$($(2)_SRCDIR) HARE="$$(HARE_RUN_CMD)" HAREFLAGS="$$(HARE_ARCH_OPT) $$(HARE_LIBDIR_OPT)" all; \
 > -	else \
 > -		$(MAKE) -C $$($(2)_SRCDIR) HARE="$$(HARE_RUN_CMD)" HAREFLAGS="$$(HARE_ARCH_OPT) $$(HARE_LIBDIR_OPT)"; \
 > -	fi
 > +		target=all ; \
 > +	fi ; \
 > +	$(MAKE) -C $$($(2)_SRCDIR) HARE="$$(HARE_RUN_CMD)" HAREFLAGS="$$(HARE_ARCH_OPT) $$(HARE_LIBDIR_OPT)" $${target}

I already pushed, but this presumably will break quite badly if someone
has a 'target' environment variable set. Maybe we should explicitly set
it to the empty string in the else clause?

-- 
Bye, Peter Korsgaard
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot