diff mbox

toolchain-external: add a specific check to avoid Angstrom toolchains

Message ID 1381572882-22592-1-git-send-email-thomas.petazzoni@free-electrons.com
State Superseded
Headers show

Commit Message

Thomas Petazzoni Oct. 12, 2013, 10:14 a.m. UTC
The Angstrom toolchains available at
http://www.angstrom-distribution.org/toolchains/ are not usable as
external toolchains in Buildroot, because they are not pure toolchains
with just the C library, but instead complete SDKs with many
cross-compiled libraries (Gtk, Qt, glib, neon, sqlite, X.org, and many
more, approximately 200 MB of libraries).

Buildroot cannot use such toolchains, and while this is documented in
our manual, some users still try to do this. Today, one such user came
on the IRC channel, reporting a build problem, which we started
investigating, only to realize after a long time that he was using an
Angstrom toolchain.

To avoid this problem in the future, we explicitly check if the
toolchain is from Angstrom by looking at the vendor part of the tuple
exposed by the toolchain: as soon as it is
<something>-angstrom-<something-else>, we reject the toolchain with an
explanation.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
---
 toolchain/helpers.mk                               | 18 ++++++++++++++++++
 toolchain/toolchain-external/toolchain-external.mk |  1 +
 2 files changed, 19 insertions(+)

Comments

Yann E. MORIN Oct. 12, 2013, 10:19 a.m. UTC | #1
Thomas, All,

On 2013-10-12 12:14 +0200, Thomas Petazzoni spake thusly:
> The Angstrom toolchains available at
> http://www.angstrom-distribution.org/toolchains/ are not usable as
> external toolchains in Buildroot, because they are not pure toolchains
> with just the C library, but instead complete SDKs with many
> cross-compiled libraries (Gtk, Qt, glib, neon, sqlite, X.org, and many
> more, approximately 200 MB of libraries).
> 
> Buildroot cannot use such toolchains, and while this is documented in
> our manual, some users still try to do this. Today, one such user came
> on the IRC channel, reporting a build problem, which we started
> investigating, only to realize after a long time that he was using an
> Angstrom toolchain.
> 
> To avoid this problem in the future, we explicitly check if the
> toolchain is from Angstrom by looking at the vendor part of the tuple
> exposed by the toolchain: as soon as it is
> <something>-angstrom-<something-else>, we reject the toolchain with an
> explanation.
> 
> Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>

> ---
>  toolchain/helpers.mk                               | 18 ++++++++++++++++++
>  toolchain/toolchain-external/toolchain-external.mk |  1 +
>  2 files changed, 19 insertions(+)
> 
> diff --git a/toolchain/helpers.mk b/toolchain/helpers.mk
> index 95120cd..d81e561 100644
> --- a/toolchain/helpers.mk
> +++ b/toolchain/helpers.mk
> @@ -324,3 +324,21 @@ check_cross_compiler_exists = \
>  		echo "Cannot execute cross-compiler '$${__CROSS_CC}'" ; \
>  		exit 1 ; \
>  	fi
> +
> +#
> +# Check for toolchains known to not work with Buildroot. To check for
> +# Angstrom toolchains, we're taking the "Target: " line in the output
> +# of gcc -v, and look at the vendor part of the tuple.
> +#
> +# $1: cross-gcc path
> +#
> +check_wrong_toolchain = \
> +	__CROSS_CC=$(strip $1) ; \
> +	vendor=`$${__CROSS_CC} -v 2>&1 | sed -n '/^Target/s/Target: [^-]*-\([^-]*\)-.*/\1/p'` ; \
> +	if test $${vendor} = "angstrom" ; then \
> +		echo "Angstrom toolchains are not pure toolchains: they contain" ; \
> +		echo "many other libraries than just the C library, which makes" ; \
> +		echo "them unsuitable as external toolchains for build systems" ; \
> +		echo "such as Buildroot" ; \
> +		exit 1 ; \
> +	fi
> diff --git a/toolchain/toolchain-external/toolchain-external.mk b/toolchain/toolchain-external/toolchain-external.mk
> index b5b1ce7..102bcfd 100644
> --- a/toolchain/toolchain-external/toolchain-external.mk
> +++ b/toolchain/toolchain-external/toolchain-external.mk
> @@ -371,6 +371,7 @@ endif
>  # type of C library and all C library features.
>  define TOOLCHAIN_EXTERNAL_CONFIGURE_CMDS
>  	$(Q)$(call check_cross_compiler_exists,$(TOOLCHAIN_EXTERNAL_CC))
> +	$(Q)$(call check_wrong_toolchain,$(TOOLCHAIN_EXTERNAL_CC))
>  	$(Q)LIBC_A_LOCATION=`readlink -f $$(LANG=C $(TOOLCHAIN_EXTERNAL_CC) -print-file-name=libc.a)` ; \
>  	SYSROOT_DIR=`echo $${LIBC_A_LOCATION} | sed -r -e 's:(usr/)?lib(32|64)?/(.*/)?libc\.a::'` ; \
>  	if test -z "$${SYSROOT_DIR}" ; then \
> -- 
> 1.8.1.2
> 
> _______________________________________________
> buildroot mailing list
> buildroot@busybox.net
> http://lists.busybox.net/mailman/listinfo/buildroot
Thomas De Schampheleire Oct. 13, 2013, 7:29 a.m. UTC | #2
Hi Thomas,

On Sat, Oct 12, 2013 at 12:14 PM, Thomas Petazzoni
<thomas.petazzoni@free-electrons.com> wrote:
> The Angstrom toolchains available at
> http://www.angstrom-distribution.org/toolchains/ are not usable as
> external toolchains in Buildroot, because they are not pure toolchains
> with just the C library, but instead complete SDKs with many
> cross-compiled libraries (Gtk, Qt, glib, neon, sqlite, X.org, and many
> more, approximately 200 MB of libraries).
>
> Buildroot cannot use such toolchains, and while this is documented in
> our manual, some users still try to do this. Today, one such user came
> on the IRC channel, reporting a build problem, which we started
> investigating, only to realize after a long time that he was using an
> Angstrom toolchain.
>
> To avoid this problem in the future, we explicitly check if the
> toolchain is from Angstrom by looking at the vendor part of the tuple
> exposed by the toolchain: as soon as it is
> <something>-angstrom-<something-else>, we reject the toolchain with an
> explanation.
>
> Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
> ---
>  toolchain/helpers.mk                               | 18 ++++++++++++++++++
>  toolchain/toolchain-external/toolchain-external.mk |  1 +
>  2 files changed, 19 insertions(+)
>
> diff --git a/toolchain/helpers.mk b/toolchain/helpers.mk
> index 95120cd..d81e561 100644
> --- a/toolchain/helpers.mk
> +++ b/toolchain/helpers.mk
> @@ -324,3 +324,21 @@ check_cross_compiler_exists = \
>                 echo "Cannot execute cross-compiler '$${__CROSS_CC}'" ; \
>                 exit 1 ; \
>         fi
> +
> +#
> +# Check for toolchains known to not work with Buildroot. To check for

I think it is more common to say 'known not to work' (nit)

> +# Angstrom toolchains, we're taking the "Target: " line in the output
> +# of gcc -v, and look at the vendor part of the tuple.

Is there a reason why you're not using 'gcc -dumpmachine'?
This directly prints the tuple...

> +#
> +# $1: cross-gcc path
> +#
> +check_wrong_toolchain = \

This is just nitting about names, but I find 'wrong' odd here. What
about 'unusable'?

> +       __CROSS_CC=$(strip $1) ; \
> +       vendor=`$${__CROSS_CC} -v 2>&1 | sed -n '/^Target/s/Target: [^-]*-\([^-]*\)-.*/\1/p'` ; \
> +       if test $${vendor} = "angstrom" ; then \

I think that you should put $${vendor}} in quotes here. Suppose it
returns an empty string for one reason or the other, then you'll get a
syntax failure.

> +               echo "Angstrom toolchains are not pure toolchains: they contain" ; \
> +               echo "many other libraries than just the C library, which makes" ; \
> +               echo "them unsuitable as external toolchains for build systems" ; \
> +               echo "such as Buildroot" ; \
> +               exit 1 ; \
> +       fi
> diff --git a/toolchain/toolchain-external/toolchain-external.mk b/toolchain/toolchain-external/toolchain-external.mk
> index b5b1ce7..102bcfd 100644
> --- a/toolchain/toolchain-external/toolchain-external.mk
> +++ b/toolchain/toolchain-external/toolchain-external.mk
> @@ -371,6 +371,7 @@ endif
>  # type of C library and all C library features.
>  define TOOLCHAIN_EXTERNAL_CONFIGURE_CMDS
>         $(Q)$(call check_cross_compiler_exists,$(TOOLCHAIN_EXTERNAL_CC))
> +       $(Q)$(call check_wrong_toolchain,$(TOOLCHAIN_EXTERNAL_CC))
>         $(Q)LIBC_A_LOCATION=`readlink -f $$(LANG=C $(TOOLCHAIN_EXTERNAL_CC) -print-file-name=libc.a)` ; \
>         SYSROOT_DIR=`echo $${LIBC_A_LOCATION} | sed -r -e 's:(usr/)?lib(32|64)?/(.*/)?libc\.a::'` ; \
>         if test -z "$${SYSROOT_DIR}" ; then \


Best regards,
Thomas
Eric Benard Oct. 13, 2013, 10:48 a.m. UTC | #3
Hi Thomas,

Le Sat, 12 Oct 2013 12:14:42 +0200,
Thomas Petazzoni <thomas.petazzoni@free-electrons.com> a écrit :
> To avoid this problem in the future, we explicitly check if the
> toolchain is from Angstrom by looking at the vendor part of the tuple
> exposed by the toolchain: as soon as it is
> <something>-angstrom-<something-else>, we reject the toolchain with an
> explanation.
> 
maybe you should also check for at least "oe" and "poky" as these are
the vendor part of toolchains generated with plain OE-Core or
Yocto/Poky.

Eric
Thomas Petazzoni Oct. 13, 2013, 1:41 p.m. UTC | #4
Dear Eric Bénard,

On Sun, 13 Oct 2013 12:48:03 +0200, Eric Bénard wrote:

> Le Sat, 12 Oct 2013 12:14:42 +0200,
> Thomas Petazzoni <thomas.petazzoni@free-electrons.com> a écrit :
> > To avoid this problem in the future, we explicitly check if the
> > toolchain is from Angstrom by looking at the vendor part of the tuple
> > exposed by the toolchain: as soon as it is
> > <something>-angstrom-<something-else>, we reject the toolchain with an
> > explanation.
> > 
> maybe you should also check for at least "oe" and "poky" as these are
> the vendor part of toolchains generated with plain OE-Core or
> Yocto/Poky.

Hum, right. Are those available in the form of tarballs, like the
angstrom toolchain available at
http://www.angstrom-distribution.org/toolchains/ ? I know you can
generate such tarballs with OE or Yocto, but unless these tarballs are
available pre-built on the web, I guess it's pretty unlikely that such
toolchains get used as external toolchains in Buildroot.

Ideally, we would need to check if the toolchain is not a pure
toolchain, but it's quite hard to check. We would have to verify if the
toolchain does *not* have more than what it should have, but doing that
in an exhaustive way is quite difficult.

Thomas
Eric Benard Oct. 13, 2013, 1:50 p.m. UTC | #5
Le Sun, 13 Oct 2013 15:41:06 +0200,
Thomas Petazzoni <thomas.petazzoni@free-electrons.com> a écrit :

> Dear Eric Bénard,
> 
> On Sun, 13 Oct 2013 12:48:03 +0200, Eric Bénard wrote:
> 
> > Le Sat, 12 Oct 2013 12:14:42 +0200,
> > Thomas Petazzoni <thomas.petazzoni@free-electrons.com> a écrit :
> > > To avoid this problem in the future, we explicitly check if the
> > > toolchain is from Angstrom by looking at the vendor part of the tuple
> > > exposed by the toolchain: as soon as it is
> > > <something>-angstrom-<something-else>, we reject the toolchain with an
> > > explanation.
> > > 
> > maybe you should also check for at least "oe" and "poky" as these are
> > the vendor part of toolchains generated with plain OE-Core or
> > Yocto/Poky.
> 
> Hum, right. Are those available in the form of tarballs, like the
> angstrom toolchain available at
> http://www.angstrom-distribution.org/toolchains/ ? I know you can
> generate such tarballs with OE or Yocto, but unless these tarballs are
> available pre-built on the web, I guess it's pretty unlikely that such
> toolchains get used as external toolchains in Buildroot.
> 
http://downloads.yoctoproject.org/releases/yocto/yocto-1.4/toolchain/

Eric
diff mbox

Patch

diff --git a/toolchain/helpers.mk b/toolchain/helpers.mk
index 95120cd..d81e561 100644
--- a/toolchain/helpers.mk
+++ b/toolchain/helpers.mk
@@ -324,3 +324,21 @@  check_cross_compiler_exists = \
 		echo "Cannot execute cross-compiler '$${__CROSS_CC}'" ; \
 		exit 1 ; \
 	fi
+
+#
+# Check for toolchains known to not work with Buildroot. To check for
+# Angstrom toolchains, we're taking the "Target: " line in the output
+# of gcc -v, and look at the vendor part of the tuple.
+#
+# $1: cross-gcc path
+#
+check_wrong_toolchain = \
+	__CROSS_CC=$(strip $1) ; \
+	vendor=`$${__CROSS_CC} -v 2>&1 | sed -n '/^Target/s/Target: [^-]*-\([^-]*\)-.*/\1/p'` ; \
+	if test $${vendor} = "angstrom" ; then \
+		echo "Angstrom toolchains are not pure toolchains: they contain" ; \
+		echo "many other libraries than just the C library, which makes" ; \
+		echo "them unsuitable as external toolchains for build systems" ; \
+		echo "such as Buildroot" ; \
+		exit 1 ; \
+	fi
diff --git a/toolchain/toolchain-external/toolchain-external.mk b/toolchain/toolchain-external/toolchain-external.mk
index b5b1ce7..102bcfd 100644
--- a/toolchain/toolchain-external/toolchain-external.mk
+++ b/toolchain/toolchain-external/toolchain-external.mk
@@ -371,6 +371,7 @@  endif
 # type of C library and all C library features.
 define TOOLCHAIN_EXTERNAL_CONFIGURE_CMDS
 	$(Q)$(call check_cross_compiler_exists,$(TOOLCHAIN_EXTERNAL_CC))
+	$(Q)$(call check_wrong_toolchain,$(TOOLCHAIN_EXTERNAL_CC))
 	$(Q)LIBC_A_LOCATION=`readlink -f $$(LANG=C $(TOOLCHAIN_EXTERNAL_CC) -print-file-name=libc.a)` ; \
 	SYSROOT_DIR=`echo $${LIBC_A_LOCATION} | sed -r -e 's:(usr/)?lib(32|64)?/(.*/)?libc\.a::'` ; \
 	if test -z "$${SYSROOT_DIR}" ; then \