[Buildroot] [PATCH 0/2 for 2015.11.x] rpath: fix fallout of check-host-rpath (branch yem/host-gawk)

[Buildroot] [PATCH 0/2 for 2015.11.x] rpath: fix fallout of check-host-rpath (branch yem/host-gawk)

[Yann E. MORIN]

Hello All!

This series, intended for master *and* a 2015.11.x maitenance branch,
fixes two bugs related to the RPATH handling that we introduced in
2015.11.

The first is a trivial oversight in sanitising the host directory in the
check-host-rpath script.

The second is more complex. It simply disables optional features of our
host gawk when the user has gmp and mpfr or readline development files
installed on its host system.

See the second commit log, the topic is fully explained there, and is
not trivial to sumarise.


Regards,
Yann E. MORIN.


The following changes since commit 3f90e532ff48669873f682a4f0c3b6f0910cb2e9:

  Update for 2015.11 (2015-11-30 23:13:41 +0100)

are available in the git repository at:

  git://git.busybox.net/~ymorin/git/buildroot yem/host-gawk

for you to fetch changes up to 9cd7b6bdd9790d4105829b278c0e6acba9f103f3:

  package/gawk: do not use readeline ormpfr in host variant (2015-12-01 22:56:50 +0100)

----------------------------------------------------------------
Yann E. MORIN (2):
      support/check-host-rpath: remove trailing '/' in host dir
      package/gawk: do not use readeline ormpfr in host variant

 package/gawk/gawk.mk             | 4 +++-
 support/scripts/check-host-rpath | 2 +-
 2 files changed, 4 insertions(+), 2 deletions(-)

-- 
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 223 225 172 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'

[Buildroot] [PATCH 1/2 for 2015.11.x] support/check-host-rpath: remove trailing '/' in host dir

[Yann E. MORIN]

Despite the comment saying so, the trailing '/' in the host directory is
not removed. note however that it is properly removed from extracted
RPATH tags.

This is not visible when the host directory is our default $(O)/host
location, but breaks for user-supplied external host ditrectory, when
the user leaves a trailing slash in the path.

Fix that.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Peter Korsgaard <jacmet@uclibc.org>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

---
Peter, for the maintenance branch over 2015.11.
Sorry for the mess, it's a case I did not test...
---
 support/scripts/check-host-rpath | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/support/scripts/check-host-rpath b/support/scripts/check-host-rpath
index dc9e17f..48d69da 100755
--- a/support/scripts/check-host-rpath
+++ b/support/scripts/check-host-rpath
@@ -14,7 +14,7 @@ main() {
     local file ret
 
     # Remove duplicate and trailing '/' for proper match
-    hostdir="$( sed -r -e 's:/+:/:g;' <<<"${hostdir}" )"
+    hostdir="$( sed -r -e 's:/+:/:g; s:/$::;' <<<"${hostdir}" )"
 
     ret=0
     while read file; do
-- 
1.9.1

[Buildroot] [PATCH 2/2 for 2015.11.x] package/gawk: do not use readeline ormpfr in host variant

[Yann E. MORIN]

gawk has an optional dependency on mpfr (and thus gmp) and readline,
and will probe for them. If they are present, they are used; if they
are missing, that's not an error. mpfr (and gmp) is used for "BIGNUM"
support on gawk; readline is used by the gawk debugger.

However, mpfr (bringing gmp) are also host-packages in Buildroot, but in
the standard build order (i.e. a plain 'make'), they are built after
gawk. Ditto readline (from ncurses).

If the user has the development files for gmp and mpfr, then gawk is
linked to them. Ditto readline.

Now, further on in the build, we build gmp and mpfr (for gcci or guile),
so we install them in the host dir. Ditto readline (for gdb, ncurses
itself and a few other packages...)

But because we forcibly set an RPATH tag on all our host binaries, our
host gawk will now dynamically link with our versions, when it was in
fact built against the host ones.

This did not seem to cause any harm dso far, but is far from ideal.

Since we do not really need BIGNUM or the debugger in our host gawk, we
just forcibly disable them and configure gawk without readline or mpfr
(there's no switch for gmp, but it;s not a direct dependency, it comes
just with mpfr).

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Peter Korsgaard <jacmet@uclibc.org>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

---
Peter, that's also for the 2015.11.x maintenance branch.
---
 package/gawk/gawk.mk | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/package/gawk/gawk.mk b/package/gawk/gawk.mk
index 1820689..661ec1e 100644
--- a/package/gawk/gawk.mk
+++ b/package/gawk/gawk.mk
@@ -16,9 +16,11 @@ ifeq ($(BR2_PACKAGE_BUSYBOX),y)
 GAWK_DEPENDENCIES += busybox
 endif
 
-# We don't have a host-busybox
+# We don't have a host-busybox, and we don't want to use mpfr
 HOST_GAWK_DEPENDENCIES =
 
+HOST_GAWK_CONF_OPTS = --without-readline --without-mpfr
+
 define GAWK_CREATE_SYMLINK
 	ln -sf /usr/bin/gawk $(TARGET_DIR)/usr/bin/awk
 endef
-- 
1.9.1

[Buildroot] [PATCH 2/2 for 2015.11.x] package/gawk: do not use readeline ormpfr in host variant

[Thomas Petazzoni]

Dear Yann E. MORIN,

Two typos in the commit title.

On Tue,  1 Dec 2015 23:19:07 +0100, Yann E. MORIN wrote:
> gawk has an optional dependency on mpfr (and thus gmp) and readline,
> and will probe for them. If they are present, they are used; if they
> are missing, that's not an error. mpfr (and gmp) is used for "BIGNUM"
> support on gawk; readline is used by the gawk debugger.
> 
> However, mpfr (bringing gmp) are also host-packages in Buildroot, but in
> the standard build order (i.e. a plain 'make'), they are built after
> gawk. Ditto readline (from ncurses).
> 
> If the user has the development files for gmp and mpfr, then gawk is
> linked to them. Ditto readline.
> 
> Now, further on in the build, we build gmp and mpfr (for gcci or guile),
> so we install them in the host dir. Ditto readline (for gdb, ncurses
> itself and a few other packages...)
> 
> But because we forcibly set an RPATH tag on all our host binaries, our
> host gawk will now dynamically link with our versions, when it was in
> fact built against the host ones.
> 
> This did not seem to cause any harm dso far, but is far from ideal.

dso -> so

> 
> Since we do not really need BIGNUM or the debugger in our host gawk, we
> just forcibly disable them and configure gawk without readline or mpfr
> (there's no switch for gmp, but it;s not a direct dependency, it comes

it;s -> it's

> -# We don't have a host-busybox
> +# We don't have a host-busybox, and we don't want to use mpfr

Is the comment really important, especially when it talks about mpfr
but not about readline ?

>  HOST_GAWK_DEPENDENCIES =
>  
> +HOST_GAWK_CONF_OPTS = --without-readline --without-mpfr
> +

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com

[Buildroot] [PATCH 2/2 for 2015.11.x] package/gawk: do not use readeline ormpfr in host variant

[Yann E. MORIN]

Thomas, All,

On 2015-12-01 23:22 +0100, Thomas Petazzoni spake thusly:
> On Tue,  1 Dec 2015 23:19:07 +0100, Yann E. MORIN wrote:
> > gawk has an optional dependency on mpfr (and thus gmp) and readline,
> > and will probe for them. If they are present, they are used; if they
> > are missing, that's not an error. mpfr (and gmp) is used for "BIGNUM"
> > support on gawk; readline is used by the gawk debugger.
> > 
> > However, mpfr (bringing gmp) are also host-packages in Buildroot, but in
> > the standard build order (i.e. a plain 'make'), they are built after
> > gawk. Ditto readline (from ncurses).
> > 
> > If the user has the development files for gmp and mpfr, then gawk is
> > linked to them. Ditto readline.
> > 
> > Now, further on in the build, we build gmp and mpfr (for gcci or guile),
> > so we install them in the host dir. Ditto readline (for gdb, ncurses
> > itself and a few other packages...)
> > 
> > But because we forcibly set an RPATH tag on all our host binaries, our
> > host gawk will now dynamically link with our versions, when it was in
> > fact built against the host ones.
> > 
> > This did not seem to cause any harm dso far, but is far from ideal.
> 
> dso -> so
> 
> > 
> > Since we do not really need BIGNUM or the debugger in our host gawk, we
> > just forcibly disable them and configure gawk without readline or mpfr
> > (there's no switch for gmp, but it;s not a direct dependency, it comes
> 
> it;s -> it's
> 
> > -# We don't have a host-busybox
> > +# We don't have a host-busybox, and we don't want to use mpfr
> 
> Is the comment really important, especially when it talks about mpfr
> but not about readline ?

Or just append 'or readline' to the comment.

Thanks!

Regards,
Yann E. MORIN.

> >  HOST_GAWK_DEPENDENCIES =
> >  
> > +HOST_GAWK_CONF_OPTS = --without-readline --without-mpfr
> > +
> 
> Thomas
> -- 
> Thomas Petazzoni, CTO, Free Electrons
> Embedded Linux, Kernel and Android engineering
> http://free-electrons.com

-- 
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 223 225 172 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'

[Buildroot] [PATCH 2/2 for 2015.11.x] package/gawk: do not use readeline ormpfr in host variant

[Peter Korsgaard]

>>>>> "Yann" == Yann E MORIN <yann.morin.1998@free.fr> writes:

Hi,

>> > -# We don't have a host-busybox
 >> > +# We don't have a host-busybox, and we don't want to use mpfr
 >> 
 >> Is the comment really important, especially when it talks about mpfr
 >> but not about readline ?

 > Or just append 'or readline' to the comment.

But we _DO_ have a host-readline?

-- 
Bye, Peter Korsgaard

[Buildroot] [PATCH 2/2 for 2015.11.x] package/gawk: do not use readeline ormpfr in host variant

[Yann E. MORIN]

Peter, All,

On 2015-12-02 23:34 +0100, Peter Korsgaard spake thusly:
> >>>>> "Yann" == Yann E MORIN <yann.morin.1998@free.fr> writes:
> 
> Hi,
> 
> >> > -# We don't have a host-busybox
>  >> > +# We don't have a host-busybox, and we don't want to use mpfr
>  >> 
>  >> Is the comment really important, especially when it talks about mpfr
>  >> but not about readline ?
> 
>  > Or just append 'or readline' to the comment.
> 
> But we _DO_ have a host-readline?

Yes, but we do not need it for the our host-gawk (or glibc would have
been broken quite a long time ago). And since it is a host package, we
do not usually enable unneeded features.

Regards,
Yann E. MORIN.

-- 
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 223 225 172 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'

[Buildroot] [PATCH 1/2 for 2015.11.x] support/check-host-rpath: remove trailing '/' in host dir

[Peter Korsgaard]

>>>>> "Yann" == Yann E MORIN <yann.morin.1998@free.fr> writes:

 > Despite the comment saying so, the trailing '/' in the host directory is
 > not removed. note however that it is properly removed from extracted
 > RPATH tags.

 > This is not visible when the host directory is our default $(O)/host
 > location, but breaks for user-supplied external host ditrectory, when
 > the user leaves a trailing slash in the path.

 > Fix that.

 > Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
 > Cc: Peter Korsgaard <jacmet@uclibc.org>
 > Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

 > ---
 > Peter, for the maintenance branch over 2015.11.
 > Sorry for the mess, it's a case I did not test...

Committed, thanks.

I'll create a 2015.11.x branch and cherry pick it there as well.

-- 
Bye, Peter Korsgaard

[Buildroot] [PATCH 2/2 for 2015.11.x] package/gawk: do not use readeline ormpfr in host variant

[Peter Korsgaard]

>>>>> "Yann" == Yann E MORIN <yann.morin.1998@free.fr> writes:

 > gawk has an optional dependency on mpfr (and thus gmp) and readline,
 > and will probe for them. If they are present, they are used; if they
 > are missing, that's not an error. mpfr (and gmp) is used for "BIGNUM"
 > support on gawk; readline is used by the gawk debugger.

 > However, mpfr (bringing gmp) are also host-packages in Buildroot, but in
 > the standard build order (i.e. a plain 'make'), they are built after
 > gawk. Ditto readline (from ncurses).

 > If the user has the development files for gmp and mpfr, then gawk is
 > linked to them. Ditto readline.

 > Now, further on in the build, we build gmp and mpfr (for gcci or guile),
 > so we install them in the host dir. Ditto readline (for gdb, ncurses
 > itself and a few other packages...)

 > But because we forcibly set an RPATH tag on all our host binaries, our
 > host gawk will now dynamically link with our versions, when it was in
 > fact built against the host ones.

 > This did not seem to cause any harm dso far, but is far from ideal.

 > Since we do not really need BIGNUM or the debugger in our host gawk, we
 > just forcibly disable them and configure gawk without readline or mpfr
 > (there's no switch for gmp, but it;s not a direct dependency, it comes
 > just with mpfr).

 > Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
 > Cc: Peter Korsgaard <jacmet@uclibc.org>
 > Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

 > ---
 > Peter, that's also for the 2015.11.x maintenance branch.

Committed, thanks - And I'll also cherry pick this.

Not directly related to this, but we should probably also explictly
handle the readline/mpfr dependencies for the target gawk.

-- 
Bye, Peter Korsgaard