[PATCH] no-static-libs.inc: add gdb

[PATCH] no-static-libs.inc: add gdb

[Martin Jansa]

* gdb has own bundled version of readline and do_compile fails with:
  | config.status: creating config.h
  | make[2]: *** No rule to make target '../readline/libreadline.a', needed by 'gdb'.  Stop.
  | make[2]: *** Waiting for unfinished jobs....

Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
---
 meta/conf/distro/include/no-static-libs.inc | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/meta/conf/distro/include/no-static-libs.inc b/meta/conf/distro/include/no-static-libs.inc
index 6c0f859..0bb6647 100644
--- a/meta/conf/distro/include/no-static-libs.inc
+++ b/meta/conf/distro/include/no-static-libs.inc
@@ -13,6 +13,8 @@ DISABLE_STATIC_pn-libcap-native = ""
 DISABLE_STATIC_pn-libpcap = ""
 # needed by gdb
 DISABLE_STATIC_pn-readline = ""
+# builds it's own static readline
+DISABLE_STATIC_pn-gdb = ""
 # needed by pseudo
 DISABLE_STATIC_pn-sqlite3 = ""
 DISABLE_STATIC_pn-sqlite3-native = ""
-- 
2.7.2

Re: [PATCH] no-static-libs.inc: add gdb

[Richard Purdie]

On Mon, 2016-02-29 at 20:10 +0100, Martin Jansa wrote:
> * gdb has own bundled version of readline and do_compile fails with:
>   | config.status: creating config.h
>   | make[2]: *** No rule to make target '../readline/libreadline.a',
> needed by 'gdb'.  Stop.
>   | make[2]: *** Waiting for unfinished jobs....
> 
> Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>

I'm a bit puzzled as our autobuilders build this all the time and we
don't see failures. Any idea what the configuration difference might
be?

Cheers,

Richard

Re: [PATCH] no-static-libs.inc: add gdb

[Burton, Ross]

[-- Attachment #1: Type: text/plain, Size: 534 bytes --]

On 29 February 2016 at 20:40, Richard Purdie <
richard.purdie@linuxfoundation.org> wrote:

> I'm a bit puzzled as our autobuilders build this all the time and we
> don't see failures. Any idea what the configuration difference might
> be?
>

Since oe-core 94e1b917078bedf73830e54278af77f742c93581 gdb has a
packageconfig for system readline, which is enabled by default (and has
been enabled explicitly for a long time).  Do you have some local patches
adding more PACKAGECONFIG options but not enabling readline?

Ross

[-- Attachment #2: Type: text/html, Size: 1001 bytes --]

Re: [PATCH] no-static-libs.inc: add gdb

[Andre McCurdy]

On Mon, Feb 29, 2016 at 12:43 PM, Burton, Ross <ross.burton@intel.com> wrote:
>
> On 29 February 2016 at 20:40, Richard Purdie
> <richard.purdie@linuxfoundation.org> wrote:
>>
>> I'm a bit puzzled as our autobuilders build this all the time and we
>> don't see failures. Any idea what the configuration difference might
>> be?
>
> Since oe-core 94e1b917078bedf73830e54278af77f742c93581 gdb has a
> packageconfig for system readline, which is enabled by default (and has been
> enabled explicitly for a long time).  Do you have some local patches adding
> more PACKAGECONFIG options but not enabling readline?

gdb 7.10.1 depends on readline features which are not present in
readline 5.2 (ie the GPLv2 version). Using the gdb internal readline
is a workaround for people who don't want to switch to the GPLv3
version of readline.

(Although gdb is itself GPLv3, it's self contained and easy to exclude
from release builds).

> Ross
>
> --
> _______________________________________________
> Openembedded-core mailing list
> Openembedded-core@lists.openembedded.org
> http://lists.openembedded.org/mailman/listinfo/openembedded-core
>

Re: [PATCH] no-static-libs.inc: add gdb

[Martin Jansa]

[-- Attachment #1: Type: text/plain, Size: 1656 bytes --]

No special configuration IIRC (will check tomorrow and send whole
config.log).

But it failed the same in my minimal chroot env with just oe-core and on
Ubuntu jenkins slave doing bitbake world with many layers, here is full log:
http://errors.yoctoproject.org/Errors/Details/38149/

and it shows --without-system-readline parameter.

On Mon, Feb 29, 2016 at 9:50 PM, Andre McCurdy <armccurdy@gmail.com> wrote:

> On Mon, Feb 29, 2016 at 12:43 PM, Burton, Ross <ross.burton@intel.com>
> wrote:
> >
> > On 29 February 2016 at 20:40, Richard Purdie
> > <richard.purdie@linuxfoundation.org> wrote:
> >>
> >> I'm a bit puzzled as our autobuilders build this all the time and we
> >> don't see failures. Any idea what the configuration difference might
> >> be?
> >
> > Since oe-core 94e1b917078bedf73830e54278af77f742c93581 gdb has a
> > packageconfig for system readline, which is enabled by default (and has
> been
> > enabled explicitly for a long time).  Do you have some local patches
> adding
> > more PACKAGECONFIG options but not enabling readline?
>
> gdb 7.10.1 depends on readline features which are not present in
> readline 5.2 (ie the GPLv2 version). Using the gdb internal readline
> is a workaround for people who don't want to switch to the GPLv3
> version of readline.
>
> (Although gdb is itself GPLv3, it's self contained and easy to exclude
> from release builds).
>
> > Ross
> >
> > --
> > _______________________________________________
> > Openembedded-core mailing list
> > Openembedded-core@lists.openembedded.org
> > http://lists.openembedded.org/mailman/listinfo/openembedded-core
> >
>

[-- Attachment #2: Type: text/html, Size: 2767 bytes --]

Re: [PATCH] no-static-libs.inc: add gdb

[Martin Jansa]

[-- Attachment #1: Type: text/plain, Size: 2177 bytes --]

On Mon, Feb 29, 2016 at 10:30:17PM +0100, Martin Jansa wrote:
> No special configuration IIRC (will check tomorrow and send whole
> config.log).
> 
> But it failed the same in my minimal chroot env with just oe-core and on
> Ubuntu jenkins slave doing bitbake world with many layers, here is full log:
> http://errors.yoctoproject.org/Errors/Details/38149/
> 
> and it shows --without-system-readline parameter.

It's caused by bbappend in meta-qt5

OE qemux86@ ~/build/oe-core $ cat meta-qt5/recipes-devtools/gdb/gdb_%.bbappend
PACKAGECONFIG ??= "python"

But on the other hand does this change make more harm than confusing
error for people who try to build without readline in PACKAGECONFIG?

Regards,

> 
> On Mon, Feb 29, 2016 at 9:50 PM, Andre McCurdy <armccurdy@gmail.com> wrote:
> 
> > On Mon, Feb 29, 2016 at 12:43 PM, Burton, Ross <ross.burton@intel.com>
> > wrote:
> > >
> > > On 29 February 2016 at 20:40, Richard Purdie
> > > <richard.purdie@linuxfoundation.org> wrote:
> > >>
> > >> I'm a bit puzzled as our autobuilders build this all the time and we
> > >> don't see failures. Any idea what the configuration difference might
> > >> be?
> > >
> > > Since oe-core 94e1b917078bedf73830e54278af77f742c93581 gdb has a
> > > packageconfig for system readline, which is enabled by default (and has
> > been
> > > enabled explicitly for a long time).  Do you have some local patches
> > adding
> > > more PACKAGECONFIG options but not enabling readline?
> >
> > gdb 7.10.1 depends on readline features which are not present in
> > readline 5.2 (ie the GPLv2 version). Using the gdb internal readline
> > is a workaround for people who don't want to switch to the GPLv3
> > version of readline.
> >
> > (Although gdb is itself GPLv3, it's self contained and easy to exclude
> > from release builds).
> >
> > > Ross
> > >
> > > --
> > > _______________________________________________
> > > Openembedded-core mailing list
> > > Openembedded-core@lists.openembedded.org
> > > http://lists.openembedded.org/mailman/listinfo/openembedded-core
> > >
> >

-- 
Martin 'JaMa' Jansa     jabber: Martin.Jansa@gmail.com

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 188 bytes --]

Re: [PATCH] no-static-libs.inc: add gdb

[Richard Purdie]

On Tue, 2016-03-01 at 01:43 +0100, Martin Jansa wrote:
> On Mon, Feb 29, 2016 at 10:30:17PM +0100, Martin Jansa wrote:
> > No special configuration IIRC (will check tomorrow and send whole
> > config.log).
> > 
> > But it failed the same in my minimal chroot env with just oe-core
> > and on
> > Ubuntu jenkins slave doing bitbake world with many layers, here is
> > full log:
> > http://errors.yoctoproject.org/Errors/Details/38149/
> > 
> > and it shows --without-system-readline parameter.
> 
> It's caused by bbappend in meta-qt5
> 
> OE qemux86@ ~/build/oe-core $ cat meta-qt5/recipes
> -devtools/gdb/gdb_%.bbappend
> PACKAGECONFIG ??= "python"
> 
> But on the other hand does this change make more harm than confusing
> error for people who try to build without readline in PACKAGECONFIG?

I mainly wanted to understand what configuration caused this, which we
now understand/ I agree we should probably do something about the
error.

I do wonder what a gdb bbappend is doing in meta-qt5 though. I'd say
that is really "distro" configuration.

Cheers,

Richard

Re: [PATCH] no-static-libs.inc: add gdb

[Martin Jansa]

[-- Attachment #1: Type: text/plain, Size: 1475 bytes --]

On Tue, Mar 01, 2016 at 08:04:27AM +0000, Richard Purdie wrote:
> On Tue, 2016-03-01 at 01:43 +0100, Martin Jansa wrote:
> > On Mon, Feb 29, 2016 at 10:30:17PM +0100, Martin Jansa wrote:
> > > No special configuration IIRC (will check tomorrow and send whole
> > > config.log).
> > > 
> > > But it failed the same in my minimal chroot env with just oe-core
> > > and on
> > > Ubuntu jenkins slave doing bitbake world with many layers, here is
> > > full log:
> > > http://errors.yoctoproject.org/Errors/Details/38149/
> > > 
> > > and it shows --without-system-readline parameter.
> > 
> > It's caused by bbappend in meta-qt5
> > 
> > OE qemux86@ ~/build/oe-core $ cat meta-qt5/recipes
> > -devtools/gdb/gdb_%.bbappend
> > PACKAGECONFIG ??= "python"
> > 
> > But on the other hand does this change make more harm than confusing
> > error for people who try to build without readline in PACKAGECONFIG?
> 
> I mainly wanted to understand what configuration caused this, which we
> now understand/ I agree we should probably do something about the
> error.
> 
> I do wonder what a gdb bbappend is doing in meta-qt5 though. I'd say
> that is really "distro" configuration.

python support is needed for qt-creator:
http://patchwork.openembedded.org/patch/87523/

I've changed it to use _append to respect default value from oe-core:
http://patchwork.openembedded.org/patch/116889/

-- 
Martin 'JaMa' Jansa     jabber: Martin.Jansa@gmail.com

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 188 bytes --]

Re: [PATCH] no-static-libs.inc: add gdb

[Burton, Ross]

[-- Attachment #1: Type: text/plain, Size: 325 bytes --]

On 1 March 2016 at 00:43, Martin Jansa <martin.jansa@gmail.com> wrote:

> OE qemux86@ ~/build/oe-core $ cat
> meta-qt5/recipes-devtools/gdb/gdb_%.bbappend
> PACKAGECONFIG ??= "python"
>

Having fixed number of these myself it's definitely best practise to use
PACKAGECONFIG_append and _remove in a bbappend.

Ross

[-- Attachment #2: Type: text/html, Size: 746 bytes --]

Re: [PATCH] no-static-libs.inc: add gdb

[Martin Jansa]

[-- Attachment #1: Type: text/plain, Size: 654 bytes --]

On Tue, Mar 01, 2016 at 08:40:13AM +0000, Burton, Ross wrote:
> On 1 March 2016 at 00:43, Martin Jansa <martin.jansa@gmail.com> wrote:
> 
> > OE qemux86@ ~/build/oe-core $ cat
> > meta-qt5/recipes-devtools/gdb/gdb_%.bbappend
> > PACKAGECONFIG ??= "python"
> >
> 
> Having fixed number of these myself it's definitely best practise to use
> PACKAGECONFIG_append and _remove in a bbappend.

That's what I did, but on other hand it's harder for DISTRO to remove
python if meta-qt5 bbappend uses _append.

So I'm thinking about changing it to
PACKAGECONFIG ??= "readline python"

-- 
Martin 'JaMa' Jansa     jabber: Martin.Jansa@gmail.com

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 188 bytes --]

Re: [PATCH] no-static-libs.inc: add gdb

[Martin Jansa]

[-- Attachment #1: Type: text/plain, Size: 2043 bytes --]

On Mon, Feb 29, 2016 at 12:50:03PM -0800, Andre McCurdy wrote:
> On Mon, Feb 29, 2016 at 12:43 PM, Burton, Ross <ross.burton@intel.com> wrote:
> >
> > On 29 February 2016 at 20:40, Richard Purdie
> > <richard.purdie@linuxfoundation.org> wrote:
> >>
> >> I'm a bit puzzled as our autobuilders build this all the time and we
> >> don't see failures. Any idea what the configuration difference might
> >> be?
> >
> > Since oe-core 94e1b917078bedf73830e54278af77f742c93581 gdb has a
> > packageconfig for system readline, which is enabled by default (and has been
> > enabled explicitly for a long time).  Do you have some local patches adding
> > more PACKAGECONFIG options but not enabling readline?
> 
> gdb 7.10.1 depends on readline features which are not present in
> readline 5.2 (ie the GPLv2 version). Using the gdb internal readline
> is a workaround for people who don't want to switch to the GPLv3
> version of readline.

Thanks!

After fixing the default in meta-qt5 I've started to see this issue in
other builds which were indeed built with
PREFERRED_VERSION_readline ?= "5.2"
to prevent GPLv3 readline being used by other components.

For gdb which isn't part of our release builds we can use bundled
readline, so I had to apply this no-static-libs.inc change and
explicitly remove readline from gdb's PACKAGECONFIG.

So to recap, meta-qt5 change fixed my jenkins "bitbake world" builds by
enabling system readline.
For internal builds (without GPLv3 where possible) on the other hand I
need to remove readline from PACKAGECONFIG and apply this fix to build
static readline inside gdb's build.

> (Although gdb is itself GPLv3, it's self contained and easy to exclude
> from release builds).
> 
> > Ross
> >
> > --
> > _______________________________________________
> > Openembedded-core mailing list
> > Openembedded-core@lists.openembedded.org
> > http://lists.openembedded.org/mailman/listinfo/openembedded-core
> >

-- 
Martin 'JaMa' Jansa     jabber: Martin.Jansa@gmail.com

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 188 bytes --]

Re: [PATCH] no-static-libs.inc: add gdb

[Burton, Ross]

[-- Attachment #1: Type: text/plain, Size: 453 bytes --]

On 29 February 2016 at 20:50, Andre McCurdy <armccurdy@gmail.com> wrote:

> gdb 7.10.1 depends on readline features which are not present in
> readline 5.2 (ie the GPLv2 version). Using the gdb internal readline
> is a workaround for people who don't want to switch to the GPLv3
> version of readline.
>

Would it be possible to force a -static argument into the link of the
internal readline to ensure that a static library is built?

Ross

[-- Attachment #2: Type: text/html, Size: 863 bytes --]

Re: [PATCH] no-static-libs.inc: add gdb

[Burton, Ross]

[-- Attachment #1: Type: text/plain, Size: 300 bytes --]

On 3 March 2016 at 17:19, Burton, Ross <ross.burton@intel.com> wrote:

> Would it be possible to force a -static argument into the link of the
> internal readline to ensure that a static library is built?
>

Answer: yes.  And it's upstreamable as it's obviously right. Patch incoming.

Ross

[-- Attachment #2: Type: text/html, Size: 670 bytes --]