diff mbox

[v2,2/4] libfslcodec: add missing configure options

Message ID 1449755560-2911-3-git-send-email-gary.bisson@boundarydevices.com
State Superseded
Headers show

Commit Message

Gary Bisson Dec. 10, 2015, 1:52 p.m. UTC
Enable VPU and hard float support when the proper configuration is setup.
BR2_ARM_EABIHF=y
BR2_PACKAGE_IMX_VPU=y

See Yocto equivalent:
https://github.com/Freescale/meta-fsl-arm/blob/fido/recipes-
multimedia/libfslcodec/libfslcodec.inc

Tested with the Linaro ARM 2014.09 toolchain.

Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
---

Thanks to those options, I could see that:
- lib_jpeg_dec_arm11_elinux_novpu.so isn't copied
- lib_peq_arm11_elinux.so.1.fhw is copied instead of the fsw version

Regards,
Gary

---
 package/libfslcodec/libfslcodec.mk | 8 ++++++++
 1 file changed, 8 insertions(+)

Comments

Yann E. MORIN Dec. 13, 2015, 8:32 p.m. UTC | #1
Gary, All,

On 2015-12-10 14:52 +0100, Gary Bisson spake thusly:
> Enable VPU and hard float support when the proper configuration is setup.
> BR2_ARM_EABIHF=y
> BR2_PACKAGE_IMX_VPU=y
> 
> See Yocto equivalent:
> https://github.com/Freescale/meta-fsl-arm/blob/fido/recipes-
> multimedia/libfslcodec/libfslcodec.inc
> 
> Tested with the Linaro ARM 2014.09 toolchain.
> 
> Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
> ---
> 
> Thanks to those options, I could see that:
> - lib_jpeg_dec_arm11_elinux_novpu.so isn't copied
> - lib_peq_arm11_elinux.so.1.fhw is copied instead of the fsw version
> 
> Regards,
> Gary
> 
> ---
>  package/libfslcodec/libfslcodec.mk | 8 ++++++++
>  1 file changed, 8 insertions(+)
> 
> diff --git a/package/libfslcodec/libfslcodec.mk b/package/libfslcodec/libfslcodec.mk
> index f8c6aa8..05e5edb 100644
> --- a/package/libfslcodec/libfslcodec.mk
> +++ b/package/libfslcodec/libfslcodec.mk
> @@ -13,6 +13,14 @@ LIBFSLCODEC_LICENSE = Freescale Semiconductor Software License Agreement, BSD-3c
>  LIBFSLCODEC_LICENSE_FILES = EULA COPYING
>  LIBFSLCODEC_REDISTRIBUTE = NO
>  
> +ifeq ($(BR2_ARM_EABIHF),y)
> +LIBFSLCODEC_CONF_OPTS += --enable-fhw
> +endif

What about adding --disable-fhw when not EABIhf?

> +ifeq ($(BR2_PACKAGE_IMX_VPU),y)
> +LIBFSLCODEC_CONF_OPTS += --enable-vpu
> +endif

Ditto when imx-vpu is not enabled?

Regards,
Yann E. MORIN.

>  define LIBFSLCODEC_EXTRACT_CMDS
>  	$(call FREESCALE_IMX_EXTRACT_HELPER,$(DL_DIR)/$(LIBFSLCODEC_SOURCE))
>  endef
> -- 
> 2.6.2
> 
> _______________________________________________
> buildroot mailing list
> buildroot@busybox.net
> http://lists.busybox.net/mailman/listinfo/buildroot
Gary Bisson Dec. 14, 2015, 9:28 p.m. UTC | #2
Yann, All,

On Sun, Dec 13, 2015 at 09:32:24PM +0100, Yann E. MORIN wrote:
> Gary, All,
> 
> On 2015-12-10 14:52 +0100, Gary Bisson spake thusly:
> > Enable VPU and hard float support when the proper configuration is setup.
> > BR2_ARM_EABIHF=y
> > BR2_PACKAGE_IMX_VPU=y
> > 
> > See Yocto equivalent:
> > https://github.com/Freescale/meta-fsl-arm/blob/fido/recipes-
> > multimedia/libfslcodec/libfslcodec.inc
> > 
> > Tested with the Linaro ARM 2014.09 toolchain.
> > 
> > Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
> > ---
> > 
> > Thanks to those options, I could see that:
> > - lib_jpeg_dec_arm11_elinux_novpu.so isn't copied
> > - lib_peq_arm11_elinux.so.1.fhw is copied instead of the fsw version
> > 
> > Regards,
> > Gary
> > 
> > ---
> >  package/libfslcodec/libfslcodec.mk | 8 ++++++++
> >  1 file changed, 8 insertions(+)
> > 
> > diff --git a/package/libfslcodec/libfslcodec.mk b/package/libfslcodec/libfslcodec.mk
> > index f8c6aa8..05e5edb 100644
> > --- a/package/libfslcodec/libfslcodec.mk
> > +++ b/package/libfslcodec/libfslcodec.mk
> > @@ -13,6 +13,14 @@ LIBFSLCODEC_LICENSE = Freescale Semiconductor Software License Agreement, BSD-3c
> >  LIBFSLCODEC_LICENSE_FILES = EULA COPYING
> >  LIBFSLCODEC_REDISTRIBUTE = NO
> >  
> > +ifeq ($(BR2_ARM_EABIHF),y)
> > +LIBFSLCODEC_CONF_OPTS += --enable-fhw
> > +endif
> 
> What about adding --disable-fhw when not EABIhf?

The configure only looks for enable-fhw and therefore considers the
features is disabled by default.

> > +ifeq ($(BR2_PACKAGE_IMX_VPU),y)
> > +LIBFSLCODEC_CONF_OPTS += --enable-vpu
> > +endif
> 
> Ditto when imx-vpu is not enabled?

Same here for the vpu option, configure only looks for the enable
option.

Regards,
Gary
Gary Bisson March 4, 2016, 9:17 a.m. UTC | #3
Yann, All,

On Mon, Dec 14, 2015 at 10:28:54PM +0100, Gary Bisson wrote:
> Yann, All,
> 
> On Sun, Dec 13, 2015 at 09:32:24PM +0100, Yann E. MORIN wrote:
> > Gary, All,
> > 
> > On 2015-12-10 14:52 +0100, Gary Bisson spake thusly:
> > > Enable VPU and hard float support when the proper configuration is setup.
> > > BR2_ARM_EABIHF=y
> > > BR2_PACKAGE_IMX_VPU=y
> > > 
> > > See Yocto equivalent:
> > > https://github.com/Freescale/meta-fsl-arm/blob/fido/recipes-
> > > multimedia/libfslcodec/libfslcodec.inc
> > > 
> > > Tested with the Linaro ARM 2014.09 toolchain.
> > > 
> > > Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
> > > ---
> > > 
> > > Thanks to those options, I could see that:
> > > - lib_jpeg_dec_arm11_elinux_novpu.so isn't copied
> > > - lib_peq_arm11_elinux.so.1.fhw is copied instead of the fsw version
> > > 
> > > Regards,
> > > Gary
> > > 
> > > ---
> > >  package/libfslcodec/libfslcodec.mk | 8 ++++++++
> > >  1 file changed, 8 insertions(+)
> > > 
> > > diff --git a/package/libfslcodec/libfslcodec.mk b/package/libfslcodec/libfslcodec.mk
> > > index f8c6aa8..05e5edb 100644
> > > --- a/package/libfslcodec/libfslcodec.mk
> > > +++ b/package/libfslcodec/libfslcodec.mk
> > > @@ -13,6 +13,14 @@ LIBFSLCODEC_LICENSE = Freescale Semiconductor Software License Agreement, BSD-3c
> > >  LIBFSLCODEC_LICENSE_FILES = EULA COPYING
> > >  LIBFSLCODEC_REDISTRIBUTE = NO
> > >  
> > > +ifeq ($(BR2_ARM_EABIHF),y)
> > > +LIBFSLCODEC_CONF_OPTS += --enable-fhw
> > > +endif
> > 
> > What about adding --disable-fhw when not EABIhf?
> 
> The configure only looks for enable-fhw and therefore considers the
> features is disabled by default.
> 
> > > +ifeq ($(BR2_PACKAGE_IMX_VPU),y)
> > > +LIBFSLCODEC_CONF_OPTS += --enable-vpu
> > > +endif
> > 
> > Ditto when imx-vpu is not enabled?
> 
> Same here for the vpu option, configure only looks for the enable
> option.

Sorry for reviving this thread (again), but I was thinking that maybe we
could merge the two "missing configure options" patches from the series
since they do not depend on the libraries location fix.

Then I could offer another version of the series with only the path fix.
I'm still unclear on which option would get those patches approved. I
understand the symbolic links aren't great. I'm considering removing
gst-fsl-plugins for GStreamer0.10 since Yocto did it:
https://github.com/Freescale/meta-fsl-arm/commit/83b240e7

Let me know your thoughts.

Thanks,
Gary
diff mbox

Patch

diff --git a/package/libfslcodec/libfslcodec.mk b/package/libfslcodec/libfslcodec.mk
index f8c6aa8..05e5edb 100644
--- a/package/libfslcodec/libfslcodec.mk
+++ b/package/libfslcodec/libfslcodec.mk
@@ -13,6 +13,14 @@  LIBFSLCODEC_LICENSE = Freescale Semiconductor Software License Agreement, BSD-3c
 LIBFSLCODEC_LICENSE_FILES = EULA COPYING
 LIBFSLCODEC_REDISTRIBUTE = NO
 
+ifeq ($(BR2_ARM_EABIHF),y)
+LIBFSLCODEC_CONF_OPTS += --enable-fhw
+endif
+
+ifeq ($(BR2_PACKAGE_IMX_VPU),y)
+LIBFSLCODEC_CONF_OPTS += --enable-vpu
+endif
+
 define LIBFSLCODEC_EXTRACT_CMDS
 	$(call FREESCALE_IMX_EXTRACT_HELPER,$(DL_DIR)/$(LIBFSLCODEC_SOURCE))
 endef