[Buildroot] [PATCH] package/openocd: disable on blackfin

[Buildroot] [PATCH] package/openocd: disable on blackfin

[Samuel Martin]

The "_" symbol prefix added by the blackfin toolchain causes "undefined
symbol" failures during the configuration.

Fixes:
  http://autobuild.buildroot.org/results/e90/e90b4d5ad79d99487f21c9d18581e8eba7034501/

Signed-off-by: Samuel Martin <s.martin49@gmail.com>
---
 package/openocd/Config.in | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/package/openocd/Config.in b/package/openocd/Config.in
index 7c140c7..9543c03 100644
--- a/package/openocd/Config.in
+++ b/package/openocd/Config.in
@@ -1,5 +1,9 @@
 config BR2_PACKAGE_OPENOCD
 	bool "openocd"
+	# disable openocd on blackfin because it triggers "undefined symbol" error
+	# becasue of the "_" prefix:
+	# http://autobuild.buildroot.org/results/e90/e90b4d5ad79d99487f21c9d18581e8eba7034501/
+	depends on !BR2_bfin
 	help
 	  OpenOCD - Open On-Chip Debugger
 
-- 
2.2.1

[Buildroot] [PATCH] package/openocd: disable on blackfin

[Thomas Petazzoni]

Dear Samuel Martin,

On Mon,  5 Jan 2015 21:33:09 +0100, Samuel Martin wrote:
> The "_" symbol prefix added by the blackfin toolchain causes "undefined
> symbol" failures during the configuration.
> 
> Fixes:
>   http://autobuild.buildroot.org/results/e90/e90b4d5ad79d99487f21c9d18581e8eba7034501/
> 
> Signed-off-by: Samuel Martin <s.martin49@gmail.com>
> ---
>  package/openocd/Config.in | 4 ++++
>  1 file changed, 4 insertions(+)
> 
> diff --git a/package/openocd/Config.in b/package/openocd/Config.in
> index 7c140c7..9543c03 100644
> --- a/package/openocd/Config.in
> +++ b/package/openocd/Config.in
> @@ -1,5 +1,9 @@
>  config BR2_PACKAGE_OPENOCD
>  	bool "openocd"
> +	# disable openocd on blackfin because it triggers "undefined symbol" error
> +	# becasue of the "_" prefix:
> +	# http://autobuild.buildroot.org/results/e90/e90b4d5ad79d99487f21c9d18581e8eba7034501/
> +	depends on !BR2_bfin
>  	help
>  	  OpenOCD - Open On-Chip Debugger
>  

This seems weird because the "_" prefix normally only causes issues
when linking C code against assembly code. If you're only linking C
code against C code, it *should* be fully transparent.

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

[Buildroot] [PATCH] package/openocd: disable on blackfin

[Yann E. MORIN]

Samuel, Thomas, All,

On 2015-01-06 09:26 +0100, Thomas Petazzoni spake thusly:
> On Mon,  5 Jan 2015 21:33:09 +0100, Samuel Martin wrote:
> > The "_" symbol prefix added by the blackfin toolchain causes "undefined
> > symbol" failures during the configuration.
> > 
> > Fixes:
> >   http://autobuild.buildroot.org/results/e90/e90b4d5ad79d99487f21c9d18581e8eba7034501/
> > 
> > Signed-off-by: Samuel Martin <s.martin49@gmail.com>
> > ---
> >  package/openocd/Config.in | 4 ++++
> >  1 file changed, 4 insertions(+)
> > 
> > diff --git a/package/openocd/Config.in b/package/openocd/Config.in
> > index 7c140c7..9543c03 100644
> > --- a/package/openocd/Config.in
> > +++ b/package/openocd/Config.in
> > @@ -1,5 +1,9 @@
> >  config BR2_PACKAGE_OPENOCD
> >  	bool "openocd"
> > +	# disable openocd on blackfin because it triggers "undefined symbol" error
> > +	# becasue of the "_" prefix:
> > +	# http://autobuild.buildroot.org/results/e90/e90b4d5ad79d99487f21c9d18581e8eba7034501/
> > +	depends on !BR2_bfin
> >  	help
> >  	  OpenOCD - Open On-Chip Debugger
> >  
> 
> This seems weird because the "_" prefix normally only causes issues
> when linking C code against assembly code. If you're only linking C
> code against C code, it *should* be fully transparent.

The real culprit seems to be libusb-compat. See config.log:

configure:14893: [...]/bfin-uclinux-gcc -o conftest [...] -static [...] -lftdi -lusb
/home/ymorin/dev/buildroot/O/host/usr/bfin-buildroot-uclinux-uclibc/sysroot/usr/lib/libusb.a(libusb_la-core.o):
In function `_usb_detach_kernel_driver_np':
core.c:(.text+0x96): undefined reference to `_libusb_detach_kernel_driver'

So, it means libusb.a as an undefined reference to _libusb_detach_kernel_driver
which is in fact defined in libus-1.0.a.

If I add -lusb-1.0 -pthread to the link command, it all works great! :-)

So, it looks like 'just' a missing dependency in a static case.

So, marking this patch as Rejected in Patchwork: the real fix is to link
with the appropriate libraries.

Hint: the .pc files seem to be correct. But line 1060 in configure.ac is
where the error originates from. It is the last clause in an AC_RUN_IFELSE
block starting on line 1042... And the rest is left as an exercise. ;-)

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.  |
'------------------------------^-------^------------------^--------------------'