[PATCH 0/4] Small fixes in barebox

[PATCH 0/4] Small fixes in barebox

[Alexander Aring]

First I tried to change getopt optstring argument to "const char *"
and I test it with an allyesconfig for compiler warnings.

Then there popup some compile errors, which I tried to fix.

I am not happy with some of these solutions. Maybe someone have an idea
how we can make it better.

hush: add getopt only if it enabled
 - I don't know if this needs a fix. But if we disable getopt there is a
   no-op getopt implementation in the command list. This patch don't add
   getopt to the command list at runtime, but the no-op implementation is
   still there. Maybe a compiletime solution is better to remove the
   no-op implementation.

sata-imx: fix depends on ARCH_IMX
 - I am not sure but this looks like ARCH_IMX related things. I got a error
   that some imx51 header files could'nt be found.

net: fix cpsw depends on ARCH_OMAP
 - Same thing like sata-imx. I got a compile error that mach/cpsw.h wasn't
   found. I only found that in the omap implementation.


Another (maybe toolchain related issue) is:
barebox/lib/libubigen.c:105: undefined reference to `__divdi3'

Is it in this line:

tmp = (vi->bytes + ui->leb_size - 1) / ui->leb_size;

If we know that ui->leb_size is a power-of-two, we can use a shift for this
instead of division. But I don't know if leb_size a power-of-two number.

Alexander Aring (4):
  hush: add getopt only if it enabled
  getopt: change optstring to const char*
  sata-imx: fix depends on ARCH_IMX
  net: fix cpsw depends on ARCH_OMAP

 common/hush.c       | 3 ++-
 drivers/ata/Kconfig | 1 +
 drivers/net/Kconfig | 1 +
 include/getopt.h    | 2 +-
 lib/getopt.c        | 4 ++--
 5 files changed, 7 insertions(+), 4 deletions(-)

-- 
1.8.1.2


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

[PATCH 1/4] hush: add getopt only if it enabled

[Alexander Aring]

This patch add getopt to the command list if it enabled via
Kconfig. With this patch we get a 'command not found' error.
Otherwise getopt doing nothing.

Signed-off-by: Alexander Aring <alex.aring@gmail.com>
---
 common/hush.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/common/hush.c b/common/hush.c
index 1f468f6..602f8f1 100644
--- a/common/hush.c
+++ b/common/hush.c
@@ -782,7 +782,8 @@ static int run_pipe_real(struct p_context *ctx, struct pipe *pi)
 
 	remove_quotes(globbuf.gl_pathc, globbuf.gl_pathv);
 
-	if (!strcmp(globbuf.gl_pathv[0], "getopt")) {
+	if (!strcmp(globbuf.gl_pathv[0], "getopt") &&
+			IS_ENABLED(CONFIG_HUSH_GETOPT)) {
 		ret = builtin_getopt(ctx, child, globbuf.gl_pathc, globbuf.gl_pathv);
 	} else if (!strcmp(globbuf.gl_pathv[0], "exit")) {
 		ret = builtin_exit(ctx, child, globbuf.gl_pathc, globbuf.gl_pathv);
-- 
1.8.1.2


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

[PATCH 2/4] getopt: change optstring to const char*

[Alexander Aring]

Change getopt optstring parameter type to const char *.
Also change type to const char * of tmp variable which
pointed to optstring. This will only handle readonly.

Signed-off-by: Alexander Aring <alex.aring@gmail.com>
---
 include/getopt.h | 2 +-
 lib/getopt.c     | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/include/getopt.h b/include/getopt.h
index f23175f..4f48ba8 100644
--- a/include/getopt.h
+++ b/include/getopt.h
@@ -35,7 +35,7 @@ extern char *optarg;
  * - options can be mixed with nonoptions (like ls /bin -R)
  */
 
-int getopt(int argc, char *argv[], char *optstring);
+int getopt(int argc, char *argv[], const char *optstring);
 
 struct getopt_context {
 	int opterr;
diff --git a/lib/getopt.c b/lib/getopt.c
index ead9150..fd12a88 100644
--- a/lib/getopt.c
+++ b/lib/getopt.c
@@ -56,10 +56,10 @@ void getopt_context_restore(struct getopt_context *gc)
 }
 EXPORT_SYMBOL(getopt_context_restore);
 
-int getopt(int argc, char *argv[], char *optstring)
+int getopt(int argc, char *argv[], const char *optstring)
 {
 	char curopt;   /* current option character */
-	char *curoptp; /* pointer to the current option in optstring */
+	const char *curoptp; /* pointer to the current option in optstring */
 
 	while(1) {
 		debug("optindex: %d nonopts: %d optind: %d\n", optindex, nonopts, optind);
-- 
1.8.1.2


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

[PATCH 3/4] sata-imx: fix depends on ARCH_IMX

[Alexander Aring]

Signed-off-by: Alexander Aring <alex.aring@gmail.com>
---
 drivers/ata/Kconfig | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/ata/Kconfig b/drivers/ata/Kconfig
index ff6528a..42f2065 100644
--- a/drivers/ata/Kconfig
+++ b/drivers/ata/Kconfig
@@ -38,6 +38,7 @@ config DISK_AHCI
 	select DISK_DRIVE
 
 config DISK_AHCI_IMX
+	depends on ARCH_IMX
 	depends on DISK_AHCI
 	bool "i.MX AHCI support"
 
-- 
1.8.1.2


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

[PATCH 4/4] net: fix cpsw depends on ARCH_OMAP

[Alexander Aring]

drivers/net/cpsw.c:34:23: fatal error: mach/cpsw.h: No such file or directory

Signed-off-by: Alexander Aring <alex.aring@gmail.com>
---
 drivers/net/Kconfig | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index cf6ceee..2b7decd 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -33,6 +33,7 @@ config DRIVER_NET_CS8900
 
 config DRIVER_NET_CPSW
 	bool "CPSW ethernet driver"
+	depends on ARCH_OMAP
 	select PHYLIB
 
 config DRIVER_NET_SMC911X
-- 
1.8.1.2


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH 4/4] net: fix cpsw depends on ARCH_OMAP

[Alexander Aring]

Hi,

On Thu, Feb 07, 2013 at 09:36:20PM +0100, Alexander Aring wrote:
> drivers/net/cpsw.c:34:23: fatal error: mach/cpsw.h: No such file or directory

grml, I should resend them with an empty commit msg.
Sorry.

Regards
Alex

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH 0/4] Small fixes in barebox

[Sascha Hauer]

On Thu, Feb 07, 2013 at 09:36:16PM +0100, Alexander Aring wrote:
> First I tried to change getopt optstring argument to "const char *"
> and I test it with an allyesconfig for compiler warnings.
> 
> Then there popup some compile errors, which I tried to fix.
> 
> I am not happy with some of these solutions. Maybe someone have an idea
> how we can make it better.
> 
> hush: add getopt only if it enabled
>  - I don't know if this needs a fix. But if we disable getopt there is a
>    no-op getopt implementation in the command list. This patch don't add
>    getopt to the command list at runtime, but the no-op implementation is
>    still there. Maybe a compiletime solution is better to remove the
>    no-op implementation.

I don't understand. getopt is not there if CONFIG_HUSH_GETOPT is
disabled. It's ifdeffed out.

> 
> sata-imx: fix depends on ARCH_IMX
>  - I am not sure but this looks like ARCH_IMX related things. I got a error
>    that some imx51 header files could'nt be found.

That's fine you did this. Another option would be to move the header
file to include/, but currently this is not worth it.

> 
> net: fix cpsw depends on ARCH_OMAP
>  - Same thing like sata-imx. I got a compile error that mach/cpsw.h wasn't
>    found. I only found that in the omap implementation.

Also fine.

> 
> 
> Another (maybe toolchain related issue) is:
> barebox/lib/libubigen.c:105: undefined reference to `__divdi3'
> 
> Is it in this line:
> 
> tmp = (vi->bytes + ui->leb_size - 1) / ui->leb_size;
> 
> If we know that ui->leb_size is a power-of-two, we can use a shift for this
> instead of division. But I don't know if leb_size a power-of-two number.

leb_size (logical erase block size) is never a power of two. This is the
peb size (physical erase block size) minus the ubi metadata.

Sascha

-- 
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH 0/4] Small fixes in barebox

[Sascha Hauer]

On Thu, Feb 07, 2013 at 10:35:24PM +0100, Sascha Hauer wrote:
> On Thu, Feb 07, 2013 at 09:36:16PM +0100, Alexander Aring wrote:
> > First I tried to change getopt optstring argument to "const char *"
> > and I test it with an allyesconfig for compiler warnings.
> > 
> > Then there popup some compile errors, which I tried to fix.
> > 
> > I am not happy with some of these solutions. Maybe someone have an idea
> > how we can make it better.
> > 
> > hush: add getopt only if it enabled
> >  - I don't know if this needs a fix. But if we disable getopt there is a
> >    no-op getopt implementation in the command list. This patch don't add
> >    getopt to the command list at runtime, but the no-op implementation is
> >    still there. Maybe a compiletime solution is better to remove the
> >    no-op implementation.
> 
> I don't understand. getopt is not there if CONFIG_HUSH_GETOPT is
> disabled. It's ifdeffed out.

Ah, now I got it. You miss the 'unknown command... message'. Then the
patch seems fine.

Sascha

-- 
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH 0/4] Small fixes in barebox

[Alexander Aring]

Hi Sascha,

On Thu, Feb 07, 2013 at 10:35:24PM +0100, Sascha Hauer wrote:
> On Thu, Feb 07, 2013 at 09:36:16PM +0100, Alexander Aring wrote:
> > First I tried to change getopt optstring argument to "const char *"
> > and I test it with an allyesconfig for compiler warnings.
> > 
> > Then there popup some compile errors, which I tried to fix.
> > 
> > I am not happy with some of these solutions. Maybe someone have an idea
> > how we can make it better.
> > 
> > hush: add getopt only if it enabled
> >  - I don't know if this needs a fix. But if we disable getopt there is a
> >    no-op getopt implementation in the command list. This patch don't add
> >    getopt to the command list at runtime, but the no-op implementation is
> >    still there. Maybe a compiletime solution is better to remove the
> >    no-op implementation.
> 
> I don't understand. getopt is not there if CONFIG_HUSH_GETOPT is
> disabled. It's ifdeffed out.
> 

Yes it's ifdeffed out with an empty no-op getopt implementation. But
this will be add to the barebox command list.

So if we type "getopt" we don't get a "Unknown command" message, barebox
run only a no-op getopt implementation.

This patch will don't add getopt to the command list if we don't enable
it. I let the no-op getopt implementation there, otherwise we get a
compiler warning.

Regards
Alex


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH 0/4] Small fixes in barebox

[Alexander Aring]

Hi,

On Thu, Feb 07, 2013 at 10:41:56PM +0100, Sascha Hauer wrote:
> On Thu, Feb 07, 2013 at 10:35:24PM +0100, Sascha Hauer wrote:
> > On Thu, Feb 07, 2013 at 09:36:16PM +0100, Alexander Aring wrote:
> > > First I tried to change getopt optstring argument to "const char *"
> > > and I test it with an allyesconfig for compiler warnings.
> > > 
> > > Then there popup some compile errors, which I tried to fix.
> > > 
> > > I am not happy with some of these solutions. Maybe someone have an idea
> > > how we can make it better.
> > > 
> > > hush: add getopt only if it enabled
> > >  - I don't know if this needs a fix. But if we disable getopt there is a
> > >    no-op getopt implementation in the command list. This patch don't add
> > >    getopt to the command list at runtime, but the no-op implementation is
> > >    still there. Maybe a compiletime solution is better to remove the
> > >    no-op implementation.
> > 
> > I don't understand. getopt is not there if CONFIG_HUSH_GETOPT is
> > disabled. It's ifdeffed out.
> 
> Ah, now I got it. You miss the 'unknown command... message'. Then the
> patch seems fine.
>

Yea, it's only a small thing... I don't know if this is necessary. :-)
Maybe there is a special reason for that no error occurs.

Regards
Alex

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH 0/4] Small fixes in barebox

[Sascha Hauer]

On Thu, Feb 07, 2013 at 10:50:54PM +0100, Alexander Aring wrote:
> Hi,
> 
> On Thu, Feb 07, 2013 at 10:41:56PM +0100, Sascha Hauer wrote:
> > On Thu, Feb 07, 2013 at 10:35:24PM +0100, Sascha Hauer wrote:
> > > On Thu, Feb 07, 2013 at 09:36:16PM +0100, Alexander Aring wrote:
> > > > First I tried to change getopt optstring argument to "const char *"
> > > > and I test it with an allyesconfig for compiler warnings.
> > > > 
> > > > Then there popup some compile errors, which I tried to fix.
> > > > 
> > > > I am not happy with some of these solutions. Maybe someone have an idea
> > > > how we can make it better.
> > > > 
> > > > hush: add getopt only if it enabled
> > > >  - I don't know if this needs a fix. But if we disable getopt there is a
> > > >    no-op getopt implementation in the command list. This patch don't add
> > > >    getopt to the command list at runtime, but the no-op implementation is
> > > >    still there. Maybe a compiletime solution is better to remove the
> > > >    no-op implementation.
> > > 
> > > I don't understand. getopt is not there if CONFIG_HUSH_GETOPT is
> > > disabled. It's ifdeffed out.
> > 
> > Ah, now I got it. You miss the 'unknown command... message'. Then the
> > patch seems fine.
> >
> 
> Yea, it's only a small thing... I don't know if this is necessary. :-)
> Maybe there is a special reason for that no error occurs.

Well the patch will result in dropping the unnecessary strcmp, so a
slightly smaller binary.

Sascha

-- 
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

[PATCH 4/4] net: fix cpsw depends on ARCH_OMAP

[Alexander Aring]

Signed-off-by: Alexander Aring <alex.aring@gmail.com>
---
 drivers/net/Kconfig | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index cf6ceee..2b7decd 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -33,6 +33,7 @@ config DRIVER_NET_CS8900
 
 config DRIVER_NET_CPSW
 	bool "CPSW ethernet driver"
+	depends on ARCH_OMAP
 	select PHYLIB
 
 config DRIVER_NET_SMC911X
-- 
1.8.1.2


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox