Patchwork [U-Boot] nds32: Add a basic errno.h

login
register
mail settings
Submitter Simon Glass
Date Jan. 5, 2013, 8:48 p.m.
Message ID <1357418903-26736-1-git-send-email-sjg@chromium.org>
Download mbox | patch
Permalink /patch/209697/
State Accepted, archived
Delegated to: Macpaul Lin
Headers show

Comments

Simon Glass - Jan. 5, 2013, 8:48 p.m.
This is available on other architectures, and nds32 will start to break
without it as code starts to use error numbers more.

Signed-off-by: Simon Glass <sjg@chromium.org>
---
 arch/nds32/include/asm/errno.h |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
 create mode 100644 arch/nds32/include/asm/errno.h
Macpaul Lin - Jan. 6, 2013, 11:37 a.m.
Hi Simon,

2013/1/6 Simon Glass <sjg@chromium.org>

> This is available on other architectures, and nds32 will start to break
> without it as code starts to use error numbers more.
>
> +++ b/arch/nds32/include/asm/errno.h
> @@ -0,0 +1 @@
> +#include <asm-generic/errno.h


Thanks for your patch.

I've also noticed that NDS32 currently has compiling error because of
-fstack-usage option has been enabled by default.
However, the gcc-4.6 compiler for nds32 is still under developing.
I've told compiler team to speed up developing work on  gcc-4.6 and later.
If there were fail on compiling test of nds32 architecture, please ignore
it.
Sorry for the inconvenince.
Simon Glass - Jan. 6, 2013, 4:28 p.m.
Hi,

On Sun, Jan 6, 2013 at 3:37 AM, Macpaul Lin <macpaul@gmail.com> wrote:
> Hi Simon,
>
> 2013/1/6 Simon Glass <sjg@chromium.org>
>>
>> This is available on other architectures, and nds32 will start to break
>> without it as code starts to use error numbers more.
>>
>> +++ b/arch/nds32/include/asm/errno.h
>> @@ -0,0 +1 @@
>> +#include <asm-generic/errno.h
>
>
> Thanks for your patch.
>
> I've also noticed that NDS32 currently has compiling error because of
> -fstack-usage option has been enabled by default.
> However, the gcc-4.6 compiler for nds32 is still under developing.
> I've told compiler team to speed up developing work on  gcc-4.6 and later.
> If there were fail on compiling test of nds32 architecture, please ignore
> it.
> Sorry for the inconvenince.

I don't see that one at present, but I do get these with gcc 4.4.4:

     nds32:   + adp-ag101p  + adp-ag102  + adp-ag101
+ns16550.c: In function 'NS16550_init':
+ns16550.c:47: warning: passing argument 2 of 'writeb' makes pointer
from integer without a cast
+/c/cosarm/src/third_party/u-boot/us-vboot7/.bm-work/27/build/include2/asm/io.h:109:
note: expected 'unsigned char *' but argument is of type 'long
unsigned int'
+cmd_ide.c: In function '__ide_output_data':
+cmd_ide.c:546: warning: passing argument 2 of 'ide_write_data'
discards qualifiers from pointer target type
+/c/cosarm/src/third_party/u-boot/us-vboot7/.bm-work/03/include/ide.h:76:
note: expected 'ulong *' but argument is of type 'const ulong *'
+/c/cosarm/src/third_party/u-boot/us-vboot7/.bm-work/03/build/include2/asm/io.h:109:
note: expected 'unsigned char *' but argument is of type 'long
unsigned int'
+In file included from ext4fs.c:39:
+ext4_common.h:39:23: error: asm/errno.h: No such file or directory
+make[1]: *** No rule to make target
`/c/cosarm/src/third_party/u-boot/us-vboot7/.bm-work/03/build/fs/ext4/.depend',
needed by `/c/cosarm/src/third_party/u-boot/us-vboot7/.bm-work/03/build/fs/ext4/libext4fs.o'.
 Stop.
+make: *** [/c/cosarm/src/third_party/u-boot/us-vboot7/.bm-work/03/build/fs/ext4/libext4fs.o]
Error 2
+/c/cosarm/src/third_party/u-boot/us-vboot7/.bm-work/30/build/include2/asm/io.h:109:
note: expected 'unsigned char *' but argument is of type 'long
unsigned int'


If you don't support a gcc option it should be possible to disable it
in U-Boot for your toolchain as a temporary workaround. It would be
good to get a pointer to your toolchain at some point.

Regards,
Simon

>
>
> --
> Best regards,
> Macpaul Lin
Tom Rini - Jan. 6, 2013, 5:16 p.m.
On Sun, Jan 06, 2013 at 08:28:39AM -0800, Simon Glass wrote:

[snip]
> If you don't support a gcc option it should be possible to disable it
> in U-Boot for your toolchain as a temporary workaround. It would be
> good to get a pointer to your toolchain at some point.

Unfortunately gcc doesn't error nicely when you try and test for
-fstack-usage support.  The 'cc-option' check says that -fstack-usage
works but then fails in real usage.  It's possible that a more
heavy-weight test would correctly detect the failure however.
Simon Glass - Jan. 6, 2013, 5:24 p.m.
Hi Tom,

On Sun, Jan 6, 2013 at 9:16 AM, Tom Rini <trini@ti.com> wrote:
> On Sun, Jan 06, 2013 at 08:28:39AM -0800, Simon Glass wrote:
>
> [snip]
>> If you don't support a gcc option it should be possible to disable it
>> in U-Boot for your toolchain as a temporary workaround. It would be
>> good to get a pointer to your toolchain at some point.
>
> Unfortunately gcc doesn't error nicely when you try and test for
> -fstack-usage support.  The 'cc-option' check says that -fstack-usage
> works but then fails in real usage.  It's possible that a more
> heavy-weight test would correctly detect the failure however.

Well, if the problem is going to exist for a while, perhaps we could
put up with explicit nds32 check?

>
> --
> Tom

Regards,
Simon
Tom Rini - Jan. 7, 2013, 1:54 p.m.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 01/06/2013 12:24 PM, Simon Glass wrote:
> Hi Tom,
> 
> On Sun, Jan 6, 2013 at 9:16 AM, Tom Rini <trini@ti.com> wrote:
>> On Sun, Jan 06, 2013 at 08:28:39AM -0800, Simon Glass wrote:
>> 
>> [snip]
>>> If you don't support a gcc option it should be possible to
>>> disable it in U-Boot for your toolchain as a temporary
>>> workaround. It would be good to get a pointer to your toolchain
>>> at some point.
>> 
>> Unfortunately gcc doesn't error nicely when you try and test for 
>> -fstack-usage support.  The 'cc-option' check says that
>> -fstack-usage works but then fails in real usage.  It's possible
>> that a more heavy-weight test would correctly detect the failure
>> however.
> 
> Well, if the problem is going to exist for a while, perhaps we
> could put up with explicit nds32 check?

OK with me.

- -- 
Tom
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with undefined - http://www.enigmail.net/
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=no2O
-----END PGP SIGNATURE-----
Macpaul Lin - Feb. 18, 2013, 7:37 a.m.
Hi Simon,

2013/1/6 Simon Glass <sjg@chromium.org>
>
> This is available on other architectures, and nds32 will start to break
> without it as code starts to use error numbers more.
>
> Signed-off-by: Simon Glass <sjg@chromium.org>
> ---
>  arch/nds32/include/asm/errno.h |    1 +
>  1 files changed, 1 insertions(+), 0 deletions(-)
>  create mode 100644 arch/nds32/include/asm/errno.h
>

This bug fix has been applied to u-boot-nds32.git/master
Thanks for your help!

--
Best regards,
Macpaul Lin

Patch

diff --git a/arch/nds32/include/asm/errno.h b/arch/nds32/include/asm/errno.h
new file mode 100644
index 0000000..4c82b50
--- /dev/null
+++ b/arch/nds32/include/asm/errno.h
@@ -0,0 +1 @@ 
+#include <asm-generic/errno.h>