opkg-sdk_svn build failure

opkg-sdk_svn build failure

[Dmitry Vinokurov]

Hello,

I'm trying to build QTE SDK according to OE manual chapter "Creating and 
Using a Qt Embedded SDK". Command "bitbake meta-toolchain-qte" fails with:
---------------------------------------
ERROR: function do_compile failed
ERROR: see log in 
/home/raydan/work/oe/build/tmp/work/x86_64-armv5te-sdk-oe-linux-gnueabi/opkg-sdk-0.1.6+svnr363-r16/temp/log.do_compile.19556
---------------------------------------

Tail of log contains following lines:
---------------------------------------
cc1: warnings being treated as errors
hash_table.c: In function 'hash_print_stats':
hash_table.c:81: error: format '%d' expects type 'int', but argument 3 
has type 'long unsigned int'
---------------------------------------

Ready to provide any additional information, if needed.

-- 
Best Regards, 
Dmitry Vinokurov 
<d.vinokuroff@gmail.com>

Re: opkg-sdk_svn build failure

[Dmitry Vinokurov]

[-- Attachment #1: Type: text/plain, Size: 1430 bytes --]

Dmitry Vinokurov wrote:
> Hello,
>
> I'm trying to build QTE SDK according to OE manual chapter "Creating 
> and Using a Qt Embedded SDK". Command "bitbake meta-toolchain-qte" 
> fails with:
> ---------------------------------------
> ERROR: function do_compile failed
> ERROR: see log in 
> /home/raydan/work/oe/build/tmp/work/x86_64-armv5te-sdk-oe-linux-gnueabi/opkg-sdk-0.1.6+svnr363-r16/temp/log.do_compile.19556 
>
> ---------------------------------------
>
> Tail of log contains following lines:
> ---------------------------------------
> cc1: warnings being treated as errors
> hash_table.c: In function 'hash_print_stats':
> hash_table.c:81: error: format '%d' expects type 'int', but argument 3 
> has type 'long unsigned int'
> ---------------------------------------
>
> Ready to provide any additional information, if needed.
>
Possible patch to include in opkg.inc attached.

--
Best Regards, 
Dmitry Vinokurov 
<d.vinokuroff@gmail.com>

--- opkg/libopkg/hash_table.c.orig	2009-11-25 00:48:49.000000000 +0500
+++ opkg/libopkg/hash_table.c	2009-11-25 00:49:16.000000000 +0500
@@ -64,7 +64,7 @@
 void
 hash_print_stats(hash_table_t *hash)
 {
-	printf("hash_table: %s, %d bytes\n"
+	printf("hash_table: %s, %ld bytes\n"
 		"\tn_buckets=%d, n_elements=%d, n_collisions=%d\n"
 		"\tmax_bucket_len=%d, n_used_buckets=%d, ave_bucket_len=%.2f\n"
 		"\tn_hits=%d, n_misses=%d\n",

Re: opkg-sdk_svn build failure

[Graham Gower]

sizeof returns long unsigned int on x86_64. This is fixed in r365.

Re: opkg-sdk_svn build failure

[Khem Raj]

On (25/11/09 07:49), Graham Gower wrote:
> sizeof returns long unsigned int on x86_64. This is fixed in r365.

sizeof return size_t which is a type in itself so if you want
portability between 32-bit and 64-bit systems you should probably
use 'z' length modifier in printf formats. '%zd' would work correctly
irrespective of word lenght of the processor. You need C99 compliant
compiler though.

Thanks

-Khem

> 
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel

Re: opkg-sdk_svn build failure

[Graham Gower]

2009/12/27 Khem Raj <raj.khem@gmail.com>:
> On (25/11/09 07:49), Graham Gower wrote:
>> sizeof returns long unsigned int on x86_64. This is fixed in r365.
>
> sizeof return size_t which is a type in itself so if you want
> portability between 32-bit and 64-bit systems you should probably
> use 'z' length modifier in printf formats. '%zd' would work correctly
> irrespective of word lenght of the processor. You need C99 compliant
> compiler though.
>
> Thanks
>
> -Khem

You're right. r365 casts it to an int instead, in an attempt to
maintain portability.