https://redports.org/buildarchive/20140817161100-53255

Doesn't break amd64 or i386. So that's good.

This is th buildfailure I get without this patch:

http://chips.ysv.freebsd.org/data/11armv632-default/2014-08-16_23h54m58s/logs/errors/nspr-4.10.6.log

In D617#6, @sbruno wrote:

I'm not 100% sure how to do that in this file, so maybe this works, maybe it doesn't?

#elif defined(__arm__)
#if defined(__armeb__)
#undef IS_BIG_ENDIAN
#define IS_LITTLE_ENDIAN 1
#else
#undef IS_LITTLE_ENDIAN
#define IS_BIG_ENDIAN 1
#endif /*armeb */

Upper case is needed. ARMEB. However, that's just for gcc. If you want to future proof this against clang supporting arm big endian, the Arm C Language Extention (ACLE) specifies that this is ARM_BIG_ENDIAN, so I'd test for both for now.

The web page is rendering double underscore as an underscored thing. There are two leading and trailing underscores for the symbols in my last comment.

In D617#8, @imp wrote:

The web page is rendering double underscore as an underscored thing. There are two leading and trailing underscores for the symbols in my last comment.

This is what you guys are going for right?

#elif defined(__arm__)
  
#if defined(__ARMEB__) || defined(__ARM_BIG_ENDIAN__)
#undef  IS_LITTLE_ENDIAN
#define IS_BIG_ENDIAN    1
#else
#undef  IS_BIG_ENDIAN
#define IS_LITTLE_ENDIAN 1
#endif

Yes. the quoted code looks good.

https://redports.org/buildarchive/20140818002502-52216/

All groovy on amd64/i386

Confirmed builds on armv6 qemu-user enabled jail