Re: [head tinderbox] failure on amd64/amd64

Garrett Cooper wrote:
>    Is there a reason why this line of code is invalid for amd64 (from
> /usr/src/contrib/bind9/lib/isc/x86_64/include/isc/atomic.h:25-28)?
> 
> #ifdef ISC_PLATFORM_USEGCCASM
> 
> /* We share the gcc-version with x86_32 */
> #error "impossible case.  check build configuration"
> 
>    After doing some searching, all that that macro does is use says the
> OS can use gcc asm instructions.

Yeah, I looked at that header and asked myself the same question. I'm
going to follow up with the ISC folks on that issue.

Meanwhile I finally got access to an amd64 machine that I could build
the port on (our reference system is literally FUBAR), and it looks
like things are in better shape than I feared. Ruslan's suggestion of
using the x86_32 atomic.h header that I already committed turns out to
be how the port does it, and the other generated headers look good, so
the build should be fine now. If anyone could confirm that with
sources that include v. 1.18 of src/lib/bind/config.mk the world
actually builds, I'd appreciate it. If you could do some rudimentary
testing of the results, I'd be eternally grateful. :)

FWIW, I relied on the fact that the bind94 port builds just fine on
our platforms to convince me that things would be fine for the base.
While that seems to have worked for most of the cases, it turns out to
have not been a good assumption to make. I apologize for the
inconvenience that I created.


Mea culpa,

Doug

-- 

    This .signature sanitized for your protection