svn commit: r217213 - head/lib/bind

Warner Losh imp at bsdimp.com
Wed Jan 12 00:05:20 UTC 2011


On 01/11/2011 06:47, John Baldwin wrote:
> On Tuesday, January 11, 2011 12:04:56 am Doug Barton wrote:
>> I have no objection to putting it back to the state that it was in at
>> r209886, although frankly less diffs to RELENG_[78] without good reason
>> make my life easier.
> 209886 did not use MACHINE_CPUARCH for ISC_ATOMIC_ARCH, and in fact the file
> is now back to the 209886 state (nwhitehorn did not change ISC_ATOMIC_ARCH
> to use MACHINE_CPUARCH).
>
> Fixing this does create diffs to [78] because MACHINE_CPUARCH is not present
> at all in [78], but that is because [78]'s handling of different endians for
> mips, arm, etc. is deficient.  You are likely stuck with using MACHINE_CPUARCH
> for 9+ and MACHINE_ARCH for<= 8.

I've created compatibility shims for MACHINE_CPUARCH in RELENG_8, but 
not for RELENG_7 (at least in some places, now that I go looking for 
it).  I'm thinking seriously of MFCing it to make future MFCs easier, 
especially for some MIPS work that I've been doing.

In -current, Doug's likely will operate correctly for quite some time.  
We're not going to change MACHINE_ARCH of i386 or amd64.  At most, we'll 
change MACHINE_CPUARCH to x86, but that would be a lot of work to 
undertake in the whole tree...  The only down-side is that it increases 
by two the number of MACHINE_ARCH instances that I need to audit when I 
sweep the tree looking for problems.  If there's no good reason for the 
chaff, I change it.  If there is, like this case, I note it as good and 
move on.

While I'd prefer to have everything converted, I'm willing to be 
flexible when necessary...

Warner


More information about the svn-src-all mailing list