svn commit: r268055 - head/sys/sys

[Eitan Adler]

On 30 June 2014 14:03, John Baldwin <jhb at freebsd.org> wrote:
> On Monday, June 30, 2014 4:15:55 pm Marius Strobl wrote:
>> On Mon, Jun 30, 2014 at 02:47:21PM -0400, John Baldwin wrote:
>> > On Monday, June 30, 2014 2:43:42 pm Hans Petter Selasky wrote:
>> > > On 06/30/14 20:37, John Baldwin wrote:
>> > > > Reindenting the whitespace made this diff harder to read.  Why hasn't this
>> > > > been a problem before on powerpc64?
>> > >
>> > > Hi,
>> > >
>> > > This has not been a problem before, because producers of DATA_SET()'s
>> > > were using syntax along these lines:
>> > >
>> > > static int test2;
>> > > static int test3;
>> > >
>> > > DATA_SET(test, test2);
>> > > DATA_SET(test, test3);
>> > >
>> > > Now if you change this simple code to:
>> > >
>> > > static int test2;
>> > > int test3;
>> > >
>> > > DATA_SET(test, test2);
>> > > DATA_SET(test, test3);
>> > >
>> > > It breaks on powerpc64. Should be very easy to reproduce.
>> > >
>> > > The problem with SYSCTL's is that some are global and some are not.
>> > > Before an indirect pointer was hiding this problem from appearing.
>> > >
>> > > Do you see? Or do you want me to explain more.
>> >
>> > Ahh, ok.  Seems odd that this sort of thing would be ppc64-specific however.
>> >
>>
>> Apparently it isn't and has also has shown up on at least ia64:
>> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=31490
>
> Ugh.  Maybe this should be conditional on the GCC version rather than ppc64?

There should also be a GCC bug id in a comment.



-- 
Eitan Adler
Source, Ports, Doc committer
Bugmeister, Ports Security teams