This is the mail archive of the libstdc++@gcc.gnu.org mailing list for the libstdc++ project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH][RFC] Remove volatile from data members in libstdc++


Mike Stump wrote:
On Jul 18, 2006, at 10:37 AM, Martin Sebor wrote:

we were told by the compiler vendor


Which compiler, which vendor?

We reported it against IBM XLC++ 8.0 (PMR 02254). The code in the test case was actually slightly different. It went something like this:

    extern int init;
    for (volatile int *p = &init; *p < 1000; );


[ I'd really like to know if it was tech supporting batting the bug away, or if an engineer thought long and hard about it, but, I'd not expect that much visibility ]

From the support engineer's response it sounded like he was communicating someone else's rationale for the "optimization." They suggested either making init volatile or declaring the pointer itself volatile (i.e., int* volatile p). The latter doesn't really make sense to me.

Martin


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]