This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: PATCH:[darwin] fix load of a misaligned double word
On Dec 31, 2003, at 2:10 PM, David Edelsohn wrote:
You seem to be defining "quality of implementation" and "bug" to
mean whether the compiler does what you want with "implementation
defined"
behavior.
The standard defines all conversions between ints and pointers (other
than the four exceptions I listed) to be "implementation defined". How
gcc handles "implementation defined" behavior is by definition a
"quality of implementation" issue. That is the sense in which I'm
using the term.
I'm saying it's a bug because gcc is using an incorrect offset for a
64-bit integer register. Are you claiming that this is *not* a bug?
The GCC PowerPC backend could bend over backward to handle the
unusual pointer you are computing, as I have mentioned in another
thread.
What's unusual? The bit pattern of the pointer has three 0s in the
low-order bits so it's properly aligned, the memory is properly
allocated.
Brad