> > I selected 0xdf a long time ago as an alternative to the 0xd0 (Duh) > > byte used for new chunks. Both as a mnemonic for "free" and because it > > is likely to cause segfaults. A pointer ending in 0xdf often will be > > unaligned. Of course that won't work on all archs or all pointers. > > > > Random patterns are also likely to produce segfaults, using them as a > > pointer has a big chance of being unaligned or pointing to an unmapped > > page. > > Did you ever try filling all allocations and frees with random data > to see if it exposes new bugs? Having that option could be useful for > testing & development. > > The d0 & df patterns can hide problems, I've found the hard way. E.g. > in integer comparisons there's a good chance they will consistently > take you to the same branch, and that could be a branch where your > use of freed or unitialized memory doesn't manifest in any observable > malbehavior. Random data would likely trigger such snafus.
Well, a diff has been sent which does what you speak of. So please show a bug of the class you just described, using it. You found out the hard way, so show it being discovered here. Looking forward to your mail. The problem is the cost of this change is *far from free*. If it is 3x as expensive at runtime, and results in fewer people using the option as a result, then it will discovered fewer results than it finds now.
