search for: _prove_

...X = ... SCEV *Y = ... if (X == Y) ... will have to be changed to do SCEV *X = ... SCEV *Y = ... if (X->equals(Y)) ... This has potential for compile-time regressions. Hopefully they'll all be addressable. There are cases in which SCEV (via trip count analysis, say) can _prove_ that a certain expression does not overflow. In those cases we will mutate the SCEV expression to indicate no-wrap; since the no-wrap flag is just a "cache" of a proof based on the structure of the SCEV expression, and _does_ apply to all SCEV expressions with the same shapes. Concretel...

...>> >> SCEV *X = ... >> SCEV *Y = ... >> if (X->equals(Y)) >> ... >> >> This has potential for compile-time regressions. Hopefully they'll all be addressable. >> >> There are cases in which SCEV (via trip count analysis, say) can _prove_ that a certain expression does not overflow. In those cases we will mutate the SCEV expression to indicate no-wrap; since the no-wrap flag is just a "cache" of a proof based on the structure of the SCEV expression, and _does_ apply to all SCEV expressions with the same shapes. >>...

Hi Andy, Andrew Trick wrote: >> On Sep 23, 2016, at 7:50 AM, Christof Douma <Christof.Douma at arm.com <mailto:Christof.Douma at arm.com>> wrote: >> >> I can’t help to ask. Why not define a wrapping nsw instruction as UB, instead of “delayed UB” aka poison? I believe we >> have the notion of poison solely to ease the movement of instructions. In my example

...a non-reversible way. This kind of sounds like the inverse situation where the default is AD and it can fall back to NT4. I'm wagering if you expose that Windows 10 machine to an AD DC domain and then tried to switch back again, this trick would no longer work. That test wouldn't actually _prove_ anything other than, "you really should retire your NT4 domain a year ago", but that's been painfully obvious for a while now. If I get bored I might snapshot some VMs and replay the interaction a couple of different ways. Marc, I'm assuming your test was a clean Samba install wi...

Just checking if anybody has Samba's classic DC functioning with Windows 10 domain member clients? In particular, I'm interested in any tests with git master or 4.2. I'm not talking about samba4 DCs (our AD DC), but with the NT4-like mode. The reason I ask is that I've got reports it doesn't work, and I've checked with Microsoft who basically say 'NT4 support ended

...X = ... SCEV *Y = ... if (X == Y) ... will have to be changed to do SCEV *X = ... SCEV *Y = ... if (X->equals(Y)) ... This has potential for compile-time regressions. Hopefully they'll all be addressable. There are cases in which SCEV (via trip count analysis, say) can _prove_ that a certain expression does not overflow. In those cases we will mutate the SCEV expression to indicate no-wrap; since the no-wrap flag is just a "cache" of a proof based on the structure of the SCEV expression, and _does_ apply to all SCEV expressions with the same shapes. Concretel...

...s the strong post-dominator analysis to a new pass that propagate nsw flags in the IR, but that will likely be only a small part now that SCEV is going to add NSW to many more expressions (which is the whole point). >> There are cases in which SCEV (via trip count analysis, say) can >> _prove_ that a certain expression does not overflow. In those cases >> we will mutate the SCEV expression to indicate no-wrap; since the >> no-wrap flag is just a "cache" of a proof based on the structure of >> the SCEV expression, and _does_ apply to all SCEV expressions with...