search for: asanison

...st need a clean way to express this in the IR. > > Could clang be aware of asan being on and introduce a "please don't > widen" metadata on local variable accesses? Yes, "we just need a clean way to express this in the IR." LLVM can't have a global "bool ASANIsOn;" that the optimizers listen to. -Chris

...his in the IR. > > > > Could clang be aware of asan being on and introduce a "please don't > > widen" metadata on local variable accesses? > > Yes, "we just need a clean way to express this in the IR." > > LLVM can't have a global "bool ASANIsOn;" that the optimizers listen to. > A global is bad. What about a metadata attached to a Function saying that transformations which will read out of bounds (even "safely") are illegal? asan and SAFEcode will add this metadata, optimizers will listen to it. Any other suggestion?...

...ust need a clean way to express this in the IR. > > Could clang be aware of asan being on and introduce a "please don't > widen" metadata on local variable accesses? Yes, "we just need a clean way to express this in the IR." LLVM can't have a global "bool ASANIsOn;" that the optimizers listen to. A global is bad. What about a metadata attached to a Function saying that transformations which will read out of bounds (even "safely") are illegal? asan and SAFEcode will add this metadata, optimizers will listen to it. Any other suggestion? --k...

...>> > Could clang be aware of asan being on and introduce a "please don't >> > widen" metadata on local variable accesses? >> >> Yes, "we just need a clean way to express this in the IR." >> >> LLVM can't have a global "bool ASANIsOn;" that the optimizers listen to. >> > > > A global is bad. > What about a metadata attached to a Function saying that transformations > which will read out of bounds (even "safely") are illegal? > asan and SAFEcode will add this metadata, optimizers will list...

...ust need a clean way to express this in the IR. > > Could clang be aware of asan being on and introduce a "please don't > widen" metadata on local variable accesses? Yes, "we just need a clean way to express this in the IR." LLVM can't have a global "bool ASANIsOn;" that the optimizers listen to. A global is bad. What about a metadata attached to a Function saying that transformations which will read out of bounds (even "safely") are illegal? asan and SAFEcode will add this metadata, optimizers will listen to it. Any other suggestion? --k...

...clang be aware of asan being on and introduce a "please don't >>> > widen" metadata on local variable accesses? >>> >>> Yes, "we just need a clean way to express this in the IR." >>> >>> LLVM can't have a global "bool ASANIsOn;" that the optimizers listen to. >>> >> >> >> A global is bad. >> What about a metadata attached to a Function saying that transformations >> which will read out of bounds (even "safely") are illegal? >> asan and SAFEcode will add this me...

I'm not opposed to disabling this transformation when asan is on, we just need a clean way to express this in the IR. -Chris On Dec 16, 2011, at 3:08 PM, Kostya Serebryany <kcc at google.com> wrote: > Why you don't like the idea to disable or restrict the widening when asan is on? > > --kcc

On 16/12/11 08:46 PM, Chris Lattner wrote: > I'm not opposed to disabling this transformation when asan is on, we just need a clean way to express this in the IR. Could clang be aware of asan being on and introduce a "please don't widen" metadata on local variable accesses? Alternatively, could asan analyze the IL to see which bytes are used? > -Chris > Cheers,

...uld clang be aware of asan being on and introduce a "please don't > > widen" metadata on local variable accesses? > > Yes, "we just need a clean way to express this in the IR." > > LLVM can't have a global "bool ASANIsOn;" that the optimizers listen to. > > > > A global is bad. > What about a metadata attached to a Function saying that transformations > which will read out of bounds (even "safely") are illegal? > asan and SAFEcode will add this met...