[ImplicitNulls] Add some clarifying comments; NFC

llvm-svn: 253020
This commit is contained in:
Sanjoy Das 2015-11-13 08:14:00 +00:00
parent b3c97271e3
commit ac9c5b1901
1 changed files with 25 additions and 1 deletions

View File

@ -281,7 +281,7 @@ bool ImplicitNullChecks::analyzeBlockForNullChecks(
//
// we want to end up with
//
// Def = TrappingLoad (%RAX + <offset>), LblNull
// Def = FaultingLoad (%RAX + <offset>), LblNull
// jmp LblNotNull ;; explicit or fallthrough
//
// LblNotNull:
@ -292,6 +292,30 @@ bool ImplicitNullChecks::analyzeBlockForNullChecks(
// LblNull:
// callq throw_NullPointerException
//
//
// To see why this is legal, consider the two possibilities:
//
// 1. %RAX is null: since we constrain <offset> to be less than PageSize, the
// load instruction dereferences the null page, causing a segmentation
// fault.
//
// 2. %RAX is not null: in this case we know that the load cannot fault, as
// otherwise the load would've faulted in the original program too and the
// original program would've been undefined.
//
// This reasoning cannot be extended to justify hoisting through arbitrary
// control flow. For instance, in the example below (in pseudo-C)
//
// if (ptr == null) { throw_npe(); unreachable; }
// if (some_cond) { return 42; }
// v = ptr->field; // LD
// ...
//
// we cannot (without code duplication) use the load marked "LD" to null check
// ptr -- clause (2) above does not apply in this case. In the above program
// the safety of ptr->field can be dependent on some_cond; and, for instance,
// ptr could be some non-null invalid reference that never gets loaded from
// because some_cond is always true.
unsigned PointerReg = MBP.LHS.getReg();