[cfe-commits] r126342 - in /cfe/trunk: include/clang/Basic/DiagnosticSemaKinds.td include/clang/Sema/Sema.h lib/Sema/SemaExpr.cpp test/Analysis/retain-release-basic-store.m test/Analysis/retain-release-gc-only.m test/Analysis/retain-release-region-store.m test/Analysis/retain-release.m test/Sema/shift.c

[Chris Lattner]

On Feb 23, 2011, at 3:34 PM, Chandler Carruth wrote:

> Author: chandlerc
> Date: Wed Feb 23 17:34:11 2011
> New Revision: 126342
> 
> URL: http://llvm.org/viewvc/llvm-project?rev=126342&view=rev
> Log:
> Implement a warning for known shift overflows on constant shift
> expressions. Consider the code:
> 
>  int64_t i = 10 << 30;
> 
> This compiles fine, but most developers expect it to produce the value
> for 10 gigs, not -2 gigs. This is actually undefined behavior because
> the LHS is a signed integer type.
> 
> The warning is currently gated behind -Wshift-overflow.

Cool, thanks Chandler!  The wording on these is a bit awkward though:

+def warn_shift_result_gt_typewidth : Warning<
+  "shift result (%0) requires %1 bits to represent, but the promoted type of "
+  "the shift expression is %2 with only %3 bits">,
+   InGroup<DiagGroup<"shift-overflow">>;
+def warn_shift_result_overrides_sign_bit : Warning<
+  "shift result (%0) overrides the sign bit of the promoted type of the shift "
+  "expression (%1) and becomes negative">,
+  InGroup<DiagGroup<"shift-sign-overflow">>, DefaultIgnore;

Mentioning the "promoted type" doesn't make sense in a lot of cases (when there is no promotion happening).  It seems completely fine to simplify it to:

"shift result (%0) requires %1 bits to represent, but %2 only has %3 bits"

What do you think?

-Chris