<div dir="ltr">FE warning is good, but the case with the constant is for demonstration purposes only :)<br><br>The standard doesn't make any connection between left and right hand side types, neither limitations for rhs type size. And it matches to how clang interprets this:<div><br></div><div><div> | `-VarDecl 0x7ff53d0de998 <col:5, col:22> col:9 used res 'int' cinit</div><div> | `-BinaryOperator 0x7ff53d0dea38 <col:15, col:22> 'int' '>>'</div><div> | |-IntegerLiteral 0x7ff53d0de9f8 <col:15> 'int' 123</div><div> | `-IntegerLiteral 0x7ff53d0dea18 <col:22> 'unsigned long long' 4294967296</div><div><br></div><div>Here's 5.8.2 (from C++17):<br>The operands shall be of integral or unscoped enumeration type and integral promotions are performed. The type of the result is that of the promoted left operand. The behavior is undefined if the right operand is negative, or greater than or equal to the length in bits of the promoted left operand.<br><div><div class="gmail-page" title="Page 146"><div class="gmail-layoutArea"><div class="gmail-column">
</div>
</div>
</div></div></div></div><div><br></div><div>So I treat it as ubsan missing UB in this case by working on the IR with truncation for RHS, as the standard doesn't assume this truncation.</div><div><br></div><div>Dmitry.</div></div><div class="gmail_extra"><br><div class="gmail_quote">On Tue, Dec 13, 2016 at 2:27 AM, Kostya Serebryany <span dir="ltr"><<a href="mailto:kcc@google.com" target="_blank">kcc@google.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div>Is it actually undefined behavior? </div><div>What does the standard say about " int32_value >> int64_value" ?</div><div>If int64_value should be first cast to int32, then there is no UB in:</div><div><div style="font-size:12.8px"> res += 123 >> l;</div></div><div style="font-size:12.8px"><br></div><div>As for the first buggy line, there is a FE warning: </div><div><div>b.c:6:19: warning: shift count >= width of type [-Wshift-count-overflow]</div><span class=""><div> int res = 123 >> 0x100000000ULL;</div></span></div><div><br></div><div>--kcc </div><div><br></div><div><br></div><div><br></div></div><div class="HOEnZb"><div class="h5"><div class="gmail_extra"><br><div class="gmail_quote">On Mon, Dec 12, 2016 at 2:46 PM, Dmitry Babokin <span dir="ltr"><<a href="mailto:babokin@gmail.com" target="_blank">babokin@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">Hello,<div><br></div><div>I'm using ubsan together with creduce to ensure that reduced test doesn't have UB. And I've noticed that ubsan is not detecting UB in case of large shift amount, particularly when rhs of shift has larger type that lhs. To reproduce, rhs also has to have the value that is non-zero, but after truncation to lhs type the value becomes zero.</div><div><br></div><div>Consider the following example.</div><div><br></div><div><div>#include <stdio.h></div><div><br></div><div>// 1 is in 33rd bit.</div><div>unsigned long long l = 0x100000000ULL;</div><div><br></div><div>int main() {</div><div> // Ubsan doesn't fire</div><div> int res = 123 >> 0x100000000ULL;</div><div> // Ubsan doesn't fire</div><div> res += 123 >> l;</div><div> // Ubsan does fire</div><div> res += 123ULL >> l;</div><div> printf("result = %d\n", res);</div><div> return 0;</div><div>}</div></div><div><br></div><div>Changing the constant to the value, which fit 32 bits makes ubsan firing.</div><div><br></div><div>I understand where the problem comes from - LLVM IR definition requires both operands of the shift to be of the same integer type. And ubsan actually check already truncated value. But it doesn't match C/C++ standard definition of UB in shift operation.</div><div><br></div><div>Is it possible to fix in current infrastructure? Should I file a bug for this problem?</div><div><br></div><div>Note, it's not a theoretical problem, it's very practical one, which pops up during automatic test reduction relatively frequently for me.</div><span class="m_-1119274409795271357HOEnZb"><font color="#888888"><div><br></div><div>Dmitry.</div></font></span></div>
</blockquote></div><br></div>
</div></div></blockquote></div><br></div>