Bug #72335 [Opn->Csd]: Misoptimize due to type narrowing

Edit report at https://bugs.php.net/bug.php?id=72335&edit=1 ID: 72335 Updated by: dmitry@php.net Reported by: nikic@php.net Summary: Misoptimize due to type narrowing -Status: Open +Status: Closed Type: Bug Package: opcache PHP Version: master-Git-2016-06-04 (Git) -Assigned To: +Assigned To: dmitry Block user comment: N Private report: N Previous Comments: ------------------------------------------------------------------------ [2016-06-06 06:33:47] laruence@php.net Added negative case: <?php function test() { $b = false; $x = -(1<<53) -1; do { $x = 1.0 * ($x + (1<<53)); } while ($b); return $x; } var_dump(test()); ------------------------------------------------------------------------ [2016-06-06 06:02:06] laruence@php.net something like: https://gist.github.com/laruence/f1d62308ff7dc9557f934b8c2e444646 ------------------------------------------------------------------------ [2016-06-06 05:40:34] laruence@php.net we should check if there is any precision losing while converting long to double. ------------------------------------------------------------------------ [2016-06-04 21:31:47] nikic@php.net Description: ------------ We perform (long|double)->double narrowing even if chains of operations that only eventually result in a double are involved. This can lead to wrong results because parts of the calculation that would have been performed on integers are performed on doubles instead, with the corresponding loss in precision. We need to either not narrow for such chains or ensure that the transformation is safe. Test script: --------------- <?php function test() { $b = false; $x = (1<<53)+1; do { $x = 1.0 * ($x - (1<<53)); } while ($b); return $x; } var_dump(test()); Expected result: ---------------- float(1) Actual result: -------------- float(0) ------------------------------------------------------------------------ -- Edit this bug report at https://bugs.php.net/bug.php?id=72335&edit=1