[light] Fix uint16_t overflow in color_uncorrect_channel_

When max_brightness and local_brightness_ are small but non-zero, the
intermediate (uncorrected / max_brightness) * 255 can exceed 65535
before the std::min(255) clamp runs, producing an incorrect low result.
Widen intermediates to uint32_t. Copilot review catch on #15727.
This commit is contained in:
J. Nick Koston
2026-04-13 15:08:36 -10:00
parent 827afb0e98
commit 76c8eeede9
@@ -31,9 +31,11 @@ Color ESPColorCorrection::color_uncorrect(Color color) const {
uint8_t ESPColorCorrection::color_uncorrect_channel_(uint8_t value, uint8_t max_brightness) const {
if (max_brightness == 0 || this->local_brightness_ == 0)
return 0;
uint16_t uncorrected = this->gamma_uncorrect_(value) * 255UL;
uint16_t res = ((uncorrected / max_brightness) * 255UL) / this->local_brightness_;
return (uint8_t) std::min(res, uint16_t(255));
// Use 32-bit intermediates: when max_brightness and local_brightness_ are small but non-zero,
// (uncorrected / max_brightness) * 255 can exceed 65535 before the std::min(255) clamp runs.
uint32_t uncorrected = this->gamma_uncorrect_(value) * 255UL;
uint32_t res = ((uncorrected / max_brightness) * 255UL) / this->local_brightness_;
return static_cast<uint8_t>(std::min(res, uint32_t(255)));
}
} // namespace esphome::light