Hello, I have been testing the latest RTC drivers on an OpenWrt build on a MTK7688 SOC and I got an infinite loop on boot that would block the entire system. Basically, if I don't have a RTC connected, any read would return 0xFF. That may be also be a problem of the I2C driver, that instead of generating an error just returns that value. But this may happen in other cases, like if we add an EEPROM or another IC with the same address as the RTC, or the RTC IC fails in some unpredictable way. The current implementation of that piece of code has some bad practices that lead to infinite loops due to this error. Since on these RTCs the byte at address 0x00 will never be 0xFF (check the datasheet), I added a sanity check for this special case. A better way to do this would be to add a retry counter when setting the CH bit, but that's an even more edge case that will probably never happen. The same infinite loop will happen on other RTC ICs that do a similar check in other parts of code, but I don't have them so I can't edit the code and test it. The code has been tested and works; please consider merging it. Thanks, Best regards Cristiano Griletti - Mastro Gippo
Attachment:
0001-Fixed-infinite-loop-if-RTC-is-broken-disconnected.patch
Description: Binary data
