On Fri Sep 6 15:55:34 2024 +0000, Grigory Vasilyev wrote:
@nsivov This is exactly the correct behavior, in case of interruption by the kernel or blocking of the file, we try to read it again. It is unlikely that we will get stuck in an infinite blocking. I would prefer to keep the current behavior.
EAGAIN only related to nonblocking reads and file read is not interruptable. You might also look at other usages of read on the Unix side, we certainly don't make a loop for each file read() and don't have to.