On 3/21/06, Jan Engelhardt <[EMAIL PROTECTED]> wrote: > I have found out that when linking a userspace application with -lpthread, > then for some reason, a device driver's read() routine suddenly has > IO_NDELAY set in flags. Bug?
the current pthread library fakes out concurrency by re-implementing read/write/... and trying to switch threads whenever a syscall would block. so it's intentional, but also has unfortunate side effects for some programs.