Every version of the man-pages read(2) page in the git repository appears to have the following language:

"POSIX allows a read() that is interrupted after reading some data to return -1 (with errno set to EINTR) or to return the number of bytes already read."

But POSIX.1-2001 [1] and later versions of the standard [2] explicitly prohibit returning -1 EINTR in this case:

"If a read() is interrupted by a signal after it has successfully read some data, it shall return the number of bytes read."

The rationale explains:

"Previous versions of IEEE Std 1003.1-2001 allowed two very different behaviors with regard to the handling of interrupts. In order to minimize the resulting confusion, it was decided that IEEE Std 1003.1-2001 should support only one of these behaviors. Historical practice on AT&T-derived systems was to have read() and write() return -1 and set errno to [EINTR] when interrupted after some, but not all, of the data requested had been transferred. However, the U.S. Department of Commerce FIPS 151-1 and FIPS 151-2 require the historical BSD behavior, in which read() and write() return the number of bytes actually transferred before the interrupt. If -1 is returned when any data is transferred, it is difficult to recover from the error on a seekable device and impossible on a non-seekable device. Most new implementations support this behavior. The behavior required by IEEE Std 1003.1-2001 is to return the number of bytes transferred."

Either this text in the manpage should be deleted, or it should be updated to reflect that the standard has changed (and perhaps to note which historical implementations return -1 EINTR when interrupted after a partial read).

[1] http://pubs.opengroup.org/onlinepubs/009695399/functions/read.html
[2] http://pubs.opengroup.org/onlinepubs/9699919799/functions/read.html