rhia...@falu.nl (Rhialto) writes: >On Thu 03 Jun 2021 at 09:12:52 -0000, Michael van Elst wrote: >> What happens: >>=20 >> namei() return EEXIST when it works on a CREATE operation and >> crosses a mountpoint. >>=20 >> procfs cheats in that the directory nodes like /proc/$pid/fd are >> served by procfs and have a v_mount pointing to /proc. But the >> file descriptors /proc/$pid/fd/$fd are magic aliases and you get >> the real vnode of that open file which belongs to a different >> mount, e.g. /dev/pts for a pty or a ffs filesystem for a regular file. >>=20 >> For namei() that's crossing a mountpoint -> EEXIST.
>I guess the question for the short term is: will it work as intended >when the O_CREAT flag is omitted from the call (since it is redundant >in this case anyway)? I guess so. But AFAIK Samba has refactored the code so that O_CREAT is always used when writing a file. You'd need to add a hack that detects the procfs writes to strip O_CREAT for only that case.
