On Thu, Jan 27, 2022 at 9:15 AM Matthew Wilcox <willy@xxxxxxxxxxxxx> wrote: > > On Thu, Jan 27, 2022 at 09:03:36AM +1100, Daniel Black wrote: > > Is it going to be reasonable to expect fcntl(fd, F_SETFL, O_DIRECT) to > > return EINVAL if O_DIRECT isn't supported? > > That is a reasonable expectation. I can't guarantee that we won't have > bugs, of course ... Ha, sure. I've begun to https://kernelci.org/ options to try to catch at least some of them pre-release. > > My problem it seems, I'll see what I can do to get back to using real > > filesystems more. > > Heh. I know Hugh is looking at "supporting" O_DIRECT on tmpfs, at least > for his internal testing. Not sure what his plans are for merging > that support. I'd be happy to see it in that it will remove a long standing cludge bit of weakly commented user space code in the fullness of supported kernel lifespans, but no great urgency. Thanks Matthew.
