On Tue, May 2, 2017 at 8:09 PM, Amir Goldstein <amir73il@xxxxxxxxx> wrote: ... >>> > Non-uniqueue uuids are an absolute no-go. >>> >>> I'm not sure I follow your specific concern here. >>> Surely you are not proposing to get rid of the nouuid >>> mount option, are you? So what's the point of hiding >>> the fact that there are 2 mounted filesystems with the >>> same uuid from VFS? >> >> Because it breaks people using s_uuid. ... >> >> It's not. The whole point of exporting s_uuid is to have a uniqueue >> identifier for a superblock. If it's not actually uniqueue there is >> no point in having or using it. > > Well it's useful for my use case. I need s_uuid as a sanity check > and IIUC, so does cleancache. > I don't know about EVM/IMA, but apparently, it did well > with pseudo uniqueness so far. > Mimi, Can you please comment on the implication of starting to publish sb->s_uuid by xfs on EVM/IMA. Are there any persistent integrity signatures that could be broken, because an xfs filesystem once published zeroes for sb->s_uuid and will start to publish non-zero sb->s_uuid? BTW, the same question applies w.r.t ubifs, which are also going to start publishing sb->s_uuid, although I don't if there are ubifs+evm users. Thanks, Amir.
