On 25/03/16 06:31AM, Aditya Garg wrote: > > > > On 16 Mar 2025, at 9:01 AM, Ernesto A. Fernández <ernesto.mnd.fernandez@xxxxxxxxx> wrote: > > > > Hi Ethan, > > > > I'm happy to see your enthusiasm for my driver but, if you want to help, I > > think you should simply send the changes you have in mind to the out-of-tree > > repo. That way you'll start learning the codebase while I can review your > > work and run xfstests for you. Filesystems are very dangerous things; I've > > probably done a lot of damage myself back in the day trying to help out with > > the hfs drivers. > > > > As for upstreaming, the driver still has a few rough edges, but I don't > > think that's the real reason I never tried to submit. I'm just no longer > > confident that filesystem compatibility is a reasonable goal, and I don't > > expect much interest from reviewers. There are too many risks, and too many > > hardware restrictions these days; regular users have much easier (even if > > slower) ways to move their files around. Other uses exist of course (like > > Aditya can explain), but they are a bit esoteric. Of course if upstream > > people disagree, and they do want the apfs support, I will be glad to > > prepare a patch series. > > As far as I can tell, in case of upstreaming, making the FS readonly is worth it. Definitely. I agree. From my understanding, and Ernesto, correct me if I am wrong, but write is not explicity enabled unless mounted as so, unless the module is compiled with CONFIG_APFS_RW_ALWAYS. I enabled this by default in the previous patch set, but we could definitely add a Kconfig /Kbuild option for it. > > Writes, I won’t comment. Maybe add option to force them, just like it is rn, old just remove > the whole code. The second option IMO would require quite a lot of work from your side. >
