On Mon, Sep 29, 2014 at 10:20:57PM +0100, Al Viro wrote: > On Tue, Sep 30, 2014 at 12:26:59AM +0400, Dmitry V. Levin wrote: > > > > Again, we do not break userland. > > > > Occasionally, we do. If you say there was no chroot prefix removal in the > > kernel before, and there seems to be a prefix removal for at least 5 years, > > then there must have been a break some years ago that went unnoticed. > > D'oh! Yes, we did. And yes, it was quite a few years ago, actually - back > in 2000. I plead being low on sleep and lower on caffeine... > > For a moment I was even afraid that it was my own doing, but no - it came > in https://lkml.org/lkml/2000/1/16/54. Werner Almesberger. IIRC, I hadn't > realized that it was going to cause fun problems back then - basically, > by that point 2.3 and 2.2 had already diverged a lot and quite a few things > got written off as "oh, well - upgrade from 2.2 to 2.4 is going to be not > far from building the system from scratch anyway, and nobody sane would be > using 2.3 on anything other than scratch filesystem - IDE broken more often > than not, memory corruption aplenty, etc." Thanks, I won't be able to locate the change that fast. It also means that the phrase "show mountpoints relative to the root directory" I used in the commit message is irrelevant and should be removed to avoid confusion. > My apologies. Hmm... Filtering the out-of-root ones out is probably OK, > but let's put that into -next after 3.18-rc1 and see if anyone yells. BTW, the cleanup patch (vfs: cleanup show_mountinfo) could be safely applied anytime - it's completely harmless. -- ldv -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
