On 4/21/22 07:43, Rich Felker wrote: > On Thu, Apr 21, 2022 at 08:52:59AM +0200, Geert Uytterhoeven wrote: >> On Thu, Apr 21, 2022 at 1:53 AM Greg Ungerer <gerg@xxxxxxxxxxxxxx> wrote: >> > On 21/4/22 00:58, Eric W. Biederman wrote: >> > > In a recent discussion[1] it was reported that the binfmt_flat library >> > > support was only ever used on m68k and even on m68k has not been used >> > > in a very long time. >> > > >> > > The structure of binfmt_flat is different from all of the other binfmt >> > > implementations becasue of this shared library support and it made >> > > life and code review more effort when I refactored the code in fs/exec.c. >> > > >> > > Since in practice the code is dead remove the binfmt_flat shared libarary >> > > support and make maintenance of the code easier. >> > > >> > > [1] https://lkml.kernel.org/r/81788b56-5b15-7308-38c7-c7f2502c4e15@xxxxxxxxxxxxxx >> > > Signed-off-by: "Eric W. Biederman" <ebiederm@xxxxxxxxxxxx> >> > > --- >> > > >> > > Can the binfmt_flat folks please verify that the shared library support >> > > really isn't used? >> > >> > I can definitely confirm I don't use it on m68k. And I don't know of >> > anyone that has used it in many years. >> > >> > >> > > Was binfmt_flat being enabled on arm and sh the mistake it looks like? >> >> I think the question was intended to be >> >> Was *binfmt_flat_shared_flat* being enabled on arm and sh the >> mistake it looks like? > > Early in my work on j2, I tried to research the history of shared flat > support on sh, and it turned out the mainline tooling never even > supported it, and the out-of-line tooling I eventually found was using > all sorts of wrong conditionals for how it did the linking and elf2flt > conversion, e.g. mere presence of any PIC-like relocation in any file > made it assume the whole program was PIC-compatible. There's no way > that stuf was ever used in any meaningful way. It just didn't work. > > Quickly dropped that and got plain ELF (no shared text/xip, but no > worse than the existing flat support) working, and soon after, FDPIC. > > The whole binfmt_flat ecosystem is a mess with no good reason to > exist. FYI when I had to come up to speed on this in 2014 I did a writeup on my own research: https://landley.net/notes-2014.html#07-12-2014 The lack of a canonical "upstream" elf2flt repository was probably the biggest problem at the time. (There's a reason I grabbed fdpic hard and tried to make that work everywhere.) > Rich Rob
