Messages in this thread | | | Date | Fri, 06 Apr 2012 08:36:03 +0400 | From | Konstantin Khlebnikov <> | Subject | Re: [PATCH 6/7] mm: kill vma flag VM_EXECUTABLE |
| |
Linus Torvalds wrote: > On Thu, Apr 5, 2012 at 2:44 PM, Matt Helsley<matthltc@us.ibm.com> wrote: >> >> I don't think the definition of an ABI is whether there's documentation >> for it. It's whether the interface is used or not. At least that's the >> impression I've gotten from reading Linus' rants over the years. > > Yes. > > That said, I *do* have some very dim memory of us having had real > issues with the /proc/<pid>/exe thing and having regressions due to > holding refcounts to executables that were a.out binaries and not > demand-loaded. And people wanting to unmount filesystems despite the > binaries being live. > > That said, I suspect that whatever issues we used to have with that > are pretty long gone. I don't think people use non-mmap'ed binaries > any more. So I think we can try it and see. And revert if somebody > actually notices and has problems.
Instead of tracking count of vma with VM_EXECUTABLE bit we can track count of vma with vma->vm_file == mm->exe_file, this will be nearly the same behaviour. This was in early version of my patch, but I prefer to go deeper. So, we can revert it without introducing VM_EXECUTABLE again.
> > Linus
| |