[Cryptography] Is glibc right on randomness

John Gilmore gnu at toad.com
Wed Dec 7 15:29:57 EST 2016


> Yes, there is no getrandom in libc.  That's because glibc developers
> are being.... glibc, as I said before.  There has been discussion
> about maybe the kernel developers should support a libinux.a library
> that would allow us to bypass glibc when they are being non-helpful.

Here is the glibc bug report:

  https://sourceware.org/bugzilla/show_bug.cgi?id=17252

Perhaps some real users who want to call getrandom() in their portable
code should pile on?  The holdup seems to be:

  joseph at codesourcery.com 2014-08-11 15:46:58 UTC

  If you wish to propose a new API for glibc, please make the proposal on 
  libc-alpha, taking an active lead in driving the discussion to consensus.

  joseph at codesourcery.com 2015-02-24 16:27:07 UTC

  If you want progress on this, take a lead in the general discussion
  on libc-alpha of when glibc should provide bindings to Linux kernel
  syscalls, seeking to understand the differences of views expressed,
  find common ground and drive the discussion to consensus.  Once we
  have agreed principles on bindings for syscalls, then we can
  consider which new or old syscalls should have such bindings added
  under those principles.

However, the bug is assigned to Florian Weimer, not to Joseph.  Still,
no progress has been made among these "maintainers" in the two years
since the kernel syscall was added.

	John


More information about the cryptography mailing list