linux/arch/arm64
Linus Torvalds 81429eb8d9 arm64 fix for 5.4
- Ensure PAN is re-enabled following user fault in uaccess routines
 -----BEGIN PGP SIGNATURE-----
 
 iQFEBAABCgAuFiEEPxTL6PPUbjXGY88ct6xw3ITBYzQFAl3WdTEQHHdpbGxAa2Vy
 bmVsLm9yZwAKCRC3rHDchMFjNE43B/9ow5C7aV1Jmb9dk3zfkDkVuviCkv9STpNW
 YosKSAHwGvsfmBd/f+kd/ipqnt/fBLbXHqRderbayDW7xundCP8Qgl51N2RwJ92O
 si3rPHW62xjbv0U3XqY9h74wH4jfox3KAT8c5EBcLuZtmV3gTb1txWDQ0PfQZMpN
 0cmfTRPHB7LKJ/hupEyq6iVPYx+mV1ajenW1uzBOQhu+c8jC5Uf9kDn/D93rCYCg
 7NL88odhP0Oyt6HDcR5LB+nl7fz+q0jXzxyF8qHDWNitV4aovU9rLMpDpn4YVvYd
 DcFEyGdwYSxExLFrBuvmevZ7+2biT3qWQcPuNQGCcqWQSu38ULFK
 =r7nI
 -----END PGP SIGNATURE-----

Merge tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux

Pull arm64 fix from Will Deacon:
 "Ensure PAN is re-enabled following user fault in uaccess routines.

  After I thought we were done for 5.4, we had a report this week of a
  nasty issue that has been shown to leak data between different user
  address spaces thanks to corruption of entries in the TLB. In
  hindsight, we should have spotted this in review when the PAN code was
  merged back in v4.3, but hindsight is 20/20 and I'm trying not to beat
  myself up too much about it despite being fairly miserable.

  Anyway, the fix is "obvious" but the actual failure is more more
  subtle, and is described in the commit message. I've included a fairly
  mechanical follow-up patch here as well, which moves this checking out
  into the C wrappers which is what we do for {get,put}_user() already
  and allows us to remove these bloody assembly macros entirely. The
  patches have passed kernelci [1] [2] [3] and CKI [4] tests over night,
  as well as some targetted testing [5] for this particular issue.

  The first patch is tagged for stable and should be applied to 4.14,
  4.19 and 5.3. I have separate backports for 4.4 and 4.9, which I'll
  send out once this has landed in your tree (although the original
  patch applies cleanly, it won't build for those two trees).

  Thanks to Pavel Tatashin for reporting this and Mark Rutland for
  helping to diagnose the issue and review/test the solution"

* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
  arm64: uaccess: Remove uaccess_*_not_uao asm macros
  arm64: uaccess: Ensure PAN is re-enabled after unhandled uaccess fault
2019-11-21 12:15:24 -08:00
..
boot ARM: SoC fixes 2019-11-10 13:41:59 -08:00
configs arm64/ARM: configs: Change CONFIG_REMOTEPROC from m to y 2019-09-30 11:48:00 -07:00
crypto crypto: arm64/aes-neonbs - implement ciphertext stealing for XTS 2019-09-09 17:35:39 +10:00
include arm64 fix for 5.4 2019-11-21 12:15:24 -08:00
kernel arm64: apply ARM64_ERRATUM_843419 workaround for Brahma-B53 core 2019-11-01 10:47:37 +00:00
kvm KVM/arm fixes for 5.4, take #2 2019-10-22 13:31:29 +02:00
lib arm64: uaccess: Remove uaccess_*_not_uao asm macros 2019-11-20 18:51:54 +00:00
mm arm64: mm: fix inverted PAR_EL1.F check 2019-10-16 09:58:03 -07:00
net arm64: bpf: optimize modulo operation 2019-09-03 15:44:40 +02:00
xen xen/efi: have a common runtime setup function 2019-10-02 10:31:07 -04:00
Kbuild arm64: add arch/arm64/Kbuild 2019-08-21 18:47:15 +01:00
Kconfig Merge branch 'errata/tx2-219' into for-next/fixes 2019-10-17 13:42:42 -07:00
Kconfig.debug treewide: Add SPDX license identifier - Makefile/Kconfig 2019-05-21 10:50:46 +02:00
Kconfig.platforms arm64: exynos: Enable exynos-chipid driver 2019-09-04 22:43:26 +02:00
Makefile arm64: Kconfig: Make CONFIG_COMPAT_VDSO a proper Kconfig option 2019-10-07 13:32:15 +01:00