Note: the issue is only reproducible on a system with more than 1 CPU.
It fails on the "Cross-CPU LDT invalidation" testcase.

-- 
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/1755817

Title:
  Segmentation fault in ldt_gdt_64

Status in linux package in Ubuntu:
  Incomplete
Status in linux source package in Trusty:
  Fix Committed

Bug description:
  == SRU Justification ==
  The ldt_gdt_64 x86 selftest segfaults with the currently released Trusty 3.13 
kernel. The commit that introduced the segfault is aeb315d60afe ("x86/ldt: Make 
modify_ldt synchronous").

  == Fix ==
  Upstream commit 8ff5bd2e1e27 ("x86/signal/64: Fix SS if needed when 
delivering a 64-bit signal"). This commit was found by doing a reverse git 
bisect of the upstream kernel (i.e., when did the test stop segfaulting).
  The backport of the commit is a simple context adjustment. The second commit 
is a pre-requisite which simply renames some defines (no functional changes).

  == Regression Potential ==
  Low. The commit is very small and isolated and the code path is only executed 
in special circumstances (and for x86 only). I built a test kernel and ran the 
whole set of x86 selftests and perf NMI test for several hours to verify 
stability.

  == Test Case ==
  Run the ldt_gdt_64 x86 selftets from a current upstream kernel source. The 
test segfaults consistently.

  Original bug description:

  Trusty 3.13 segfaults when running ldt_gdt_64 from the kernel's x86
  selftests.

  git bisect revealed that the following commit introduced the issue:

  commit aeb315d60afee129d32558f4a4b356eec2e7da7b
  Author: Andy Lutomirski <l...@kernel.org>
  Date:   Thu Jul 30 14:31:32 2015 -0700

      x86/ldt: Make modify_ldt synchronous

      CVE-2017-5754

      commit 37868fe113ff2ba814b3b4eb12df214df555f8dc upstream.

      modify_ldt() has questionable locking and does not synchronize
      threads.  Improve it: redesign the locking and synchronize all
      threads' LDTs using an IPI on all modifications.

      This will dramatically slow down modify_ldt in multithreaded
      programs, but there shouldn't be any multithreaded programs that
      care about modify_ldt's performance in the first place.

      This fixes some fallout from the CVE-2015-5157 fixes.

      Signed-off-by: Andy Lutomirski <l...@kernel.org>
      Reviewed-by: Borislav Petkov <b...@suse.de>
      Cc: Andrew Cooper <andrew.coop...@citrix.com>
      Cc: Andy Lutomirski <l...@amacapital.net>
      Cc: Boris Ostrovsky <boris.ostrov...@oracle.com>
      Cc: Borislav Petkov <b...@alien8.de>
      Cc: Brian Gerst <brge...@gmail.com>
      Cc: Denys Vlasenko <dvlas...@redhat.com>
      Cc: H. Peter Anvin <h...@zytor.com>
      Cc: Jan Beulich <jbeul...@suse.com>
      Cc: Konrad Rzeszutek Wilk <konrad.w...@oracle.com>
      Cc: Linus Torvalds <torva...@linux-foundation.org>
      Cc: Peter Zijlstra <pet...@infradead.org>
      Cc: Sasha Levin <sasha.le...@oracle.com>
      Cc: Steven Rostedt <rost...@goodmis.org>
      Cc: Thomas Gleixner <t...@linutronix.de>
      Link: 
http://lkml.kernel.org/r/4c6978476782160600471bd865b318db34c7b628.1438291540.git.l...@kernel.org
      Signed-off-by: Ingo Molnar <mi...@kernel.org>
      Signed-off-by: Jiri Slaby <jsl...@suse.cz>
      (cherry picked from commit 62fc7228f8cc8c89ecbd37008a0495ac28e41c5c)
      Signed-off-by: Juerg Haefliger <juerg.haefli...@canonical.com>
      Signed-off-by: Stefan Bader <stefan.ba...@canonical.com>

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1755817/+subscriptions

-- 
Mailing list: https://launchpad.net/~kernel-packages
Post to     : kernel-packages@lists.launchpad.net
Unsubscribe : https://launchpad.net/~kernel-packages
More help   : https://help.launchpad.net/ListHelp

Reply via email to