I'm glad to see that commit:
c9bfb460c3e4 c9bfb460c3e4da2462e16b0f0b200990b36b1dd2 "s390/perf: obtain 
sie_block from the right address"
is tagged (upstream) for stable updates with:
Cc: sta...@vger.kernel.org

With that, this commit will automatically be picked up by the Ubuntu kernel 
teams
"Focal update: v5.4.xxx upstream stable release" process.

This hasn't happened yet with the latest ticket:
"Focal update: v5.4.191 upstream stable release" - LP#1976116
but will be soon.

This LP bug will be used for tracking the status.

** Tags added: tracking-upstream-stable

** Changed in: linux (Ubuntu)
   Importance: Undecided => High

** Tags removed: tracking-upstream-stable
** Tags added: upstream-stable

-- 
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/1977962

Title:
  [UBUNTU 21.10] s390/perf: obtain sie_block from the right address

Status in Ubuntu on IBM z Systems:
  New
Status in linux package in Ubuntu:
  New

Bug description:
  Description:   s390/perf: obtain sie_block from the right address

  Symptom:       Read from unitialized stack memory, incorrect guest 
                 perf data reported, possibly system crash.

  Problem:       When perf is instructed to record a software-based event and
                 to trace the guest, i.e. using the following command:

                 perf kvm --guest --guestvmlinux=endless.elf top -e cpu-clock
                 
                 The reported guest instruction pointer is inaccurate or
                 the system may crash.

                 This is caused by the sie_block being taken from empty1[0]
                 in sie_block() in arch/s390/kernel/perf_event.c, but since 
                 1179f170b6f0 ("s390: fix fpu restore in entry.S")
                 it resides at empty1[1].

  Solution:      Obtain sie_block from the right address in sie_block().

  Reproduction:  1. Start a KVM guest executing an endless loop.
                 2. Determine the instruction address of the loop.
                 3. Start perf recording for any software event and trace 
                    the guest, for example using this command:

                    perf kvm --guest --guestvmlinux=endless.elf top -e
  cpu-clock

                 4. The reported guest instruction addresses don't match
                    the loop's instruction address or the system crashes.

  Upstream-ID:   c9bfb460c3e4da2462e16b0f0b200990b36b1dd2

  BBPF Probl.ID: 198450
  Preventive:    yes
  Reported:      -
  SupportTicket: -
  Reference:     -
  Date:          2022-06-02
  Author:        n...@linux.ibm.com
  Component:     kernel
  Link:          
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=c9bfb460c3e4da2462e16b0f0b200990b36b1dd2

  Problem occurs / affects impish and jammy

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu-z-systems/+bug/1977962/+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