[SRU][Trusty][PULL] Cleanup Meltdown/Spectre implementation (LP: #1779848)

classic Classic list List threaded Threaded
9 messages Options
Reply | Threaded
Open this post in threaded view
|

[SRU][Trusty][PULL] Cleanup Meltdown/Spectre implementation (LP: #1779848)

Juerg Haefliger
BugLink: https://bugs.launchpad.net/bugs/1779848

== SRU Justification ==

Ubuntu kernels contain Meltdown and Spectre mitigations that are largely based on embargoed patches but what eventually landed in upstream is different in some places. We should clean up the different kernels to bring them closer in line with upstream.

== Fix ==

Add missing upstream patches.

== Regression Potential ==

Medium. The patches have been in upstream for quite a while now and are baked in but some of the backporting is not completely trivial and without risk.

== Test Case ==

TBD.

*** BLURB HERE ***

Signed-off-by: Juerg Haefliger <[hidden email]>
---

The following changes since commit cefa65a1685282dcbadad7c914d4d8e89428a5eb:

  UBUNTU: Ubuntu-3.13.0-161.211 (2018-10-03 16:41:43 +0200)

are available in the Git repository at:

  git://git.launchpad.net/~juergh/+git/trusty-linux lp1779848

for you to fetch changes up to 8cccec097d1fb83c610f9dddd47827fb6d3237aa:

  UBUNTU: SAUCE: x86/speculation: Cleanup IBRS runtime control handling (2018-10-09 10:46:55 +0200)

----------------------------------------------------------------
Arnd Bergmann (1):
      x86/pti: Mark constant arrays as __initconst

Ben Hutchings (1):
      xen: Add xen_arch_suspend()

Borislav Petkov (2):
      x86/nospec: Fix header guards names
      x86/bugs: Drop one "mitigation" from dmesg

Dan Carpenter (1):
      x86/spectre: Fix an error message

Dave Hansen (1):
      x86/Documentation: Add PTI description

David Woodhouse (5):
      x86/cpufeatures: Add AMD feature bits for Speculation Control
      x86/speculation: Add basic IBPB (Indirect Branch Prediction Barrier) support
      x86/cpufeatures: Clean up Spectre v2 related CPUID flags
      x86/cpuid: Fix up "virtual" IBRS/IBPB/STIBP feature bits on Intel
      x86/speculation: Use IBRS if available before calling into firmware

Dou Liyang (1):
      x86/spectre: Check CONFIG_RETPOLINE in command line parser

Ingo Molnar (1):
      x86/speculation: Move firmware_restrict_branch_speculation_*() from C to CPP

Juerg Haefliger (14):
      Revert "x86/cpu/AMD: Make the LFENCE instruction serialized"
      UBUNTU: SAUCE: x86/cpufeatures: Reorder spectre-related feature bits
      UBUNTU: SAUCE: x86/msr: Fix formatting of msr-index.h
      UBUNTU: SAUCE: x86/msr: Rename MSR spec control feature bits
      UBUNTU: SAUCE: x86/speculation: Use synthetic bits for IBRS/IBPB/STIBP
      UBUNTU: SAUCE: x86/cpufeatures: Disentangle MSR_SPEC_CTRL enumeration from IBRS
      UBUNTU: SAUCE: x86/bugs: Fix re-use of SPEC_CTRL MSR boot value
      UBUNTU: SAUCE: Move SSBD feature detection to common code
      UBUNTU: SAUCE: x86/speculation: Move vendor specific IBRS/IBPB control code
      UBUNTU: SAUCE: x86/speculation: Query individual feature flags when reloading microcode
      UBUNTU: SAUCE: x86/pti: Evaluate X86_BUG_CPU_MELTDOWN when pti=auto
      UBUNTU: SAUCE: x86/speculation: Make use of indirect_branch_prediction_barrier()
      UBUNTU: SAUCE: x86/speculation: Cleanup IBPB runtime control handling
      UBUNTU: SAUCE: x86/speculation: Cleanup IBRS runtime control handling

Juergen Gross (1):
      x86/xen: Zero MSR_IA32_SPEC_CTRL before suspend

KarimAllah Ahmed (1):
      x86/spectre: Simplify spectre_v2 command line parsing

Tom Lendacky (2):
      x86/cpu/AMD: Make LFENCE a serializing instruction
      x86/cpu/AMD: Use LFENCE_RDTSC in preference to MFENCE_RDTSC

zhenwei.pi (1):
      x86/pti: Document fix wrong index

 Documentation/kernel-parameters.txt   |  21 ++--
 Documentation/x86/pti.txt             | 186 +++++++++++++++++++++++++++++++++
 arch/x86/include/asm/apm.h            |   6 ++
 arch/x86/include/asm/cpufeature.h     |  31 ++++--
 arch/x86/include/asm/efi.h            |   7 ++
 arch/x86/include/asm/nospec-branch.h  |  62 ++++++++++-
 arch/x86/include/asm/spec_ctrl.h      |  13 ++-
 arch/x86/include/uapi/asm/msr-index.h |  34 +++---
 arch/x86/kernel/acpi/cstate.c         |   6 +-
 arch/x86/kernel/cpu/amd.c             |  92 ++++++++--------
 arch/x86/kernel/cpu/bugs.c            | 169 ++++++++++++++++++++----------
 arch/x86/kernel/cpu/common.c          |  40 ++++++-
 arch/x86/kernel/cpu/intel.c           |  19 ----
 arch/x86/kernel/microcode_core.c      |  20 ----
 arch/x86/kernel/process.c             |  11 +-
 arch/x86/kernel/smpboot.c             |   6 +-
 arch/x86/kvm/svm.c                    |  11 +-
 arch/x86/kvm/vmx.c                    |   3 +-
 arch/x86/mm/kaiser.c                  |   7 +-
 arch/x86/mm/tlb.c                     |   5 +-
 arch/x86/xen/suspend.c                |  24 +++++
 drivers/xen/manage.c                  |   9 +-
 include/linux/smp.h                   |  87 ----------------
 include/xen/xen-ops.h                 |   1 +
 kernel/smp.c                          |  46 --------
 kernel/sysctl.c                       | 191 ++++++++++++++++++++--------------
 26 files changed, 672 insertions(+), 435 deletions(-)
 create mode 100644 Documentation/x86/pti.txt

--
kernel-team mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/kernel-team
Reply | Threaded
Open this post in threaded view
|

Re: [SRU][Trusty][PULL] Cleanup Meltdown/Spectre implementation (LP: #1779848)

Juerg Haefliger
On Tue,  9 Oct 2018 15:37:14 +0200
Juerg Haefliger <[hidden email]> wrote:

> BugLink: https://bugs.launchpad.net/bugs/1779848
>
> == SRU Justification ==
>
> Ubuntu kernels contain Meltdown and Spectre mitigations that are
> largely based on embargoed patches but what eventually landed in
> upstream is different in some places. We should clean up the
> different kernels to bring them closer in line with upstream.
>
> == Fix ==
>
> Add missing upstream patches.
>
> == Regression Potential ==
>
> Medium. The patches have been in upstream for quite a while now and
> are baked in but some of the backporting is not completely trivial
> and without risk.
>
> == Test Case ==
>
> TBD.
>
> *** BLURB HERE ***
Saving before sending would help :-( The blurb is:

Compile-tested all architectures. Ran Trusty ADT tests successfully
against a test kernel.

...Juerg
 

> Signed-off-by: Juerg Haefliger <[hidden email]>
> ---
>
> The following changes since commit
> cefa65a1685282dcbadad7c914d4d8e89428a5eb:
>
>   UBUNTU: Ubuntu-3.13.0-161.211 (2018-10-03 16:41:43 +0200)
>
> are available in the Git repository at:
>
>   git://git.launchpad.net/~juergh/+git/trusty-linux lp1779848
>
> for you to fetch changes up to
> 8cccec097d1fb83c610f9dddd47827fb6d3237aa:
>
>   UBUNTU: SAUCE: x86/speculation: Cleanup IBRS runtime control
> handling (2018-10-09 10:46:55 +0200)
>
> ----------------------------------------------------------------
> Arnd Bergmann (1):
>       x86/pti: Mark constant arrays as __initconst
>
> Ben Hutchings (1):
>       xen: Add xen_arch_suspend()
>
> Borislav Petkov (2):
>       x86/nospec: Fix header guards names
>       x86/bugs: Drop one "mitigation" from dmesg
>
> Dan Carpenter (1):
>       x86/spectre: Fix an error message
>
> Dave Hansen (1):
>       x86/Documentation: Add PTI description
>
> David Woodhouse (5):
>       x86/cpufeatures: Add AMD feature bits for Speculation Control
>       x86/speculation: Add basic IBPB (Indirect Branch Prediction
> Barrier) support x86/cpufeatures: Clean up Spectre v2 related CPUID
> flags x86/cpuid: Fix up "virtual" IBRS/IBPB/STIBP feature bits on
> Intel x86/speculation: Use IBRS if available before calling into
> firmware
>
> Dou Liyang (1):
>       x86/spectre: Check CONFIG_RETPOLINE in command line parser
>
> Ingo Molnar (1):
>       x86/speculation: Move firmware_restrict_branch_speculation_*()
> from C to CPP
>
> Juerg Haefliger (14):
>       Revert "x86/cpu/AMD: Make the LFENCE instruction serialized"
>       UBUNTU: SAUCE: x86/cpufeatures: Reorder spectre-related feature
> bits UBUNTU: SAUCE: x86/msr: Fix formatting of msr-index.h
>       UBUNTU: SAUCE: x86/msr: Rename MSR spec control feature bits
>       UBUNTU: SAUCE: x86/speculation: Use synthetic bits for
> IBRS/IBPB/STIBP UBUNTU: SAUCE: x86/cpufeatures: Disentangle
> MSR_SPEC_CTRL enumeration from IBRS UBUNTU: SAUCE: x86/bugs: Fix
> re-use of SPEC_CTRL MSR boot value UBUNTU: SAUCE: Move SSBD feature
> detection to common code UBUNTU: SAUCE: x86/speculation: Move vendor
> specific IBRS/IBPB control code UBUNTU: SAUCE: x86/speculation: Query
> individual feature flags when reloading microcode UBUNTU: SAUCE:
> x86/pti: Evaluate X86_BUG_CPU_MELTDOWN when pti=auto UBUNTU: SAUCE:
> x86/speculation: Make use of indirect_branch_prediction_barrier()
> UBUNTU: SAUCE: x86/speculation: Cleanup IBPB runtime control handling
> UBUNTU: SAUCE: x86/speculation: Cleanup IBRS runtime control handling
>
> Juergen Gross (1):
>       x86/xen: Zero MSR_IA32_SPEC_CTRL before suspend
>
> KarimAllah Ahmed (1):
>       x86/spectre: Simplify spectre_v2 command line parsing
>
> Tom Lendacky (2):
>       x86/cpu/AMD: Make LFENCE a serializing instruction
>       x86/cpu/AMD: Use LFENCE_RDTSC in preference to MFENCE_RDTSC
>
> zhenwei.pi (1):
>       x86/pti: Document fix wrong index
>
>  Documentation/kernel-parameters.txt   |  21 ++--
>  Documentation/x86/pti.txt             | 186
> +++++++++++++++++++++++++++++++++
> arch/x86/include/asm/apm.h            |   6 ++
> arch/x86/include/asm/cpufeature.h     |  31 ++++--
> arch/x86/include/asm/efi.h            |   7 ++
> arch/x86/include/asm/nospec-branch.h  |  62 ++++++++++-
> arch/x86/include/asm/spec_ctrl.h      |  13 ++-
> arch/x86/include/uapi/asm/msr-index.h |  34 +++---
> arch/x86/kernel/acpi/cstate.c         |   6 +-
> arch/x86/kernel/cpu/amd.c             |  92 ++++++++--------
> arch/x86/kernel/cpu/bugs.c            | 169
> ++++++++++++++++++++---------- arch/x86/kernel/cpu/common.c
> |  40 ++++++- arch/x86/kernel/cpu/intel.c           |  19 ----
> arch/x86/kernel/microcode_core.c      |  20 ----
> arch/x86/kernel/process.c             |  11 +-
> arch/x86/kernel/smpboot.c             |   6 +-
> arch/x86/kvm/svm.c                    |  11 +-
> arch/x86/kvm/vmx.c                    |   3 +-
> arch/x86/mm/kaiser.c                  |   7 +-
> arch/x86/mm/tlb.c                     |   5 +-
> arch/x86/xen/suspend.c                |  24 +++++
> drivers/xen/manage.c                  |   9 +-
> include/linux/smp.h                   |  87 ----------------
> include/xen/xen-ops.h                 |   1 +
> kernel/smp.c                          |  46 --------
> kernel/sysctl.c                       | 191
> ++++++++++++++++++++-------------- 26 files changed, 672
> insertions(+), 435 deletions(-) create mode 100644
> Documentation/x86/pti.txt

--
kernel-team mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/kernel-team

attachment0 (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: [SRU][Trusty][PULL] Cleanup Meltdown/Spectre implementation (LP: #1779848)

Stefan Bader-2
In reply to this post by Juerg Haefliger
On 09.10.2018 15:37, Juerg Haefliger wrote:
> BugLink: https://bugs.launchpad.net/bugs/1779848
>
> == SRU Justification ==
>
> Ubuntu kernels contain Meltdown and Spectre mitigations that are largely based on embargoed patches but what eventually landed in upstream is different in some places. We should clean up the different kernels to bring them closer in line with upstream.
>

The testing for this is as good as one can get. Just for clarity, I believe the
main difference we had and possibly still have in some cases is the handling of
IBRS/IBPB and how things are exposed as knobs to twiddle.
How did you do those and (that might be for Tyler) do we have a somehow
documented decision of the goal. I made some decision when working on Xenial and
I fear those might be different than yours for the lack of guidance.

-Stefan

> == Fix ==
>
> Add missing upstream patches.
>
> == Regression Potential ==
>
> Medium. The patches have been in upstream for quite a while now and are baked in but some of the backporting is not completely trivial and without risk.
>
> == Test Case ==
>
> TBD.
>
> *** BLURB HERE ***
>
> Signed-off-by: Juerg Haefliger <[hidden email]>
> ---
>
> The following changes since commit cefa65a1685282dcbadad7c914d4d8e89428a5eb:
>
>   UBUNTU: Ubuntu-3.13.0-161.211 (2018-10-03 16:41:43 +0200)
>
> are available in the Git repository at:
>
>   git://git.launchpad.net/~juergh/+git/trusty-linux lp1779848
>
> for you to fetch changes up to 8cccec097d1fb83c610f9dddd47827fb6d3237aa:
>
>   UBUNTU: SAUCE: x86/speculation: Cleanup IBRS runtime control handling (2018-10-09 10:46:55 +0200)
>
> ----------------------------------------------------------------
> Arnd Bergmann (1):
>       x86/pti: Mark constant arrays as __initconst
>
> Ben Hutchings (1):
>       xen: Add xen_arch_suspend()
>
> Borislav Petkov (2):
>       x86/nospec: Fix header guards names
>       x86/bugs: Drop one "mitigation" from dmesg
>
> Dan Carpenter (1):
>       x86/spectre: Fix an error message
>
> Dave Hansen (1):
>       x86/Documentation: Add PTI description
>
> David Woodhouse (5):
>       x86/cpufeatures: Add AMD feature bits for Speculation Control
>       x86/speculation: Add basic IBPB (Indirect Branch Prediction Barrier) support
>       x86/cpufeatures: Clean up Spectre v2 related CPUID flags
>       x86/cpuid: Fix up "virtual" IBRS/IBPB/STIBP feature bits on Intel
>       x86/speculation: Use IBRS if available before calling into firmware
>
> Dou Liyang (1):
>       x86/spectre: Check CONFIG_RETPOLINE in command line parser
>
> Ingo Molnar (1):
>       x86/speculation: Move firmware_restrict_branch_speculation_*() from C to CPP
>
> Juerg Haefliger (14):
>       Revert "x86/cpu/AMD: Make the LFENCE instruction serialized"
>       UBUNTU: SAUCE: x86/cpufeatures: Reorder spectre-related feature bits
>       UBUNTU: SAUCE: x86/msr: Fix formatting of msr-index.h
>       UBUNTU: SAUCE: x86/msr: Rename MSR spec control feature bits
>       UBUNTU: SAUCE: x86/speculation: Use synthetic bits for IBRS/IBPB/STIBP
>       UBUNTU: SAUCE: x86/cpufeatures: Disentangle MSR_SPEC_CTRL enumeration from IBRS
>       UBUNTU: SAUCE: x86/bugs: Fix re-use of SPEC_CTRL MSR boot value
>       UBUNTU: SAUCE: Move SSBD feature detection to common code
>       UBUNTU: SAUCE: x86/speculation: Move vendor specific IBRS/IBPB control code
>       UBUNTU: SAUCE: x86/speculation: Query individual feature flags when reloading microcode
>       UBUNTU: SAUCE: x86/pti: Evaluate X86_BUG_CPU_MELTDOWN when pti=auto
>       UBUNTU: SAUCE: x86/speculation: Make use of indirect_branch_prediction_barrier()
>       UBUNTU: SAUCE: x86/speculation: Cleanup IBPB runtime control handling
>       UBUNTU: SAUCE: x86/speculation: Cleanup IBRS runtime control handling
>
> Juergen Gross (1):
>       x86/xen: Zero MSR_IA32_SPEC_CTRL before suspend
>
> KarimAllah Ahmed (1):
>       x86/spectre: Simplify spectre_v2 command line parsing
>
> Tom Lendacky (2):
>       x86/cpu/AMD: Make LFENCE a serializing instruction
>       x86/cpu/AMD: Use LFENCE_RDTSC in preference to MFENCE_RDTSC
>
> zhenwei.pi (1):
>       x86/pti: Document fix wrong index
>
>  Documentation/kernel-parameters.txt   |  21 ++--
>  Documentation/x86/pti.txt             | 186 +++++++++++++++++++++++++++++++++
>  arch/x86/include/asm/apm.h            |   6 ++
>  arch/x86/include/asm/cpufeature.h     |  31 ++++--
>  arch/x86/include/asm/efi.h            |   7 ++
>  arch/x86/include/asm/nospec-branch.h  |  62 ++++++++++-
>  arch/x86/include/asm/spec_ctrl.h      |  13 ++-
>  arch/x86/include/uapi/asm/msr-index.h |  34 +++---
>  arch/x86/kernel/acpi/cstate.c         |   6 +-
>  arch/x86/kernel/cpu/amd.c             |  92 ++++++++--------
>  arch/x86/kernel/cpu/bugs.c            | 169 ++++++++++++++++++++----------
>  arch/x86/kernel/cpu/common.c          |  40 ++++++-
>  arch/x86/kernel/cpu/intel.c           |  19 ----
>  arch/x86/kernel/microcode_core.c      |  20 ----
>  arch/x86/kernel/process.c             |  11 +-
>  arch/x86/kernel/smpboot.c             |   6 +-
>  arch/x86/kvm/svm.c                    |  11 +-
>  arch/x86/kvm/vmx.c                    |   3 +-
>  arch/x86/mm/kaiser.c                  |   7 +-
>  arch/x86/mm/tlb.c                     |   5 +-
>  arch/x86/xen/suspend.c                |  24 +++++
>  drivers/xen/manage.c                  |   9 +-
>  include/linux/smp.h                   |  87 ----------------
>  include/xen/xen-ops.h                 |   1 +
>  kernel/smp.c                          |  46 --------
>  kernel/sysctl.c                       | 191 ++++++++++++++++++++--------------
>  26 files changed, 672 insertions(+), 435 deletions(-)
>  create mode 100644 Documentation/x86/pti.txt
>


--
kernel-team mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/kernel-team

signature.asc (836 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: [SRU][Trusty][PULL] Cleanup Meltdown/Spectre implementation (LP: #1779848)

Juerg Haefliger
On Thu, 11 Oct 2018 09:49:33 +0200
Stefan Bader <[hidden email]> wrote:

> On 09.10.2018 15:37, Juerg Haefliger wrote:
> > BugLink: https://bugs.launchpad.net/bugs/1779848
> >
> > == SRU Justification ==
> >
> > Ubuntu kernels contain Meltdown and Spectre mitigations that are
> > largely based on embargoed patches but what eventually landed in
> > upstream is different in some places. We should clean up the
> > different kernels to bring them closer in line with upstream.
>
> The testing for this is as good as one can get. Just for clarity, I
> believe the main difference we had and possibly still have in some
> cases is the handling of IBRS/IBPB and how things are exposed as
> knobs to twiddle. How did you do those and (that might be for Tyler)
> do we have a somehow documented decision of the goal. I made some
> decision when working on Xenial and I fear those might be different
> than yours for the lack of guidance.
Sorry, your reply slipped through the cracks. I'm not sure what you
mean. AFAIK our specific runtime control knobs and boottime options
and their implementations are identical for Trusty and Xenial. I just
rearranged things to integrate it more logically with all the other
modifications but there should be no functional changes. These are the
last 2 commits in the PR. As for testing those changes, I just verified
that I can twiddle the knobs as before and that the kernel boots.

...Juerg

--
kernel-team mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/kernel-team

attachment0 (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: [SRU][Trusty][PULL] Cleanup Meltdown/Spectre implementation (LP: #1779848)

Stefan Bader-2
On 18.10.18 08:14, Juerg Haefliger wrote:

> On Thu, 11 Oct 2018 09:49:33 +0200
> Stefan Bader <[hidden email]> wrote:
>
>> On 09.10.2018 15:37, Juerg Haefliger wrote:
>>> BugLink: https://bugs.launchpad.net/bugs/1779848
>>>
>>> == SRU Justification ==
>>>
>>> Ubuntu kernels contain Meltdown and Spectre mitigations that are
>>> largely based on embargoed patches but what eventually landed in
>>> upstream is different in some places. We should clean up the
>>> different kernels to bring them closer in line with upstream.
>>
>> The testing for this is as good as one can get. Just for clarity, I
>> believe the main difference we had and possibly still have in some
>> cases is the handling of IBRS/IBPB and how things are exposed as
>> knobs to twiddle. How did you do those and (that might be for Tyler)
>> do we have a somehow documented decision of the goal. I made some
>> decision when working on Xenial and I fear those might be different
>> than yours for the lack of guidance.
>
> Sorry, your reply slipped through the cracks. I'm not sure what you
> mean. AFAIK our specific runtime control knobs and boottime options
> and their implementations are identical for Trusty and Xenial. I just
> rearranged things to integrate it more logically with all the other
> modifications but there should be no functional changes. These are the
> last 2 commits in the PR. As for testing those changes, I just verified
> that I can twiddle the knobs as before and that the kernel boots.
OK, maybe its better than I feared. It is just very hard to keep track of things
with other stuff going on in parallel. I just very vaguely remember that for
Bionic/4.15 and possibly later, IBRS/IBPB was only decided based on cpu features
while some of what we did there was an additional sysfs interface.

-Stefan
>
> ...Juerg
>



--
kernel-team mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/kernel-team

signature.asc (836 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: [SRU][Trusty][PULL] Cleanup Meltdown/Spectre implementation (LP: #1779848)

Juerg Haefliger
On Thu, 18 Oct 2018 11:47:02 +0200
Stefan Bader <[hidden email]> wrote:

> On 18.10.18 08:14, Juerg Haefliger wrote:
> > On Thu, 11 Oct 2018 09:49:33 +0200
> > Stefan Bader <[hidden email]> wrote:
> >  
> >> On 09.10.2018 15:37, Juerg Haefliger wrote:  
> >>> BugLink: https://bugs.launchpad.net/bugs/1779848
> >>>
> >>> == SRU Justification ==
> >>>
> >>> Ubuntu kernels contain Meltdown and Spectre mitigations that are
> >>> largely based on embargoed patches but what eventually landed in
> >>> upstream is different in some places. We should clean up the
> >>> different kernels to bring them closer in line with upstream.  
> >>
> >> The testing for this is as good as one can get. Just for clarity, I
> >> believe the main difference we had and possibly still have in some
> >> cases is the handling of IBRS/IBPB and how things are exposed as
> >> knobs to twiddle. How did you do those and (that might be for
> >> Tyler) do we have a somehow documented decision of the goal. I
> >> made some decision when working on Xenial and I fear those might
> >> be different than yours for the lack of guidance.  
> >
> > Sorry, your reply slipped through the cracks. I'm not sure what you
> > mean. AFAIK our specific runtime control knobs and boottime options
> > and their implementations are identical for Trusty and Xenial. I
> > just rearranged things to integrate it more logically with all the
> > other modifications but there should be no functional changes.
> > These are the last 2 commits in the PR. As for testing those
> > changes, I just verified that I can twiddle the knobs as before and
> > that the kernel boots.  
>
> OK, maybe its better than I feared. It is just very hard to keep
> track of things with other stuff going on in parallel. I just very
> vaguely remember that for Bionic/4.15 and possibly later, IBRS/IBPB
> was only decided based on cpu features while some of what we did
> there was an additional sysfs interface.
Yes, the sysfs knobs and noib{rs,pb} commandline options were part of
the embargoed patches but that never went upstream. We only have this in
Xenial 4.4 and Trusty 3.13. And since it's user exposed we probably
can't rip it out :-(

...Juerg

> -Stefan
> >
> > ...Juerg
> >  
>
>


--
kernel-team mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/kernel-team

attachment0 (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

ACK/Cmnt: [SRU][Trusty][PULL] Cleanup Meltdown/Spectre implementation (LP: #1779848)

Stefan Bader-2
In reply to this post by Juerg Haefliger
On 09.10.18 15:37, Juerg Haefliger wrote:

> BugLink: https://bugs.launchpad.net/bugs/1779848
>
> == SRU Justification ==
>
> Ubuntu kernels contain Meltdown and Spectre mitigations that are largely based on embargoed patches but what eventually landed in upstream is different in some places. We should clean up the different kernels to bring them closer in line with upstream.
>
> == Fix ==
>
> Add missing upstream patches.
>
> == Regression Potential ==
>
> Medium. The patches have been in upstream for quite a while now and are baked in but some of the backporting is not completely trivial and without risk.
>
> == Test Case ==
>
> TBD.
>
> *** BLURB HERE ***
>
> Signed-off-by: Juerg Haefliger <[hidden email]>
> ---
>
> The following changes since commit cefa65a1685282dcbadad7c914d4d8e89428a5eb:
>
>   UBUNTU: Ubuntu-3.13.0-161.211 (2018-10-03 16:41:43 +0200)
>
> are available in the Git repository at:
>
>   git://git.launchpad.net/~juergh/+git/trusty-linux lp1779848
>
> for you to fetch changes up to 8cccec097d1fb83c610f9dddd47827fb6d3237aa:
>
>   UBUNTU: SAUCE: x86/speculation: Cleanup IBRS runtime control handling (2018-10-09 10:46:55 +0200)
>
> ----------------------------------------------------------------
> Arnd Bergmann (1):
>       x86/pti: Mark constant arrays as __initconst
>
> Ben Hutchings (1):
>       xen: Add xen_arch_suspend()
>
> Borislav Petkov (2):
>       x86/nospec: Fix header guards names
>       x86/bugs: Drop one "mitigation" from dmesg
>
> Dan Carpenter (1):
>       x86/spectre: Fix an error message
>
> Dave Hansen (1):
>       x86/Documentation: Add PTI description
>
> David Woodhouse (5):
>       x86/cpufeatures: Add AMD feature bits for Speculation Control
>       x86/speculation: Add basic IBPB (Indirect Branch Prediction Barrier) support
>       x86/cpufeatures: Clean up Spectre v2 related CPUID flags
>       x86/cpuid: Fix up "virtual" IBRS/IBPB/STIBP feature bits on Intel
>       x86/speculation: Use IBRS if available before calling into firmware
>
> Dou Liyang (1):
>       x86/spectre: Check CONFIG_RETPOLINE in command line parser
>
> Ingo Molnar (1):
>       x86/speculation: Move firmware_restrict_branch_speculation_*() from C to CPP
>
> Juerg Haefliger (14):
>       Revert "x86/cpu/AMD: Make the LFENCE instruction serialized"
>       UBUNTU: SAUCE: x86/cpufeatures: Reorder spectre-related feature bits
>       UBUNTU: SAUCE: x86/msr: Fix formatting of msr-index.h
>       UBUNTU: SAUCE: x86/msr: Rename MSR spec control feature bits
>       UBUNTU: SAUCE: x86/speculation: Use synthetic bits for IBRS/IBPB/STIBP
>       UBUNTU: SAUCE: x86/cpufeatures: Disentangle MSR_SPEC_CTRL enumeration from IBRS
>       UBUNTU: SAUCE: x86/bugs: Fix re-use of SPEC_CTRL MSR boot value
>       UBUNTU: SAUCE: Move SSBD feature detection to common code
>       UBUNTU: SAUCE: x86/speculation: Move vendor specific IBRS/IBPB control code
>       UBUNTU: SAUCE: x86/speculation: Query individual feature flags when reloading microcode
>       UBUNTU: SAUCE: x86/pti: Evaluate X86_BUG_CPU_MELTDOWN when pti=auto
>       UBUNTU: SAUCE: x86/speculation: Make use of indirect_branch_prediction_barrier()
>       UBUNTU: SAUCE: x86/speculation: Cleanup IBPB runtime control handling
>       UBUNTU: SAUCE: x86/speculation: Cleanup IBRS runtime control handling
>
> Juergen Gross (1):
>       x86/xen: Zero MSR_IA32_SPEC_CTRL before suspend
>
> KarimAllah Ahmed (1):
>       x86/spectre: Simplify spectre_v2 command line parsing
>
> Tom Lendacky (2):
>       x86/cpu/AMD: Make LFENCE a serializing instruction
>       x86/cpu/AMD: Use LFENCE_RDTSC in preference to MFENCE_RDTSC
>
> zhenwei.pi (1):
>       x86/pti: Document fix wrong index
>
>  Documentation/kernel-parameters.txt   |  21 ++--
>  Documentation/x86/pti.txt             | 186 +++++++++++++++++++++++++++++++++
>  arch/x86/include/asm/apm.h            |   6 ++
>  arch/x86/include/asm/cpufeature.h     |  31 ++++--
>  arch/x86/include/asm/efi.h            |   7 ++
>  arch/x86/include/asm/nospec-branch.h  |  62 ++++++++++-
>  arch/x86/include/asm/spec_ctrl.h      |  13 ++-
>  arch/x86/include/uapi/asm/msr-index.h |  34 +++---
>  arch/x86/kernel/acpi/cstate.c         |   6 +-
>  arch/x86/kernel/cpu/amd.c             |  92 ++++++++--------
>  arch/x86/kernel/cpu/bugs.c            | 169 ++++++++++++++++++++----------
>  arch/x86/kernel/cpu/common.c          |  40 ++++++-
>  arch/x86/kernel/cpu/intel.c           |  19 ----
>  arch/x86/kernel/microcode_core.c      |  20 ----
>  arch/x86/kernel/process.c             |  11 +-
>  arch/x86/kernel/smpboot.c             |   6 +-
>  arch/x86/kvm/svm.c                    |  11 +-
>  arch/x86/kvm/vmx.c                    |   3 +-
>  arch/x86/mm/kaiser.c                  |   7 +-
>  arch/x86/mm/tlb.c                     |   5 +-
>  arch/x86/xen/suspend.c                |  24 +++++
>  drivers/xen/manage.c                  |   9 +-
>  include/linux/smp.h                   |  87 ----------------
>  include/xen/xen-ops.h                 |   1 +
>  kernel/smp.c                          |  46 --------
>  kernel/sysctl.c                       | 191 ++++++++++++++++++++--------------
>  26 files changed, 672 insertions(+), 435 deletions(-)
>  create mode 100644 Documentation/x86/pti.txt
>
Ok, my main concern was to keep Xenial and Trusty functionally in sync which I
think was confirmed we do, so I guess we should go ahead and pull this in
carefully observe how it goes.

-Stefan


Acked-by: Stefan Bader <[hidden email]>


--
kernel-team mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/kernel-team

signature.asc (836 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

ACK: [SRU][Trusty][PULL] Cleanup Meltdown/Spectre implementation (LP: #1779848)

Kleber Souza
In reply to this post by Juerg Haefliger
On 10/09/18 15:37, Juerg Haefliger wrote:

> BugLink: https://bugs.launchpad.net/bugs/1779848
>
> == SRU Justification ==
>
> Ubuntu kernels contain Meltdown and Spectre mitigations that are largely based on embargoed patches but what eventually landed in upstream is different in some places. We should clean up the different kernels to bring them closer in line with upstream.
>
> == Fix ==
>
> Add missing upstream patches.
>
> == Regression Potential ==
>
> Medium. The patches have been in upstream for quite a while now and are baked in but some of the backporting is not completely trivial and without risk.
>
> == Test Case ==
>
> TBD.
>
> *** BLURB HERE ***
>
> Signed-off-by: Juerg Haefliger <[hidden email]>
> ---
>
> The following changes since commit cefa65a1685282dcbadad7c914d4d8e89428a5eb:
>
>   UBUNTU: Ubuntu-3.13.0-161.211 (2018-10-03 16:41:43 +0200)
>
> are available in the Git repository at:
>
>   git://git.launchpad.net/~juergh/+git/trusty-linux lp1779848
>
> for you to fetch changes up to 8cccec097d1fb83c610f9dddd47827fb6d3237aa:
>
>   UBUNTU: SAUCE: x86/speculation: Cleanup IBRS runtime control handling (2018-10-09 10:46:55 +0200)
>
> ----------------------------------------------------------------
> Arnd Bergmann (1):
>       x86/pti: Mark constant arrays as __initconst
>
> Ben Hutchings (1):
>       xen: Add xen_arch_suspend()
>
> Borislav Petkov (2):
>       x86/nospec: Fix header guards names
>       x86/bugs: Drop one "mitigation" from dmesg
>
> Dan Carpenter (1):
>       x86/spectre: Fix an error message
>
> Dave Hansen (1):
>       x86/Documentation: Add PTI description
>
> David Woodhouse (5):
>       x86/cpufeatures: Add AMD feature bits for Speculation Control
>       x86/speculation: Add basic IBPB (Indirect Branch Prediction Barrier) support
>       x86/cpufeatures: Clean up Spectre v2 related CPUID flags
>       x86/cpuid: Fix up "virtual" IBRS/IBPB/STIBP feature bits on Intel
>       x86/speculation: Use IBRS if available before calling into firmware
>
> Dou Liyang (1):
>       x86/spectre: Check CONFIG_RETPOLINE in command line parser
>
> Ingo Molnar (1):
>       x86/speculation: Move firmware_restrict_branch_speculation_*() from C to CPP
>
> Juerg Haefliger (14):
>       Revert "x86/cpu/AMD: Make the LFENCE instruction serialized"
>       UBUNTU: SAUCE: x86/cpufeatures: Reorder spectre-related feature bits
>       UBUNTU: SAUCE: x86/msr: Fix formatting of msr-index.h
>       UBUNTU: SAUCE: x86/msr: Rename MSR spec control feature bits
>       UBUNTU: SAUCE: x86/speculation: Use synthetic bits for IBRS/IBPB/STIBP
>       UBUNTU: SAUCE: x86/cpufeatures: Disentangle MSR_SPEC_CTRL enumeration from IBRS
>       UBUNTU: SAUCE: x86/bugs: Fix re-use of SPEC_CTRL MSR boot value
>       UBUNTU: SAUCE: Move SSBD feature detection to common code
>       UBUNTU: SAUCE: x86/speculation: Move vendor specific IBRS/IBPB control code
>       UBUNTU: SAUCE: x86/speculation: Query individual feature flags when reloading microcode
>       UBUNTU: SAUCE: x86/pti: Evaluate X86_BUG_CPU_MELTDOWN when pti=auto
>       UBUNTU: SAUCE: x86/speculation: Make use of indirect_branch_prediction_barrier()
>       UBUNTU: SAUCE: x86/speculation: Cleanup IBPB runtime control handling
>       UBUNTU: SAUCE: x86/speculation: Cleanup IBRS runtime control handling
>
> Juergen Gross (1):
>       x86/xen: Zero MSR_IA32_SPEC_CTRL before suspend
>
> KarimAllah Ahmed (1):
>       x86/spectre: Simplify spectre_v2 command line parsing
>
> Tom Lendacky (2):
>       x86/cpu/AMD: Make LFENCE a serializing instruction
>       x86/cpu/AMD: Use LFENCE_RDTSC in preference to MFENCE_RDTSC
>
> zhenwei.pi (1):
>       x86/pti: Document fix wrong index
>
>  Documentation/kernel-parameters.txt   |  21 ++--
>  Documentation/x86/pti.txt             | 186 +++++++++++++++++++++++++++++++++
>  arch/x86/include/asm/apm.h            |   6 ++
>  arch/x86/include/asm/cpufeature.h     |  31 ++++--
>  arch/x86/include/asm/efi.h            |   7 ++
>  arch/x86/include/asm/nospec-branch.h  |  62 ++++++++++-
>  arch/x86/include/asm/spec_ctrl.h      |  13 ++-
>  arch/x86/include/uapi/asm/msr-index.h |  34 +++---
>  arch/x86/kernel/acpi/cstate.c         |   6 +-
>  arch/x86/kernel/cpu/amd.c             |  92 ++++++++--------
>  arch/x86/kernel/cpu/bugs.c            | 169 ++++++++++++++++++++----------
>  arch/x86/kernel/cpu/common.c          |  40 ++++++-
>  arch/x86/kernel/cpu/intel.c           |  19 ----
>  arch/x86/kernel/microcode_core.c      |  20 ----
>  arch/x86/kernel/process.c             |  11 +-
>  arch/x86/kernel/smpboot.c             |   6 +-
>  arch/x86/kvm/svm.c                    |  11 +-
>  arch/x86/kvm/vmx.c                    |   3 +-
>  arch/x86/mm/kaiser.c                  |   7 +-
>  arch/x86/mm/tlb.c                     |   5 +-
>  arch/x86/xen/suspend.c                |  24 +++++
>  drivers/xen/manage.c                  |   9 +-
>  include/linux/smp.h                   |  87 ----------------
>  include/xen/xen-ops.h                 |   1 +
>  kernel/smp.c                          |  46 --------
>  kernel/sysctl.c                       | 191 ++++++++++++++++++++--------------
>  26 files changed, 672 insertions(+), 435 deletions(-)
>  create mode 100644 Documentation/x86/pti.txt
>

Acked-by: Kleber Sacilotto de Souza <[hidden email]>

--
kernel-team mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/kernel-team
Reply | Threaded
Open this post in threaded view
|

APPLIED: [SRU][Trusty][PULL] Cleanup Meltdown/Spectre implementation (LP: #1779848)

Kleber Souza
In reply to this post by Juerg Haefliger
On 10/09/18 15:37, Juerg Haefliger wrote:

> BugLink: https://bugs.launchpad.net/bugs/1779848
>
> == SRU Justification ==
>
> Ubuntu kernels contain Meltdown and Spectre mitigations that are largely based on embargoed patches but what eventually landed in upstream is different in some places. We should clean up the different kernels to bring them closer in line with upstream.
>
> == Fix ==
>
> Add missing upstream patches.
>
> == Regression Potential ==
>
> Medium. The patches have been in upstream for quite a while now and are baked in but some of the backporting is not completely trivial and without risk.
>
> == Test Case ==
>
> TBD.
>
> *** BLURB HERE ***
>
> Signed-off-by: Juerg Haefliger <[hidden email]>
> ---
>
> The following changes since commit cefa65a1685282dcbadad7c914d4d8e89428a5eb:
>
>   UBUNTU: Ubuntu-3.13.0-161.211 (2018-10-03 16:41:43 +0200)
>
> are available in the Git repository at:
>
>   git://git.launchpad.net/~juergh/+git/trusty-linux lp1779848
>
> for you to fetch changes up to 8cccec097d1fb83c610f9dddd47827fb6d3237aa:
>
>   UBUNTU: SAUCE: x86/speculation: Cleanup IBRS runtime control handling (2018-10-09 10:46:55 +0200)
>
> ----------------------------------------------------------------
> Arnd Bergmann (1):
>       x86/pti: Mark constant arrays as __initconst
>
> Ben Hutchings (1):
>       xen: Add xen_arch_suspend()
>
> Borislav Petkov (2):
>       x86/nospec: Fix header guards names
>       x86/bugs: Drop one "mitigation" from dmesg
>
> Dan Carpenter (1):
>       x86/spectre: Fix an error message
>
> Dave Hansen (1):
>       x86/Documentation: Add PTI description
>
> David Woodhouse (5):
>       x86/cpufeatures: Add AMD feature bits for Speculation Control
>       x86/speculation: Add basic IBPB (Indirect Branch Prediction Barrier) support
>       x86/cpufeatures: Clean up Spectre v2 related CPUID flags
>       x86/cpuid: Fix up "virtual" IBRS/IBPB/STIBP feature bits on Intel
>       x86/speculation: Use IBRS if available before calling into firmware
>
> Dou Liyang (1):
>       x86/spectre: Check CONFIG_RETPOLINE in command line parser
>
> Ingo Molnar (1):
>       x86/speculation: Move firmware_restrict_branch_speculation_*() from C to CPP
>
> Juerg Haefliger (14):
>       Revert "x86/cpu/AMD: Make the LFENCE instruction serialized"
>       UBUNTU: SAUCE: x86/cpufeatures: Reorder spectre-related feature bits
>       UBUNTU: SAUCE: x86/msr: Fix formatting of msr-index.h
>       UBUNTU: SAUCE: x86/msr: Rename MSR spec control feature bits
>       UBUNTU: SAUCE: x86/speculation: Use synthetic bits for IBRS/IBPB/STIBP
>       UBUNTU: SAUCE: x86/cpufeatures: Disentangle MSR_SPEC_CTRL enumeration from IBRS
>       UBUNTU: SAUCE: x86/bugs: Fix re-use of SPEC_CTRL MSR boot value
>       UBUNTU: SAUCE: Move SSBD feature detection to common code
>       UBUNTU: SAUCE: x86/speculation: Move vendor specific IBRS/IBPB control code
>       UBUNTU: SAUCE: x86/speculation: Query individual feature flags when reloading microcode
>       UBUNTU: SAUCE: x86/pti: Evaluate X86_BUG_CPU_MELTDOWN when pti=auto
>       UBUNTU: SAUCE: x86/speculation: Make use of indirect_branch_prediction_barrier()
>       UBUNTU: SAUCE: x86/speculation: Cleanup IBPB runtime control handling
>       UBUNTU: SAUCE: x86/speculation: Cleanup IBRS runtime control handling
>
> Juergen Gross (1):
>       x86/xen: Zero MSR_IA32_SPEC_CTRL before suspend
>
> KarimAllah Ahmed (1):
>       x86/spectre: Simplify spectre_v2 command line parsing
>
> Tom Lendacky (2):
>       x86/cpu/AMD: Make LFENCE a serializing instruction
>       x86/cpu/AMD: Use LFENCE_RDTSC in preference to MFENCE_RDTSC
>
> zhenwei.pi (1):
>       x86/pti: Document fix wrong index
>
>  Documentation/kernel-parameters.txt   |  21 ++--
>  Documentation/x86/pti.txt             | 186 +++++++++++++++++++++++++++++++++
>  arch/x86/include/asm/apm.h            |   6 ++
>  arch/x86/include/asm/cpufeature.h     |  31 ++++--
>  arch/x86/include/asm/efi.h            |   7 ++
>  arch/x86/include/asm/nospec-branch.h  |  62 ++++++++++-
>  arch/x86/include/asm/spec_ctrl.h      |  13 ++-
>  arch/x86/include/uapi/asm/msr-index.h |  34 +++---
>  arch/x86/kernel/acpi/cstate.c         |   6 +-
>  arch/x86/kernel/cpu/amd.c             |  92 ++++++++--------
>  arch/x86/kernel/cpu/bugs.c            | 169 ++++++++++++++++++++----------
>  arch/x86/kernel/cpu/common.c          |  40 ++++++-
>  arch/x86/kernel/cpu/intel.c           |  19 ----
>  arch/x86/kernel/microcode_core.c      |  20 ----
>  arch/x86/kernel/process.c             |  11 +-
>  arch/x86/kernel/smpboot.c             |   6 +-
>  arch/x86/kvm/svm.c                    |  11 +-
>  arch/x86/kvm/vmx.c                    |   3 +-
>  arch/x86/mm/kaiser.c                  |   7 +-
>  arch/x86/mm/tlb.c                     |   5 +-
>  arch/x86/xen/suspend.c                |  24 +++++
>  drivers/xen/manage.c                  |   9 +-
>  include/linux/smp.h                   |  87 ----------------
>  include/xen/xen-ops.h                 |   1 +
>  kernel/smp.c                          |  46 --------
>  kernel/sysctl.c                       | 191 ++++++++++++++++++++--------------
>  26 files changed, 672 insertions(+), 435 deletions(-)
>  create mode 100644 Documentation/x86/pti.txt
>

Applied to trusty/master-next branch.

Thanks,
Kleber

--
kernel-team mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/kernel-team