Re: [PATCH v4 00/10] Introduce SMT level and add PowerPC support

From: Laurent Dufour
Date: Mon Jul 10 2023 - 05:09:04 EST




Le 09/07/2023 à 17:25, Zhang, Rui a écrit :
Hi, Laurent,

I ran into a boot hang regression with latest upstream code, and it
took me a while to bisect the offending commit and workaround it.

Now I have tested this patch series on an Intel RaptorLake Hybrid
platform (4 Pcores with HT and 4 Ecores without HT), and it works as
expected.

So, for patch 1~7 in this series,

Tested-by: Zhang Rui <rui.zhang@xxxxxxxxx>

Thanks Rui!

thanks,
rui

On Wed, 2023-07-05 at 16:51 +0200, Laurent Dufour wrote:
I'm taking over the series Michael sent previously [1] which is
smartly
reviewing the initial series I sent [2].  This series is addressing
the
comments sent by Thomas and me on the Michael's one.

Here is a short introduction to the issue this series is addressing:

When a new CPU is added, the kernel is activating all its threads.
This
leads to weird, but functional, result when adding CPU on a SMT 4
system
for instance.

Here the newly added CPU 1 has 8 threads while the other one has 4
threads
active (system has been booted with the 'smt-enabled=4' kernel
option):

ltcden3-lp12:~ # ppc64_cpu --info
Core   0:    0*    1*    2*    3*    4     5     6     7
Core   1:    8*    9*   10*   11*   12*   13*   14*   15*

This mixed SMT level may confused end users and/or some applications.

There is no SMT level recorded in the kernel (common code), neither
in user
space, as far as I know. Such a level is helpful when adding new CPU
or
when optimizing the energy efficiency (when reactivating CPUs).

When SMP and HOTPLUG_SMT are defined, this series is adding a new SMT
level
(cpu_smt_num_threads) and few callbacks allowing the architecture
code to
fine control this value, setting a max and a "at boot" level, and
controling whether a thread should be onlined or not.

v4:
  Rebase on top of 6.5's updates
  Remove a dependancy against the X86's symbol
cpu_primary_thread_mask
v3:
  Fix a build error in the patch 6/9
v2:
  As Thomas suggested,
    Reword some commit's description
    Remove topology_smt_supported()
    Remove topology_smt_threads_supported()
    Introduce CONFIG_SMT_NUM_THREADS_DYNAMIC
    Remove switch() in __store_smt_control()
  Update kernel-parameters.txt

[1]
https://lore.kernel.org/linuxppc-dev/20230524155630.794584-1-mpe@xxxxxxxxxxxxxx/
[2]
https://lore.kernel.org/linuxppc-dev/20230331153905.31698-1-ldufour@xxxxxxxxxxxxx/


Laurent Dufour (2):
  cpu/hotplug: remove dependancy against cpu_primary_thread_mask
  cpu/SMT: Remove topology_smt_supported()

Michael Ellerman (8):
  cpu/SMT: Move SMT prototypes into cpu_smt.h
  cpu/SMT: Move smt/control simple exit cases earlier
  cpu/SMT: Store the current/max number of threads
  cpu/SMT: Create topology_smt_thread_allowed()
  cpu/SMT: Allow enabling partial SMT states via sysfs
  powerpc/pseries: Initialise CPU hotplug callbacks earlier
  powerpc: Add HOTPLUG_SMT support
  powerpc/pseries: Honour current SMT state when DLPAR onlining CPUs

 .../ABI/testing/sysfs-devices-system-cpu      |   1 +
 .../admin-guide/kernel-parameters.txt         |   4 +-
 arch/Kconfig                                  |   3 +
 arch/powerpc/Kconfig                          |   2 +
 arch/powerpc/include/asm/topology.h           |  15 ++
 arch/powerpc/kernel/smp.c                     |   8 +-
 arch/powerpc/platforms/pseries/hotplug-cpu.c  |  30 ++--
 arch/powerpc/platforms/pseries/pseries.h      |   2 +
 arch/powerpc/platforms/pseries/setup.c        |   2 +
 arch/x86/include/asm/topology.h               |   4 +-
 arch/x86/kernel/cpu/common.c                  |   2 +-
 arch/x86/kernel/smpboot.c                     |   8 -
 include/linux/cpu.h                           |  25 +--
 include/linux/cpu_smt.h                       |  33 ++++
 kernel/cpu.c                                  | 142 +++++++++++++---
--
 15 files changed, 196 insertions(+), 85 deletions(-)
 create mode 100644 include/linux/cpu_smt.h