linux-kernel - [PATCH v6 0/3] x86/fpu Measure the Latency of XSAVES and

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-Id: <20230901143414.1664368-1-yi.sun@intel.com>
Date:   Fri,  1 Sep 2023 22:34:11 +0800
From:   Yi Sun <yi.sun@...el.com>
To:     dave.hansen@...el.com, tglx@...utronix.de,
        linux-kernel@...r.kernel.org, x86@...nel.org
Cc:     sohil.mehta@...el.com, ak@...ux.intel.com,
        ilpo.jarvinen@...ux.intel.com, heng.su@...el.com,
        tony.luck@...el.com, dave.hansen@...ux.intel.com,
        yi.sun@...el.intel.com, Yi Sun <yi.sun@...el.com>
Subject: [PATCH v6 0/3] x86/fpu Measure the Latency of XSAVES and

This series implements a solution for measuring the latency of
XSAVES/XRSTORS instructions on x86 processors. The motivation for This
solution is to provide a way to identify potential bottlenecks and
anomalies in the implementation of these instructions on different X86
hardware platforms and identify when the cost of XSAVES/XRSTORS is
increasing or decreasing. And Get the RFBM and XINUSE to figure out reason
if there are any issues.

The solution consists of two parts:
1. Two new tracepoints that calculate delta TSC as the latency.
2. A user-space tool that consumes trace logs and shows results.

We conducted extensive internal testing and compared several latency
computation methodologies. We believe it could be improved. Also the script
for consuming trace logs uses sqlite3 to display statistics data such as
max and min latency, and 97% tail latency, etc.

- Change from v5 to v6:
  - Rebase patch to the most recent tag v6.5-rc7, no conflict.
  - Add a colomn 'count' to the output of the script.

- Change from v4 to v5:
  - Rebase patch to the most recent tag v6.1-rc2, no conflict.

- Change from v3 to v4:
  - Rebase patch set to the latest tag v6.0-rc3, replacing
    ALTERNATIVE_2 with ALTERNATIVE_3

- Change from v2 to v3:
  - Move the script xsave-latency-trace.sh to folder tools/testing/fpu/
    (Ingo Molnar)

- Change from v1 to v2:
  - Reword the comments. (Laight, David; Mehta, Sohil)
  - Rename all variable 'dtsc' to 'latency'. (Mehta, Sohil)
Yi Sun (3):
  x86/fpu: Measure the Latency of XSAVE and XRSTOR
  tools/testing/fpu: Add script to consume trace log of xsave latency
  tools/testing/fpu: Add a 'count' column.

 arch/x86/include/asm/trace/fpu.h         |  35 ++++
 arch/x86/kernel/fpu/xstate.h             |  49 ++++-
 tools/testing/fpu/xsave-latency-trace.sh | 234 +++++++++++++++++++++++
 3 files changed, 313 insertions(+), 5 deletions(-)
 create mode 100755 tools/testing/fpu/xsave-latency-trace.sh

-- 
2.34.1