linux-kernel - [PATCH 0/1] x86/kprobes: Classify opcode while preparing kprobe

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: <160726388853.3413805.4854581312983421622.stgit@devnote2>
Date:   Sun,  6 Dec 2020 23:11:28 +0900
From:   Masami Hiramatsu <mhiramat@...nel.org>
To:     Ingo Molnar <mingo@...nel.org>,
        Steven Rostedt <rostedt@...dmis.org>
Cc:     mhiramat@...nel.org, linux-kernel@...r.kernel.org,
        "Naveen N . Rao" <naveen.n.rao@...ux.ibm.com>,
        Anil S Keshavamurthy <anil.s.keshavamurthy@...el.com>,
        David Miller <davem@...emloft.net>,
        Borislav Petkov <bp@...en8.de>, x86@...nel.org
Subject: [PATCH  0/1] x86/kprobes: Classify opcode while preparing kprobe

Hi,

I noticed that the kprobe x86 still has an inefficient code.

Currently kprobes x86 decodes opcode right after single
stepping in resume_execution(), which is kprobe's hot path.
But it already decoded the opcode while preparing
arch_specific_insn in arch_copy_kprobe(), so this is
inefficient.

So decode the opcode while preparing kprobes (arch_copy_kprobe())
instead of resume_execution() and set some flags for resuming from
single stepping. This also removes the custom instruction prefix
decoding, which should be done in x86 instruction decoder.

Thank you,

---

Masami Hiramatsu (1):
      x86/kprobes: Do not decode opcode in resume_execution()

 arch/x86/include/asm/kprobes.h |   11 ++-
 arch/x86/kernel/kprobes/core.c |  166 ++++++++++++++++++----------------------
 2 files changed, 80 insertions(+), 97 deletions(-)

--
Masami Hiramatsu (Linaro) <mhiramat@...nel.org>