linux-kernel - [PATCH V2 0/7] x86/entry: Convert error

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: <20220516131739.521817-1-jiangshanlai@gmail.com>
Date:   Mon, 16 May 2022 21:17:31 +0800
From:   Lai Jiangshan <jiangshanlai@...il.com>
To:     linux-kernel@...r.kernel.org
Cc:     Borislav Petkov <bp@...en8.de>,
        Peter Zijlstra <peterz@...radead.org>,
        Josh Poimboeuf <jpoimboe@...hat.com>,
        Andy Lutomirski <luto@...nel.org>,
        Thomas Gleixner <tglx@...utronix.de>, x86@...nel.org,
        Lai Jiangshan <jiangshan.ljs@...group.com>
Subject: [PATCH V2 0/7] x86/entry: Convert error_entry() to C code

From: Lai Jiangshan <jiangshan.ljs@...group.com>

Add some C equivalent functions of the ASM macros and implement the whole
error_entry() as C code.

The patches are picked and re-made from the huge patchset
https://lore.kernel.org/lkml/20211126101209.8613-1-jiangshanlai@gmail.com/
which converts a large chunk of ASM code to C code.

The C version generally has better readability and easier to be
updated/improved.

This smaller patchset converts error_entry() only.
The equivalent ASM macros are not removed because they are still used by
the IST exceptions.

No functional change intended and comments are also copied.

The complier generates very similar code for the C code and the original
ASM code except minor differences.

The complier uses tail-call-optimization for calling sync_regs().  It
uses "JMP sync_regs" while the ASM code uses "CALL+RET".

The compiler generates "AND $0xe7,%ah" (3 bytes) for the code
"cr3 = user_cr3 & ~PTI_USER_PGTABLE_AND_PCID_MASK" while the ASM code in
SWITCH_TO_KERNEL_CR3() results "AND $0xffffffffffffe7ff,%rax" (6 bytes).

The compiler generates lengthier code for "cr3 |= X86_CR3_PCID_NOFLUSH"
because it uses "MOVABS+OR" (13 bytes)  rather than a single
"BTS" (5 bytes).

ALTERNATIVE and static_cpu_has() are also different which depends on
what alternative instructions for ALTERNATIVE are.

[V1]: https://lore.kernel.org/lkml/20220511072747.3960-1-jiangshanlai@gmail.com/

Changed from V1:
	remove unneeded cleanup in patch2

Changed from the old huge patchset:
	squash some patches

Lai Jiangshan (7):
  x86/entry: Introduce __entry_text for entry code written in C
  x86/entry: Move PTI_USER_* to arch/x86/include/asm/processor-flags.h
  x86: Mark __native_read_cr3() & native_write_cr3() as __always_inline
  x86/entry: Add arch/x86/entry/entry64.c for C entry code
  x86/entry: Add the C verion of SWITCH_TO_KERNEL_CR3 as
    switch_to_kernel_cr3()
  x86/traps: Add fence_swapgs_{user,kernel}_entry() and
    user_entry_swapgs_and_fence()
  x86/entry: Implement the whole error_entry() as C code

 arch/x86/entry/Makefile                |   3 +-
 arch/x86/entry/calling.h               |  10 --
 arch/x86/entry/entry64.c               | 137 +++++++++++++++++++++++++
 arch/x86/entry/entry_64.S              |  85 +--------------
 arch/x86/include/asm/idtentry.h        |   3 +
 arch/x86/include/asm/processor-flags.h |  15 +++
 arch/x86/include/asm/proto.h           |   1 +
 arch/x86/include/asm/special_insns.h   |   4 +-
 arch/x86/include/asm/traps.h           |   1 +
 arch/x86/kernel/traps.c                |   2 -
 include/linux/compiler_types.h         |   8 +-
 11 files changed, 169 insertions(+), 100 deletions(-)
 create mode 100644 arch/x86/entry/entry64.c

-- 
2.19.1.6.gb485710b