| 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
| ||
|
Message-ID: <202408052138.EB7B9788E3@keescook>
Date: Mon, 5 Aug 2024 21:39:05 -0700
From: Kees Cook <kees@...nel.org>
To: Gatlin Newhouse <gatlin.newhouse@...il.com>
Cc: Thomas Gleixner <tglx@...utronix.de>, Ingo Molnar <mingo@...hat.com>,
Borislav Petkov <bp@...en8.de>,
Dave Hansen <dave.hansen@...ux.intel.com>, x86@...nel.org,
"H. Peter Anvin" <hpa@...or.com>, Marco Elver <elver@...gle.com>,
Andrey Konovalov <andreyknvl@...il.com>,
Andrey Ryabinin <ryabinin.a.a@...il.com>,
Nathan Chancellor <nathan@...nel.org>,
Nick Desaulniers <ndesaulniers@...gle.com>,
Bill Wendling <morbo@...gle.com>,
Justin Stitt <justinstitt@...gle.com>,
Andrew Morton <akpm@...ux-foundation.org>,
Baoquan He <bhe@...hat.com>,
"Mike Rapoport (IBM)" <rppt@...nel.org>,
Josh Poimboeuf <jpoimboe@...nel.org>,
Changbin Du <changbin.du@...wei.com>,
Rick Edgecombe <rick.p.edgecombe@...el.com>,
Pengfei Xu <pengfei.xu@...el.com>, Jason Gunthorpe <jgg@...pe.ca>,
Xin Li <xin3.li@...el.com>, Uros Bizjak <ubizjak@...il.com>,
"Kirill A. Shutemov" <kirill.shutemov@...ux.intel.com>,
linux-kernel@...r.kernel.org, kasan-dev@...glegroups.com,
linux-hardening@...r.kernel.org, llvm@...ts.linux.dev
Subject: Re: [PATCH v5] x86/traps: Enable UBSAN traps on x86
On Wed, Jul 24, 2024 at 12:01:55AM +0000, Gatlin Newhouse wrote:
> Currently ARM architectures extract which specific sanitizer
> has caused a trap via encoded data in the trap instruction.[1]
> Clang on x86 currently encodes the same data in ud1 instructions
> but the x86 handle_bug() and is_valid_bugaddr() functions
> currently only look at ud2s.
>
> Bring x86 to parity with arm64, similar to commit 25b84002afb9
> ("arm64: Support Clang UBSAN trap codes for better reporting").
> Enable the reporting of UBSAN sanitizer detail on x86 architectures
> compiled with clang when CONFIG_UBSAN_TRAP=y.
>
> [1] Details are in llvm/lib/Target/X86/X86MCInstLower.cpp. See:
> https://github.com/llvm/llvm-project/commit/c5978f42ec8e9#diff-bb68d7cd885f41cfc35843998b0f9f534adb60b415f647109e597ce448e92d9f
>
> EmitAndCountInstruction() uses the UD1Lm template, which uses a
> OpSize32. See:
> https://github.com/llvm/llvm-project/blob/main/llvm/lib/Target/X86/X86InstrSystem.td#L27
>
> Signed-off-by: Gatlin Newhouse <gatlin.newhouse@...il.com>
> ---
> Changes in v5:
> - Added references to the LLVM commits in the commit message from
> Kees and Marco's feedback
> - Renamed incorrect defines, and removed handle_ubsan_failure()'s
> duplicated work per Peter's feedback
Ping to the x86 maintainers... can someone pick this up? It looks like
all the feedback has been addressed.
Thanks!
-Kees
>
> Changes in v4:
> - Implement Peter's suggestions for decode_bug(), and fix
> inconsistent capitalization in hex values.
>
> Changes in v3:
> - Address Thomas's remarks about: change log structure,
> get_ud_type() instead of is_valid_bugaddr(), handle_bug()
> changes, and handle_ubsan_failure().
>
> Changes in v2:
> - Name the new constants 'LEN_ASOP' and 'INSN_ASOP' instead of
> 'LEN_REX' and 'INSN_REX'
> - Change handle_ubsan_failure() from enum bug_trap_type to void
> function
>
> v1: https://lore.kernel.org/linux-hardening/20240529022043.3661757-1-gatlin.newhouse@gmail.com/
> v2: https://lore.kernel.org/linux-hardening/20240601031019.3708758-1-gatlin.newhouse@gmail.com/
> v3: https://lore.kernel.org/linux-hardening/20240625032509.4155839-1-gatlin.newhouse@gmail.com/
> v4: https://lore.kernel.org/linux-hardening/20240710203250.238782-1-gatlin.newhouse@gmail.com/
> ---
> MAINTAINERS | 2 ++
> arch/x86/include/asm/bug.h | 12 ++++++++
> arch/x86/include/asm/ubsan.h | 18 ++++++++++++
> arch/x86/kernel/Makefile | 1 +
> arch/x86/kernel/traps.c | 57 ++++++++++++++++++++++++++++++++----
> arch/x86/kernel/ubsan.c | 19 ++++++++++++
> 6 files changed, 104 insertions(+), 5 deletions(-)
> create mode 100644 arch/x86/include/asm/ubsan.h
> create mode 100644 arch/x86/kernel/ubsan.c
>
> diff --git a/MAINTAINERS b/MAINTAINERS
> index 28e20975c26f..b8512887ffb1 100644
> --- a/MAINTAINERS
> +++ b/MAINTAINERS
> @@ -22635,6 +22635,8 @@ L: kasan-dev@...glegroups.com
> L: linux-hardening@...r.kernel.org
> S: Supported
> T: git git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux.git for-next/hardening
> +F: arch/x86/include/asm/ubsan.h
> +F: arch/x86/kernel/ubsan.c
> F: Documentation/dev-tools/ubsan.rst
> F: include/linux/ubsan.h
> F: lib/Kconfig.ubsan
> diff --git a/arch/x86/include/asm/bug.h b/arch/x86/include/asm/bug.h
> index a3ec87d198ac..751e45ea27ca 100644
> --- a/arch/x86/include/asm/bug.h
> +++ b/arch/x86/include/asm/bug.h
> @@ -13,6 +13,18 @@
> #define INSN_UD2 0x0b0f
> #define LEN_UD2 2
>
> +/*
> + * In clang we have UD1s reporting UBSAN failures on X86, 64 and 32bit.
> + */
> +#define INSN_ASOP 0x67
> +#define OPCODE_ESCAPE 0x0f
> +#define SECOND_BYTE_OPCODE_UD1 0xb9
> +#define SECOND_BYTE_OPCODE_UD2 0x0b
> +
> +#define BUG_NONE 0xffff
> +#define BUG_UD1 0xfffe
> +#define BUG_UD2 0xfffd
> +
> #ifdef CONFIG_GENERIC_BUG
>
> #ifdef CONFIG_X86_32
> diff --git a/arch/x86/include/asm/ubsan.h b/arch/x86/include/asm/ubsan.h
> new file mode 100644
> index 000000000000..1d7c2b4129de
> --- /dev/null
> +++ b/arch/x86/include/asm/ubsan.h
> @@ -0,0 +1,18 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +#ifndef _ASM_X86_UBSAN_H
> +#define _ASM_X86_UBSAN_H
> +
> +/*
> + * Clang Undefined Behavior Sanitizer trap mode support.
> + */
> +#include <linux/bug.h>
> +#include <linux/ubsan.h>
> +#include <asm/ptrace.h>
> +
> +#ifdef CONFIG_UBSAN_TRAP
> +void handle_ubsan_failure(struct pt_regs *regs, u32 type);
> +#else
> +static inline void handle_ubsan_failure(struct pt_regs *regs, u32 type) { return; }
> +#endif /* CONFIG_UBSAN_TRAP */
> +
> +#endif /* _ASM_X86_UBSAN_H */
> diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile
> index 74077694da7d..fe1d9db27500 100644
> --- a/arch/x86/kernel/Makefile
> +++ b/arch/x86/kernel/Makefile
> @@ -145,6 +145,7 @@ obj-$(CONFIG_UNWINDER_GUESS) += unwind_guess.o
> obj-$(CONFIG_AMD_MEM_ENCRYPT) += sev.o
>
> obj-$(CONFIG_CFI_CLANG) += cfi.o
> +obj-$(CONFIG_UBSAN_TRAP) += ubsan.o
>
> obj-$(CONFIG_CALL_THUNKS) += callthunks.o
>
> diff --git a/arch/x86/kernel/traps.c b/arch/x86/kernel/traps.c
> index 4fa0b17e5043..6350d00a6555 100644
> --- a/arch/x86/kernel/traps.c
> +++ b/arch/x86/kernel/traps.c
> @@ -67,6 +67,7 @@
> #include <asm/vdso.h>
> #include <asm/tdx.h>
> #include <asm/cfi.h>
> +#include <asm/ubsan.h>
>
> #ifdef CONFIG_X86_64
> #include <asm/x86_init.h>
> @@ -91,6 +92,45 @@ __always_inline int is_valid_bugaddr(unsigned long addr)
> return *(unsigned short *)addr == INSN_UD2;
> }
>
> +/*
> + * Check for UD1 or UD2, accounting for Address Size Override Prefixes.
> + * If it's a UD1, get the ModRM byte to pass along to UBSan.
> + */
> +__always_inline int decode_bug(unsigned long addr, u32 *imm)
> +{
> + u8 v;
> +
> + if (addr < TASK_SIZE_MAX)
> + return BUG_NONE;
> +
> + v = *(u8 *)(addr++);
> + if (v == INSN_ASOP)
> + v = *(u8 *)(addr++);
> + if (v != OPCODE_ESCAPE)
> + return BUG_NONE;
> +
> + v = *(u8 *)(addr++);
> + if (v == SECOND_BYTE_OPCODE_UD2)
> + return BUG_UD2;
> + if (v != SECOND_BYTE_OPCODE_UD1)
> + return BUG_NONE;
> +
> + v = *(u8 *)(addr++);
> + if (X86_MODRM_RM(v) == 4)
> + addr++;
> +
> + *imm = 0;
> + if (X86_MODRM_MOD(v) == 1)
> + *imm = *(u8 *)addr;
> + else if (X86_MODRM_MOD(v) == 2)
> + *imm = *(u32 *)addr;
> + else
> + WARN_ONCE(1, "Unexpected MODRM_MOD: %u\n", X86_MODRM_MOD(v));
> +
> + return BUG_UD1;
> +}
> +
> +
> static nokprobe_inline int
> do_trap_no_signal(struct task_struct *tsk, int trapnr, const char *str,
> struct pt_regs *regs, long error_code)
> @@ -216,6 +256,8 @@ static inline void handle_invalid_op(struct pt_regs *regs)
> static noinstr bool handle_bug(struct pt_regs *regs)
> {
> bool handled = false;
> + int ud_type;
> + u32 imm;
>
> /*
> * Normally @regs are unpoisoned by irqentry_enter(), but handle_bug()
> @@ -223,7 +265,8 @@ static noinstr bool handle_bug(struct pt_regs *regs)
> * irqentry_enter().
> */
> kmsan_unpoison_entry_regs(regs);
> - if (!is_valid_bugaddr(regs->ip))
> + ud_type = decode_bug(regs->ip, &imm);
> + if (ud_type == BUG_NONE)
> return handled;
>
> /*
> @@ -236,10 +279,14 @@ static noinstr bool handle_bug(struct pt_regs *regs)
> */
> if (regs->flags & X86_EFLAGS_IF)
> raw_local_irq_enable();
> - if (report_bug(regs->ip, regs) == BUG_TRAP_TYPE_WARN ||
> - handle_cfi_failure(regs) == BUG_TRAP_TYPE_WARN) {
> - regs->ip += LEN_UD2;
> - handled = true;
> + if (ud_type == BUG_UD2) {
> + if (report_bug(regs->ip, regs) == BUG_TRAP_TYPE_WARN ||
> + handle_cfi_failure(regs) == BUG_TRAP_TYPE_WARN) {
> + regs->ip += LEN_UD2;
> + handled = true;
> + }
> + } else {
> + handle_ubsan_failure(regs, imm);
> }
> if (regs->flags & X86_EFLAGS_IF)
> raw_local_irq_disable();
> diff --git a/arch/x86/kernel/ubsan.c b/arch/x86/kernel/ubsan.c
> new file mode 100644
> index 000000000000..63f819928820
> --- /dev/null
> +++ b/arch/x86/kernel/ubsan.c
> @@ -0,0 +1,19 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Clang Undefined Behavior Sanitizer trap mode support.
> + */
> +#include <linux/bug.h>
> +#include <linux/string.h>
> +#include <linux/printk.h>
> +#include <linux/ubsan.h>
> +#include <asm/ptrace.h>
> +#include <asm/ubsan.h>
> +
> +/*
> + * Checks for the information embedded in the UD1 trap instruction
> + * for the UB Sanitizer in order to pass along debugging output.
> + */
> +void handle_ubsan_failure(struct pt_regs *regs, u32 type)
> +{
> + pr_crit("%s at %pS\n", report_ubsan_failure(regs, type), (void *)regs->ip);
> +}
> --
> 2.25.1
>
--
Kees Cook
Powered by blists - more mailing lists