| 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: <alpine.LSU.2.21.1810171051560.21847@pobox.suse.cz>
Date: Wed, 17 Oct 2018 15:39:31 +0200 (CEST)
From: Miroslav Benes <mbenes@...e.cz>
To: Torsten Duwe <duwe@....de>
cc: Will Deacon <will.deacon@....com>,
Catalin Marinas <catalin.marinas@....com>,
Julien Thierry <julien.thierry@....com>,
Steven Rostedt <rostedt@...dmis.org>,
Josh Poimboeuf <jpoimboe@...hat.com>,
Ingo Molnar <mingo@...hat.com>,
Ard Biesheuvel <ard.biesheuvel@...aro.org>,
Arnd Bergmann <arnd@...db.de>,
AKASHI Takahiro <takahiro.akashi@...aro.org>,
linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org,
live-patching@...r.kernel.org, jeyu@...nel.org
Subject: Re: [PATCH v3 3/4] arm64: implement live patching
On Mon, 1 Oct 2018, Torsten Duwe wrote:
> Based on ftrace with regs, do the usual thing. Also allocate a
> task flag for whatever consistency handling will be used.
> Watch out for interactions with the graph tracer.
Similar to what Mark wrote about 2/4, I'd appreciate a better commit log.
Could you explain traditional "what/why/how", please?
> Signed-off-by: Torsten Duwe <duwe@...e.de>
>
> --- a/arch/arm64/Kconfig
> +++ b/arch/arm64/Kconfig
> @@ -119,6 +119,7 @@ config ARM64
> select HAVE_GENERIC_DMA_COHERENT
> select HAVE_HW_BREAKPOINT if PERF_EVENTS
> select HAVE_IRQ_TIME_ACCOUNTING
> + select HAVE_LIVEPATCH
> select HAVE_MEMBLOCK
> select HAVE_MEMBLOCK_NODE_MAP if NUMA
> select HAVE_NMI
> @@ -1349,4 +1350,6 @@ if CRYPTO
> source "arch/arm64/crypto/Kconfig"
> endif
>
> +source "kernel/livepatch/Kconfig"
> +
> source "lib/Kconfig"
> --- a/arch/arm64/include/asm/thread_info.h
> +++ b/arch/arm64/include/asm/thread_info.h
> @@ -76,6 +76,7 @@ void arch_release_task_struct(struct tas
> #define TIF_FOREIGN_FPSTATE 3 /* CPU's FP state is not current's */
> #define TIF_UPROBE 4 /* uprobe breakpoint or singlestep */
> #define TIF_FSCHECK 5 /* Check FS is USER_DS on return */
> +#define TIF_PATCH_PENDING 6
> #define TIF_NOHZ 7
> #define TIF_SYSCALL_TRACE 8
> #define TIF_SYSCALL_AUDIT 9
> @@ -94,6 +95,7 @@ void arch_release_task_struct(struct tas
> #define _TIF_NEED_RESCHED (1 << TIF_NEED_RESCHED)
> #define _TIF_NOTIFY_RESUME (1 << TIF_NOTIFY_RESUME)
> #define _TIF_FOREIGN_FPSTATE (1 << TIF_FOREIGN_FPSTATE)
> +#define _TIF_PATCH_PENDING (1 << TIF_PATCH_PENDING)
> #define _TIF_NOHZ (1 << TIF_NOHZ)
> #define _TIF_SYSCALL_TRACE (1 << TIF_SYSCALL_TRACE)
> #define _TIF_SYSCALL_AUDIT (1 << TIF_SYSCALL_AUDIT)
> @@ -106,7 +108,8 @@ void arch_release_task_struct(struct tas
>
> #define _TIF_WORK_MASK (_TIF_NEED_RESCHED | _TIF_SIGPENDING | \
> _TIF_NOTIFY_RESUME | _TIF_FOREIGN_FPSTATE | \
> - _TIF_UPROBE | _TIF_FSCHECK)
> + _TIF_UPROBE | _TIF_FSCHECK | \
> + _TIF_PATCH_PENDING)
Could you add a note to the changelog what this means? My ability to read
arm64 entry.S is very limited, but I can see that _TIF_WORK_MASK is
process in a syscall exit and irq return paths. That's good. It is also
called (via "b ret_to_user") in a couple of different places (el0_*
labels). I guess those are returns from exception handling. A comment
about it in the changelog would be appreciated.
> #define _TIF_SYSCALL_WORK (_TIF_SYSCALL_TRACE | _TIF_SYSCALL_AUDIT | \
> _TIF_SYSCALL_TRACEPOINT | _TIF_SECCOMP | \
> --- /dev/null
> +++ b/arch/arm64/include/asm/livepatch.h
> @@ -0,0 +1,36 @@
> +/* SPDX-License-Identifier: GPL-2.0
> + *
> + * livepatch.h - arm64-specific Kernel Live Patching Core
> + *
> + * Copyright (C) 2016,2018 SUSE
> + *
> + * This program is free software; you can redistribute it and/or
> + * modify it under the terms of the GNU General Public License
> + * as published by the Free Software Foundation; either version 2
> + * of the License, or (at your option) any later version.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> + * GNU General Public License for more details.
> + *
> + * You should have received a copy of the GNU General Public License
> + * along with this program; if not, see <http://www.gnu.org/licenses/>.
> + */
> +#ifndef _ASM_ARM64_LIVEPATCH_H
> +#define _ASM_ARM64_LIVEPATCH_H
> +
> +#include <linux/module.h>
> +#include <linux/ftrace.h>
I think that only
#include <asm/ptrace.h>
is in fact needed, because of "struct pt_regs".
Ad relocations. I checked that everything in struct mod_arch_specific
stays after the module is load. Both core and init get SHF_ALLOC set
(mod->arch.core.plt->sh_flags in module_frob_arch_sections(). It is
important because apply_relocate_add() may use those sections
through module_emit_plt_entry() call.
ftrace_trampoline section gets SHF_ALLOC as well. Btw, don't you want to
do something similar for new ftrace_regs_trampoline in
module_frob_arch_sections()? Perhaps even edit
arch/arm64/kernel/module.lds? I'm completely clueless here, so it may be
ok. And it applies to 2/4 patch.
The last thing is count_plts() function called from
module_frob_arch_sections(). It needed to be changed in 2016 as well. See
Jessica's patch
(20160713001113.GA30925@...ker-debian-8-amd64.digitalocean.com). The logic
in the function has changed since then. If I am not mistaken,
count_plts() is fine as it is right now. It does not consider SHN_UNDEF
anymore, it looks at the destination section (where a symbol should
resolved to) only.
Jessica, could you doublecheck, please?
Torsten, please doublecheck all this as well and add a comment about all
this to the changelog, so we don't have to analyze it from the beginning
again in the future.
Thanks,
Miroslav
Powered by blists - more mailing lists