From: Roman Peniaev <r.peniaev@gmail.com>
To: Will Deacon <will.deacon@arm.com>
Cc: Russell King <linux@arm.linux.org.uk>,
Stefano Stabellini <stefano.stabellini@eu.citrix.com>,
Marc Zyngier <Marc.Zyngier@arm.com>,
Catalin Marinas <Catalin.Marinas@arm.com>,
Sekhar Nori <nsekhar@ti.com>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
"linux-arm-kernel@lists.infradead.org"
<linux-arm-kernel@lists.infradead.org>,
"stable@vger.kernel.org" <stable@vger.kernel.org>,
Christoffer Dall <christoffer.dall@linaro.org>,
Kees Cook <keescook@chromium.org>
Subject: Re: [PATCH 1/2] ARM: entry-common: fix forgotten set of thread_info->syscall
Date: Wed, 14 Jan 2015 11:23:06 +0900 [thread overview]
Message-ID: <CACZ9PQW9aXuCd7T7ci7PmsR7uvbXtzcs6NjNJeMKAu8SF4MfiA@mail.gmail.com> (raw)
In-Reply-To: <CACZ9PQXW9G4Vw-dcPwkdVE=uoYmDYEQnJWDnnpup1LxC3s5QPg@mail.gmail.com>
On Tue, Jan 13, 2015 at 5:35 PM, Roman Peniaev <r.peniaev@gmail.com> wrote:
> On Tue, Jan 13, 2015 at 3:39 AM, Will Deacon <will.deacon@arm.com> wrote:
>> On Sun, Jan 11, 2015 at 02:32:30PM +0000, Roman Pen wrote:
>>> thread_info->syscall is used only for ptrace, but syscall number
>>> is also used by syscall_get_nr and returned to userspace by the
>>> following proc file access:
>>>
>>> $ cat /proc/self/syscall
>>> 0 0x3 0xbe928bd8 0x1000 0x0 0xac9e0 0x3 0xbe928bb4 0xb6f5dfbc
>>> ^
>>> The first number is the syscall number, currently it is zero.
>>> Patch fixes this:
>>>
>>> $ cat /proc/self/syscall
>>> 3 0x3 0xbefc7bd8 0x1000 0x0 0xac9e0 0x3 0xbefc7bb4 0xb6e82fbc
>>> ^
>>> Right, read syscall
>>
>> Yes, it seems that despite requiring CONFIG_HAVE_ARCH_TRACEHOOK,
>> the /proc code requires syscall_get_nr to work regardless of
>> TIF_SYSCALL_TRACE.
>>
>>> Signed-off-by: Roman Pen <r.peniaev@gmail.com>
>>> Cc: Russell King <linux@arm.linux.org.uk>
>>> Cc: Marc Zyngier <marc.zyngier@arm.com>
>>> Cc: Catalin Marinas <catalin.marinas@arm.com>
>>> Cc: Christoffer Dall <christoffer.dall@linaro.org>
>>> Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
>>> Cc: Sekhar Nori <nsekhar@ti.com>
>>> Cc: linux-arm-kernel@lists.infradead.org
>>> Cc: linux-kernel@vger.kernel.org
>>> Cc: stable@vger.kernel.org
>>> ---
>>> arch/arm/kernel/asm-offsets.c | 1 +
>>> arch/arm/kernel/entry-common.S | 1 +
>>> 2 files changed, 2 insertions(+)
>>>
>>> diff --git a/arch/arm/kernel/asm-offsets.c b/arch/arm/kernel/asm-offsets.c
>>> index 2d2d608..6911bad 100644
>>> --- a/arch/arm/kernel/asm-offsets.c
>>> +++ b/arch/arm/kernel/asm-offsets.c
>>> @@ -70,6 +70,7 @@ int main(void)
>>> DEFINE(TI_CPU, offsetof(struct thread_info, cpu));
>>> DEFINE(TI_CPU_DOMAIN, offsetof(struct thread_info, cpu_domain));
>>> DEFINE(TI_CPU_SAVE, offsetof(struct thread_info, cpu_context));
>>> + DEFINE(TI_SYSCALL, offsetof(struct thread_info, syscall));
>>> DEFINE(TI_USED_CP, offsetof(struct thread_info, used_cp));
>>> DEFINE(TI_TP_VALUE, offsetof(struct thread_info, tp_value));
>>> DEFINE(TI_FPSTATE, offsetof(struct thread_info, fpstate));
>>> diff --git a/arch/arm/kernel/entry-common.S b/arch/arm/kernel/entry-common.S
>>> index f8ccc21..89452ff 100644
>>> --- a/arch/arm/kernel/entry-common.S
>>> +++ b/arch/arm/kernel/entry-common.S
>>> @@ -189,6 +189,7 @@ ENTRY(vector_swi)
>>> #endif
>>>
>>> local_restart:
>>> + str scno, [tsk, #TI_SYSCALL] @ set syscall number
>>> ldr r10, [tsk, #TI_FLAGS] @ check for syscall tracing
>>> stmdb sp!, {r4, r5} @ push fifth and sixth args
>>
>> Do we definitely want to update scno on syscall restarting?
>
>
> Good question.
>
> First thing to mention is __sys_trace will trace 'restart_syscall',
> not the real syscall we are going to restart.
>
> E.g. in test application we do infinite poll and then send STOP and
> CONT to this app:
>
> test-243 [002] ...1 1792.067726: sys_enter: NR 168 (0, 0,
> ffffffff, 0, 0, 0)
> test-243 [002] ...1 1802.299073: sys_exit: NR 168 = -516
> test-243 [004] ...1 1814.716264: sys_enter: NR 0 (0, 0,
> ffffffff, 0, 0, 0)
> test-243 [004] ...1 2183.687225: sys_exit: NR 0 = -516
>
> the poll was restarted and trace shows that we are in restart_syscall.
>
> Is that expected?
>
> And the second thing is that my next patch did some tweaks in
> 'syscall_trace_enter', where we take scno not from param we passed,
> but from thread_info->syscall we previously set.
>
> So, regarding your question, if I set scno only once - I will break
> previous behavior, and __sys_trace will trace the syscall we restarted.
>
> And I think this is what we need, because according to the
> 'syscall_trace_enter' code we do 'secure_computing' and
> 'audit_syscall_entry', which definitely expect original syscall, not
> the 'restart_syscall'.
+Kees Cook
Moreover, I tested on x86, and behavior is the same:
if we do restart_syscall userspace sees 0 as scnr.
(for me it sounds strange, because as I understand userspace
should avoid any awareness about possible syscall restart)
x86:
[root@qemu-x86 ~]# ./test.x86 &
[1] 285
[root@qemu-x86 ~]# cat /proc/285/syscall
168 0x0 0x0 0xffffffff 0x0 0xbf9387ac 0xbf9387b8 0xbf9386f8 0xffffe424
[root@qemu-x86 ~]# kill -STOP 285
[root@qemu-x86 ~]#
[1]+ Stopped ./test.x86
[root@qemu-x86 ~]# kill -CONT 285
[root@qemu-x86 ~]# cat /proc/285/syscall
0 0x0 0x0 0xffffffff 0x0 0xbf9387ac 0xbf9387b8 0xbf9386f8 0xffffe424
And getting syscall tracing on another run of this test application:
[root@qemu-x86 ~]# cat /sys/kernel/debug/tracing/trace | grep test.x86
.....
test.x86-279 [000] ...1 189.926878: sys_enter: NR 168 (0, 0,
ffffffff, 0, bfe6824c, bfe68258)
test.x86-279 [001] ...1 199.931406: sys_exit: NR 168 = -516
test.x86-279 [001] ...1 203.446946: sys_enter: NR 0 (0, 0,
ffffffff, 0, bfe6824c, bfe68258)
test.x86-279 [000] ...1 276.294314: sys_exit: NR 0 = -516
.....
ARM behavior is the same with these two patches:
sh-3.2# /tmp/test &
[1] 178
sh-3.2# cat /proc/178/syscall
168 0x0 0x0 0xffffffff 0x0 0x0 0x0 0xbdcdece4 0x4220a6cc
sh-3.2# kill -STOP 178
sh-3.2#
[1]+ Stopped(SIGSTOP) /tmp/test
sh-3.2# cat /proc/178/syscall
168 0x0 0x0 0xffffffff 0x0 0x0 0x0 0xbdcdece4 0x4220a6c8
sh-3.2# kill -CONT 178
sh-3.2# cat /proc/178/syscall
0 0x0 0x0 0xffffffff 0x0 0x0 0x0 0xbdcdece4 0x4220a6cc
--
Roman
next prev parent reply other threads:[~2015-01-14 2:23 UTC|newest]
Thread overview: 28+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-01-11 14:32 [PATCH 0/2] ARM: set thread_info->syscall just before sys_* execution Roman Pen
2015-01-11 14:32 ` [PATCH 1/2] ARM: entry-common: fix forgotten set of thread_info->syscall Roman Pen
2015-01-12 18:39 ` Will Deacon
2015-01-13 8:35 ` Roman Peniaev
2015-01-14 2:23 ` Roman Peniaev [this message]
2015-01-14 20:51 ` Kees Cook
2015-01-15 1:54 ` Roman Peniaev
2015-01-15 22:54 ` Kees Cook
2015-01-16 15:57 ` Roman Peniaev
2015-01-16 15:59 ` Russell King - ARM Linux
2015-01-16 16:08 ` Roman Peniaev
2015-01-16 16:17 ` Russell King - ARM Linux
2015-01-16 19:57 ` Kees Cook
2015-01-16 23:54 ` Kees Cook
2015-01-19 5:58 ` Roman Peniaev
2015-01-20 18:56 ` Kees Cook
2015-01-19 9:20 ` Will Deacon
2015-01-20 18:31 ` Kees Cook
2015-01-20 22:45 ` Russell King - ARM Linux
2015-01-20 23:04 ` Russell King - ARM Linux
2015-01-21 23:32 ` Kees Cook
2015-01-22 1:24 ` Roman Peniaev
2015-01-22 18:07 ` Kees Cook
2015-01-23 4:17 ` Roman Peniaev
2015-01-11 14:32 ` [PATCH 2/2] ARM: entry-common,ptrace: do not pass scno to syscall_trace_enter Roman Pen
2015-01-13 20:08 ` Kees Cook
2015-01-13 23:21 ` Roman Peniaev
2015-01-13 23:43 ` Kees Cook
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CACZ9PQW9aXuCd7T7ci7PmsR7uvbXtzcs6NjNJeMKAu8SF4MfiA@mail.gmail.com \
--to=r.peniaev@gmail.com \
--cc=Catalin.Marinas@arm.com \
--cc=Marc.Zyngier@arm.com \
--cc=christoffer.dall@linaro.org \
--cc=keescook@chromium.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux@arm.linux.org.uk \
--cc=nsekhar@ti.com \
--cc=stable@vger.kernel.org \
--cc=stefano.stabellini@eu.citrix.com \
--cc=will.deacon@arm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).