linux-kernel - Re: [PATCHv3] x86/mm: set x32 syscall bit in SET

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-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <1cf1ef57-94b5-68d2-4a44-0247f49d6990@virtuozzo.com>
Date:   Wed, 22 Mar 2017 16:40:08 +0300
From:   Dmitry Safonov <dsafonov@...tuozzo.com>
To:     Thomas Gleixner <tglx@...utronix.de>, <hpa@...or.com>
CC:     <linux-kernel@...r.kernel.org>, <0x7f454c46@...il.com>,
        Adam Borowski <kilobyte@...band.pl>, <linux-mm@...ck.org>,
        Andrei Vagin <avagin@...il.com>,
        Cyrill Gorcunov <gorcunov@...nvz.org>,
        Borislav Petkov <bp@...e.de>,
        "Kirill A. Shutemov" <kirill.shutemov@...ux.intel.com>,
        <x86@...nel.org>, Andy Lutomirski <luto@...nel.org>,
        Ingo Molnar <mingo@...hat.com>
Subject: Re: [PATCHv3] x86/mm: set x32 syscall bit in SET_PERSONALITY()

On 03/22/2017 01:34 AM, Thomas Gleixner wrote:
> On Tue, 21 Mar 2017, hpa@...or.com wrote:
>
>> On March 21, 2017 3:21:13 PM PDT, Thomas Gleixner <tglx@...utronix.de> wrote:
>>> On Tue, 21 Mar 2017, Dmitry Safonov wrote:
>>>> v3:
>>>> - clear x32 syscall flag during x32 -> x86-64 exec() (thanks, HPA).
>>>
>>> For correctness sake, this wants to be cleared in the IA32 path as
>>> well. It's not causing any harm, but ....
>>>
>>> I'll amend the patch.

Indeed, thanks!

>> Since the i386 syscall namespace is totally separate (and different),
>> should we simply change the system call number to the appropriate
>> sys_execve number?
>
> That should work as well and would be more intuitive.

Not sure that I got the idea correctly, something like this?
I haven't find any easy way to get compat syscall nr like
__NR_compat_execve, so I defined it there.
I'll resend v4 with the fixup if that's what was expected.

--->8---
diff --git a/arch/x86/kernel/process_64.c b/arch/x86/kernel/process_64.c
index b03f186369eb..c58ac0bff2f1 100644
--- a/arch/x86/kernel/process_64.c
+++ b/arch/x86/kernel/process_64.c
@@ -507,6 +507,8 @@ void set_personality_64bit(void)
  	current->personality &= ~READ_IMPLIES_EXEC;
  }

+#define __NR_ia32_execve	11
+
  void set_personality_ia32(bool x32)
  {
  	/* inherit personality from parent */
@@ -537,6 +539,7 @@ void set_personality_ia32(bool x32)
  			current->mm->context.ia32_compat = TIF_IA32;
  		current->personality |= force_personality32;
  		/* Prepare the first "return" to user space */
+		task_pt_regs(current)->orig_ax = __NR_ia32_execve;
  		current->thread.status |= TS_COMPAT;
  	}
  }