linux-kernel - Re: [RFC PATCH v2 2/2] selftests/x86: sysret_rip: Add more syscall tests with respect to `%rcx` and `%r11`

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: <BC632CA8-D2CB-4781-82E5-9810347293B0@zytor.com>
Date:   Tue, 24 Jan 2023 01:07:38 -0800
From:   "H. Peter Anvin" <hpa@...or.com>
To:     Ammar Faizi <ammarfaizi2@...weeb.org>
CC:     x86 Mailing List <x86@...nel.org>,
        Dave Hansen <dave.hansen@...el.com>,
        Dave Hansen <dave.hansen@...ux.intel.com>,
        Xin Li <xin3.li@...el.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Andrew Cooper <Andrew.Cooper3@...rix.com>,
        Brian Gerst <brgerst@...il.com>,
        Ingo Molnar <mingo@...hat.com>, Borislav Petkov <bp@...en8.de>,
        Peter Zijlstra <peterz@...radead.org>,
        Shuah Khan <shuah@...nel.org>, Ingo Molnar <mingo@...nel.org>,
        Andy Lutomirski <luto@...nel.org>,
        "Kirill A. Shutemov" <kirill.shutemov@...ux.intel.com>,
        Linux Kselftest Mailing List 
        <linux-kselftest@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [RFC PATCH v2 2/2] selftests/x86: sysret_rip: Add more syscall tests with respect to `%rcx` and `%r11`

On January 23, 2023 10:41:20 PM PST, Ammar Faizi <ammarfaizi2@...weeb.org> wrote:
>On Mon, Jan 23, 2023 at 10:16:01PM -0800, H. Peter Anvin wrote:
>> On 1/23/23 18:27, Ammar Faizi wrote:
>> > Test that:
>> > 
>> >   - "syscall" in a FRED system doesn't clobber %rcx and %r11.
>> >   - "syscall" in a non-FRED system sets %rcx=%rip and %r11=%rflags.
>> > 
>> > Test them out with a trivial system call like __NR_getppid and friends
>> > which are extremely likely to return with SYSRET on an IDT system; check
>> > that it returns a nonnegative value and then save the result.
>> > 
>> 
>> "Nonnegative" here should be "valid"; it is an implementation detail that
>> the error value is -1.
>
>Copy-paste error, will fix that!
>
>> However, you are not checking that you don't get a mix of REGS_SAVED and
>> REGS_SYSRET, which is a major part of the point.
>
>Good point!
>
>What do you think of adding this on top of patch #1?
>
>diff --git a/tools/testing/selftests/x86/sysret_rip.c b/tools/testing/selftests/x86/sysret_rip.c
>index 75c72d34dbc5840c..3da827713831acbc 100644
>--- a/tools/testing/selftests/x86/sysret_rip.c
>+++ b/tools/testing/selftests/x86/sysret_rip.c
>@@ -47,11 +47,14 @@ static const unsigned long rcx_sentinel = 0x5ca1ab1e0b57ac1e;
> static const unsigned long rflags_sentinel = 0x200a93;
> 
> enum regs_ok {
>-	REGS_ERROR  = -1,	/* Invalid register contents */
>-	REGS_SAVED  =  0,	/* Registers properly preserved */
>-	REGS_SYSRET =  1	/* Registers match syscall/sysret */
>+	REGS_INIT_VAL	= -2,	/* For init value checker, never returned */
>+	REGS_ERROR 	= -1,	/* Invalid register contents */
>+	REGS_SAVED 	=  0,	/* Registers properly preserved */
>+	REGS_SYSRET	=  1	/* Registers match syscall/sysret */
> };
> 
>+static enum regs_ok regs_ok_state = REGS_INIT_VAL;
>+
> /*
>  * Returns:
>  *  0 = %rcx and %r11 preserved.
>@@ -86,6 +89,7 @@ static long do_syscall(long nr_syscall, unsigned long arg1, unsigned long arg2,
> 	register unsigned long r9 asm("%r9");
> 	register void *rsp asm("%rsp");
> 	unsigned long rcx, rbx;
>+	enum regs_ok ret;
> 
> 	r11 = r11_sentinel;
> 	rcx = rcx_sentinel;
>@@ -124,7 +128,14 @@ static long do_syscall(long nr_syscall, unsigned long arg1, unsigned long arg2,
> 	 * - "syscall" in a non-FRED system sets %rcx=%rip and %r11=%rflags.
> 	 *
> 	 */
>-	assert(check_regs_result(r11, rcx, rbx) != REGS_ERROR);
>+	ret = check_regs_result(r11, rcx, rbx);
>+	assert(ret != REGS_ERROR);
>+
>+	if (regs_ok_state == REGS_INIT_VAL)
>+		regs_ok_state = ret;
>+	else
>+		assert(ret == regs_ok_state);
>+
> 	return nr_syscall;
> }
> 

Works for me, although I would call it REGS_UNDEFINED myself.