ARM64: Disabling warnings about deprecated armv8 instructions

Michael Zoran mzoran at crowfest.net
Sun Jan 22 03:05:15 PST 2017


On Sun, 2017-01-22 at 10:54 +0000, Ard Biesheuvel wrote:
> On 22 January 2017 at 09:52, Michael Zoran <mzoran at crowfest.net>
> wrote:
> > On Sun, 2017-01-22 at 09:43 +0000, Ard Biesheuvel wrote:
> > > On 22 January 2017 at 09:38, Michael Zoran <mzoran at crowfest.net>
> > > wrote:
> > > > On Sun, 2017-01-22 at 17:05 +0800, Jisheng Zhang wrote:
> > > > > On Sun, 22 Jan 2017 00:58:48 -0800 Michael Zoran wrote:
> > > > > 
> > > > > > On Sun, 2017-01-22 at 09:52 +0100, Alexander Stein wrote:
> > > > > > > Hi Michael,
> > > > > > > 
> > > > > > > On Sunday, January 22, 2017, 12:07:04 AM CET Michael
> > > > > > > Zoran
> > > > > > > wrote:
> > > > > > > > I'm not sure if this if the correct place to be asking
> > > > > > > > this.   The
> > > > > > > > RPI
> > > > > > > > 3 running ARM64 is slowly reaching the point of being
> > > > > > > > about
> > > > > > > > to
> > > > > > > > seriously run a 32 bit vender OS like Raspbian.  When
> > > > > > > > running
> > > > > > > > Raspbian,
> > > > > > > > I'm seeing a very large number(thousands) of kernel log
> > > > > > > > messages
> > > > > > > > about
> > > > > > > > deprecated instructions especially setend and barrier
> > > > > > > > instuctions.
> > > > > > > > This can be very annoying and is completely filling the
> > > > > > > > kernel
> > > > > > > > log.
> > > > > > > > 
> > > > > > > > I'm considering submitting a patch to add a Kconfig
> > > > > > > > option
> > > > > > > > to
> > > > > > > > disable
> > > > > > > > these warnings with the default being to keep the
> > > > > > > > warnings
> > > > > > > > enabled.  I
> > > > > > > > was wondering if such a patch could be seriously
> > > > > > > > considered.
> > > > > > > 
> > > > > > > Could you please provide an example of those warning an
> > > > > > > what
> > > > > > > is
> > > > > > > trigging
> > > > > > > those?
> > > > > > > 
> > > > > > > Thanks and best regards,
> > > > > > > Alexander
> > > > > > 
> > > > > > Sure, here is a snipped from dmesg.  I think this is
> > > > > > happening
> > > > > > because
> > > > > > the entire Raspbian OS is compiled with a custom gcc
> > > > > > compiler
> > > > > > that
> > > > > > is
> > > > > > targeting arm6+VFP.
> > > > > 
> > > > > IMHO, this is the root cause. I didn't see below warnings
> > > > > from
> > > > > dmesg
> > > > > with
> > > > > armhf debian and ubuntu If the underlying HW is armv8
> > > > > compatible,
> > > > > it
> > > > > doesn't make sense to target the compiler to armv6
> > > > > 
> > > > > 
> > > > 
> > > > Advanced users can install a different OS. Beginners can't and
> > > > will
> > > > need a single OS image for simple install.
> > > > 
> > > 
> > > That single OS image can simply test for the presence of
> > > /proc/sys/abi, and perform the writes if it is there.
> > > 
> > > > Does anybody know if these instructions are detected at
> > > > runtime?
> > > 
> > > What do you mean by 'detected'?
> > > 
> > > > I'm
> > > > also looking into an issue with Mathmatica due to broken OS
> > > > detection.
> > > > It appears that arm64 is currently missing some of the CPU
> > > > spoofing
> > > > that happens on other architectures...
> > > > 
> > > 
> > > Could you elaborate? What do you think should be there that
> > > isn't?
> > 
> > I'm saying that when running a ELF32 executable, the system needs
> > to
> > fake the CPU architecture according to the uname system API and
> > /proc/cpuinfo.
> > 
> > Today, I run uname -a in Raspbian and I get aarch64.  I want it to
> > display arm6l or whatever to get Mathematica to load.
> > 
> > I see some code to make this work on other architectures such as
> > X86,
> > but most of this appears to be unimplemented on arm64.
> > 
> 
> Did you try linux32?
> 
> pi at raspberrypi:~$ uname -a
> Linux raspberrypi 4.10.0-rc3-v8+ #3 SMP PREEMPT Sun Jan 15 15:22:12
> WET 2017 aarch64 GNU/Linux
> pi at raspberrypi:~ $ linux32 uname -a
> Linux raspberrypi 4.10.0-rc3-v8+ #3 SMP PREEMPT Sun Jan 15 15:22:12
> WET 2017 armv8l GNU/Linux
> 
> If the app in question cannot deal with the armv8l personality, the
> app should be fixed instead.

Thanks, I didn't know about the tool.

I'm looking at the code in arch/arm64/include/asm/elf.h. I'm looking at
the COMPAT_SET_PERSONALITY macro.  It appears to be broken.  It seems 
like it should have a call to set_personality with PER_LINUX32.  

If I look in include/uapi/linux/personality.h, I see some very
interesting things. Quite the complex set of options I would say.

/*
 * Flags for bug emulation.
 *
 * These occupy the top three bytes.
 */
enum {
	UNAME26	=               0x0020000,
	ADDR_NO_RANDOMIZE = 	0x0040000,	/* disable
randomization of VA space */
	FDPIC_FUNCPTRS =	0x0080000,	/* userspace function
ptrs point to descriptors
						 * (signal handling)
						 */
	MMAP_PAGE_ZERO =	0x0100000,
	ADDR_COMPAT_LAYOUT =	0x0200000,
	READ_IMPLIES_EXEC =	0x0400000,
	ADDR_LIMIT_32BIT =	0x0800000,
	SHORT_INODE =		0x1000000,
	WHOLE_SECONDS =		0x2000000,
	STICKY_TIMEOUTS	=	0x4000000,
	ADDR_LIMIT_3GB = 	0x8000000,
};

/*
 * Security-relevant compatibility flags that must be
 * cleared upon setuid or setgid exec:
 */
#define PER_CLEAR_ON_SETID (READ_IMPLIES_EXEC  | \
			    ADDR_NO_RANDOMIZE  | \
			    ADDR_COMPAT_LAYOUT | \
			    MMAP_PAGE_ZERO)

/*
 * Personality types.
 *
 * These go in the low byte.  Avoid using the top bit, it will
 * conflict with error returns.
 */
enum {
	PER_LINUX =		0x0000,
	PER_LINUX_32BIT =	0x0000 | ADDR_LIMIT_32BIT,
	PER_LINUX_FDPIC =	0x0000 | FDPIC_FUNCPTRS,
	PER_SVR4 =		0x0001 | STICKY_TIMEOUTS |
MMAP_PAGE_ZERO,
	PER_SVR3 =		0x0002 | STICKY_TIMEOUTS |
SHORT_INODE,
	PER_SCOSVR3 =		0x0003 | STICKY_TIMEOUTS |
					 WHOLE_SECONDS | SHORT_INODE,
	PER_OSR5 =		0x0003 | STICKY_TIMEOUTS |
WHOLE_SECONDS,
	PER_WYSEV386 =		0x0004 | STICKY_TIMEOUTS |
SHORT_INODE,
	PER_ISCR4 =		0x0005 | STICKY_TIMEOUTS,
	PER_BSD =		0x0006,
	PER_SUNOS =		0x0006 | STICKY_TIMEOUTS,
	PER_XENIX =		0x0007 | STICKY_TIMEOUTS |
SHORT_INODE,
	PER_LINUX32 =		0x0008,
	PER_LINUX32_3GB =	0x0008 | ADDR_LIMIT_3GB,
	PER_IRIX32 =		0x0009 | STICKY_TIMEOUTS,/* IRIX5
32-bit */
	PER_IRIXN32 =		0x000a | STICKY_TIMEOUTS,/* IRIX6
new 32-bit */
	PER_IRIX64 =		0x000b | STICKY_TIMEOUTS,/* IRIX6
64-bit */
	PER_RISCOS =		0x000c,
	PER_SOLARIS =		0x000d | STICKY_TIMEOUTS,
	PER_UW7 =		0x000e | STICKY_TIMEOUTS |
MMAP_PAGE_ZERO,
	PER_OSF4 =		0x000f,			 /*
OSF/1 v4 */
	PER_HPUX =		0x0010,
	PER_MASK =		0x00ff,
};





More information about the linux-rpi-kernel mailing list