On Wed, Apr 24, 2019 at 05:40:33PM +0100, Mark Rutland wrote:

On Wed, Apr 24, 2019 at 11:25:00AM -0400, Mathieu Desnoyers wrote:

...

Handle compiling with -mbig-endian on aarch64, which generates binaries with mixed code vs data endianness (little endian code, big endian data).

Else mismatch between code endianness for the generated signatures and data endianness for the RSEQ_SIG parameter passed to the rseq registration will trigger application segmentation faults when the kernel try to abort rseq critical sections.

Signed-off-by: Mathieu Desnoyers mathieu.desnoyers@efficios.com CC: Peter Zijlstra peterz@infradead.org CC: Thomas Gleixner tglx@linutronix.de CC: Joel Fernandes joelaf@google.com CC: Catalin Marinas catalin.marinas@arm.com CC: Dave Watson davejwatson@fb.com CC: Will Deacon will.deacon@arm.com CC: Shuah Khan shuah@kernel.org CC: Andi Kleen andi@firstfloor.org CC: linux-kselftest@vger.kernel.org CC: "H . Peter Anvin" hpa@zytor.com CC: Chris Lameter cl@linux.com CC: Russell King linux@arm.linux.org.uk CC: Michael Kerrisk mtk.manpages@gmail.com CC: "Paul E . McKenney" paulmck@linux.vnet.ibm.com CC: Paul Turner pjt@google.com CC: Boqun Feng boqun.feng@gmail.com CC: Josh Triplett josh@joshtriplett.org CC: Steven Rostedt rostedt@goodmis.org CC: Ben Maurer bmaurer@fb.com CC: linux-api@vger.kernel.org CC: Andy Lutomirski luto@amacapital.net CC: Andrew Morton akpm@linux-foundation.org CC: Linus Torvalds torvalds@linux-foundation.org

---

tools/testing/selftests/rseq/rseq-arm64.h | 17 +++++++++++++++-- 1 file changed, 15 insertions(+), 2 deletions(-)

diff --git a/tools/testing/selftests/rseq/rseq-arm64.h b/tools/testing/selftests/rseq/rseq-arm64.h index b41a2a48e965..200dae9e4208 100644 --- a/tools/testing/selftests/rseq/rseq-arm64.h +++ b/tools/testing/selftests/rseq/rseq-arm64.h @@ -6,7 +6,20 @@

• (C) Copyright 2018 - Will Deacon will.deacon@arm.com

*/ -#define RSEQ_SIG 0xd428bc00 /* BRK #0x45E0 */ +/*

• • aarch64 -mbig-endian generates mixed endianness code vs data:
• • little-endian code and big-endian data. Ensure the RSEQ_SIG signature
• • matches code endianness.
• */

+#define RSEQ_SIG_CODE 0xd428bc00 /* BRK #0x45E0. */

+#ifdef __AARCH64EB__ +#define RSEQ_SIG_DATA 0x00bc28d4 /* BRK #0x45E0. */ +#else +#define RSEQ_SIG_DATA RSEQ_SIG_CODE +#endif

+#define RSEQ_SIG RSEQ_SIG_DATA #define rseq_smp_mb() __asm__ __volatile__ ("dmb ish" ::: "memory") #define rseq_smp_rmb() __asm__ __volatile__ ("dmb ishld" ::: "memory") @@ -121,7 +134,7 @@ do { \ #define RSEQ_ASM_DEFINE_ABORT(label, abort_label) \ " b 222f\n" \

• " .inst " __rseq_str(RSEQ_SIG) "\n" \

• " .inst " __rseq_str(RSEQ_SIG_CODE) "\n" \

I don't think this is right; the .inst directive _should_ emit the value in the instruction stream endianness (i.e. LE, regardless of the data endianness).

That's certainly the case with the kernel.org crosstool GCC:

[mark@lakrids:/mnt/data/tests/inst-test]% cat test.c void func(void) { asm volatile(".inst 0xd4000001"); } [mark@lakrids:/mnt/data/tests/inst-test]% usekorg 8.1.0 aarch64-linux-gcc -c test.c [mark@lakrids:/mnt/data/tests/inst-test]% usekorg 8.1.0 aarch64-linux-objdump -d test.o

test.o: file format elf64-littleaarch64

Disassembly of section .text:

0000000000000000 <func>: 0: d4000001 svc #0x0 4: d503201f nop 8: d65f03c0 ret [mark@lakrids:/mnt/data/tests/inst-test]% usekorg 8.1.0 aarch64-linux-gcc -mbig-endian -c test.c [mark@lakrids:/mnt/data/tests/inst-test]% usekorg 8.1.0 aarch64-linux-objdump -d test.o

test.o: file format elf64-bigaarch64

Disassembly of section .text:

0000000000000000 <func>: 0: d4000001 svc #0x0 4: d503201f nop 8: d65f03c0 ret

Have you tested this? Is there some toolchain that doesn't get this right?

I think that the issue is that the kernel loads the thing to check the signature. RSEQ_SIG_CODE isn't byte-swapped explicitly and is used with .inst. RSEG_SIG_DATA is byte-swapped to ensure that the value passed into the syscall is consistent with what the kernel will load.

But yeah, I've just spent the last ten minutes confusing myself with this.

Will