[ARM] Fix alignment fault handling for ARMv6 and later CPUs

On ARMv6 and later CPUs, it is possible for userspace processes to
get stuck on a misaligned load or store due to the "ignore fault"
setting; unlike previous CPUs, retrying the instruction without
the 'A' bit set does not always cause the load to succeed.

We have no real option but to default to fixing up alignment faults
on these CPUs, and having the CPU fix up those misaligned accesses
which it can.

Reported-by: Wolfgang Grandegger <wg@grandegger.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

1 files changed, 23 insertions, 3 deletions

diff --git a/arch/arm/mm/alignment.c b/arch/arm/mm/alignment.c
index 133e65d..2d5884c 100644
--- a/arch/arm/mm/alignment.c
+++ b/arch/arm/mm/alignment.c
@@ -70,6 +70,10 @@ static unsigned long ai_dword;
 static unsigned long ai_multi;
 static int ai_usermode;
 
+#define UM_WARN		(1 << 0)
+#define UM_FIXUP	(1 << 1)
+#define UM_SIGNAL	(1 << 2)
+
 #ifdef CONFIG_PROC_FS
 static const char *usermode_action[] = {
 	"ignored",
@@ -754,7 +758,7 @@ do_alignment(unsigned long addr, unsigned int fsr, struct pt_regs *regs)
  user:
 	ai_user += 1;
 
-	if (ai_usermode & 1)
+	if (ai_usermode & UM_WARN)
 		printk("Alignment trap: %s (%d) PC=0x%08lx Instr=0x%0*lx "
 		       "Address=0x%08lx FSR 0x%03x\n", current->comm,
 			task_pid_nr(current), instrptr,
@@ -762,10 +766,10 @@ do_alignment(unsigned long addr, unsigned int fsr, struct pt_regs *regs)
 		        thumb_mode(regs) ? tinstr : instr,
 		        addr, fsr);
 
-	if (ai_usermode & 2)
+	if (ai_usermode & UM_FIXUP)
 		goto fixup;
 
-	if (ai_usermode & 4)
+	if (ai_usermode & UM_SIGNAL)
 		force_sig(SIGBUS, current);
 	else
 		set_cr(cr_no_alignment);
@@ -796,6 +800,22 @@ static int __init alignment_init(void)
 	res->write_proc = proc_alignment_write;
 #endif
 
+	/*
+	 * ARMv6 and later CPUs can perform unaligned accesses for
+	 * most single load and store instructions up to word size.
+	 * LDM, STM, LDRD and STRD still need to be handled.
+	 *
+	 * Ignoring the alignment fault is not an option on these
+	 * CPUs since we spin re-faulting the instruction without
+	 * making any progress.
+	 */
+	if (cpu_architecture() >= CPU_ARCH_ARMv6 && (cr_alignment & CR_U)) {
+		cr_alignment &= ~CR_A;
+		cr_no_alignment &= ~CR_A;
+		set_cr(cr_alignment);
+		ai_usermode = UM_FIXUP;
+	}
+
 	hook_fault_code(1, do_alignment, SIGILL, "alignment exception");
 	hook_fault_code(3, do_alignment, SIGILL, "alignment exception");