This was broken by the fix for 9b6a3ea7a699594162ed3d11e4e04b98568dc5c0.
Specifically a movs pc,lr in the kernels ret_fast_syscall returning to
some thumb mode user space code but store_reg unconditionally aligned
the return PC instead of treating the return as an "interworking"

I suspect we need to audit all calls to store_reg that might involve the
PC to ensure "interworking" branches are correctly handled. Also I'm not
quite sure how the code worked before 9b6a3e as the store_reg path
wouldn't have triggered the store_cpu_field(var, thumb) to set the
processor mode back to thumb.

Signed-off-by: Alex Bennée <>
 target-arm/translate.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/target-arm/translate.c b/target-arm/translate.c
index 5e21b52..373d68a 100644
--- a/target-arm/translate.c
+++ b/target-arm/translate.c
@@ -4318,7 +4318,7 @@ static void gen_mrs_banked(DisasContext *s, int r, int 
sysm, int rn)
 static void gen_exception_return(DisasContext *s, TCGv_i32 pc)
     TCGv_i32 tmp;
-    store_reg(s, 15, pc);
+    store_reg_bx(s, 15, pc);
     tmp = load_cpu_field(spsr);
     gen_helper_cpsr_write_eret(cpu_env, tmp);

Reply via email to