sparc64: Fix accidental syscall restart on child return from clone/fork/vfork.

This fixes a regression added by
238468b2ac76020c192a7402c92df5097916bf4a ("[SPARC64]: Use trap type
stored in pt_regs to handle syscall restart.")

Because we now encode the "returning from syscall" status in the
pt_regs area, we have to be mindful to zap it out in the child
of a fork.

During a parallel kernel build I saw an accidental -EINTR return
from vfork() in 'make' because of this bug.

Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/arch/sparc64/kernel/process.c b/arch/sparc64/kernel/process.c
index 500ac6d..0560137 100644 (file)
--- a/arch/sparc64/kernel/process.c
+++ b/arch/sparc64/kernel/process.c
@@ -591,6 +591,12 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long sp,
        if (clone_flags & CLONE_SETTLS)
                t->kregs->u_regs[UREG_G7] = regs->u_regs[UREG_I3];
 
+       /* We do not want to accidently trigger system call restart
+        * handling in the new thread.  Therefore, clear out the trap
+        * type, which will make pt_regs_regs_is_syscall() return false.
+        */
+       pt_regs_clear_trap_type(t->kregs);
+
        return 0;
 }