sparc: Fix debugger syscall restart interactions.

So, forever, we've had this ptrace_signal_deliver implementation
which tries to handle all of the nasties that can occur when the
debugger looks at a process about to take a signal.  It's meant
to address all of these issues inside of the kernel so that the
debugger need not be mindful of such things.

Problem is, this doesn't work.

The idea was that we should do the syscall restart business first, so
that the debugger captures that state.  Otherwise, if the debugger for
example saves the child's state, makes the child execute something
else, then restores the saved state, we won't handle the syscall
restart properly because we lose the "we're in a syscall" state.

The code here worked for most cases, but if the debugger actually
passes the signal through to the child unaltered, it's possible that
we would do a syscall restart when we shouldn't have.

In particular this breaks the case of debugging a process under a gdb
which is being debugged by yet another gdb.  gdb uses sigsuspend
to wait for SIGCHLD of the inferior, but if gdb itself is being
debugged by a top-level gdb we get a ptrace_stop().  The top-level gdb
does a PTRACE_CONT with SIGCHLD to let the inferior gdb see the
signal.  But ptrace_signal_deliver() assumed the debugger would cancel
out the signal and therefore did a syscall restart, because the return
error was ERESTARTNOHAND.

Fix this by simply making ptrace_signal_deliver() a nop, and providing
a way for the debugger to control system call restarting properly:

1) Report a "in syscall" software bit in regs->{tstate,psr}.
   It is set early on in trap entry to a system call and is fully
   visible to the debugger via ptrace() and regsets.

2) Test this bit right before doing a syscall restart.  We have
   to do a final recheck right after get_signal_to_deliver() in
   case the debugger cleared the bit during ptrace_stop().

3) Clear the bit in trap return so we don't accidently try to set
   that bit in the real register.

As a result we also get a ptrace_{is,clear}_syscall() for sparc32 just
like sparc64 has.

M68K has this same exact bug, and is now the only other user of the
ptrace_signal_deliver hook.  It needs to be fixed in the same exact
way as sparc.

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

9 files changed, 91 insertions, 96 deletions

diff --git a/arch/sparc/kernel/entry.S b/arch/sparc/kernel/entry.S
index 57d1bbdd0bd..4bcfe54f878 100644
--- a/arch/sparc/kernel/entry.S
+++ b/arch/sparc/kernel/entry.S
@@ -1306,6 +1306,8 @@ ret_from_fork:
 	.align	4
 	.globl	linux_sparc_syscall
 linux_sparc_syscall:
+	sethi	%hi(PSR_SYSCALL), %l4
+	or	%l0, %l4, %l0
 	/* Direct access to user regs, must faster. */
 	cmp	%g1, NR_SYSCALLS
 	bgeu	linux_sparc_ni_syscall
diff --git a/arch/sparc/kernel/ptrace.c b/arch/sparc/kernel/ptrace.c
index 60dfc65549d..81f3b929743 100644
--- a/arch/sparc/kernel/ptrace.c
+++ b/arch/sparc/kernel/ptrace.c
@@ -170,8 +170,8 @@ static int genregs32_set(struct task_struct *target,
 		switch (pos) {
 		case 32: /* PSR */
 			psr = regs->psr;
-			psr &= ~PSR_ICC;
-			psr |= (reg & PSR_ICC);
+			psr &= ~(PSR_ICC | PSR_SYSCALL);
+			psr |= (reg & (PSR_ICC | PSR_SYSCALL));
 			regs->psr = psr;
 			break;
 		case 33: /* PC */
diff --git a/arch/sparc/kernel/rtrap.S b/arch/sparc/kernel/rtrap.S
index 77ca6fd8125..b27b5b56f77 100644
--- a/arch/sparc/kernel/rtrap.S
+++ b/arch/sparc/kernel/rtrap.S
@@ -50,8 +50,9 @@ rtrap_7win_patch5:	and	%g1, 0x7f, %g1
 ret_trap_entry:
 ret_trap_lockless_ipi:
 	andcc	%t_psr, PSR_PS, %g0
+	sethi	%hi(PSR_SYSCALL), %g1
 	be	1f
-	 nop
+	 andn	%t_psr, %g1, %t_psr
 
 	wr	%t_psr, 0x0, %psr
 	b	ret_trap_kernel
@@ -73,7 +74,6 @@ signal_p:
 	 ld	[%sp + STACKFRAME_SZ + PT_PSR], %t_psr
 
 	mov	%l5, %o1
-	mov	%l6, %o2
 	call	do_signal
 	 add	%sp, STACKFRAME_SZ, %o0	! pt_regs ptr
 
@@ -81,6 +81,8 @@ signal_p:
 	ld	[%sp + STACKFRAME_SZ + PT_PSR], %t_psr
 	clr	%l6
 ret_trap_continue:
+	sethi	%hi(PSR_SYSCALL), %g1
+	andn	%t_psr, %g1, %t_psr
 	wr	%t_psr, 0x0, %psr
 	WRITE_PAUSE
 
@@ -137,8 +139,9 @@ ret_trap_userwins_ok:
 	LOAD_PT_PRIV(sp, t_psr, t_pc, t_npc)
 	or	%t_pc, %t_npc, %g2
 	andcc	%g2, 0x3, %g0
+	sethi	%hi(PSR_SYCALL), %g2
 	be	1f
-	 nop
+	 andn	%t_psr, %g2, %t_psr
 
 	b	ret_trap_unaligned_pc
 	 add	%sp, STACKFRAME_SZ, %o0
@@ -201,6 +204,8 @@ rtrap_patch5:	and	%g1, 0xff, %g1
 1:
 	LOAD_PT_ALL(sp, t_psr, t_pc, t_npc, g1)
 2:
+	sethi	%hi(PSR_SYSCALL), %twin_tmp1
+	andn	%t_psr, %twin_tmp1, %t_psr
 	wr	%t_psr, 0x0, %psr
 	WRITE_PAUSE
 
diff --git a/arch/sparc/kernel/signal.c b/arch/sparc/kernel/signal.c
index 368157926d2..3fd1df9f9ba 100644
--- a/arch/sparc/kernel/signal.c
+++ b/arch/sparc/kernel/signal.c
@@ -145,6 +145,9 @@ asmlinkage void do_sigreturn(struct pt_regs *regs)
 	regs->psr = (up_psr & ~(PSR_ICC | PSR_EF))
 		  | (regs->psr & (PSR_ICC | PSR_EF));
 
+	/* Prevent syscall restart.  */
+	pt_regs_clear_syscall(regs);
+
 	err |= __get_user(fpu_save, &sf->fpu_save);
 
 	if (fpu_save)
@@ -199,6 +202,9 @@ asmlinkage void do_rt_sigreturn(struct pt_regs *regs)
 
 	regs->psr = (regs->psr & ~PSR_ICC) | (psr & PSR_ICC);
 
+	/* Prevent syscall restart.  */
+	pt_regs_clear_syscall(regs);
+
 	err |= __get_user(fpu_save, &sf->fpu_save);
 
 	if (fpu_save)
@@ -507,26 +513,36 @@ static inline void syscall_restart(unsigned long orig_i0, struct pt_regs *regs,
  * want to handle. Thus you cannot kill init even with a SIGKILL even by
  * mistake.
  */
-asmlinkage void do_signal(struct pt_regs * regs, unsigned long orig_i0, int restart_syscall)
+asmlinkage void do_signal(struct pt_regs * regs, unsigned long orig_i0)
 {
-	siginfo_t info;
-	struct sparc_deliver_cookie cookie;
 	struct k_sigaction ka;
-	int signr;
+	int restart_syscall;
 	sigset_t *oldset;
+	siginfo_t info;
+	int signr;
 
-	cookie.restart_syscall = restart_syscall;
-	cookie.orig_i0 = orig_i0;
+	if (pt_regs_is_syscall(regs) && (regs->psr & PSR_C))
+		restart_syscall = 1;
+	else
+		restart_syscall = 0;
 
 	if (test_thread_flag(TIF_RESTORE_SIGMASK))
 		oldset = &current->saved_sigmask;
 	else
 		oldset = &current->blocked;
 
-	signr = get_signal_to_deliver(&info, &ka, regs, &cookie);
+	signr = get_signal_to_deliver(&info, &ka, regs, NULL);
+
+	/* If the debugger messes with the program counter, it clears
+	 * the software "in syscall" bit, directing us to not perform
+	 * a syscall restart.
+	 */
+	if (restart_syscall && !pt_regs_is_syscall(regs))
+		restart_syscall = 0;
+
 	if (signr > 0) {
-		if (cookie.restart_syscall)
-			syscall_restart(cookie.orig_i0, regs, &ka.sa);
+		if (restart_syscall)
+			syscall_restart(orig_i0, regs, &ka.sa);
 		handle_signal(signr, &ka, &info, oldset, regs);
 
 		/* a signal was successfully delivered; the saved
@@ -538,16 +554,16 @@ asmlinkage void do_signal(struct pt_regs * regs, unsigned long orig_i0, int rest
 			clear_thread_flag(TIF_RESTORE_SIGMASK);
 		return;
 	}
-	if (cookie.restart_syscall &&
+	if (restart_syscall &&
 	    (regs->u_regs[UREG_I0] == ERESTARTNOHAND ||
 	     regs->u_regs[UREG_I0] == ERESTARTSYS ||
 	     regs->u_regs[UREG_I0] == ERESTARTNOINTR)) {
 		/* replay the system call when we are done */
-		regs->u_regs[UREG_I0] = cookie.orig_i0;
+		regs->u_regs[UREG_I0] = orig_i0;
 		regs->pc -= 4;
 		regs->npc -= 4;
 	}
-	if (cookie.restart_syscall &&
+	if (restart_syscall &&
 	    regs->u_regs[UREG_I0] == ERESTART_RESTARTBLOCK) {
 		regs->u_regs[UREG_G1] = __NR_restart_syscall;
 		regs->pc -= 4;
@@ -599,27 +615,3 @@ do_sys_sigstack(struct sigstack __user *ssptr, struct sigstack __user *ossptr,
 out:
 	return ret;
 }
-
-void ptrace_signal_deliver(struct pt_regs *regs, void *cookie)
-{
-	struct sparc_deliver_cookie *cp = cookie;
-
-	if (cp->restart_syscall &&
-	    (regs->u_regs[UREG_I0] == ERESTARTNOHAND ||
-	     regs->u_regs[UREG_I0] == ERESTARTSYS ||
-	     regs->u_regs[UREG_I0] == ERESTARTNOINTR)) {
-		/* replay the system call when we are done */
-		regs->u_regs[UREG_I0] = cp->orig_i0;
-		regs->pc -= 4;
-		regs->npc -= 4;
-		cp->restart_syscall = 0;
-	}
-
-	if (cp->restart_syscall &&
-	    regs->u_regs[UREG_I0] == ERESTART_RESTARTBLOCK) {
-		regs->u_regs[UREG_G1] = __NR_restart_syscall;
-		regs->pc -= 4;
-		regs->npc -= 4;
-		cp->restart_syscall = 0;
-	}
-}
diff --git a/arch/sparc64/kernel/etrap.S b/arch/sparc64/kernel/etrap.S
index b49d3b60bc0..f25e1da3fd0 100644
--- a/arch/sparc64/kernel/etrap.S
+++ b/arch/sparc64/kernel/etrap.S
@@ -27,11 +27,12 @@
 
 		.text		
 		.align	64
-		.globl	etrap, etrap_irq, etraptl1
+		.globl	etrap_syscall, etrap, etrap_irq, etraptl1
 etrap:		rdpr	%pil, %g2
-etrap_irq:
-		TRAP_LOAD_THREAD_REG(%g6, %g1)
+etrap_irq:	clr	%g3
+etrap_syscall:	TRAP_LOAD_THREAD_REG(%g6, %g1)
 		rdpr	%tstate, %g1
+		or	%g1, %g3, %g1
 		sllx	%g2, 20, %g3
 		andcc	%g1, TSTATE_PRIV, %g0
 		or	%g1, %g3, %g1
diff --git a/arch/sparc64/kernel/ptrace.c b/arch/sparc64/kernel/ptrace.c
index b803fe9b2c8..f6c9fc92921 100644
--- a/arch/sparc64/kernel/ptrace.c
+++ b/arch/sparc64/kernel/ptrace.c
@@ -287,11 +287,11 @@ static int genregs64_set(struct task_struct *target,
 					 32 * sizeof(u64),
 					 33 * sizeof(u64));
 		if (!ret) {
-			/* Only the condition codes can be modified
-			 * in the %tstate register.
+			/* Only the condition codes and the "in syscall"
+			 * state can be modified in the %tstate register.
 			 */
-			tstate &= (TSTATE_ICC | TSTATE_XCC);
-			regs->tstate &= ~(TSTATE_ICC | TSTATE_XCC);
+			tstate &= (TSTATE_ICC | TSTATE_XCC | TSTATE_SYSCALL);
+			regs->tstate &= ~(TSTATE_ICC | TSTATE_XCC | TSTATE_SYSCALL);
 			regs->tstate |= tstate;
 		}
 	}
@@ -657,8 +657,10 @@ static int genregs32_set(struct task_struct *target,
 		switch (pos) {
 		case 32: /* PSR */
 			tstate = regs->tstate;
-			tstate &= ~(TSTATE_ICC | TSTATE_XCC);
+			tstate &= ~(TSTATE_ICC | TSTATE_XCC | TSTATE_SYSCALL);
 			tstate |= psr_to_tstate_icc(reg);
+			if (reg & PSR_SYSCALL)
+				tstate |= TSTATE_SYSCALL;
 			regs->tstate = tstate;
 			break;
 		case 33: /* PC */
diff --git a/arch/sparc64/kernel/rtrap.S b/arch/sparc64/kernel/rtrap.S
index ecf6753b204..b9b785fd8b4 100644
--- a/arch/sparc64/kernel/rtrap.S
+++ b/arch/sparc64/kernel/rtrap.S
@@ -257,6 +257,7 @@ rt_continue:	ldx			[%sp + PTREGS_OFF + PT_V9_G1], %g1
 		wr			%o3, %g0, %y
 		wrpr			%l4, 0x0, %pil
 		wrpr			%g0, 0x1, %tl
+		andn			%l1, TSTATE_SYSCALL, %l1
 		wrpr			%l1, %g0, %tstate
 		wrpr			%l2, %g0, %tpc
 		wrpr			%o2, %g0, %tnpc
diff --git a/arch/sparc64/kernel/signal.c b/arch/sparc64/kernel/signal.c
index 07c0443ea3f..2378482c2aa 100644
--- a/arch/sparc64/kernel/signal.c
+++ b/arch/sparc64/kernel/signal.c
@@ -333,7 +333,7 @@ void do_rt_sigreturn(struct pt_regs *regs)
 	regs->tnpc = tnpc;
 
 	/* Prevent syscall restart.  */
-	pt_regs_clear_trap_type(regs);
+	pt_regs_clear_syscall(regs);
 
 	sigdelsetmask(&set, ~_BLOCKABLE);
 	spin_lock_irq(&current->sighand->siglock);
@@ -499,7 +499,7 @@ static inline void handle_signal(unsigned long signr, struct k_sigaction *ka,
 }
 
 static inline void syscall_restart(unsigned long orig_i0, struct pt_regs *regs,
-				     struct sigaction *sa)
+				   struct sigaction *sa)
 {
 	switch (regs->u_regs[UREG_I0]) {
 	case ERESTART_RESTARTBLOCK:
@@ -525,19 +525,17 @@ static inline void syscall_restart(unsigned long orig_i0, struct pt_regs *regs,
  */
 static void do_signal(struct pt_regs *regs, unsigned long orig_i0)
 {
-	struct signal_deliver_cookie cookie;
 	struct k_sigaction ka;
+	int restart_syscall;
 	sigset_t *oldset;
 	siginfo_t info;
 	int signr;
 	
 	if (pt_regs_is_syscall(regs) &&
 	    (regs->tstate & (TSTATE_XCARRY | TSTATE_ICARRY))) {
-		pt_regs_clear_trap_type(regs);
-		cookie.restart_syscall = 1;
+		restart_syscall = 1;
 	} else
-		cookie.restart_syscall = 0;
-	cookie.orig_i0 = orig_i0;
+		restart_syscall = 0;
 
 	if (test_thread_flag(TIF_RESTORE_SIGMASK))
 		oldset = &current->saved_sigmask;
@@ -547,16 +545,25 @@ static void do_signal(struct pt_regs *regs, unsigned long orig_i0)
 #ifdef CONFIG_COMPAT
 	if (test_thread_flag(TIF_32BIT)) {
 		extern void do_signal32(sigset_t *, struct pt_regs *,
-					struct signal_deliver_cookie *);
-		do_signal32(oldset, regs, &cookie);
+					int restart_syscall,
+					unsigned long orig_i0);
+		do_signal32(oldset, regs, restart_syscall, orig_i0);
 		return;
 	}
 #endif	
 
-	signr = get_signal_to_deliver(&info, &ka, regs, &cookie);
+	signr = get_signal_to_deliver(&info, &ka, regs, NULL);
+
+	/* If the debugger messes with the program counter, it clears
+	 * the software "in syscall" bit, directing us to not perform
+	 * a syscall restart.
+	 */
+	if (restart_syscall && !pt_regs_is_syscall(regs))
+		restart_syscall = 0;
+
 	if (signr > 0) {
-		if (cookie.restart_syscall)
-			syscall_restart(cookie.orig_i0, regs, &ka.sa);
+		if (restart_syscall)
+			syscall_restart(orig_i0, regs, &ka.sa);
 		handle_signal(signr, &ka, &info, oldset, regs);
 
 		/* a signal was successfully delivered; the saved
@@ -568,16 +575,16 @@ static void do_signal(struct pt_regs *regs, unsigned long orig_i0)
 			clear_thread_flag(TIF_RESTORE_SIGMASK);
 		return;
 	}
-	if (cookie.restart_syscall &&
+	if (restart_syscall &&
 	    (regs->u_regs[UREG_I0] == ERESTARTNOHAND ||
 	     regs->u_regs[UREG_I0] == ERESTARTSYS ||
 	     regs->u_regs[UREG_I0] == ERESTARTNOINTR)) {
 		/* replay the system call when we are done */
-		regs->u_regs[UREG_I0] = cookie.orig_i0;
+		regs->u_regs[UREG_I0] = orig_i0;
 		regs->tpc -= 4;
 		regs->tnpc -= 4;
 	}
-	if (cookie.restart_syscall &&
+	if (restart_syscall &&
 	    regs->u_regs[UREG_I0] == ERESTART_RESTARTBLOCK) {
 		regs->u_regs[UREG_G1] = __NR_restart_syscall;
 		regs->tpc -= 4;
@@ -598,26 +605,3 @@ void do_notify_resume(struct pt_regs *regs, unsigned long orig_i0, unsigned long
 	if (thread_info_flags & (_TIF_SIGPENDING | _TIF_RESTORE_SIGMASK))
 		do_signal(regs, orig_i0);
 }
-
-void ptrace_signal_deliver(struct pt_regs *regs, void *cookie)
-{
-	struct signal_deliver_cookie *cp = cookie;
-
-	if (cp->restart_syscall &&
-	    (regs->u_regs[UREG_I0] == ERESTARTNOHAND ||
-	     regs->u_regs[UREG_I0] == ERESTARTSYS ||
-	     regs->u_regs[UREG_I0] == ERESTARTNOINTR)) {
-		/* replay the system call when we are done */
-		regs->u_regs[UREG_I0] = cp->orig_i0;
-		regs->tpc -= 4;
-		regs->tnpc -= 4;
-		cp->restart_syscall = 0;
-	}
-	if (cp->restart_syscall &&
-	    regs->u_regs[UREG_I0] == ERESTART_RESTARTBLOCK) {
-		regs->u_regs[UREG_G1] = __NR_restart_syscall;
-		regs->tpc -= 4;
-		regs->tnpc -= 4;
-		cp->restart_syscall = 0;
-	}
-}
diff --git a/arch/sparc64/kernel/signal32.c b/arch/sparc64/kernel/signal32.c
index 0f6b7b156ef..3f19e9af3d1 100644
--- a/arch/sparc64/kernel/signal32.c
+++ b/arch/sparc64/kernel/signal32.c
@@ -269,7 +269,7 @@ void do_sigreturn32(struct pt_regs *regs)
 	regs->tstate |= psr_to_tstate_icc(psr);
 
 	/* Prevent syscall restart.  */
-	pt_regs_clear_trap_type(regs);
+	pt_regs_clear_syscall(regs);
 
 	err |= __get_user(fpu_save, &sf->fpu_save);
 	if (fpu_save)
@@ -355,7 +355,7 @@ asmlinkage void do_rt_sigreturn32(struct pt_regs *regs)
 	regs->tstate |= psr_to_tstate_icc(psr);
 
 	/* Prevent syscall restart.  */
-	pt_regs_clear_trap_type(regs);
+	pt_regs_clear_syscall(regs);
 
 	err |= __get_user(fpu_save, &sf->fpu_save);
 	if (fpu_save)
@@ -768,16 +768,24 @@ static inline void syscall_restart32(unsigned long orig_i0, struct pt_regs *regs
  * mistake.
  */
 void do_signal32(sigset_t *oldset, struct pt_regs * regs,
-		 struct signal_deliver_cookie *cookie)
+		 int restart_syscall, unsigned long orig_i0)
 {
 	struct k_sigaction ka;
 	siginfo_t info;
 	int signr;
 	
-	signr = get_signal_to_deliver(&info, &ka, regs, cookie);
+	signr = get_signal_to_deliver(&info, &ka, regs, NULL);
+
+	/* If the debugger messes with the program counter, it clears
+	 * the "in syscall" bit, directing us to not perform a syscall
+	 * restart.
+	 */
+	if (restart_syscall && !pt_regs_is_syscall(regs))
+		restart_syscall = 0;
+
 	if (signr > 0) {
-		if (cookie->restart_syscall)
-			syscall_restart32(cookie->orig_i0, regs, &ka.sa);
+		if (restart_syscall)
+			syscall_restart32(orig_i0, regs, &ka.sa);
 		handle_signal32(signr, &ka, &info, oldset, regs);
 
 		/* a signal was successfully delivered; the saved
@@ -789,16 +797,16 @@ void do_signal32(sigset_t *oldset, struct pt_regs * regs,
 			clear_thread_flag(TIF_RESTORE_SIGMASK);
 		return;
 	}
-	if (cookie->restart_syscall &&
+	if (restart_syscall &&
 	    (regs->u_regs[UREG_I0] == ERESTARTNOHAND ||
 	     regs->u_regs[UREG_I0] == ERESTARTSYS ||
 	     regs->u_regs[UREG_I0] == ERESTARTNOINTR)) {
 		/* replay the system call when we are done */
-		regs->u_regs[UREG_I0] = cookie->orig_i0;
+		regs->u_regs[UREG_I0] = orig_i0;
 		regs->tpc -= 4;
 		regs->tnpc -= 4;
 	}
-	if (cookie->restart_syscall &&
+	if (restart_syscall &&
 	    regs->u_regs[UREG_I0] == ERESTART_RESTARTBLOCK) {
 		regs->u_regs[UREG_G1] = __NR_restart_syscall;
 		regs->tpc -= 4;