Skip to contentSkip to conversation

powerpc: Make alignment exception always check exception table

PATCH In review Benjamin Herrenschmidt 2 participants 5 messages Nov 1, 2006 PowerPC

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 powerpc: Make alignment exception always check exception table

The alignment exception used to only check the exception table for -EFAULT, not for other errors. That opens an oops window if we can coerce the kernel into getting an alignment exception for other reasons in what would normally be a user-protected accessor, which can be done via some of the futex ops. This fixes it by always checking the exception tables.

---

This is 2.6.19 material and should probably go into stable as well. (Greg: take it if paulus acks it and it applies :)

arch/powerpc/kernel/traps.c +6 -10
Index: linux-work/arch/powerpc/kernel/traps.c
===================================================================
@@ -843,7 +843,7 @@ void __kprobes program_check_exception(s
 
 void alignment_exception(struct pt_regs *regs)
 {
-	int fixed = 0;
+	int sig, fixed = 0;
 
 	/* we don't implement logging of alignment exceptions */
 	if (!(current->thread.align_ctl & PR_UNALIGN_SIGBUS))
@@ -856,15 +856,11 @@ void alignment_exception(struct pt_regs 
 	}
 
 	/* Operand address was bad */
-	if (fixed == -EFAULT) {
-		if (user_mode(regs))
-			_exception(SIGSEGV, regs, SEGV_ACCERR, regs->dar);
-		else
-			/* Search exception table */
-			bad_page_fault(regs, regs->dar, SIGSEGV);
-		return;
-	}
-	_exception(SIGBUS, regs, BUS_ADRALN, regs->dar);
+	sig = fixed == -EFAULT ? SIGSEGV : SIGBUS;
+	if (user_mode(regs))
+		_exception(sig, regs, SEGV_ACCERR, regs->dar);
+	else
+		bad_page_fault(regs, regs->dar, sig);
 }
 
 void StackOverflow(struct pt_regs *regs)
#1 lore raw

PATCH 1/1 powerpc: Make alignment exception always check exception table (#2)

The alignment exception used to only check the exception table for -EFAULT, not for other errors. That opens an oops window if we can coerce the kernel into getting an alignment exception for other reasons in what would normally be a user-protected accessor, which can be done via some of the futex ops. This fixes it by always checking the exception tables.

---

Paul: this version also fixes arch/ppc

This is 2.6.19 material and should probably go into stable as well. (Greg: take it if paulus acks it and it applies :)

arch/powerpc/kernel/traps.c +6 -10
Index: linux-work/arch/powerpc/kernel/traps.c
===================================================================
@@ -843,7 +843,7 @@ void __kprobes program_check_exception(s
 
 void alignment_exception(struct pt_regs *regs)
 {
-	int fixed = 0;
+	int sig, fixed = 0;
 
 	/* we don't implement logging of alignment exceptions */
 	if (!(current->thread.align_ctl & PR_UNALIGN_SIGBUS))
@@ -856,15 +856,11 @@ void alignment_exception(struct pt_regs 
 	}
 
 	/* Operand address was bad */
-	if (fixed == -EFAULT) {
-		if (user_mode(regs))
-			_exception(SIGSEGV, regs, SEGV_ACCERR, regs->dar);
-		else
-			/* Search exception table */
-			bad_page_fault(regs, regs->dar, SIGSEGV);
-		return;
-	}
-	_exception(SIGBUS, regs, BUS_ADRALN, regs->dar);
+	sig = fixed == -EFAULT ? SIGSEGV : SIGBUS;
+	if (user_mode(regs))
+		_exception(sig, regs, SEGV_ACCERR, regs->dar);
+	else
+		bad_page_fault(regs, regs->dar, sig);
 }
 
 void StackOverflow(struct pt_regs *regs)
arch/ppc/kernel/traps.c +6 -10
Index: linux-work/arch/ppc/kernel/traps.c
===================================================================
@@ -708,7 +708,7 @@ void single_step_exception(struct pt_reg
 
 void alignment_exception(struct pt_regs *regs)
 {
-	int fixed;
+	int sig, fixed = 0;
 
 	fixed = fix_alignment(regs);
 	if (fixed == 1) {
@@ -716,15 +716,11 @@ void alignment_exception(struct pt_regs 
 		emulate_single_step(regs);
 		return;
 	}
-	if (fixed == -EFAULT) {
-		/* fixed == -EFAULT means the operand address was bad */
-		if (user_mode(regs))
-			_exception(SIGSEGV, regs, SEGV_ACCERR, regs->dar);
-		else
-			bad_page_fault(regs, regs->dar, SIGSEGV);
-		return;
-	}
-	_exception(SIGBUS, regs, BUS_ADRALN, regs->dar);
+	sig = fixed == -EFAULT ? SIGSEGV : SIGBUS;
+	if (user_mode(regs))
+		_exception(sig, regs, SEGV_ACCERR, regs->dar);
+	else
+		bad_page_fault(regs, regs->dar, sig);
 }
 
 void StackOverflow(struct pt_regs *regs)
#2 lore raw

PATCH 1/1 powerpc: Make alignment exception always check exception table (#3)

In reply to Benjamin Herrenschmidt

The alignment exception used to only check the exception table for -EFAULT, not for other errors. That opens an oops window if we can coerce the kernel into getting an alignment exception for other reasons in what would normally be a user-protected accessor, which can be done via some of the futex ops. This fixes it by always checking the exception tables.

---

Paul: this version sets the correct si_code

This is 2.6.19 material and should probably go into stable as well. (Greg: take it if paulus acks it and it applies :)

arch/powerpc/kernel/traps.c +10 -8
Index: linux-work/arch/powerpc/kernel/traps.c
===================================================================
@@ -843,7 +843,7 @@ void __kprobes program_check_exception(s
 
 void alignment_exception(struct pt_regs *regs)
 {
-	int fixed = 0;
+	int sig, code, fixed = 0;
 
 	/* we don't implement logging of alignment exceptions */
 	if (!(current->thread.align_ctl & PR_UNALIGN_SIGBUS))
@@ -857,14 +857,16 @@ void alignment_exception(struct pt_regs 
 
 	/* Operand address was bad */
 	if (fixed == -EFAULT) {
-		if (user_mode(regs))
-			_exception(SIGSEGV, regs, SEGV_ACCERR, regs->dar);
-		else
-			/* Search exception table */
-			bad_page_fault(regs, regs->dar, SIGSEGV);
-		return;
+		sig = SIGSEGV;
+		code = SEGV_ACCERR;
+	} else {
+		sig = SIGBUS;
+		code = BUS_ADRALN;
 	}
-	_exception(SIGBUS, regs, BUS_ADRALN, regs->dar);
+	if (user_mode(regs))
+		_exception(sig, regs, code, regs->dar);
+	else
+		bad_page_fault(regs, regs->dar, sig);
 }
 
 void StackOverflow(struct pt_regs *regs)
arch/ppc/kernel/traps.c +10 -8
Index: linux-work/arch/ppc/kernel/traps.c
===================================================================
@@ -708,7 +708,7 @@ void single_step_exception(struct pt_reg
 
 void alignment_exception(struct pt_regs *regs)
 {
-	int fixed;
+	int sig, code, fixed = 0;
 
 	fixed = fix_alignment(regs);
 	if (fixed == 1) {
@@ -717,14 +717,16 @@ void alignment_exception(struct pt_regs 
 		return;
 	}
 	if (fixed == -EFAULT) {
-		/* fixed == -EFAULT means the operand address was bad */
-		if (user_mode(regs))
-			_exception(SIGSEGV, regs, SEGV_ACCERR, regs->dar);
-		else
-			bad_page_fault(regs, regs->dar, SIGSEGV);
-		return;
+		sig = SIGSEGV;
+		code = SEGV_ACCERR;
+	} else {
+		sig = SIGBUS;
+		code = BUS_ADRALN;
 	}
-	_exception(SIGBUS, regs, BUS_ADRALN, regs->dar);
+	if (user_mode(regs))
+		_exception(sig, regs, code, regs->dar);
+	else
+		bad_page_fault(regs, regs->dar, sig);
 }
 
 void StackOverflow(struct pt_regs *regs)
#3 lore raw
In reply to Benjamin Herrenschmidt
Quoted 7 lines · Benjamin Herrenschmidt

The alignment exception used to only check the exception table for -EFAULT, not for other errors. That opens an oops window if we can coerce the kernel into getting an alignment exception for other reasons in what would normally be a user-protected accessor, which can be done via some of the futex ops. This fixes it by always checking the exception tables.

Quoted 5 lines

---

This is 2.6.19 material and should probably go into stable as well. (Greg: take it if paulus acks it and it applies :)

Please forward the final patch (which ever version it is :) to the [email protected] address when this goes into mainline.

thanks,

greg k-h

#4 lore raw
Quoted 8 lines · Greg KH
Quoted 7 lines · Benjamin Herrenschmidt

The alignment exception used to only check the exception table for -EFAULT, not for other errors. That opens an oops window if we can coerce the kernel into getting an alignment exception for other reasons in what would normally be a user-protected accessor, which can be done via some of the futex ops. This fixes it by always checking the exception tables.

Quoted 8 lines
Quoted 5 lines

---

This is 2.6.19 material and should probably go into stable as well. (Greg: take it if paulus acks it and it applies :)

Please forward the final patch (which ever version it is :) to the [email protected] address when this goes into mainline.

Ok, will do. I wan't sure what the proper procedure was.

Cheers,
Ben.