Merge branch 'sparc-perf-stack'

David Ahern says:

====================
sparc64: perf fixes for userspace stacks

Coming back to the perf userspace callchain problem. As a reminder there are
a series of problems trying to use perf to collect callchains with scheduling
tracepoints, e.g., perf sched record -g -- <cmd>.

The first patch disables pagefaults while walking the user stack. As discussed
a couple of months ago this is the right fix, but I was puzzled as to why
processes were terminating with sigbus (and sometimes sigsegv). I believe the
root of this problem is bad addresses trying to walk the frames using frame
pointers. The bad addresses lead to faults that get handled by do_sparc64_fault
and it aborts the task though I am still puzzled as to why it gets past this
check in do_sparc64_fault:

        if (in_atomic() || !mm)
                goto intr_or_no_mm;

pagefault_disable bumps the preempt_count which should make in_atomic return != 0
(building kernels with preemption set to voluntar, CONFIG_PREEMPT_VOLUNTARY=y).

While this set does not fully solve the problem it does prevent a number of
pain points with the current code, most notably able to lock up the system.
====================

Signed-off-by: David S. Miller <[email protected]>

arch/sparc/include/asm/uaccess_64.h

@@ -49,6 +49,28 @@ do {										\
 	__asm__ __volatile__ ("wr %%g0, %0, %%asi" : : "r" ((val).seg));	\
 } while(0)
 
+/*
+ * Test whether a block of memory is a valid user space address.
+ * Returns 0 if the range is valid, nonzero otherwise.
+ */
+static inline bool __chk_range_not_ok(unsigned long addr, unsigned long size, unsigned long limit)
+{
+	if (__builtin_constant_p(size))
+		return addr > limit - size;
+
+	addr += size;
+	if (addr < size)
+		return true;
+
+	return addr > limit;
+}
+
+#define __range_not_ok(addr, size, limit)                               \
+({                                                                      \
+	__chk_user_ptr(addr);                                           \
+	__chk_range_not_ok((unsigned long __force)(addr), size, limit); \
+})
+
 static inline int __access_ok(const void __user * addr, unsigned long size)
 {
 	return 1;

arch/sparc/kernel/perf_event.c

@@ -21,7 +21,7 @@
 
 #include <asm/stacktrace.h>
 #include <asm/cpudata.h>
-#include <asm/uaccess.h>
+#include <linux/uaccess.h>
 #include <linux/atomic.h>
 #include <asm/nmi.h>
 #include <asm/pcr.h>
@@ -1741,18 +1741,31 @@ void perf_callchain_kernel(struct perf_callchain_entry *entry,
 	} while (entry->nr < PERF_MAX_STACK_DEPTH);
 }
 
+static inline int
+valid_user_frame(const void __user *fp, unsigned long size)
+{
+	/* addresses should be at least 4-byte aligned */
+	if (((unsigned long) fp) & 3)
+		return 0;
+
+	return (__range_not_ok(fp, size, TASK_SIZE) == 0);
+}
+
 static void perf_callchain_user_64(struct perf_callchain_entry *entry,
 				   struct pt_regs *regs)
 {
 	unsigned long ufp;
 
-	ufp = regs->u_regs[UREG_I6] + STACK_BIAS;
+	ufp = regs->u_regs[UREG_FP] + STACK_BIAS;
 	do {
 		struct sparc_stackf __user *usf;
 		struct sparc_stackf sf;
 		unsigned long pc;
 
 		usf = (struct sparc_stackf __user *)ufp;
+		if (!valid_user_frame(usf, sizeof(sf)))
+			break;
+
 		if (__copy_from_user_inatomic(&sf, usf, sizeof(sf)))
 			break;
 
@@ -1767,7 +1780,7 @@ static void perf_callchain_user_32(struct perf_callchain_entry *entry,
 {
 	unsigned long ufp;
 
-	ufp = regs->u_regs[UREG_I6] & 0xffffffffUL;
+	ufp = regs->u_regs[UREG_FP] & 0xffffffffUL;
 	do {
 		unsigned long pc;
 
@@ -1803,8 +1816,13 @@ perf_callchain_user(struct perf_callchain_entry *entry, struct pt_regs *regs)
 		return;
 
 	flushw_user();
+
+	pagefault_disable();
+
 	if (test_thread_flag(TIF_32BIT))
 		perf_callchain_user_32(entry, regs);
 	else
 		perf_callchain_user_64(entry, regs);
+
+	pagefault_enable();
 }

arch/sparc/mm/fault_64.c

@@ -413,8 +413,9 @@ asmlinkage void __kprobes do_sparc64_fault(struct pt_regs *regs)
 	 * that here.
 	 */
 	if ((fault_code & FAULT_CODE_ITLB) && !(vma->vm_flags & VM_EXEC)) {
-		BUG_ON(address != regs->tpc);
-		BUG_ON(regs->tstate & TSTATE_PRIV);
+		WARN(address != regs->tpc,
+		     "address (%lx) != regs->tpc (%lx)\n", address, regs->tpc);
+		WARN_ON(regs->tstate & TSTATE_PRIV);
 		goto bad_area;
 	}