Skip to contentSkip to conversation

lkdtm/stackleak: Make the test more verbose

PATCH In review Alexander Popov 3 participants 6 messages Jan 2, 2020 – Jan 28, 2020

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 lkdtm/stackleak: Make the test more verbose

Make the stack erasing test more verbose about the errors that it can detect.

---
 drivers/misc/lkdtm/stackleak.c | 25 +++++++++++++++++--------
 1 file changed, 17 insertions(+), 8 deletions(-)
drivers/misc/lkdtm/stackleak.c +17 -8
index d5a084475abc..d1a5c0705be3 100644
@@ -16,6 +16,7 @@ void lkdtm_STACKLEAK_ERASING(void)
 	unsigned long *sp, left, found, i;
 	const unsigned long check_depth =
 			STACKLEAK_SEARCH_DEPTH / sizeof(unsigned long);
+	bool test_failed = false;
 
 	/*
 	 * For the details about the alignment of the poison values, see
@@ -34,7 +35,8 @@ void lkdtm_STACKLEAK_ERASING(void)
 		left--;
 	} else {
 		pr_err("FAIL: not enough stack space for the test\n");
-		return;
+		test_failed = true;
+		goto end;
 	}
 
 	pr_info("checking unused part of the thread stack (%lu bytes)...\n",
@@ -52,22 +54,29 @@ void lkdtm_STACKLEAK_ERASING(void)
 	}
 
 	if (found <= check_depth) {
-		pr_err("FAIL: thread stack is not erased (checked %lu bytes)\n",
+		pr_err("FAIL: the erased part is not found (checked %lu bytes)\n",
 						i * sizeof(unsigned long));
-		return;
+		test_failed = true;
+		goto end;
 	}
 
-	pr_info("first %lu bytes are unpoisoned\n",
+	pr_info("the erased part begins after %lu not poisoned bytes\n",
 				(i - found) * sizeof(unsigned long));
 
 	/* The rest of thread stack should be erased */
 	for (; i < left; i++) {
 		if (*(sp - i) != STACKLEAK_POISON) {
-			pr_err("FAIL: thread stack is NOT properly erased\n");
-			return;
+			pr_err("FAIL: bad value number %lu in the erased part: 0x%lx\n",
+								i, *(sp - i));
+			test_failed = true;
 		}
 	}
 
-	pr_info("OK: the rest of the thread stack is properly erased\n");
-	return;
+end:
+	if (test_failed) {
+		pr_err("FAIL: the thread stack is NOT properly erased\n");
+		dump_stack();
+	} else {
+		pr_info("OK: the rest of the thread stack is properly erased\n");
+	}
 }
Signature · 1 line
2.23.0
#1 lore raw

Hello!

Let me remind of this patch.

Best regards,
Alexander

Quoted 68 lines · Alexander Popov

Make the stack erasing test more verbose about the errors that it can detect.

---
 drivers/misc/lkdtm/stackleak.c | 25 +++++++++++++++++--------
 1 file changed, 17 insertions(+), 8 deletions(-)
drivers/misc/lkdtm/stackleak.c +17 -8
index d5a084475abc..d1a5c0705be3 100644
@@ -16,6 +16,7 @@ void lkdtm_STACKLEAK_ERASING(void)
 	unsigned long *sp, left, found, i;
 	const unsigned long check_depth =
 			STACKLEAK_SEARCH_DEPTH / sizeof(unsigned long);
+	bool test_failed = false;
 
 	/*
 	 * For the details about the alignment of the poison values, see
@@ -34,7 +35,8 @@ void lkdtm_STACKLEAK_ERASING(void)
 		left--;
 	} else {
 		pr_err("FAIL: not enough stack space for the test\n");
-		return;
+		test_failed = true;
+		goto end;
 	}
 
 	pr_info("checking unused part of the thread stack (%lu bytes)...\n",
@@ -52,22 +54,29 @@ void lkdtm_STACKLEAK_ERASING(void)
 	}
 
 	if (found <= check_depth) {
-		pr_err("FAIL: thread stack is not erased (checked %lu bytes)\n",
+		pr_err("FAIL: the erased part is not found (checked %lu bytes)\n",
 						i * sizeof(unsigned long));
-		return;
+		test_failed = true;
+		goto end;
 	}
 
-	pr_info("first %lu bytes are unpoisoned\n",
+	pr_info("the erased part begins after %lu not poisoned bytes\n",
 				(i - found) * sizeof(unsigned long));
 
 	/* The rest of thread stack should be erased */
 	for (; i < left; i++) {
 		if (*(sp - i) != STACKLEAK_POISON) {
-			pr_err("FAIL: thread stack is NOT properly erased\n");
-			return;
+			pr_err("FAIL: bad value number %lu in the erased part: 0x%lx\n",
+								i, *(sp - i));
+			test_failed = true;
 		}
 	}
 
-	pr_info("OK: the rest of the thread stack is properly erased\n");
-	return;
+end:
+	if (test_failed) {
+		pr_err("FAIL: the thread stack is NOT properly erased\n");
+		dump_stack();
+	} else {
+		pr_info("OK: the rest of the thread stack is properly erased\n");
+	}
 }
#2 lore raw
In reply to Alexander Popov
Quoted 7 lines · Alexander Popov

Make the stack erasing test more verbose about the errors that it can detect.

---
 drivers/misc/lkdtm/stackleak.c | 25 +++++++++++++++++--------
 1 file changed, 17 insertions(+), 8 deletions(-)

Hello!

Pinging about this version of the patch.

Kees, it uses dump_stack() instead of BUG(), as we negotiated.

Thanks!
Alexander

#3 lore raw
Quoted 9 lines · Alexander Popov
Quoted 7 lines · Alexander Popov

Make the stack erasing test more verbose about the errors that it can detect.

---
 drivers/misc/lkdtm/stackleak.c | 25 +++++++++++++++++--------
 1 file changed, 17 insertions(+), 8 deletions(-)
Quoted 5 lines

Hello!

Pinging about this version of the patch.

Kees, it uses dump_stack() instead of BUG(), as we negotiated.

Yup, this is in my queue -- I've just gotten back from travelling and will get to it shortly. :)

Greg, feel free to take this directly if you want, too.

-Kees

Signature · 1 line
Kees Cook
#4 lore raw
Quoted 12 lines · Kees Cook
Quoted 11 lines · Alexander Popov
Quoted 7 lines · Alexander Popov

Make the stack erasing test more verbose about the errors that it can detect.

---
 drivers/misc/lkdtm/stackleak.c | 25 +++++++++++++++++--------
 1 file changed, 17 insertions(+), 8 deletions(-)

Hello!

Quoted 8 lines
Quoted 3 lines

Pinging about this version of the patch.

Kees, it uses dump_stack() instead of BUG(), as we negotiated.

Yup, this is in my queue -- I've just gotten back from travelling and will get to it shortly. :)

Greg, feel free to take this directly if you want, too.

If I were to take it, it would have to wait until after 5.6-rc1, sorry.

thanks,

greg k-h

#5 lore raw
Quoted 15 lines · Greg Kroah-Hartman
Quoted 14 lines · Kees Cook
Quoted 13 lines · Alexander Popov
Quoted 7 lines · Alexander Popov

Make the stack erasing test more verbose about the errors that it can detect.

---
 drivers/misc/lkdtm/stackleak.c | 25 +++++++++++++++++--------
 1 file changed, 17 insertions(+), 8 deletions(-)

Hello!

Pinging about this version of the patch.

Quoted 8 lines
Quoted 6 lines
Quoted 1 line

Kees, it uses dump_stack() instead of BUG(), as we negotiated.

Yup, this is in my queue -- I've just gotten back from travelling and will get to it shortly. :)

Greg, feel free to take this directly if you want, too.

If I were to take it, it would have to wait until after 5.6-rc1, sorry.

Well, it has to go via drivers/misc anyway, so I think timing is no different. I would consider this a feature patch, not a bug fix, too.

Signature · 1 line
Kees Cook