Skip to contentSkip to conversation

livepatch: Followup changes for the atomic replace patchset

PATCH Applied Petr Mladek 4 participants 10 messages Feb 4, 2019 – Feb 6, 2019 Live patching

lore
Share & tools
lore

Petr Mladek

author maintainer
#0 lore raw

v2 cover livepatch: Followup changes for the atomic replace patchset

This patchset implements ideas that were mentioned and postponed during the review of the atomic replace patchset.

The patches apply on top of livepatching.git, branch origin/for-5.1/atomic-replace.

Changes against v1:

+ Added Joe's patch that fixed ptr_id() error code [Joe]
+ Did proper error handling in the shadow variable sefttest [All]
+ Removed the controversial patch that was removing patch->enabled flag [All].
+ Fixed few typo's [Joe]
+ Added available Acks.

Joe Lawrence (1):
livepatch: return -ENOMEM on ptr_id() allocation failure

Petr Mladek (3):
livepatch: Introduce klp_for_each_patch macro
livepatch: Proper error handling in the shadow variables selftest
livepatch: Module coming and going callbacks can proceed with all
listed patches

 kernel/livepatch/core.c              | 34 ++++++++++------------------------
 kernel/livepatch/core.h              |  6 ++++++
 kernel/livepatch/transition.c        |  2 +-
 lib/livepatch/test_klp_shadow_vars.c | 24 +++++++++++++++++++++++-
 4 files changed, 40 insertions(+), 26 deletions(-)
Signature · 1 line
2.13.7
author maintainer
#1 lore raw

PATCH v2 1/4 livepatch: Introduce klp_for_each_patch macro

There are already macros to iterate over struct klp_func and klp_object.

Add also klp_for_each_patch(). But make it internal because also klp_patches list is internal.

---
 kernel/livepatch/core.c       | 8 ++++----
 kernel/livepatch/core.h       | 6 ++++++
 kernel/livepatch/transition.c | 2 +-
 3 files changed, 11 insertions(+), 5 deletions(-)
kernel/livepatch/core.c +4 -4
index adca5cf07f7e..b716a6289204 100644
@@ -554,7 +554,7 @@ static int klp_add_nops(struct klp_patch *patch)
 	struct klp_patch *old_patch;
 	struct klp_object *old_obj;
 
-	list_for_each_entry(old_patch, &klp_patches, list) {
+	klp_for_each_patch(old_patch) {
 		klp_for_each_object(old_patch, old_obj) {
 			int err;
 
@@ -1089,7 +1089,7 @@ void klp_discard_replaced_patches(struct klp_patch *new_patch)
 {
 	struct klp_patch *old_patch, *tmp_patch;
 
-	list_for_each_entry_safe(old_patch, tmp_patch, &klp_patches, list) {
+	klp_for_each_patch_safe(old_patch, tmp_patch) {
 		if (old_patch == new_patch)
 			return;
 
@@ -1133,7 +1133,7 @@ static void klp_cleanup_module_patches_limited(struct module *mod,
 	struct klp_patch *patch;
 	struct klp_object *obj;
 
-	list_for_each_entry(patch, &klp_patches, list) {
+	klp_for_each_patch(patch) {
 		if (patch == limit)
 			break;
 
@@ -1180,7 +1180,7 @@ int klp_module_coming(struct module *mod)
 	 */
 	mod->klp_alive = true;
 
-	list_for_each_entry(patch, &klp_patches, list) {
+	klp_for_each_patch(patch) {
 		klp_for_each_object(patch, obj) {
 			if (!klp_is_module(obj) || strcmp(obj->name, mod->name))
 				continue;
kernel/livepatch/core.h +6 -0
index e6200f38701f..ec43a40b853f 100644
@@ -7,6 +7,12 @@
 extern struct mutex klp_mutex;
 extern struct list_head klp_patches;
 
+#define klp_for_each_patch_safe(patch, tmp_patch)		\
+	list_for_each_entry_safe(patch, tmp_patch, &klp_patches, list)
+
+#define klp_for_each_patch(patch)	\
+	list_for_each_entry(patch, &klp_patches, list)
+
 void klp_free_patch_start(struct klp_patch *patch);
 void klp_discard_replaced_patches(struct klp_patch *new_patch);
 void klp_discard_nops(struct klp_patch *new_patch);
kernel/livepatch/transition.c +1 -1
index 300273819674..a3a6f32c6fd0 100644
@@ -642,6 +642,6 @@ void klp_force_transition(void)
 	for_each_possible_cpu(cpu)
 		klp_update_patch_state(idle_task(cpu));
 
-	list_for_each_entry(patch, &klp_patches, list)
+	klp_for_each_patch(patch)
 		patch->forced = true;
 }
Signature · 1 line
2.13.7
author maintainer
#2 lore raw

PATCH v2 2/4 livepatch: return -ENOMEM on ptr_id() allocation failure

In reply to Petr Mladek

From: Joe Lawrence <[email protected]>

Fixes the following smatch warning:

lib/livepatch/test_klp_shadow_vars.c:47 ptr_id() warn: returning -1 instead of -ENOMEM is sloppy

---
 lib/livepatch/test_klp_shadow_vars.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
lib/livepatch/test_klp_shadow_vars.c +1 -1
index 02f892f941dc..f5441c193166 100644
@@ -44,7 +44,7 @@ static int ptr_id(void *ptr)
 
 	sp = kmalloc(sizeof(*sp), GFP_ATOMIC);
 	if (!sp)
-		return -1;
+		return -ENOMEM;
 	sp->ptr = ptr;
 	sp->id = count++;
 
Signature · 1 line
2.13.7
maintainer Acked-by on v2 2/4
#3 lore raw

Re: [PATCH v2 2/4] livepatch: return -ENOMEM on ptr_id() allocation failure

Quoted 8 lines · Petr Mladek

From: Joe Lawrence <[email protected]>

Fixes the following smatch warning:

lib/livepatch/test_klp_shadow_vars.c:47 ptr_id() warn: returning -1 instead of -ENOMEM is sloppy

Miroslav

author maintainer
#4 lore raw

PATCH v2 3/4 livepatch: Proper error handling in the shadow variables selftest

In reply to Petr Mladek

Add proper error handling when allocating or getting shadow variables in the selftest. It prevents an invalid pointer access in some situations. It shows the good programming practice in the others.

The error codes are just the best guess and specific for this particular test. In general, klp_shadow_alloc() returns NULL also when the given shadow variable has already been allocated. In addition, both klp_shadow_alloc() and klp_shadow_get_or_alloc() might fail from other reasons when the constructor fails.

Note, that the error code is not really important even in the real life. The use of shadow variables should be transparent for the original livepatched code.

---
 lib/livepatch/test_klp_shadow_vars.c | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)
lib/livepatch/test_klp_shadow_vars.c +22 -0
index f5441c193166..fe5c413efe96 100644
@@ -154,22 +154,37 @@ static int test_klp_shadow_vars_init(void)
 	 * Allocate a few shadow variables with different <obj> and <id>.
 	 */
 	sv1 = shadow_alloc(obj, id, size, gfp_flags, shadow_ctor, &var1);
+	if (!sv1)
+		return -ENOMEM;
+
 	sv2 = shadow_alloc(obj + 1, id, size, gfp_flags, shadow_ctor, &var2);
+	if (!sv2)
+		return -ENOMEM;
+
 	sv3 = shadow_alloc(obj, id + 1, size, gfp_flags, shadow_ctor, &var3);
+	if (!sv3)
+		return -ENOMEM;
 
 	/*
 	 * Verify we can find our new shadow variables and that they point
 	 * to expected data.
 	 */
 	ret = shadow_get(obj, id);
+	if (!ret)
+		return -EINVAL;
 	if (ret == sv1 && *sv1 == &var1)
 		pr_info("  got expected PTR%d -> PTR%d result\n",
 			ptr_id(sv1), ptr_id(*sv1));
+
 	ret = shadow_get(obj + 1, id);
+	if (!ret)
+		return -EINVAL;
 	if (ret == sv2 && *sv2 == &var2)
 		pr_info("  got expected PTR%d -> PTR%d result\n",
 			ptr_id(sv2), ptr_id(*sv2));
 	ret = shadow_get(obj, id + 1);
+	if (!ret)
+		return -EINVAL;
 	if (ret == sv3 && *sv3 == &var3)
 		pr_info("  got expected PTR%d -> PTR%d result\n",
 			ptr_id(sv3), ptr_id(*sv3));
@@ -179,7 +194,12 @@ static int test_klp_shadow_vars_init(void)
 	 * The second invocation should return the same shadow var.
 	 */
 	sv4 = shadow_get_or_alloc(obj + 2, id, size, gfp_flags, shadow_ctor, &var4);
+	if (!sv4)
+		return -ENOMEM;
+
 	ret = shadow_get_or_alloc(obj + 2, id, size, gfp_flags, shadow_ctor, &var4);
+	if (!ret)
+		return -EINVAL;
 	if (ret == sv4 && *sv4 == &var4)
 		pr_info("  got expected PTR%d -> PTR%d result\n",
 			ptr_id(sv4), ptr_id(*sv4));
@@ -207,6 +227,8 @@ static int test_klp_shadow_vars_init(void)
 	 * We should still find an <id+1> variable.
 	 */
 	ret = shadow_get(obj, id + 1);
+	if (!ret)
+		return -EINVAL;
 	if (ret == sv3 && *sv3 == &var3)
 		pr_info("  got expected PTR%d -> PTR%d result\n",
 			ptr_id(sv3), ptr_id(*sv3));
Signature · 1 line
2.13.7
maintainer Acked-by on v2 3/4
#5 lore raw

Re: [PATCH v2 3/4] livepatch: Proper error handling in the shadow variables selftest

Quoted 10 lines · Petr Mladek

Add proper error handling when allocating or getting shadow variables in the selftest. It prevents an invalid pointer access in some situations. It shows the good programming practice in the others.

The error codes are just the best guess and specific for this particular test. In general, klp_shadow_alloc() returns NULL also when the given shadow variable has already been allocated. In addition, both klp_shadow_alloc() and klp_shadow_get_or_alloc() might fail from other reasons when the constructor fails.

Quoted 5 lines

Note, that the error code is not really important even in the real life. The use of shadow variables should be transparent for the original livepatched code.

Miroslav

maintainer Acked-by on v2 3/4
#6 lore raw

Re: [PATCH v2 3/4] livepatch: Proper error handling in the shadow variables selftest

In reply to Petr Mladek
Quoted 76 lines · Petr Mladek

Add proper error handling when allocating or getting shadow variables in the selftest. It prevents an invalid pointer access in some situations. It shows the good programming practice in the others.

The error codes are just the best guess and specific for this particular test. In general, klp_shadow_alloc() returns NULL also when the given shadow variable has already been allocated. In addition, both klp_shadow_alloc() and klp_shadow_get_or_alloc() might fail from other reasons when the constructor fails.

Note, that the error code is not really important even in the real life. The use of shadow variables should be transparent for the original livepatched code.

---
  lib/livepatch/test_klp_shadow_vars.c | 22 ++++++++++++++++++++++
  1 file changed, 22 insertions(+)
lib/livepatch/test_klp_shadow_vars.c +20 -0
index f5441c193166..fe5c413efe96 100644
@@ -154,22 +154,37 @@ static int test_klp_shadow_vars_init(void)
  	 * Allocate a few shadow variables with different <obj> and <id>.
  	 */
  	sv1 = shadow_alloc(obj, id, size, gfp_flags, shadow_ctor, &var1);
+	if (!sv1)
+		return -ENOMEM;
+
  	sv2 = shadow_alloc(obj + 1, id, size, gfp_flags, shadow_ctor, &var2);
+	if (!sv2)
+		return -ENOMEM;
+
  	sv3 = shadow_alloc(obj, id + 1, size, gfp_flags, shadow_ctor, &var3);
+	if (!sv3)
+		return -ENOMEM;
 
  	/*
  	 * Verify we can find our new shadow variables and that they point
  	 * to expected data.
  	 */
  	ret = shadow_get(obj, id);
+	if (!ret)
+		return -EINVAL;
  	if (ret == sv1 && *sv1 == &var1)
  		pr_info("  got expected PTR%d -> PTR%d result\n",
  			ptr_id(sv1), ptr_id(*sv1));
+
  	ret = shadow_get(obj + 1, id);
+	if (!ret)
+		return -EINVAL;
  	if (ret == sv2 && *sv2 == &var2)
  		pr_info("  got expected PTR%d -> PTR%d result\n",
  			ptr_id(sv2), ptr_id(*sv2));
  	ret = shadow_get(obj, id + 1);
+	if (!ret)
+		return -EINVAL;
  	if (ret == sv3 && *sv3 == &var3)
  		pr_info("  got expected PTR%d -> PTR%d result\n",
  			ptr_id(sv3), ptr_id(*sv3));
@@ -179,7 +194,12 @@ static int test_klp_shadow_vars_init(void)
  	 * The second invocation should return the same shadow var.
  	 */
  	sv4 = shadow_get_or_alloc(obj + 2, id, size, gfp_flags, shadow_ctor, &var4);
+	if (!sv4)
+		return -ENOMEM;
+
  	ret = shadow_get_or_alloc(obj + 2, id, size, gfp_flags, shadow_ctor, &var4);
+	if (!ret)
+		return -EINVAL;
  	if (ret == sv4 && *sv4 == &var4)
  		pr_info("  got expected PTR%d -> PTR%d result\n",
  			ptr_id(sv4), ptr_id(*sv4));

lib/livepatch/test_klp_shadow_vars.c static int test_klp_shadow_vars_init(void) · line 234

Quoted 8 lines
lib/livepatch/test_klp_shadow_vars.c +2 -0
@@ -207,6 +227,8 @@ static int test_klp_shadow_vars_init(void)
  	 * We should still find an <id+1> variable.
  	 */
  	ret = shadow_get(obj, id + 1);
+	if (!ret)
+		return -EINVAL;
  	if (ret == sv3 && *sv3 == &var3)
  		pr_info("  got expected PTR%d -> PTR%d result\n",
  			ptr_id(sv3), ptr_id(*sv3));

Fixes look good to me,

-- Joe

author maintainer
#7 lore raw

PATCH v2 4/4 livepatch: Module coming and going callbacks can proceed with all listed patches

In reply to Petr Mladek

Livepatches can no longer get enabled and disabled repeatedly. The list klp_patches contains only enabled patches and eventually the patch in transition.

The module coming and going callbacks do no longer need to check for these state. They have to proceed with all listed patches.

---
 kernel/livepatch/core.c | 26 ++++++--------------------
 1 file changed, 6 insertions(+), 20 deletions(-)
kernel/livepatch/core.c +6 -20
index b716a6289204..684766d306ad 100644
@@ -1141,21 +1141,14 @@ static void klp_cleanup_module_patches_limited(struct module *mod,
 			if (!klp_is_module(obj) || strcmp(obj->name, mod->name))
 				continue;
 
-			/*
-			 * Only unpatch the module if the patch is enabled or
-			 * is in transition.
-			 */
-			if (patch->enabled || patch == klp_transition_patch) {
-
-				if (patch != klp_transition_patch)
-					klp_pre_unpatch_callback(obj);
+			if (patch != klp_transition_patch)
+				klp_pre_unpatch_callback(obj);
 
-				pr_notice("reverting patch '%s' on unloading module '%s'\n",
-					  patch->mod->name, obj->mod->name);
-				klp_unpatch_object(obj);
+			pr_notice("reverting patch '%s' on unloading module '%s'\n",
+				  patch->mod->name, obj->mod->name);
+			klp_unpatch_object(obj);
 
-				klp_post_unpatch_callback(obj);
-			}
+			klp_post_unpatch_callback(obj);
 
 			klp_free_object_loaded(obj);
 			break;
@@ -1194,13 +1187,6 @@ int klp_module_coming(struct module *mod)
 				goto err;
 			}
 
-			/*
-			 * Only patch the module if the patch is enabled or is
-			 * in transition.
-			 */
-			if (!patch->enabled && patch != klp_transition_patch)
-				break;
-
 			pr_notice("applying patch '%s' to loading module '%s'\n",
 				  patch->mod->name, obj->mod->name);
 
Signature · 1 line
2.13.7
author maintainer Applied
#8 lore raw
In reply to Petr Mladek
Quoted 16 lines · Petr Mladek

This patchset implements ideas that were mentioned and postponed during the review of the atomic replace patchset.

The patches apply on top of livepatching.git, branch origin/for-5.1/atomic-replace.

Changes against v1:

+ Added Joe's patch that fixed ptr_id() error code [Joe]
+ Did proper error handling in the shadow variable sefttest [All]
+ Removed the controversial patch that was removing patch->enabled flag [All].
+ Fixed few typo's [Joe]
+ Added available Acks.

Quoted 8 lines

Joe Lawrence (1):
livepatch: return -ENOMEM on ptr_id() allocation failure

Petr Mladek (3):
livepatch: Introduce klp_for_each_patch macro
livepatch: Proper error handling in the shadow variables selftest
livepatch: Module coming and going callbacks can proceed with all
listed patches

I have applied all 4 paches into for-5.1/atomic-replace branch.

Best Regards,
Petr

Acked-by on the v2 series
#9 lore raw
In reply to Petr Mladek
Quoted 7 lines · Petr Mladek

This patchset implements ideas that were mentioned and postponed during the review of the atomic replace patchset.

The patches apply on top of livepatching.git, branch origin/for-5.1/atomic-replace.

Quoted 7 lines

Changes against v1:

+ Added Joe's patch that fixed ptr_id() error code [Joe]
+ Did proper error handling in the shadow variable sefttest [All]
+ Removed the controversial patch that was removing patch->enabled flag [All].
+ Fixed few typo's [Joe]
+ Added available Acks.

Signature · 1 line
Josh