Live Patching
 help / color / mirror / Atom feed
* [PATCH] livepatch: Fix NULL pointer dereference in klp_find_func()
@ 2026-06-23  9:16 Yafang Shao
  2026-06-25 13:24 ` Miroslav Benes
  0 siblings, 1 reply; 4+ messages in thread
From: Yafang Shao @ 2026-06-23  9:16 UTC (permalink / raw)
  To: jpoimboe, jikos, mbenes, pmladek, joe.lawrence, song
  Cc: live-patching, Yafang Shao, sashiko-bot

A NULL old_name in a newly loaded livepatch's function entry causes a
NULL pointer dereference in strcmp():

  klp_init_patch()
    klp_add_nops()
      klp_find_func()
        strcmp(old_func->old_name, func->old_name)

Add a sanity check at the beginning of klp_enable_patch() to reject
patches with NULL old_name before they reach this code path.

Reported-by: sashiko-bot <sashiko-bot@kernel.org>
Closes: https://lore.kernel.org/live-patching/20260529040130.95A9C1F00893@smtp.kernel.org/
Suggested-by: Petr Mladek <pmladek@suse.com>
Suggested-by: Miroslav Benes <mbenes@suse.cz>
Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
---
 kernel/livepatch/core.c | 34 ++++++++++++++++++++++------------
 1 file changed, 22 insertions(+), 12 deletions(-)

diff --git a/kernel/livepatch/core.c b/kernel/livepatch/core.c
index 28d15ba58a26..a240d1144e89 100644
--- a/kernel/livepatch/core.c
+++ b/kernel/livepatch/core.c
@@ -799,9 +799,6 @@ void klp_free_replaced_patches_async(struct klp_patch *new_patch)
 
 static int klp_init_func(struct klp_object *obj, struct klp_func *func)
 {
-	if (!func->old_name)
-		return -EINVAL;
-
 	/*
 	 * NOPs get the address later. The patched module must be loaded,
 	 * see klp_init_object_loaded().
@@ -1092,6 +1089,25 @@ static int __klp_enable_patch(struct klp_patch *patch)
 	return ret;
 }
 
+static int klp_check_patch(struct klp_patch *patch)
+{
+	struct klp_object *obj;
+	struct klp_func *func;
+
+	if (!patch || !patch->mod || !patch->objs)
+		return -EINVAL;
+
+	klp_for_each_object_static(patch, obj) {
+		if (!obj->funcs)
+			return -EINVAL;
+		klp_for_each_func_static(obj, func) {
+			if (!func->old_name)
+				return -EINVAL;
+		}
+	}
+	return 0;
+}
+
 /**
  * klp_enable_patch() - enable the livepatch
  * @patch:	patch to be enabled
@@ -1108,16 +1124,10 @@ static int __klp_enable_patch(struct klp_patch *patch)
 int klp_enable_patch(struct klp_patch *patch)
 {
 	int ret;
-	struct klp_object *obj;
-
-	if (!patch || !patch->mod || !patch->objs)
-		return -EINVAL;
-
-	klp_for_each_object_static(patch, obj) {
-		if (!obj->funcs)
-			return -EINVAL;
-	}
 
+	ret = klp_check_patch(patch);
+	if (ret)
+		return ret;
 
 	if (!is_livepatch_module(patch->mod)) {
 		pr_err("module %s is not marked as a livepatch module\n",
-- 
2.52.0


^ permalink raw reply related	[flat|nested] 4+ messages in thread

* Re: [PATCH] livepatch: Fix NULL pointer dereference in klp_find_func()
  2026-06-23  9:16 [PATCH] livepatch: Fix NULL pointer dereference in klp_find_func() Yafang Shao
@ 2026-06-25 13:24 ` Miroslav Benes
  2026-06-28 11:46   ` [PATCH v2] " Yafang Shao
  0 siblings, 1 reply; 4+ messages in thread
From: Miroslav Benes @ 2026-06-25 13:24 UTC (permalink / raw)
  To: Yafang Shao
  Cc: jpoimboe, jikos, pmladek, joe.lawrence, song, live-patching,
	sashiko-bot

On Tue, 23 Jun 2026, Yafang Shao wrote:

> A NULL old_name in a newly loaded livepatch's function entry causes a
> NULL pointer dereference in strcmp():
> 
>   klp_init_patch()
>     klp_add_nops()
>       klp_find_func()
>         strcmp(old_func->old_name, func->old_name)
> 
> Add a sanity check at the beginning of klp_enable_patch() to reject
> patches with NULL old_name before they reach this code path.

The changelog could also mention the introduction of klp_check_patch().

Acked-by: Miroslav Benes <mbenes@suse.cz>

M

^ permalink raw reply	[flat|nested] 4+ messages in thread

* [PATCH v2] livepatch: Fix NULL pointer dereference in klp_find_func()
  2026-06-25 13:24 ` Miroslav Benes
@ 2026-06-28 11:46   ` Yafang Shao
  2026-06-30 15:50     ` Petr Mladek
  0 siblings, 1 reply; 4+ messages in thread
From: Yafang Shao @ 2026-06-28 11:46 UTC (permalink / raw)
  To: mbenes
  Cc: jikos, joe.lawrence, jpoimboe, laoar.shao, live-patching, pmladek,
	sashiko-bot, song

A NULL old_name in a newly loaded livepatch's function entry causes a
NULL pointer dereference in strcmp():

  klp_init_patch()
    klp_add_nops()
      klp_find_func()
        strcmp(old_func->old_name, func->old_name)

Add klp_check_patch() at the beginning of klp_enable_patch() to reject
patches with NULL old_name before they reach this code path.

Reported-by: sashiko-bot <sashiko-bot@kernel.org>
Closes: https://lore.kernel.org/live-patching/20260529040130.95A9C1F00893@smtp.kernel.org/
Suggested-by: Petr Mladek <pmladek@suse.com>
Suggested-by: Miroslav Benes <mbenes@suse.cz>
Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
Acked-by: Miroslav Benes <mbenes@suse.cz>
---
 kernel/livepatch/core.c | 34 ++++++++++++++++++++++------------
 1 file changed, 22 insertions(+), 12 deletions(-)

diff --git a/kernel/livepatch/core.c b/kernel/livepatch/core.c
index 28d15ba58a26..a240d1144e89 100644
--- a/kernel/livepatch/core.c
+++ b/kernel/livepatch/core.c
@@ -799,9 +799,6 @@ void klp_free_replaced_patches_async(struct klp_patch *new_patch)
 
 static int klp_init_func(struct klp_object *obj, struct klp_func *func)
 {
-	if (!func->old_name)
-		return -EINVAL;
-
 	/*
 	 * NOPs get the address later. The patched module must be loaded,
 	 * see klp_init_object_loaded().
@@ -1092,6 +1089,25 @@ static int __klp_enable_patch(struct klp_patch *patch)
 	return ret;
 }
 
+static int klp_check_patch(struct klp_patch *patch)
+{
+	struct klp_object *obj;
+	struct klp_func *func;
+
+	if (!patch || !patch->mod || !patch->objs)
+		return -EINVAL;
+
+	klp_for_each_object_static(patch, obj) {
+		if (!obj->funcs)
+			return -EINVAL;
+		klp_for_each_func_static(obj, func) {
+			if (!func->old_name)
+				return -EINVAL;
+		}
+	}
+	return 0;
+}
+
 /**
  * klp_enable_patch() - enable the livepatch
  * @patch:	patch to be enabled
@@ -1108,16 +1124,10 @@ static int __klp_enable_patch(struct klp_patch *patch)
 int klp_enable_patch(struct klp_patch *patch)
 {
 	int ret;
-	struct klp_object *obj;
-
-	if (!patch || !patch->mod || !patch->objs)
-		return -EINVAL;
-
-	klp_for_each_object_static(patch, obj) {
-		if (!obj->funcs)
-			return -EINVAL;
-	}
 
+	ret = klp_check_patch(patch);
+	if (ret)
+		return ret;
 
 	if (!is_livepatch_module(patch->mod)) {
 		pr_err("module %s is not marked as a livepatch module\n",
-- 
2.52.0


^ permalink raw reply related	[flat|nested] 4+ messages in thread

* Re: [PATCH v2] livepatch: Fix NULL pointer dereference in klp_find_func()
  2026-06-28 11:46   ` [PATCH v2] " Yafang Shao
@ 2026-06-30 15:50     ` Petr Mladek
  0 siblings, 0 replies; 4+ messages in thread
From: Petr Mladek @ 2026-06-30 15:50 UTC (permalink / raw)
  To: Yafang Shao
  Cc: mbenes, jikos, joe.lawrence, jpoimboe, live-patching, sashiko-bot,
	song

On Sun 2026-06-28 19:46:35, Yafang Shao wrote:
> A NULL old_name in a newly loaded livepatch's function entry causes a
> NULL pointer dereference in strcmp():
> 
>   klp_init_patch()
>     klp_add_nops()
>       klp_find_func()
>         strcmp(old_func->old_name, func->old_name)
> 
> Add klp_check_patch() at the beginning of klp_enable_patch() to reject
> patches with NULL old_name before they reach this code path.
> 
> Reported-by: sashiko-bot <sashiko-bot@kernel.org>
> Closes: https://lore.kernel.org/live-patching/20260529040130.95A9C1F00893@smtp.kernel.org/
> Suggested-by: Petr Mladek <pmladek@suse.com>
> Suggested-by: Miroslav Benes <mbenes@suse.cz>
> Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
> Acked-by: Miroslav Benes <mbenes@suse.cz>

Looks good to me:

Reviewed-by: Petr Mladek <pmladek@suse.com>
Tested-by: Petr Mladek <pmladek@suse.com>

I think whether to rush this into 7.2-rcX or wait for 7.3.
On one hand, it is a possible security fix. On the other hand,
broken module could never get loaded and pass even a minimal QA.

I tend to push it for 7.2-rcX just to avoid questions from people
who ignore the real life vulnerabilty aspect.

Also see a nit below.

> --- a/kernel/livepatch/core.c
> +++ b/kernel/livepatch/core.c
> @@ -1092,6 +1089,25 @@ static int __klp_enable_patch(struct klp_patch *patch)
>  	return ret;
>  }
>  
> +static int klp_check_patch(struct klp_patch *patch)
> +{
> +	struct klp_object *obj;
> +	struct klp_func *func;
> +
> +	if (!patch || !patch->mod || !patch->objs)
> +		return -EINVAL;
> +
> +	klp_for_each_object_static(patch, obj) {
> +		if (!obj->funcs)
> +			return -EINVAL;

Nit: I would add an empty line here to make it better readable.
     Of course, this is a personal preference. But it seems that
     most contributors to this code prefer visual separators
     between misc code sections as well.

     No need for v3. I could add the empty line when pushing
     the change.

> +		klp_for_each_func_static(obj, func) {
> +			if (!func->old_name)
> +				return -EINVAL;
> +		}
> +	}
> +	return 0;
> +}
> +
>  /**
>   * klp_enable_patch() - enable the livepatch
>   * @patch:	patch to be enabled

Best Regards,
Petr

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2026-06-30 15:50 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-06-23  9:16 [PATCH] livepatch: Fix NULL pointer dereference in klp_find_func() Yafang Shao
2026-06-25 13:24 ` Miroslav Benes
2026-06-28 11:46   ` [PATCH v2] " Yafang Shao
2026-06-30 15:50     ` Petr Mladek

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox