1. Patchew
  2. linux
  3. View series

[PATCH v3] kprobes: Fix possible use-after-free issue on kprobe registration

Zheng Yejian posted 1 patch 1 year, 10 months ago 
kernel/kprobes.c | 18 ++++++++++++------
1 file changed, 12 insertions(+), 6 deletions(-)
[PATCH v3] kprobes: Fix possible use-after-free issue on kprobe registration
Posted by Zheng Yejian 1 year, 10 months ago 
When unloading a module, its state is changing MODULE_STATE_LIVE ->
 MODULE_STATE_GOING -> MODULE_STATE_UNFORMED. Each change will take
a time. `is_module_text_address()` and `__module_text_address()`
works with MODULE_STATE_LIVE and MODULE_STATE_GOING.
If we use `is_module_text_address()` and `__module_text_address()`
separately, there is a chance that the first one is succeeded but the
next one is failed because module->state becomes MODULE_STATE_UNFORMED
between those operations.

In `check_kprobe_address_safe()`, if the second `__module_text_address()`
is failed, that is ignored because it expected a kernel_text address.
But it may have failed simply because module->state has been changed
to MODULE_STATE_UNFORMED. In this case, arm_kprobe() will try to modify
non-exist module text address (use-after-free).

To fix this problem, we should not use separated `is_module_text_address()`
and `__module_text_address()`, but use only `__module_text_address()`
once and do `try_module_get(module)` which is only available with
MODULE_STATE_LIVE.

Signed-off-by: Zheng Yejian <zhengyejian1@huawei.com>
---
 kernel/kprobes.c | 18 ++++++++++++------
 1 file changed, 12 insertions(+), 6 deletions(-)

v3:
 - Update commit messages as suggested by Masami.
   Link: https://lore.kernel.org/all/20240409224922.5f192e8ace5f7a90937bfa69@kernel.org/
 - Also change to a more appropriate title.

v2:
 - Update commit messages and comments as suggested by Masami.
   Link: https://lore.kernel.org/all/20240408115038.b0c85767bf1f249eccc32fff@kernel.org/
 - Link: https://lore.kernel.org/all/20240408083403.3302274-1-zhengyejian1@huawei.com/

v1:
 - Link: https://lore.kernel.org/all/20240407035904.2556645-1-zhengyejian1@huawei.com/

diff --git a/kernel/kprobes.c b/kernel/kprobes.c
index 9d9095e81792..65adc815fc6e 100644
--- a/kernel/kprobes.c
+++ b/kernel/kprobes.c
@@ -1567,10 +1567,17 @@ static int check_kprobe_address_safe(struct kprobe *p,
 	jump_label_lock();
 	preempt_disable();
 
-	/* Ensure it is not in reserved area nor out of text */
-	if (!(core_kernel_text((unsigned long) p->addr) ||
-	    is_module_text_address((unsigned long) p->addr)) ||
-	    in_gate_area_no_mm((unsigned long) p->addr) ||
+	/* Ensure the address is in a text area, and find a module if exists. */
+	*probed_mod = NULL;
+	if (!core_kernel_text((unsigned long) p->addr)) {
+		*probed_mod = __module_text_address((unsigned long) p->addr);
+		if (!(*probed_mod)) {
+			ret = -EINVAL;
+			goto out;
+		}
+	}
+	/* Ensure it is not in reserved area. */
+	if (in_gate_area_no_mm((unsigned long) p->addr) ||
 	    within_kprobe_blacklist((unsigned long) p->addr) ||
 	    jump_label_text_reserved(p->addr, p->addr) ||
 	    static_call_text_reserved(p->addr, p->addr) ||
@@ -1580,8 +1587,7 @@ static int check_kprobe_address_safe(struct kprobe *p,
 		goto out;
 	}
 
-	/* Check if 'p' is probing a module. */
-	*probed_mod = __module_text_address((unsigned long) p->addr);
+	/* Get module refcount and reject __init functions for loaded modules. */
 	if (*probed_mod) {
 		/*
 		 * We must hold a refcount of the probed module while updating
-- 
2.25.1
Re: [PATCH v3] kprobes: Fix possible use-after-free issue on kprobe registration
Posted by Masami Hiramatsu (Google) 1 year, 10 months ago 
On Wed, 10 Apr 2024 09:58:02 +0800
Zheng Yejian <zhengyejian1@huawei.com> wrote:

> When unloading a module, its state is changing MODULE_STATE_LIVE ->
>  MODULE_STATE_GOING -> MODULE_STATE_UNFORMED. Each change will take
> a time. `is_module_text_address()` and `__module_text_address()`
> works with MODULE_STATE_LIVE and MODULE_STATE_GOING.
> If we use `is_module_text_address()` and `__module_text_address()`
> separately, there is a chance that the first one is succeeded but the
> next one is failed because module->state becomes MODULE_STATE_UNFORMED
> between those operations.
> 
> In `check_kprobe_address_safe()`, if the second `__module_text_address()`
> is failed, that is ignored because it expected a kernel_text address.
> But it may have failed simply because module->state has been changed
> to MODULE_STATE_UNFORMED. In this case, arm_kprobe() will try to modify
> non-exist module text address (use-after-free).
> 
> To fix this problem, we should not use separated `is_module_text_address()`
> and `__module_text_address()`, but use only `__module_text_address()`
> once and do `try_module_get(module)` which is only available with
> MODULE_STATE_LIVE.
> 
> Signed-off-by: Zheng Yejian <zhengyejian1@huawei.com>

Looks good to me. Let me pick this version, and it should be a bugfix.

Fixes: 28f6c37a2910 ("kprobes: Forbid probing on trampoline and BPF code areas")
Cc: stable@vger.kernel.org

Thank you!

> ---
>  kernel/kprobes.c | 18 ++++++++++++------
>  1 file changed, 12 insertions(+), 6 deletions(-)
> 
> v3:
>  - Update commit messages as suggested by Masami.
>    Link: https://lore.kernel.org/all/20240409224922.5f192e8ace5f7a90937bfa69@kernel.org/
>  - Also change to a more appropriate title.
> 
> v2:
>  - Update commit messages and comments as suggested by Masami.
>    Link: https://lore.kernel.org/all/20240408115038.b0c85767bf1f249eccc32fff@kernel.org/
>  - Link: https://lore.kernel.org/all/20240408083403.3302274-1-zhengyejian1@huawei.com/
> 
> v1:
>  - Link: https://lore.kernel.org/all/20240407035904.2556645-1-zhengyejian1@huawei.com/
> 
> diff --git a/kernel/kprobes.c b/kernel/kprobes.c
> index 9d9095e81792..65adc815fc6e 100644
> --- a/kernel/kprobes.c
> +++ b/kernel/kprobes.c
> @@ -1567,10 +1567,17 @@ static int check_kprobe_address_safe(struct kprobe *p,
>  	jump_label_lock();
>  	preempt_disable();
>  
> -	/* Ensure it is not in reserved area nor out of text */
> -	if (!(core_kernel_text((unsigned long) p->addr) ||
> -	    is_module_text_address((unsigned long) p->addr)) ||
> -	    in_gate_area_no_mm((unsigned long) p->addr) ||
> +	/* Ensure the address is in a text area, and find a module if exists. */
> +	*probed_mod = NULL;
> +	if (!core_kernel_text((unsigned long) p->addr)) {
> +		*probed_mod = __module_text_address((unsigned long) p->addr);
> +		if (!(*probed_mod)) {
> +			ret = -EINVAL;
> +			goto out;
> +		}
> +	}
> +	/* Ensure it is not in reserved area. */
> +	if (in_gate_area_no_mm((unsigned long) p->addr) ||
>  	    within_kprobe_blacklist((unsigned long) p->addr) ||
>  	    jump_label_text_reserved(p->addr, p->addr) ||
>  	    static_call_text_reserved(p->addr, p->addr) ||
> @@ -1580,8 +1587,7 @@ static int check_kprobe_address_safe(struct kprobe *p,
>  		goto out;
>  	}
>  
> -	/* Check if 'p' is probing a module. */
> -	*probed_mod = __module_text_address((unsigned long) p->addr);
> +	/* Get module refcount and reject __init functions for loaded modules. */
>  	if (*probed_mod) {
>  		/*
>  		 * We must hold a refcount of the probed module while updating
> -- 
> 2.25.1
> 


-- 
Masami Hiramatsu (Google) <mhiramat@kernel.org>

Patchew 2.1-devel-b67e4c2

© 2016 - 2026 Red Hat, Inc.