Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Tue, 16 May 2023 13:06:58 -0700
From:   Kees Cook <keescook@chromium.org>
To:     jeffxu@chromium.org
Cc:     dave.hansen@intel.com, luto@kernel.org, jorgelo@chromium.org,
        groeck@chromium.org, jannh@google.com, sroettger@google.com,
        akpm@linux-foundation.org, jeffxu@google.com,
        linux-kernel@vger.kernel.org, linux-kselftest@vger.kernel.org,
        linux-mm@kvack.org, linux-hardening@vger.kernel.org
Subject: Re: [PATCH 5/6] KEY: Apply PKEY_ENFORCE_API to munmap
Message-ID: <202305161302.16BF756DEA@keescook>
References: <20230515130553.2311248-1-jeffxu@chromium.org>
 <20230515130553.2311248-6-jeffxu@chromium.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20230515130553.2311248-6-jeffxu@chromium.org>
Precedence: bulk

On Mon, May 15, 2023 at 01:05:51PM +0000, jeffxu@chromium.org wrote:
> From: Jeff Xu <jeffxu@google.com>
> 
> This patch enables PKEY_ENFORCE_API for the munmap
> syscall.
> 
> Signed-off-by: Jeff Xu<jeffxu@google.com>
> ---
>  include/linux/mm.h |  2 +-
>  mm/mmap.c          | 34 ++++++++++++++++++++++++++--------
>  mm/mremap.c        |  6 ++++--
>  3 files changed, 31 insertions(+), 11 deletions(-)
> 
> diff --git a/include/linux/mm.h b/include/linux/mm.h
> index 27ce77080c79..48076e845d53 100644
> --- a/include/linux/mm.h
> +++ b/include/linux/mm.h
> @@ -3136,7 +3136,7 @@ extern unsigned long do_mmap(struct file *file, unsigned long addr,
>  	unsigned long pgoff, unsigned long *populate, struct list_head *uf);
>  extern int do_vmi_munmap(struct vma_iterator *vmi, struct mm_struct *mm,
>  			 unsigned long start, size_t len, struct list_head *uf,
> -			 bool downgrade);
> +			 bool downgrade, bool syscall);

For type checking and readability, I suggest using an enum instead of
"bool". Perhaps something like:

enum caller_origin {
	ON_BEHALF_OF_KERNEL = 0,
	ON_BEHALF_OF_USERSPACE,
};

 extern int do_vmi_munmap(struct vma_iterator *vmi, struct mm_struct *mm,
 			 unsigned long start, size_t len, struct list_head *uf,
-			 bool downgrade);
+			 bool downgrade, enum caller_origin called);

>  extern int do_munmap(struct mm_struct *, unsigned long, size_t,
>  		     struct list_head *uf);
>  extern int do_madvise(struct mm_struct *mm, unsigned long start, size_t len_in, int behavior);
> diff --git a/mm/mmap.c b/mm/mmap.c
> index 13678edaa22c..29329aa794a6 100644
> --- a/mm/mmap.c
> +++ b/mm/mmap.c
> @@ -2498,6 +2498,7 @@ do_vmi_align_munmap(struct vma_iterator *vmi, struct vm_area_struct *vma,
>   * @uf: The userfaultfd list_head
>   * @downgrade: set to true if the user wants to attempt to write_downgrade the
>   * mmap_lock
> + * @syscall: set to true if this is called from syscall entry
>   *
>   * This function takes a @mas that is either pointing to the previous VMA or set
>   * to MA_START and sets it up to remove the mapping(s).  The @len will be
> @@ -2507,7 +2508,7 @@ do_vmi_align_munmap(struct vma_iterator *vmi, struct vm_area_struct *vma,
>   */
>  int do_vmi_munmap(struct vma_iterator *vmi, struct mm_struct *mm,
>  		  unsigned long start, size_t len, struct list_head *uf,
> -		  bool downgrade)
> +		  bool downgrade, bool syscall)
>  {
>  	unsigned long end;
>  	struct vm_area_struct *vma;
> @@ -2519,6 +2520,19 @@ int do_vmi_munmap(struct vma_iterator *vmi, struct mm_struct *mm,
>  	if (end == start)
>  		return -EINVAL;
>  
> +	/*
> +	 * When called by syscall from userspace, check if the calling
> +	 * thread has the PKEY permission to modify the memory mapping.
> +	 */
> +	if (syscall && arch_check_pkey_enforce_api(mm, start, end) < 0) {

	if (called == ON_BEHALF_OF_USERSPACE &&
	    arch_check_pkey_enforce_api(mm, start, end) < 0) {

> +		char comm[TASK_COMM_LEN];
> +
> +		pr_warn_ratelimited(
> +			"munmap was denied on PKEY_ENFORCE_API memory, pid=%d '%s'\n",
> +			task_pid_nr(current), get_task_comm(comm, current));
> +		return -EACCES;
> +	}
> +
>  	 /* arch_unmap() might do unmaps itself.  */
>  	arch_unmap(mm, start, end);
>  
> @@ -2541,7 +2555,7 @@ int do_munmap(struct mm_struct *mm, unsigned long start, size_t len,
>  {
>  	VMA_ITERATOR(vmi, mm, start);
>  
> -	return do_vmi_munmap(&vmi, mm, start, len, uf, false);
> +	return do_vmi_munmap(&vmi, mm, start, len, uf, false, false);

+	return do_vmi_munmap(&vmi, mm, start, len, uf, false, ON_BEHALF_OF_KERNEL);

> [...]
>  SYSCALL_DEFINE2(munmap, unsigned long, addr, size_t, len)
>  {
>  	addr = untagged_addr(addr);
> -	return __vm_munmap(addr, len, true);
> +	return __vm_munmap(addr, len, true, true);

+	return __vm_munmap(addr, len, true, ON_BEHALF_OF_USERSPACE);

etc.

-- 
Kees Cook