Re: [PATCH v3] proc: add config & param to block forcing mem writes

[Ard Biesheuvel]

On Fri, 26 Jul 2024 at 11:11, Adrian Ratiu <adrian.ratiu@xxxxxxxxxxxxx> wrote:
>
> This adds a Kconfig option and boot param to allow removing
> the FOLL_FORCE flag from /proc/pid/mem write calls because
> it can be abused.
>
> The traditional forcing behavior is kept as default because
> it can break GDB and some other use cases.
>
> Previously we tried a more sophisticated approach allowing
> distributions to fine-tune /proc/pid/mem behavior, however
> that got NAK-ed by Linus [1], who prefers this simpler
> approach with semantics also easier to understand for users.
>
> Link: https://lore.kernel.org/lkml/CAHk-=wiGWLChxYmUA5HrT5aopZrB7_2VTa0NLZcxORgkUe5tEQ@xxxxxxxxxxxxxx/ [1]
> Cc: Doug Anderson <dianders@xxxxxxxxxxxx>
> Cc: Jeff Xu <jeffxu@xxxxxxxxxx>
> Cc: Jann Horn <jannh@xxxxxxxxxx>
> Cc: Kees Cook <kees@xxxxxxxxxx>
> Cc: Christian Brauner <brauner@xxxxxxxxxx>
> Suggested-by: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
> Signed-off-by: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
> Signed-off-by: Adrian Ratiu <adrian.ratiu@xxxxxxxxxxxxx>
> ---
> Changes in v3:
> * Simplified code to use shorthand ifs and a
>   lookup_constant() table.
>
> Changes in v2:
> * Added bootparam on top of Linus' patch.
> * Slightly reworded commit msg.
> ---
>  .../admin-guide/kernel-parameters.txt         | 10 ++++
>  fs/proc/base.c                                | 54 ++++++++++++++++++-
>  security/Kconfig                              | 32 +++++++++++
>  3 files changed, 95 insertions(+), 1 deletion(-)
>
> diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
> index c1134ad5f06d..793301f360ec 100644
> --- a/Documentation/admin-guide/kernel-parameters.txt
> +++ b/Documentation/admin-guide/kernel-parameters.txt
> @@ -4791,6 +4791,16 @@
>         printk.time=    Show timing data prefixed to each printk message line
>                         Format: <bool>  (1/Y/y=enable, 0/N/n=disable)
>
> +       proc_mem.force_override= [KNL]
> +                       Format: {always | ptrace | never}
> +                       Traditionally /proc/pid/mem allows users to override memory
> +                       permissions. This allows people to limit that.

Better to use passive tense here rather than referring to 'users' and 'people'.

'Traditionally, /proc/pid/mem allows memory permissions to be
overridden without restrictions.
This option may be set to restrict that'

> +                       Can be one of:
> +                       - 'always' traditional behavior always allows mem overrides.

punctuation please

> +                       - 'ptrace' only allow for active ptracers.
> +                       - 'never'  never allow mem permission overrides.

Please be consistent: 'mem overrides' or 'mem permission overrides' in
both instances.

> +                       If not specified, default is always.

'always'

> +
>         processor.max_cstate=   [HW,ACPI]
>                         Limit processor to maximum C-state
>                         max_cstate=9 overrides any DMI blacklist limit.
> diff --git a/fs/proc/base.c b/fs/proc/base.c
> index 72a1acd03675..0ca3fc3d9e0e 100644
> --- a/fs/proc/base.c
> +++ b/fs/proc/base.c
> @@ -85,6 +85,7 @@
>  #include <linux/elf.h>
>  #include <linux/pid_namespace.h>
>  #include <linux/user_namespace.h>
> +#include <linux/fs_parser.h>
>  #include <linux/fs_struct.h>
>  #include <linux/slab.h>
>  #include <linux/sched/autogroup.h>
> @@ -117,6 +118,35 @@
>  static u8 nlink_tid __ro_after_init;
>  static u8 nlink_tgid __ro_after_init;
>
> +enum proc_mem_force {
> +       PROC_MEM_FORCE_ALWAYS,
> +       PROC_MEM_FORCE_PTRACE,
> +       PROC_MEM_FORCE_NEVER
> +};
> +
> +static enum proc_mem_force proc_mem_force_override __ro_after_init =
> +       IS_ENABLED(CONFIG_PROC_MEM_ALWAYS_FORCE) ? PROC_MEM_FORCE_ALWAYS :
> +       IS_ENABLED(CONFIG_PROC_MEM_FORCE_PTRACE) ? PROC_MEM_FORCE_PTRACE :
> +       PROC_MEM_FORCE_NEVER;
> +
> +struct constant_table proc_mem_force_table[] = {

This can be static const __initconst

> +       { "always", PROC_MEM_FORCE_ALWAYS },
> +       { "ptrace", PROC_MEM_FORCE_PTRACE },
> +       { }
> +};
> +
> +static int __init early_proc_mem_force_override(char *buf)
> +{
> +       if (!buf)
> +               return -EINVAL;
> +

Can this ever happen?

> +       proc_mem_force_override = lookup_constant(proc_mem_force_table,
> +                                                 buf, PROC_MEM_FORCE_NEVER);
> +
> +       return 0;
> +}
> +early_param("proc_mem.force_override", early_proc_mem_force_override);
> +
>  struct pid_entry {
>         const char *name;
>         unsigned int len;
> @@ -835,6 +865,26 @@ static int mem_open(struct inode *inode, struct file *file)
>         return ret;
>  }
>
> +static bool proc_mem_foll_force(struct file *file, struct mm_struct *mm)
> +{
> +       switch (proc_mem_force_override) {
> +       case PROC_MEM_FORCE_NEVER:
> +               return false;
> +       case PROC_MEM_FORCE_PTRACE: {
> +               bool ptrace_active = false;
> +               struct task_struct *task = get_proc_task(file_inode(file));
> +
> +               if (task) {
> +                       ptrace_active = task->ptrace && task->mm == mm && task->parent == current;
> +                       put_task_struct(task);
> +               }
> +               return ptrace_active;
> +       }

This indentation looks dodgy. If you move the local var declarations
out of this block, and use assignments instead, you don't need to  { }
at all.


> +       default:
> +               return true;
> +       }
> +}
> +
>  static ssize_t mem_rw(struct file *file, char __user *buf,
>                         size_t count, loff_t *ppos, int write)
>  {
> @@ -855,7 +905,9 @@ static ssize_t mem_rw(struct file *file, char __user *buf,
>         if (!mmget_not_zero(mm))
>                 goto free;
>
> -       flags = FOLL_FORCE | (write ? FOLL_WRITE : 0);
> +       flags = write ? FOLL_WRITE : 0;
> +       if (proc_mem_foll_force(file, mm))
> +               flags |= FOLL_FORCE;
>
>         while (count > 0) {
>                 size_t this_len = min_t(size_t, count, PAGE_SIZE);
> diff --git a/security/Kconfig b/security/Kconfig
> index 412e76f1575d..a93c1a9b7c28 100644
> --- a/security/Kconfig
> +++ b/security/Kconfig
> @@ -19,6 +19,38 @@ config SECURITY_DMESG_RESTRICT
>
>           If you are unsure how to answer this question, answer N.
>
> +choice
> +       prompt "Allow /proc/pid/mem access override"
> +       default PROC_MEM_ALWAYS_FORCE
> +       help
> +         Traditionally /proc/pid/mem allows users to override memory
> +         permissions for users like ptrace, assuming they have ptrace
> +         capability.
> +
> +         This allows people to limit that - either never override, or
> +         require actual active ptrace attachment.
> +
> +         Defaults to the traditional behavior (for now)
> +
> +config PROC_MEM_ALWAYS_FORCE
> +       bool "Traditional /proc/pid/mem behavior"
> +       help
> +         This allows /proc/pid/mem accesses to override memory mapping
> +         permissions if you have ptrace access rights.
> +
> +config PROC_MEM_FORCE_PTRACE
> +       bool "Require active ptrace() use for access override"
> +       help
> +         This allows /proc/pid/mem accesses to override memory mapping
> +         permissions for active ptracers like gdb.
> +
> +config PROC_MEM_NO_FORCE
> +       bool "Never"
> +       help
> +         Never override memory mapping permissions
> +
> +endchoice
> +
>  config SECURITY
>         bool "Enable different security models"
>         depends on SYSFS
> --
> 2.44.2
>
>