.../admin-guide/kernel-parameters.txt | 10 ++++ fs/proc/base.c | 58 ++++++++++++++++++- security/Kconfig | 32 ++++++++++ 3 files changed, 99 insertions(+), 1 deletion(-)
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@mail.gmail.com/ [1]
Cc: Doug Anderson <dianders@chromium.org>
Cc: Jeff Xu <jeffxu@google.com>
Cc: Jann Horn <jannh@google.com>
Cc: Kees Cook <kees@kernel.org>
Cc: Christian Brauner <brauner@kernel.org>
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Adrian Ratiu <adrian.ratiu@collabora.com>
---
.../admin-guide/kernel-parameters.txt | 10 ++++
fs/proc/base.c | 58 ++++++++++++++++++-
security/Kconfig | 32 ++++++++++
3 files changed, 99 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.
+ Can be one of:
+ - 'always' traditional behavior always allows mem overrides.
+ - 'ptrace' only allow for active ptracers.
+ - 'never' never allow mem permission overrides.
+ If not specified, default is 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..5ef14ba953a2 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -117,6 +117,40 @@
static u8 nlink_tid __ro_after_init;
static u8 nlink_tgid __ro_after_init;
+enum proc_mem_force_state {
+ PROC_MEM_FORCE_ALWAYS,
+ PROC_MEM_FORCE_PTRACE,
+ PROC_MEM_FORCE_NEVER
+};
+
+#if defined(CONFIG_PROC_MEM_ALWAYS_FORCE)
+static enum proc_mem_force_state proc_mem_force_override __ro_after_init = PROC_MEM_FORCE_ALWAYS;
+#elif defined(CONFIG_PROC_MEM_FORCE_PTRACE)
+static enum proc_mem_force_state proc_mem_force_override __ro_after_init = PROC_MEM_FORCE_PTRACE;
+#else
+static enum proc_mem_force_state proc_mem_force_override __ro_after_init = PROC_MEM_FORCE_NEVER;
+#endif
+
+static int __init early_proc_mem_force_override(char *buf)
+{
+ if (!buf)
+ return -EINVAL;
+
+ if (strcmp(buf, "always") == 0) {
+ proc_mem_force_override = PROC_MEM_FORCE_ALWAYS;
+ } else if (strcmp(buf, "ptrace") == 0) {
+ proc_mem_force_override = PROC_MEM_FORCE_PTRACE;
+ } else if (strcmp(buf, "never") == 0) {
+ proc_mem_force_override = PROC_MEM_FORCE_NEVER;
+ } else {
+ pr_warn("proc_mem.force_override: ignoring unknown option '%s'\n", buf);
+ return -EINVAL;
+ }
+
+ return 0;
+}
+early_param("proc_mem.force_override", early_proc_mem_force_override);
+
struct pid_entry {
const char *name;
unsigned int len;
@@ -835,6 +869,26 @@ 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;
+ }
+ default:
+ return true;
+ }
+}
+
static ssize_t mem_rw(struct file *file, char __user *buf,
size_t count, loff_t *ppos, int write)
{
@@ -855,7 +909,9 @@ 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
--
2.44.2On Tue, 23 Jul 2024 at 10:18, Adrian Ratiu <adrian.ratiu@collabora.com> 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.
Ack, this looks much simpler.
That said, I think this can be prettied up some more:
> +enum proc_mem_force_state {
> + PROC_MEM_FORCE_ALWAYS,
> + PROC_MEM_FORCE_PTRACE,
> + PROC_MEM_FORCE_NEVER
> +};
> +
> +#if defined(CONFIG_PROC_MEM_ALWAYS_FORCE)
> +static enum proc_mem_force_state proc_mem_force_override __ro_after_init = PROC_MEM_FORCE_ALWAYS;
> +#elif defined(CONFIG_PROC_MEM_FORCE_PTRACE)
> +static enum proc_mem_force_state proc_mem_force_override __ro_after_init = PROC_MEM_FORCE_PTRACE;
> +#else
> +static enum proc_mem_force_state proc_mem_force_override __ro_after_init = PROC_MEM_FORCE_NEVER;
> +#endif
I think instead of that forest of #if defined(), we can just do
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;
I also really thought we had some parser helper for this pattern:
> +static int __init early_proc_mem_force_override(char *buf)
> +{
> + if (!buf)
> + return -EINVAL;
> +
> + if (strcmp(buf, "always") == 0) {
> + proc_mem_force_override = PROC_MEM_FORCE_ALWAYS;
....
but it turns out we only really "officially" have it for filesystem
superblock parsing.
Oh well. We could do
#include <linux/fs_parser.h>
...
struct constant_table proc_mem_force_table[] {
{ "ptrace", PROC_MEM_FORCE_PTRACE },
{ "never", PROC_MEM_FORCE_NEVER },
{ }
};
...
proc_mem_force_override = lookup_constant(
proc_mem_force_table, buf, PROC_MEM_FORCE_NEVER);
but while that looks a bit prettier, the whole "fs_parser.h" thing is
admittedly odd.
Anyway, I think the patch is ok, although I also happen to think it
could be a bit prettier.
Linus
On Tuesday, July 23, 2024 21:30 EEST, Linus Torvalds <torvalds@linux-foundation.org> wrote:
> On Tue, 23 Jul 2024 at 10:18, Adrian Ratiu <adrian.ratiu@collabora.com> 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.
>
> Ack, this looks much simpler.
>
> That said, I think this can be prettied up some more:
>
> > +enum proc_mem_force_state {
> > + PROC_MEM_FORCE_ALWAYS,
> > + PROC_MEM_FORCE_PTRACE,
> > + PROC_MEM_FORCE_NEVER
> > +};
> > +
> > +#if defined(CONFIG_PROC_MEM_ALWAYS_FORCE)
> > +static enum proc_mem_force_state proc_mem_force_override __ro_after_init = PROC_MEM_FORCE_ALWAYS;
> > +#elif defined(CONFIG_PROC_MEM_FORCE_PTRACE)
> > +static enum proc_mem_force_state proc_mem_force_override __ro_after_init = PROC_MEM_FORCE_PTRACE;
> > +#else
> > +static enum proc_mem_force_state proc_mem_force_override __ro_after_init = PROC_MEM_FORCE_NEVER;
> > +#endif
>
> I think instead of that forest of #if defined(), we can just do
>
> 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;
>
> I also really thought we had some parser helper for this pattern:
>
> > +static int __init early_proc_mem_force_override(char *buf)
> > +{
> > + if (!buf)
> > + return -EINVAL;
> > +
> > + if (strcmp(buf, "always") == 0) {
> > + proc_mem_force_override = PROC_MEM_FORCE_ALWAYS;
> ....
>
> but it turns out we only really "officially" have it for filesystem
> superblock parsing.
>
> Oh well. We could do
>
> #include <linux/fs_parser.h>
> ...
> struct constant_table proc_mem_force_table[] {
> { "ptrace", PROC_MEM_FORCE_PTRACE },
> { "never", PROC_MEM_FORCE_NEVER },
> { }
> };
> ...
> proc_mem_force_override = lookup_constant(
> proc_mem_force_table, buf, PROC_MEM_FORCE_NEVER);
>
> but while that looks a bit prettier, the whole "fs_parser.h" thing is
> admittedly odd.
>
> Anyway, I think the patch is ok, although I also happen to think it
> could be a bit prettier.
Thanks again, I am perfectly fine with using fs_parser.h.
I'll wait a few days to give others a chance to review/respond,
then apply your changes and send a v3.
(this was actually v2, however git format-patch removed my
"Changes in v2" blurb and v2 title; will add them next time)
On Tue, 23 Jul 2024 at 11:30, Linus Torvalds
<torvalds@linux-foundation.org> wrote:
>
> but while that looks a bit prettier, the whole "fs_parser.h" thing is
> admittedly odd.
.. don't get me wrong - /proc obviously *is* a filesystem, but in this
context it's a boot command line parameter, not a mount option.
The "constant_table" thing obviously does work outside of mount
options too, it's just that it's documented and used in the context of
the mount API.
Linus
© 2016 - 2024 Red Hat, Inc.