On Tue, Jul 30, 2024 at 6:25 AM 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: Ard Biesheuvel <ardb@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 v4: > * Fixed doc punctuation, used passive tense, improved > wording consistency, fixed default value wording > * Made struct constant_table a static const __initconst > * Reworked proc_mem_foll_force() indentation and var > declarations to make code clearer > * Reworked enum + struct definition so lookup_constant() > defaults to 'always'. > > 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 f1384c7b59c9..8396e015aab3 100644 > --- a/Documentation/admin-guide/kernel-parameters.txt > +++ b/Documentation/admin-guide/kernel-parameters.txt > @@ -4788,6 +4788,16 @@ > 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 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. > + - 'ptrace': only allow mem overrides for active ptracers. > + - 'never': never allow mem overrides. > + If not specified, default is the CONFIG_PROC_MEM_* choice. > + > 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..daacb8070042 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_NO_FORCE) ? PROC_MEM_FORCE_NEVER : > + IS_ENABLED(CONFIG_PROC_MEM_FORCE_PTRACE) ? PROC_MEM_FORCE_PTRACE : > + PROC_MEM_FORCE_ALWAYS; > + > +static const struct constant_table proc_mem_force_table[] __initconst = { > + { "never", PROC_MEM_FORCE_NEVER }, > + { "ptrace", PROC_MEM_FORCE_PTRACE }, > + { } > +}; > + > +static int __init early_proc_mem_force_override(char *buf) > +{ > + if (!buf) > + return -EINVAL; > + > + proc_mem_force_override = lookup_constant(proc_mem_force_table, > + buf, PROC_MEM_FORCE_ALWAYS); proc_mem_force_table has two entries, this means: if kernel cmdline has proc_mem.force_override="invalid", PROC_MEM_FORCE_ALWAYS will be used. Another option is to have 3 entries in proc_mem_force_table: adding {"aways", PROC_MEM_FORCE_ALWAYS} and let lookup_constant return -1 when not found, and not override proc_mem_force_override. This enforces the kernel cmd line must be set to one of three choices "always|ptrace|never" to be effective. If you choose this path: please modify kernel-parameters.txt to "If not specified or invalid, default is the CONFIG_PROC_MEM_* choice." or else please clarify in the kernel-parameters.text: If not specified, default is the CONFIG_PROC_MEM_* choice If invalid str or empty string, PROC_MEM_FORCE_ALWAYS will be used regardless CONFIG_PROC_MEM_* choice > + > + 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) > +{ > + struct task_struct *task; > + bool ptrace_active = false; > + > + switch (proc_mem_force_override) { > + case PROC_MEM_FORCE_NEVER: > + return false; > + case PROC_MEM_FORCE_PTRACE: > + task = get_proc_task(file_inode(file)); > + if (task) { > + ptrace_active = task->ptrace && task->mm == mm && task->parent == current; Do we need to call "read_lock(&tasklist_lock);" ? see comments in ptrace_check_attach() of kernel/ptrace.c > + 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 +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 >
