@@ -848,9 +848,18 @@ static int proc_sys_getattr(struct mnt_idmap *idmap,
return PTR_ERR(head);
generic_fillattr(&nop_mnt_idmap, request_mask, inode, stat);
- if (table)
+ if (table) {
stat->mode = (stat->mode & S_IFMT) | table->mode;
+ /*
+ * S_ISUID and S_ISGID are to temporarily represent
+ * the internal SYSCTL_FLAG_MIN and SYSCTL_FLAG_MAX,
+ * and are not visible to users
+ */
+ stat->mode &= ~SYSCTL_FLAG_MIN;
+ stat->mode &= ~SYSCTL_FLAG_MAX;
+ }
+
sysctl_head_finish(head);
return 0;
}
@@ -1163,7 +1172,8 @@ static int sysctl_check_table(const char *path, struct ctl_table_header *header)
if (!entry->proc_handler)
err |= sysctl_err(path, entry, "No proc_handler");
- if ((entry->mode & (S_IRUGO|S_IWUGO)) != entry->mode)
+ if ((entry->mode & (S_IRUGO|S_IWUGO|SYSCTL_FLAG_MIN|SYSCTL_FLAG_MAX))
+ != entry->mode)
err |= sysctl_err(path, entry, "bogus .mode 0%o",
entry->mode);
}
@@ -28,6 +28,7 @@
#include <linux/rbtree.h>
#include <linux/uidgid.h>
#include <uapi/linux/sysctl.h>
+#include <uapi/linux/stat.h>
/* For the /proc/sys support */
struct completion;
@@ -129,6 +130,9 @@ static inline void *proc_sys_poll_event(struct ctl_table_poll *poll)
#define DEFINE_CTL_TABLE_POLL(name) \
struct ctl_table_poll name = __CTL_TABLE_POLL_INITIALIZER(name)
+#define SYSCTL_FLAG_MIN S_ISUID
+#define SYSCTL_FLAG_MAX S_ISGID
+
/* A sysctl table is an array of struct ctl_table: */
struct ctl_table {
const char *procname; /* Text ID for /proc/sys */
@@ -137,8 +141,16 @@ struct ctl_table {
umode_t mode;
proc_handler *proc_handler; /* Callback for text formatting */
struct ctl_table_poll *poll;
- void *extra1;
- void *extra2;
+ union {
+ struct {
+ void *extra1;
+ void *extra2;
+ };
+ struct {
+ unsigned long min;
+ unsigned long max;
+ };
+ };
} __randomize_layout;
struct ctl_node {
@@ -210,9 +222,13 @@ struct ctl_table_root {
#define register_sysctl(path, table) \
register_sysctl_sz(path, table, ARRAY_SIZE(table))
-#define __SYSCTL_RANGE_MIN(_a, _b, _c) (((_a)->extra1) ? *(_b((_a)->extra1)) : (_c))
+#define __SYSCTL_RANGE_EXTRA1(_a, _b, _c) (((_a)->extra1) ? *(_b((_a)->extra1)) : (_c))
+#define __SYSCTL_RANGE_MIN(_a, _b, _c) ((((_a)->mode) & SYSCTL_FLAG_MIN) ? \
+ ((_a)->min) : __SYSCTL_RANGE_EXTRA1(_a, _b, _c))
-#define __SYSCTL_RANGE_MAX(_a, _b, _c) (((_a)->extra2) ? *(_b((_a)->extra2)) : (_c))
+#define __SYSCTL_RANGE_EXTRA2(_a, _b, _c) (((_a)->extra2) ? *(_b((_a)->extra2)) : (_c))
+#define __SYSCTL_RANGE_MAX(_a, _b, _c) ((((_a)->mode) & SYSCTL_FLAG_MAX) ? \
+ ((_a)->max) : __SYSCTL_RANGE_EXTRA2(_a, _b, _c))
static inline unsigned int sysctl_range_min_u8(const struct ctl_table *table)
{
Eric points out: "by turning .extra1 and .extra2 into longs instead of keeping them as pointers and needing constants to be pointed at somewhere .. The only people I can see who find a significant benefit by consolidating all of the constants into one place are people who know how to stomp kernel memory." This patch supports encoding values directly in table entries through the following work: - extra1/extra2 and min/max are placed in one union to ensure compatibility with previous code without increasing memory space, and then we could gradually remove these unnecessary extra1/extra2. - two bits were used to represent the information of the above union: SYSCTL_FLAG_MIN: 0 - using extra1, 1 - using min. SYSCTL_FLAG_MAX: 0 - using extra2, 2 - using max. - since the proc file's mode field only uses 9 bits(777), we could use the additional two bits(S_ISUID and S_ISGID) to temporarily represent SYSCTL_FLAG_MIN and SYSCTL_FLAG_MAX. - added some helper macros. Signed-off-by: Wen Yang <wen.yang@linux.dev> Cc: Luis Chamberlain <mcgrof@kernel.org> Cc: Kees Cook <keescook@chromium.org> Cc: Joel Granados <j.granados@samsung.com> Cc: Eric W. Biederman <ebiederm@xmission.com> Cc: Christian Brauner <brauner@kernel.org> Cc: Dave Young <dyoung@redhat.com> Cc: linux-kernel@vger.kernel.org --- fs/proc/proc_sysctl.c | 14 ++++++++++++-- include/linux/sysctl.h | 24 ++++++++++++++++++++---- 2 files changed, 32 insertions(+), 6 deletions(-)