From ee930fffee469d076998274a2ca55e13dc1efb67 Mon Sep 17 00:00:00 2001
From: hc <hc@nodka.com>
Date: Fri, 10 May 2024 08:50:54 +0000
Subject: [PATCH] enable tun/tap/iptables
---
kernel/drivers/misc/lkdtm/bugs.c | 287 ++++++++++++++++++++++++++++++++++++++++++++++++++++-----
1 files changed, 261 insertions(+), 26 deletions(-)
diff --git a/kernel/drivers/misc/lkdtm/bugs.c b/kernel/drivers/misc/lkdtm/bugs.c
index 7eebbdf..d39b813 100644
--- a/kernel/drivers/misc/lkdtm/bugs.c
+++ b/kernel/drivers/misc/lkdtm/bugs.c
@@ -11,6 +11,11 @@
#include <linux/sched/signal.h>
#include <linux/sched/task_stack.h>
#include <linux/uaccess.h>
+#include <linux/slab.h>
+
+#if IS_ENABLED(CONFIG_X86_32) && !IS_ENABLED(CONFIG_UML)
+#include <asm/desc.h>
+#endif
struct lkdtm_list {
struct list_head node;
@@ -22,7 +27,7 @@
* recurse past the end of THREAD_SIZE by default.
*/
#if defined(CONFIG_FRAME_WARN) && (CONFIG_FRAME_WARN > 0)
-#define REC_STACK_SIZE (CONFIG_FRAME_WARN / 2)
+#define REC_STACK_SIZE (_AC(CONFIG_FRAME_WARN, UL) / 2)
#else
#define REC_STACK_SIZE (THREAD_SIZE / 8)
#endif
@@ -32,12 +37,20 @@
static DEFINE_SPINLOCK(lock_me_up);
-static int recursive_loop(int remaining)
+/*
+ * Make sure compiler does not optimize this function or stack frame away:
+ * - function marked noinline
+ * - stack variables are marked volatile
+ * - stack variables are written (memset()) and read (pr_info())
+ * - function has external effects (pr_info())
+ * */
+static int noinline recursive_loop(int remaining)
{
- char buf[REC_STACK_SIZE];
+ volatile char buf[REC_STACK_SIZE];
- /* Make sure compiler does not optimize this away. */
- memset(buf, (remaining & 0xff) | 0x1, REC_STACK_SIZE);
+ memset((void *)buf, remaining & 0xFF, sizeof(buf));
+ pr_info("loop %d/%d ...\n", (int)buf[remaining % sizeof(buf)],
+ recur_count);
if (!remaining)
return 0;
else
@@ -67,7 +80,12 @@
void lkdtm_WARNING(void)
{
- WARN(1, "Warning message trigger count: %d\n", warn_counter++);
+ WARN_ON(++warn_counter);
+}
+
+void lkdtm_WARNING_MESSAGE(void)
+{
+ WARN(1, "Warning message trigger count: %d\n", ++warn_counter);
}
void lkdtm_EXCEPTION(void)
@@ -81,9 +99,12 @@
;
}
-void lkdtm_OVERFLOW(void)
+void lkdtm_EXHAUST_STACK(void)
{
- (void) recursive_loop(recur_count);
+ pr_info("Calling function with %lu frame size to depth %d ...\n",
+ REC_STACK_SIZE, recur_count);
+ recursive_loop(recur_count);
+ pr_info("FAIL: survived without exhausting stack?!\n");
}
static noinline void __lkdtm_CORRUPT_STACK(void *stack)
@@ -97,9 +118,8 @@
/* Use default char array length that triggers stack protection. */
char data[8] __aligned(sizeof(void *));
- __lkdtm_CORRUPT_STACK(&data);
-
- pr_info("Corrupted stack containing char array ...\n");
+ pr_info("Corrupting stack containing char array ...\n");
+ __lkdtm_CORRUPT_STACK((void *)&data);
}
/* Same as above but will only get a canary with -fstack-protector-strong */
@@ -110,9 +130,8 @@
unsigned long *ptr;
} data __aligned(sizeof(void *));
- __lkdtm_CORRUPT_STACK(&data);
-
- pr_info("Corrupted stack containing union ...\n");
+ pr_info("Corrupting stack containing union ...\n");
+ __lkdtm_CORRUPT_STACK((void *)&data);
}
void lkdtm_UNALIGNED_LOAD_STORE_WRITE(void)
@@ -125,6 +144,9 @@
if (*p == 0)
val = 0x87654321;
*p = val;
+
+ if (IS_ENABLED(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS))
+ pr_err("XFAIL: arch has CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS\n");
}
void lkdtm_SOFTLOCKUP(void)
@@ -153,6 +175,86 @@
{
set_current_state(TASK_UNINTERRUPTIBLE);
schedule();
+}
+
+volatile unsigned int huge = INT_MAX - 2;
+volatile unsigned int ignored;
+
+void lkdtm_OVERFLOW_SIGNED(void)
+{
+ int value;
+
+ value = huge;
+ pr_info("Normal signed addition ...\n");
+ value += 1;
+ ignored = value;
+
+ pr_info("Overflowing signed addition ...\n");
+ value += 4;
+ ignored = value;
+}
+
+
+void lkdtm_OVERFLOW_UNSIGNED(void)
+{
+ unsigned int value;
+
+ value = huge;
+ pr_info("Normal unsigned addition ...\n");
+ value += 1;
+ ignored = value;
+
+ pr_info("Overflowing unsigned addition ...\n");
+ value += 4;
+ ignored = value;
+}
+
+/* Intentionally using old-style flex array definition of 1 byte. */
+struct array_bounds_flex_array {
+ int one;
+ int two;
+ char data[1];
+};
+
+struct array_bounds {
+ int one;
+ int two;
+ char data[8];
+ int three;
+};
+
+void lkdtm_ARRAY_BOUNDS(void)
+{
+ struct array_bounds_flex_array *not_checked;
+ struct array_bounds *checked;
+ volatile int i;
+
+ not_checked = kmalloc(sizeof(*not_checked) * 2, GFP_KERNEL);
+ checked = kmalloc(sizeof(*checked) * 2, GFP_KERNEL);
+ if (!not_checked || !checked) {
+ kfree(not_checked);
+ kfree(checked);
+ return;
+ }
+
+ pr_info("Array access within bounds ...\n");
+ /* For both, touch all bytes in the actual member size. */
+ for (i = 0; i < sizeof(checked->data); i++)
+ checked->data[i] = 'A';
+ /*
+ * For the uninstrumented flex array member, also touch 1 byte
+ * beyond to verify it is correctly uninstrumented.
+ */
+ for (i = 0; i < sizeof(not_checked->data) + 1; i++)
+ not_checked->data[i] = 'A';
+
+ pr_info("Array access beyond bounds ...\n");
+ for (i = 0; i < sizeof(checked->data) + 1; i++)
+ checked->data[i] = 'B';
+
+ kfree(not_checked);
+ kfree(checked);
+ pr_err("FAIL: survived array bounds overflow!\n");
}
void lkdtm_CORRUPT_LIST_ADD(void)
@@ -218,16 +320,6 @@
pr_err("list_del() corruption not detected!\n");
}
-/* Test if unbalanced set_fs(KERNEL_DS)/set_fs(USER_DS) check exists. */
-void lkdtm_CORRUPT_USER_DS(void)
-{
- pr_info("setting bad task size limit\n");
- set_fs(KERNEL_DS);
-
- /* Make sure we do not keep running with a KERNEL_DS! */
- force_sig(SIGKILL, current);
-}
-
/* Test that VMAP_STACK is actually allocating with a leading guard page */
void lkdtm_STACK_GUARD_PAGE_LEADING(void)
{
@@ -239,7 +331,7 @@
byte = *ptr;
- pr_err("FAIL: accessed page before stack!\n");
+ pr_err("FAIL: accessed page before stack! (byte: %x)\n", byte);
}
/* Test that VMAP_STACK is actually allocating with a trailing guard page */
@@ -253,5 +345,148 @@
byte = *ptr;
- pr_err("FAIL: accessed page after stack!\n");
+ pr_err("FAIL: accessed page after stack! (byte: %x)\n", byte);
+}
+
+void lkdtm_UNSET_SMEP(void)
+{
+#if IS_ENABLED(CONFIG_X86_64) && !IS_ENABLED(CONFIG_UML)
+#define MOV_CR4_DEPTH 64
+ void (*direct_write_cr4)(unsigned long val);
+ unsigned char *insn;
+ unsigned long cr4;
+ int i;
+
+ cr4 = native_read_cr4();
+
+ if ((cr4 & X86_CR4_SMEP) != X86_CR4_SMEP) {
+ pr_err("FAIL: SMEP not in use\n");
+ return;
+ }
+ cr4 &= ~(X86_CR4_SMEP);
+
+ pr_info("trying to clear SMEP normally\n");
+ native_write_cr4(cr4);
+ if (cr4 == native_read_cr4()) {
+ pr_err("FAIL: pinning SMEP failed!\n");
+ cr4 |= X86_CR4_SMEP;
+ pr_info("restoring SMEP\n");
+ native_write_cr4(cr4);
+ return;
+ }
+ pr_info("ok: SMEP did not get cleared\n");
+
+ /*
+ * To test the post-write pinning verification we need to call
+ * directly into the middle of native_write_cr4() where the
+ * cr4 write happens, skipping any pinning. This searches for
+ * the cr4 writing instruction.
+ */
+ insn = (unsigned char *)native_write_cr4;
+ for (i = 0; i < MOV_CR4_DEPTH; i++) {
+ /* mov %rdi, %cr4 */
+ if (insn[i] == 0x0f && insn[i+1] == 0x22 && insn[i+2] == 0xe7)
+ break;
+ /* mov %rdi,%rax; mov %rax, %cr4 */
+ if (insn[i] == 0x48 && insn[i+1] == 0x89 &&
+ insn[i+2] == 0xf8 && insn[i+3] == 0x0f &&
+ insn[i+4] == 0x22 && insn[i+5] == 0xe0)
+ break;
+ }
+ if (i >= MOV_CR4_DEPTH) {
+ pr_info("ok: cannot locate cr4 writing call gadget\n");
+ return;
+ }
+ direct_write_cr4 = (void *)(insn + i);
+
+ pr_info("trying to clear SMEP with call gadget\n");
+ direct_write_cr4(cr4);
+ if (native_read_cr4() & X86_CR4_SMEP) {
+ pr_info("ok: SMEP removal was reverted\n");
+ } else {
+ pr_err("FAIL: cleared SMEP not detected!\n");
+ cr4 |= X86_CR4_SMEP;
+ pr_info("restoring SMEP\n");
+ native_write_cr4(cr4);
+ }
+#else
+ pr_err("XFAIL: this test is x86_64-only\n");
+#endif
+}
+
+void lkdtm_DOUBLE_FAULT(void)
+{
+#if IS_ENABLED(CONFIG_X86_32) && !IS_ENABLED(CONFIG_UML)
+ /*
+ * Trigger #DF by setting the stack limit to zero. This clobbers
+ * a GDT TLS slot, which is okay because the current task will die
+ * anyway due to the double fault.
+ */
+ struct desc_struct d = {
+ .type = 3, /* expand-up, writable, accessed data */
+ .p = 1, /* present */
+ .d = 1, /* 32-bit */
+ .g = 0, /* limit in bytes */
+ .s = 1, /* not system */
+ };
+
+ local_irq_disable();
+ write_gdt_entry(get_cpu_gdt_rw(smp_processor_id()),
+ GDT_ENTRY_TLS_MIN, &d, DESCTYPE_S);
+
+ /*
+ * Put our zero-limit segment in SS and then trigger a fault. The
+ * 4-byte access to (%esp) will fault with #SS, and the attempt to
+ * deliver the fault will recursively cause #SS and result in #DF.
+ * This whole process happens while NMIs and MCEs are blocked by the
+ * MOV SS window. This is nice because an NMI with an invalid SS
+ * would also double-fault, resulting in the NMI or MCE being lost.
+ */
+ asm volatile ("movw %0, %%ss; addl $0, (%%esp)" ::
+ "r" ((unsigned short)(GDT_ENTRY_TLS_MIN << 3)));
+
+ pr_err("FAIL: tried to double fault but didn't die\n");
+#else
+ pr_err("XFAIL: this test is ia32-only\n");
+#endif
+}
+
+#ifdef CONFIG_ARM64
+static noinline void change_pac_parameters(void)
+{
+ if (IS_ENABLED(CONFIG_ARM64_PTR_AUTH)) {
+ /* Reset the keys of current task */
+ ptrauth_thread_init_kernel(current);
+ ptrauth_thread_switch_kernel(current);
+ }
+}
+#endif
+
+noinline void lkdtm_CORRUPT_PAC(void)
+{
+#ifdef CONFIG_ARM64
+#define CORRUPT_PAC_ITERATE 10
+ int i;
+
+ if (!IS_ENABLED(CONFIG_ARM64_PTR_AUTH))
+ pr_err("FAIL: kernel not built with CONFIG_ARM64_PTR_AUTH\n");
+
+ if (!system_supports_address_auth()) {
+ pr_err("FAIL: CPU lacks pointer authentication feature\n");
+ return;
+ }
+
+ pr_info("changing PAC parameters to force function return failure...\n");
+ /*
+ * PAC is a hash value computed from input keys, return address and
+ * stack pointer. As pac has fewer bits so there is a chance of
+ * collision, so iterate few times to reduce the collision probability.
+ */
+ for (i = 0; i < CORRUPT_PAC_ITERATE; i++)
+ change_pac_parameters();
+
+ pr_err("FAIL: survived PAC changes! Kernel may be unstable from here\n");
+#else
+ pr_err("XFAIL: this test is arm64-only\n");
+#endif
}
--
Gitblit v1.6.2