Merge tag 'hardening-v6.11-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux

Pull hardening updates from Kees Cook:

 - lkdtm/bugs: add test for hung smp_call_function_single() (Mark
   Rutland)

 - gcc-plugins: Remove duplicate included header file stringpool.h
   (Thorsten Blum)

 - ARM: Remove address checking for MMUless devices (Yanjun Yang)

 - randomize_kstack: Clean up per-arch entropy and codegen

 - KCFI: Make FineIBT mode Kconfig selectable

 - fortify: Do not special-case 0-sized destinations

* tag 'hardening-v6.11-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux:
  randomize_kstack: Improve stack alignment codegen
  ARM: Remove address checking for MMUless devices
  gcc-plugins: Remove duplicate included header file stringpool.h
  randomize_kstack: Remove non-functional per-arch entropy filtering
  fortify: Do not special-case 0-sized destinations
  x86/alternatives: Make FineIBT mode Kconfig selectable
  lkdtm/bugs: add test for hung smp_call_function_single()

arch/arm/mm/fault.c

@@ -25,6 +25,8 @@
 
 #include "fault.h"
 
+#ifdef CONFIG_MMU
+
 bool copy_from_kernel_nofault_allowed(const void *unsafe_src, size_t size)
 {
 	unsigned long addr = (unsigned long)unsafe_src;
@@ -32,8 +34,6 @@ bool copy_from_kernel_nofault_allowed(const void *unsafe_src, size_t size)
 	return addr >= TASK_SIZE && ULONG_MAX - addr >= size;
 }
 
-#ifdef CONFIG_MMU
-
 /*
  * This is useful to dump out the page tables associated with
  * 'addr' in mm 'mm'.

arch/x86/Kconfig

@@ -2414,6 +2414,15 @@ config STRICT_SIGALTSTACK_SIZE
 
 	  Say 'N' unless you want to really enforce this check.
 
+config CFI_AUTO_DEFAULT
+	bool "Attempt to use FineIBT by default at boot time"
+	depends on FINEIBT
+	default y
+	help
+	  Attempt to use FineIBT by default at boot time. If enabled,
+	  this is the same as booting with "cfi=auto". If disabled,
+	  this is the same as booting with "cfi=kcfi".
+
 source "kernel/livepatch/Kconfig"
 
 endmenu

arch/x86/include/asm/cfi.h

@@ -93,7 +93,7 @@
  *
  */
 enum cfi_mode {
-	CFI_DEFAULT,	/* FineIBT if hardware has IBT, otherwise kCFI */
+	CFI_AUTO,	/* FineIBT if hardware has IBT, otherwise kCFI */
 	CFI_OFF,	/* Taditional / IBT depending on .config */
 	CFI_KCFI,	/* Optionally CALL_PADDING, IBT, RETPOLINE */
 	CFI_FINEIBT,	/* see arch/x86/kernel/alternative.c */

arch/x86/kernel/alternative.c

@@ -901,8 +901,8 @@ void __init_or_module apply_seal_endbr(s32 *start, s32 *end) { }
 
 #endif /* CONFIG_X86_KERNEL_IBT */
 
-#ifdef CONFIG_FINEIBT
-#define __CFI_DEFAULT	CFI_DEFAULT
+#ifdef CONFIG_CFI_AUTO_DEFAULT
+#define __CFI_DEFAULT	CFI_AUTO
 #elif defined(CONFIG_CFI_CLANG)
 #define __CFI_DEFAULT	CFI_KCFI
 #else
@@ -1010,7 +1010,7 @@ static __init int cfi_parse_cmdline(char *str)
 		}
 
 		if (!strcmp(str, "auto")) {
-			cfi_mode = CFI_DEFAULT;
+			cfi_mode = CFI_AUTO;
 		} else if (!strcmp(str, "off")) {
 			cfi_mode = CFI_OFF;
 			cfi_rand = false;
@@ -1270,7 +1270,7 @@ static void __apply_fineibt(s32 *start_retpoline, s32 *end_retpoline,
 		      "FineIBT preamble wrong size: %ld", fineibt_preamble_size))
 		return;
 
-	if (cfi_mode == CFI_DEFAULT) {
+	if (cfi_mode == CFI_AUTO) {
 		cfi_mode = CFI_KCFI;
 		if (HAS_KERNEL_IBT && cpu_feature_enabled(X86_FEATURE_IBT))
 			cfi_mode = CFI_FINEIBT;

drivers/misc/lkdtm/bugs.c

@@ -286,6 +286,35 @@ static void lkdtm_HARDLOCKUP(void)
 		cpu_relax();
 }
 
+static void __lkdtm_SMP_CALL_LOCKUP(void *unused)
+{
+	for (;;)
+		cpu_relax();
+}
+
+static void lkdtm_SMP_CALL_LOCKUP(void)
+{
+	unsigned int cpu, target;
+
+	cpus_read_lock();
+
+	cpu = get_cpu();
+	target = cpumask_any_but(cpu_online_mask, cpu);
+
+	if (target >= nr_cpu_ids) {
+		pr_err("FAIL: no other online CPUs\n");
+		goto out_put_cpus;
+	}
+
+	smp_call_function_single(target, __lkdtm_SMP_CALL_LOCKUP, NULL, 1);
+
+	pr_err("FAIL: did not hang\n");
+
+out_put_cpus:
+	put_cpu();
+	cpus_read_unlock();
+}
+
 static void lkdtm_SPINLOCKUP(void)
 {
 	/* Must be called twice to trigger. */
@@ -680,6 +709,7 @@ static struct crashtype crashtypes[] = {
 	CRASHTYPE(UNALIGNED_LOAD_STORE_WRITE),
 	CRASHTYPE(SOFTLOCKUP),
 	CRASHTYPE(HARDLOCKUP),
+	CRASHTYPE(SMP_CALL_LOCKUP),
 	CRASHTYPE(SPINLOCKUP),
 	CRASHTYPE(HUNG_TASK),
 	CRASHTYPE(OVERFLOW_SIGNED),

include/linux/fortify-string.h

@@ -601,11 +601,7 @@ __FORTIFY_INLINE bool fortify_memcpy_chk(__kernel_size_t size,
 	/*
 	 * Warn when writing beyond destination field size.
 	 *
-	 * We must ignore p_size_field == 0 for existing 0-element
-	 * fake flexible arrays, until they are all converted to
-	 * proper flexible arrays.
-	 *
-	 * The implementation of __builtin_*object_size() behaves
+	 * Note the implementation of __builtin_*object_size() behaves
 	 * like sizeof() when not directly referencing a flexible
 	 * array member, which means there will be many bounds checks
 	 * that will appear at run-time, without a way for them to be
@@ -613,7 +609,7 @@ __FORTIFY_INLINE bool fortify_memcpy_chk(__kernel_size_t size,
 	 * is specifically the flexible array member).
 	 * https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101832
 	 */
-	if (p_size_field != 0 && p_size_field != SIZE_MAX &&
+	if (p_size_field != SIZE_MAX &&
 	    p_size != p_size_field && p_size_field < size)
 		return true;
 

include/linux/randomize_kstack.h

@@ -32,13 +32,19 @@ DECLARE_PER_CPU(u32, kstack_offset);
 #endif
 
 /*
- * Use, at most, 10 bits of entropy. We explicitly cap this to keep the
- * "VLA" from being unbounded (see above). 10 bits leaves enough room for
- * per-arch offset masks to reduce entropy (by removing higher bits, since
- * high entropy may overly constrain usable stack space), and for
- * compiler/arch-specific stack alignment to remove the lower bits.
+ * Use, at most, 6 bits of entropy (on 64-bit; 8 on 32-bit). This cap is
+ * to keep the "VLA" from being unbounded (see above). Additionally clear
+ * the bottom 4 bits (on 64-bit systems, 2 for 32-bit), since stack
+ * alignment will always be at least word size. This makes the compiler
+ * code gen better when it is applying the actual per-arch alignment to
+ * the final offset. The resulting randomness is reasonable without overly
+ * constraining usable stack space.
  */
-#define KSTACK_OFFSET_MAX(x)	((x) & 0x3FF)
+#ifdef CONFIG_64BIT
+#define KSTACK_OFFSET_MAX(x)	((x) & 0b1111110000)
+#else
+#define KSTACK_OFFSET_MAX(x)	((x) & 0b1111111100)
+#endif
 
 /**
  * add_random_kstack_offset - Increase stack utilization by previously

lib/fortify_kunit.c

@@ -910,10 +910,9 @@ static void fortify_test_##memfunc(struct kunit *test)		\
 	memfunc(zero.buf, srcB, 0 + unconst);			\
 	KUNIT_EXPECT_EQ(test, fortify_read_overflows, 0);	\
 	KUNIT_EXPECT_EQ(test, fortify_write_overflows, 0);	\
-	/* We currently explicitly ignore zero-sized dests. */	\
 	memfunc(zero.buf, srcB, 1 + unconst);			\
 	KUNIT_EXPECT_EQ(test, fortify_read_overflows, 0);	\
-	KUNIT_EXPECT_EQ(test, fortify_write_overflows, 0);	\
+	KUNIT_EXPECT_EQ(test, fortify_write_overflows, 1);	\
 }
 __fortify_test(memcpy)
 __fortify_test(memmove)

scripts/gcc-plugins/gcc-common.h

@@ -62,11 +62,7 @@
 #include "pass_manager.h"
 #include "predict.h"
 #include "ipa-utils.h"
-
-#if BUILDING_GCC_VERSION >= 8000
 #include "stringpool.h"
-#endif
-
 #include "attribs.h"
 #include "varasm.h"
 #include "stor-layout.h"
@@ -78,7 +74,6 @@
 #include "context.h"
 #include "tree-ssa-alias.h"
 #include "tree-ssa.h"
-#include "stringpool.h"
 #if BUILDING_GCC_VERSION >= 7000
 #include "tree-vrp.h"
 #endif

tools/testing/selftests/lkdtm/tests.txt

@@ -31,6 +31,7 @@ SLAB_FREE_CROSS
 SLAB_FREE_PAGE
 #SOFTLOCKUP Hangs the system
 #HARDLOCKUP Hangs the system
+#SMP_CALL_LOCKUP Hangs the system
 #SPINLOCKUP Hangs the system
 #HUNG_TASK Hangs the system
 EXEC_DATA