s390/kernel: add support for kernel address space layout randomization (KASLR)

This patch adds support for relocating the kernel to a random address.
The random kernel offset is obtained from cpacf, using either TRNG, PRNO,
or KMC_PRNG, depending on supported MSA level.

KERNELOFFSET is added to vmcoreinfo, for crash --kaslr support.
Signed-off-by: Gerald Schaefer <gerald.schaefer@de.ibm.com>
Reviewed-by: Philipp Rudo <prudo@linux.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

arch/s390/Kconfig

@@ -637,6 +637,16 @@ config RELOCATABLE
 	  The relocations make the kernel image about 15% larger (compressed
 	  10%), but are discarded at runtime.
 
+config RANDOMIZE_BASE
+	bool "Randomize the address of the kernel image (KASLR)"
+	depends on RELOCATABLE
+	default y
+	help
+	  In support of Kernel Address Space Layout Randomization (KASLR),
+	  this randomizes the address at which the kernel image is loaded,
+	  as a security feature that deters exploit attempts relying on
+	  knowledge of the location of kernel internals.
+
 endmenu
 
 menu "Memory setup"

arch/s390/boot/Makefile

@@ -33,6 +33,7 @@ obj-y	+= string.o ebcdic.o sclp_early_core.o mem.o ipl_vmparm.o cmdline.o
 obj-y	+= ctype.o text_dma.o
 obj-$(CONFIG_PROTECTED_VIRTUALIZATION_GUEST)	+= uv.o
 obj-$(CONFIG_RELOCATABLE)	+= machine_kexec_reloc.o
+obj-$(CONFIG_RANDOMIZE_BASE)	+= kaslr.o
 targets	:= bzImage startup.a section_cmp.boot.data section_cmp.boot.preserved.data $(obj-y)
 subdir-	:= compressed
 

arch/s390/boot/boot.h

@@ -9,6 +9,9 @@ void setup_boot_command_line(void);
 void parse_boot_command_line(void);
 void setup_memory_end(void);
 void print_missing_facilities(void);
+unsigned long get_random_base(unsigned long safe_addr);
+
+extern int kaslr_enabled;
 
 unsigned long read_ipl_report(unsigned long safe_offset);
 

arch/s390/boot/ipl_parm.c

@@ -18,6 +18,8 @@ unsigned long __bootdata(memory_end);
 int __bootdata(memory_end_set);
 int __bootdata(noexec_disabled);
 
+int kaslr_enabled __section(.data);
+
 static inline int __diag308(unsigned long subcode, void *addr)
 {
 	register unsigned long _addr asm("0") = (unsigned long)addr;
@@ -214,6 +216,7 @@ void parse_boot_command_line(void)
 	char *args;
 	int rc;
 
+	kaslr_enabled = IS_ENABLED(CONFIG_RANDOMIZE_BASE);
 	args = strcpy(command_line_buf, early_command_line);
 	while (*args) {
 		args = next_arg(args, &param, &val);
@@ -231,15 +234,21 @@ void parse_boot_command_line(void)
 
 		if (!strcmp(param, "facilities"))
 			modify_fac_list(val);
+
+		if (!strcmp(param, "nokaslr"))
+			kaslr_enabled = 0;
 	}
 }
 
 void setup_memory_end(void)
 {
 #ifdef CONFIG_CRASH_DUMP
-	if (!OLDMEM_BASE && ipl_block_valid &&
-	    ipl_block.pb0_hdr.pbt == IPL_PBT_FCP &&
-	    ipl_block.fcp.opt == IPL_PB0_FCP_OPT_DUMP) {
+	if (OLDMEM_BASE) {
+		kaslr_enabled = 0;
+	} else if (ipl_block_valid &&
+		   ipl_block.pb0_hdr.pbt == IPL_PBT_FCP &&
+		   ipl_block.fcp.opt == IPL_PB0_FCP_OPT_DUMP) {
+		kaslr_enabled = 0;
 		if (!sclp_early_get_hsa_size(&memory_end) && memory_end)
 			memory_end_set = 1;
 	}

arch/s390/boot/kaslr.c

+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright IBM Corp. 2019
+ */
+#include <asm/mem_detect.h>
+#include <asm/cpacf.h>
+#include <asm/timex.h>
+#include <asm/sclp.h>
+#include "compressed/decompressor.h"
+
+#define PRNG_MODE_TDES	 1
+#define PRNG_MODE_SHA512 2
+#define PRNG_MODE_TRNG	 3
+
+struct prno_parm {
+	u32 res;
+	u32 reseed_counter;
+	u64 stream_bytes;
+	u8  V[112];
+	u8  C[112];
+};
+
+struct prng_parm {
+	u8  parm_block[32];
+	u32 reseed_counter;
+	u64 byte_counter;
+};
+
+static int check_prng(void)
+{
+	if (!cpacf_query_func(CPACF_KMC, CPACF_KMC_PRNG)) {
+		sclp_early_printk("KASLR disabled: CPU has no PRNG\n");
+		return 0;
+	}
+	if (cpacf_query_func(CPACF_PRNO, CPACF_PRNO_TRNG))
+		return PRNG_MODE_TRNG;
+	if (cpacf_query_func(CPACF_PRNO, CPACF_PRNO_SHA512_DRNG_GEN))
+		return PRNG_MODE_SHA512;
+	else
+		return PRNG_MODE_TDES;
+}
+
+static unsigned long get_random(unsigned long limit)
+{
+	struct prng_parm prng = {
+		/* initial parameter block for tdes mode, copied from libica */
+		.parm_block = {
+			0x0F, 0x2B, 0x8E, 0x63, 0x8C, 0x8E, 0xD2, 0x52,
+			0x64, 0xB7, 0xA0, 0x7B, 0x75, 0x28, 0xB8, 0xF4,
+			0x75, 0x5F, 0xD2, 0xA6, 0x8D, 0x97, 0x11, 0xFF,
+			0x49, 0xD8, 0x23, 0xF3, 0x7E, 0x21, 0xEC, 0xA0
+		},
+	};
+	unsigned long seed, random;
+	struct prno_parm prno;
+	__u64 entropy[4];
+	int mode, i;
+
+	mode = check_prng();
+	seed = get_tod_clock_fast();
+	switch (mode) {
+	case PRNG_MODE_TRNG:
+		cpacf_trng(NULL, 0, (u8 *) &random, sizeof(random));
+		break;
+	case PRNG_MODE_SHA512:
+		cpacf_prno(CPACF_PRNO_SHA512_DRNG_SEED, &prno, NULL, 0,
+			   (u8 *) &seed, sizeof(seed));
+		cpacf_prno(CPACF_PRNO_SHA512_DRNG_GEN, &prno, (u8 *) &random,
+			   sizeof(random), NULL, 0);
+		break;
+	case PRNG_MODE_TDES:
+		/* add entropy */
+		*(unsigned long *) prng.parm_block ^= seed;
+		for (i = 0; i < 16; i++) {
+			cpacf_kmc(CPACF_KMC_PRNG, prng.parm_block,
+				  (char *) entropy, (char *) entropy,
+				  sizeof(entropy));
+			memcpy(prng.parm_block, entropy, sizeof(entropy));
+		}
+		random = seed;
+		cpacf_kmc(CPACF_KMC_PRNG, prng.parm_block, (u8 *) &random,
+			  (u8 *) &random, sizeof(random));
+		break;
+	default:
+		random = 0;
+	}
+	return random % limit;
+}
+
+unsigned long get_random_base(unsigned long safe_addr)
+{
+	unsigned long base, start, end, kernel_size;
+	unsigned long block_sum, offset;
+	int i;
+
+	if (IS_ENABLED(CONFIG_BLK_DEV_INITRD) && INITRD_START && INITRD_SIZE) {
+		if (safe_addr < INITRD_START + INITRD_SIZE)
+			safe_addr = INITRD_START + INITRD_SIZE;
+	}
+	safe_addr = ALIGN(safe_addr, THREAD_SIZE);
+
+	kernel_size = vmlinux.image_size + vmlinux.bss_size;
+	block_sum = 0;
+	for_each_mem_detect_block(i, &start, &end) {
+		if (memory_end_set) {
+			if (start >= memory_end)
+				break;
+			if (end > memory_end)
+				end = memory_end;
+		}
+		if (end - start < kernel_size)
+			continue;
+		block_sum += end - start - kernel_size;
+	}
+	if (!block_sum) {
+		sclp_early_printk("KASLR disabled: not enough memory\n");
+		return 0;
+	}
+
+	base = get_random(block_sum);
+	if (base == 0)
+		return 0;
+	if (base < safe_addr)
+		base = safe_addr;
+	block_sum = offset = 0;
+	for_each_mem_detect_block(i, &start, &end) {
+		if (memory_end_set) {
+			if (start >= memory_end)
+				break;
+			if (end > memory_end)
+				end = memory_end;
+		}
+		if (end - start < kernel_size)
+			continue;
+		block_sum += end - start - kernel_size;
+		if (base <= block_sum) {
+			base = start + base - offset;
+			base = ALIGN_DOWN(base, THREAD_SIZE);
+			break;
+		}
+		offset = block_sum;
+	}
+	return base;
+}

arch/s390/boot/startup.c

@@ -12,6 +12,7 @@
 
 extern char __boot_data_start[], __boot_data_end[];
 extern char __boot_data_preserved_start[], __boot_data_preserved_end[];
+unsigned long __bootdata_preserved(__kaslr_offset);
 
 /*
  * Some code and data needs to stay below 2 GB, even when the kernel would be
@@ -113,6 +114,7 @@ static void handle_relocs(unsigned long offset)
 
 void startup_kernel(void)
 {
+	unsigned long random_lma;
 	unsigned long safe_addr;
 	void *img;
 
@@ -126,12 +128,37 @@ void startup_kernel(void)
 	parse_boot_command_line();
 	setup_memory_end();
 	detect_memory();
+
+	random_lma = __kaslr_offset = 0;
+	if (IS_ENABLED(CONFIG_RANDOMIZE_BASE) && kaslr_enabled) {
+		random_lma = get_random_base(safe_addr);
+		if (random_lma) {
+			__kaslr_offset = random_lma - vmlinux.default_lma;
+			img = (void *)vmlinux.default_lma;
+			vmlinux.default_lma += __kaslr_offset;
+			vmlinux.entry += __kaslr_offset;
+			vmlinux.bootdata_off += __kaslr_offset;
+			vmlinux.bootdata_preserved_off += __kaslr_offset;
+			vmlinux.rela_dyn_start += __kaslr_offset;
+			vmlinux.rela_dyn_end += __kaslr_offset;
+			vmlinux.dynsym_start += __kaslr_offset;
+		}
+	}
+
 	if (!IS_ENABLED(CONFIG_KERNEL_UNCOMPRESSED)) {
 		img = decompress_kernel();
 		memmove((void *)vmlinux.default_lma, img, vmlinux.image_size);
-	}
+	} else if (__kaslr_offset)
+		memcpy((void *)vmlinux.default_lma, img, vmlinux.image_size);
+
 	copy_bootdata();
 	if (IS_ENABLED(CONFIG_RELOCATABLE))
-		handle_relocs(0);
+		handle_relocs(__kaslr_offset);
+
+	if (__kaslr_offset) {
+		/* Clear non-relocated kernel */
+		if (IS_ENABLED(CONFIG_KERNEL_UNCOMPRESSED))
+			memset(img, 0, vmlinux.image_size);
+	}
 	vmlinux.entry();
 }

arch/s390/include/asm/setup.h

@@ -147,6 +147,12 @@ extern void (*_machine_restart)(char *command);
 extern void (*_machine_halt)(void);
 extern void (*_machine_power_off)(void);
 
+extern unsigned long __kaslr_offset;
+static inline unsigned long kaslr_offset(void)
+{
+	return __kaslr_offset;
+}
+
 #else /* __ASSEMBLY__ */
 
 #define IPL_DEVICE	(IPL_DEVICE_OFFSET)

arch/s390/kernel/machine_kexec.c

@@ -255,6 +255,7 @@ void arch_crash_save_vmcoreinfo(void)
 	mem_assign_absolute(S390_lowcore.vmcore_info, paddr_vmcoreinfo_note());
 	vmcoreinfo_append_str("SDMA=%lx\n", __sdma);
 	vmcoreinfo_append_str("EDMA=%lx\n", __edma);
+	vmcoreinfo_append_str("KERNELOFFSET=%lx\n", kaslr_offset());
 }
 
 void machine_shutdown(void)

arch/s390/kernel/setup.c

@@ -108,6 +108,7 @@ unsigned long __bootdata_preserved(__stext_dma);
 unsigned long __bootdata_preserved(__etext_dma);
 unsigned long __bootdata_preserved(__sdma);
 unsigned long __bootdata_preserved(__edma);
+unsigned long __bootdata_preserved(__kaslr_offset);
 
 unsigned long VMALLOC_START;
 EXPORT_SYMBOL(VMALLOC_START);