s390/kernel: build a relocatable kernel

This patch adds support for building a relocatable kernel with -fPIE.
The kernel will be relocated to 0 early in the boot process.

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

@@ -624,6 +624,19 @@ config EXPOLINE_FULL
 
 endchoice
 
+config RELOCATABLE
+	bool "Build a relocatable kernel"
+	select MODULE_REL_CRCS if MODVERSIONS
+	default y
+	help
+	  This builds a kernel image that retains relocation information
+	  so it can be loaded at an arbitrary address.
+	  The kernel is linked as a position-independent executable (PIE)
+	  and contains dynamic relocations which are processed early in the
+	  bootup process.
+	  The relocations make the kernel image about 15% larger (compressed
+	  10%), but are discarded at runtime.
+
 endmenu
 
 menu "Memory setup"

arch/s390/Makefile

@@ -16,6 +16,10 @@ KBUILD_AFLAGS_MODULE += -fPIC
 KBUILD_CFLAGS_MODULE += -fPIC
 KBUILD_AFLAGS	+= -m64
 KBUILD_CFLAGS	+= -m64
+ifeq ($(CONFIG_RELOCATABLE),y)
+KBUILD_CFLAGS	+= -fPIE
+LDFLAGS_vmlinux	:= -pie
+endif
 aflags_dwarf	:= -Wa,-gdwarf-2
 KBUILD_AFLAGS_DECOMPRESSOR := -m64 -D__ASSEMBLY__
 KBUILD_AFLAGS_DECOMPRESSOR += $(if $(CONFIG_DEBUG_INFO),$(aflags_dwarf))

arch/s390/boot/Makefile

@@ -32,6 +32,7 @@ obj-y	:= head.o als.o startup.o mem_detect.o ipl_parm.o ipl_report.o
 obj-y	+= string.o ebcdic.o sclp_early_core.o mem.o ipl_vmparm.o cmdline.o
 obj-y	+= ctype.o
 obj-$(CONFIG_PROTECTED_VIRTUALIZATION_GUEST)	+= uv.o
+obj-$(CONFIG_RELOCATABLE)	+= machine_kexec_reloc.o
 targets	:= bzImage startup.a section_cmp.boot.data section_cmp.boot.preserved.data $(obj-y)
 subdir-	:= compressed
 

arch/s390/boot/compressed/decompressor.h

@@ -19,6 +19,9 @@ struct vmlinux_info {
 	unsigned long bootdata_size;
 	unsigned long bootdata_preserved_off;
 	unsigned long bootdata_preserved_size;
+	unsigned long dynsym_start;
+	unsigned long rela_dyn_start;
+	unsigned long rela_dyn_end;
 };
 
 extern char _vmlinux_info[];

arch/s390/boot/machine_kexec_reloc.c

@@ -0,0 +1,2 @@
+// SPDX-License-Identifier: GPL-2.0
+#include "../kernel/machine_kexec_reloc.c"

arch/s390/boot/startup.c

@@ -1,6 +1,8 @@
 // SPDX-License-Identifier: GPL-2.0
 #include <linux/string.h>
+#include <linux/elf.h>
 #include <asm/setup.h>
+#include <asm/kexec.h>
 #include <asm/sclp.h>
 #include <asm/uv.h>
 #include "compressed/decompressor.h"
@@ -47,6 +49,29 @@ static void copy_bootdata(void)
 	memcpy((void *)vmlinux.bootdata_preserved_off, __boot_data_preserved_start, vmlinux.bootdata_preserved_size);
 }
 
+static void handle_relocs(unsigned long offset)
+{
+	Elf64_Rela *rela_start, *rela_end, *rela;
+	int r_type, r_sym, rc;
+	Elf64_Addr loc, val;
+	Elf64_Sym *dynsym;
+
+	rela_start = (Elf64_Rela *) vmlinux.rela_dyn_start;
+	rela_end = (Elf64_Rela *) vmlinux.rela_dyn_end;
+	dynsym = (Elf64_Sym *) vmlinux.dynsym_start;
+	for (rela = rela_start; rela < rela_end; rela++) {
+		loc = rela->r_offset + offset;
+		val = rela->r_addend + offset;
+		r_sym = ELF64_R_SYM(rela->r_info);
+		if (r_sym)
+			val += dynsym[r_sym].st_value;
+		r_type = ELF64_R_TYPE(rela->r_info);
+		rc = arch_kexec_do_relocs(r_type, (void *) loc, val, 0);
+		if (rc)
+			error("Unknown relocation type");
+	}
+}
+
 void startup_kernel(void)
 {
 	unsigned long safe_addr;
@@ -67,5 +92,7 @@ void startup_kernel(void)
 		memmove((void *)vmlinux.default_lma, img, vmlinux.image_size);
 	}
 	copy_bootdata();
+	if (IS_ENABLED(CONFIG_RELOCATABLE))
+		handle_relocs(0);
 	vmlinux.entry();
 }

arch/s390/include/asm/kexec.h

@@ -71,6 +71,8 @@ int s390_verify_sig(const char *kernel, unsigned long kernel_len);
 void *kexec_file_add_components(struct kimage *image,
 				int (*add_kernel)(struct kimage *image,
 						  struct s390_load_data *data));
+int arch_kexec_do_relocs(int r_type, void *loc, unsigned long val,
+			 unsigned long addr);
 
 extern const struct kexec_file_ops s390_kexec_image_ops;
 extern const struct kexec_file_ops s390_kexec_elf_ops;

arch/s390/kernel/Makefile

@@ -51,7 +51,7 @@ obj-y	+= debug.o irq.o ipl.o dis.o diag.o vdso.o early_nobss.o
 obj-y	+= sysinfo.o lgr.o os_info.o machine_kexec.o pgm_check.o
 obj-y	+= runtime_instr.o cache.o fpu.o dumpstack.o guarded_storage.o sthyi.o
 obj-y	+= entry.o reipl.o relocate_kernel.o kdebugfs.o alternative.o
-obj-y	+= nospec-branch.o ipl_vmparm.o
+obj-y	+= nospec-branch.o ipl_vmparm.o machine_kexec_reloc.o
 
 extra-y				+= head64.o vmlinux.lds
 

arch/s390/kernel/machine_kexec_file.c

@@ -290,7 +290,7 @@ int arch_kexec_apply_relocations_add(struct purgatory_info *pi,
 				     const Elf_Shdr *symtab)
 {
 	Elf_Rela *relas;
-	int i;
+	int i, r_type;
 
 	relas = (void *)pi->ehdr + relsec->sh_offset;
 
@@ -324,46 +324,8 @@ int arch_kexec_apply_relocations_add(struct purgatory_info *pi,
 
 		addr = section->sh_addr + relas[i].r_offset;
 
-		switch (ELF64_R_TYPE(relas[i].r_info)) {
-		case R_390_8:		/* Direct 8 bit.   */
-			*(u8 *)loc = val;
-			break;
-		case R_390_12:		/* Direct 12 bit.  */
-			*(u16 *)loc &= 0xf000;
-			*(u16 *)loc |= val & 0xfff;
-			break;
-		case R_390_16:		/* Direct 16 bit.  */
-			*(u16 *)loc = val;
-			break;
-		case R_390_20:		/* Direct 20 bit.  */
-			*(u32 *)loc &= 0xf00000ff;
-			*(u32 *)loc |= (val & 0xfff) << 16;	/* DL */
-			*(u32 *)loc |= (val & 0xff000) >> 4;	/* DH */
-			break;
-		case R_390_32:		/* Direct 32 bit.  */
-			*(u32 *)loc = val;
-			break;
-		case R_390_64:		/* Direct 64 bit.  */
-			*(u64 *)loc = val;
-			break;
-		case R_390_PC16:	/* PC relative 16 bit.	*/
-			*(u16 *)loc = (val - addr);
-			break;
-		case R_390_PC16DBL:	/* PC relative 16 bit shifted by 1.  */
-			*(u16 *)loc = (val - addr) >> 1;
-			break;
-		case R_390_PC32DBL:	/* PC relative 32 bit shifted by 1.  */
-			*(u32 *)loc = (val - addr) >> 1;
-			break;
-		case R_390_PC32:	/* PC relative 32 bit.	*/
-			*(u32 *)loc = (val - addr);
-			break;
-		case R_390_PC64:	/* PC relative 64 bit.	*/
-			*(u64 *)loc = (val - addr);
-			break;
-		default:
-			break;
-		}
+		r_type = ELF64_R_TYPE(relas[i].r_info);
+		arch_kexec_do_relocs(r_type, loc, val, addr);
 	}
 	return 0;
 }

arch/s390/kernel/machine_kexec_reloc.c

@@ -0,0 +1,53 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/elf.h>
+
+int arch_kexec_do_relocs(int r_type, void *loc, unsigned long val,
+			 unsigned long addr)
+{
+	switch (r_type) {
+	case R_390_NONE:
+		break;
+	case R_390_8:		/* Direct 8 bit.   */
+		*(u8 *)loc = val;
+		break;
+	case R_390_12:		/* Direct 12 bit.  */
+		*(u16 *)loc &= 0xf000;
+		*(u16 *)loc |= val & 0xfff;
+		break;
+	case R_390_16:		/* Direct 16 bit.  */
+		*(u16 *)loc = val;
+		break;
+	case R_390_20:		/* Direct 20 bit.  */
+		*(u32 *)loc &= 0xf00000ff;
+		*(u32 *)loc |= (val & 0xfff) << 16;	/* DL */
+		*(u32 *)loc |= (val & 0xff000) >> 4;	/* DH */
+		break;
+	case R_390_32:		/* Direct 32 bit.  */
+		*(u32 *)loc = val;
+		break;
+	case R_390_64:		/* Direct 64 bit.  */
+		*(u64 *)loc = val;
+		break;
+	case R_390_PC16:	/* PC relative 16 bit.	*/
+		*(u16 *)loc = (val - addr);
+		break;
+	case R_390_PC16DBL:	/* PC relative 16 bit shifted by 1.  */
+		*(u16 *)loc = (val - addr) >> 1;
+		break;
+	case R_390_PC32DBL:	/* PC relative 32 bit shifted by 1.  */
+		*(u32 *)loc = (val - addr) >> 1;
+		break;
+	case R_390_PC32:	/* PC relative 32 bit.	*/
+		*(u32 *)loc = (val - addr);
+		break;
+	case R_390_PC64:	/* PC relative 64 bit.	*/
+		*(u64 *)loc = (val - addr);
+		break;
+	case R_390_RELATIVE:
+		*(unsigned long *) loc = val;
+		break;
+	default:
+		return 1;
+	}
+	return 0;
+}

arch/s390/kernel/vmlinux.lds.S

@@ -144,6 +144,18 @@ SECTIONS
 	INIT_DATA_SECTION(0x100)
 
 	PERCPU_SECTION(0x100)
+
+	.dynsym ALIGN(8) : {
+		__dynsym_start = .;
+		*(.dynsym)
+		__dynsym_end = .;
+	}
+	.rela.dyn ALIGN(8) : {
+		__rela_dyn_start = .;
+		*(.rela*)
+		__rela_dyn_end = .;
+	}
+
 	. = ALIGN(PAGE_SIZE);
 	__init_end = .;		/* freed after init ends here */
 
@@ -165,6 +177,9 @@ SECTIONS
 		QUAD(__boot_data_preserved_start)		/* bootdata_preserved_off */
 		QUAD(__boot_data_preserved_end -
 		     __boot_data_preserved_start)		/* bootdata_preserved_size */
+		QUAD(__dynsym_start)				/* dynsym_start */
+		QUAD(__rela_dyn_start)				/* rela_dyn_start */
+		QUAD(__rela_dyn_end)				/* rela_dyn_end */
 	} :NONE
 
 	/* Debugging sections.	*/