Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755173Ab3HaAfM (ORCPT ); Fri, 30 Aug 2013 20:35:12 -0400 Received: from mail-vb0-f51.google.com ([209.85.212.51]:43719 "EHLO mail-vb0-f51.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752523Ab3HaAfK (ORCPT ); Fri, 30 Aug 2013 20:35:10 -0400 MIME-Version: 1.0 In-Reply-To: <20130830143631.9FC313E102A@localhost> References: <1376090777-20090-1-git-send-email-roy.franz@linaro.org> <1376090777-20090-16-git-send-email-roy.franz@linaro.org> <20130830143631.9FC313E102A@localhost> Date: Fri, 30 Aug 2013 17:35:09 -0700 Message-ID: Subject: Re: [PATCH 15/16] Add EFI stub for ARM From: Roy Franz To: Grant Likely Cc: Linux Kernel Mailing List , linux-efi@vger.kernel.org, "linux-arm-kernel@lists.infradead.org" , matt.fleming@intel.com, Russell King - ARM Linux , Leif Lindholm , Dave Martin , Mark Salter Content-Type: text/plain; charset=UTF-8 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 34410 Lines: 854 On Fri, Aug 30, 2013 at 7:36 AM, Grant Likely wrote: > On Fri, 9 Aug 2013 16:26:16 -0700, Roy Franz wrote: >> This patch adds EFI stub support for the ARM Linux kernel. The EFI stub >> operations similarly to the x86 stub: it is a shim between the EFI firmware >> and the normal zImage entry point, and sets up the environment that the >> zImage is expecting. This includes loading the initrd (optionaly) and >> device tree from the system partition based on the kernel command line. >> The stub updates the device tree as necessary, including adding reserved >> memory regions and adding entries for EFI runtime services. The PE/COFF >> "MZ" header at offset 0 results in the first instruction being an add >> that corrupts r5, which is not used by the zImage interface. >> >> Signed-off-by: Roy Franz > > Hi Roy, > > Looks like nice tight code. I've got some comments below, but in general > I'm pretty happy with it. > > g. > >> --- >> arch/arm/boot/compressed/Makefile | 15 +- >> arch/arm/boot/compressed/efi-header.S | 111 ++++++++ >> arch/arm/boot/compressed/efi-stub.c | 448 +++++++++++++++++++++++++++++++++ >> arch/arm/boot/compressed/efi-stub.h | 5 + >> arch/arm/boot/compressed/head.S | 90 ++++++- >> 5 files changed, 661 insertions(+), 8 deletions(-) >> create mode 100644 arch/arm/boot/compressed/efi-header.S >> create mode 100644 arch/arm/boot/compressed/efi-stub.c >> create mode 100644 arch/arm/boot/compressed/efi-stub.h >> >> diff --git a/arch/arm/boot/compressed/Makefile b/arch/arm/boot/compressed/Makefile >> index 7ac1610..5fad8bd 100644 >> --- a/arch/arm/boot/compressed/Makefile >> +++ b/arch/arm/boot/compressed/Makefile >> @@ -103,11 +103,22 @@ libfdt_objs := $(addsuffix .o, $(basename $(libfdt))) >> $(addprefix $(obj)/,$(libfdt) $(libfdt_hdrs)): $(obj)/%: $(srctree)/scripts/dtc/libfdt/% >> $(call cmd,shipped) >> >> -$(addprefix $(obj)/,$(libfdt_objs) atags_to_fdt.o): \ >> +$(addprefix $(obj)/,$(libfdt_objs) atags_to_fdt.o efi-stub.o): \ >> $(addprefix $(obj)/,$(libfdt_hdrs)) >> >> ifeq ($(CONFIG_ARM_ATAG_DTB_COMPAT),y) >> -OBJS += $(libfdt_objs) atags_to_fdt.o >> +OBJS += atags_to_fdt.o >> +USE_LIBFDT = y >> +endif >> + >> +ifeq ($(CONFIG_EFI_STUB),y) >> +CFLAGS_efi-stub.o += -DTEXT_OFFSET=$(TEXT_OFFSET) >> +OBJS += efi-stub.o >> +USE_LIBFDT = y >> +endif >> + >> +ifeq ($(USE_LIBFDT),y) >> +OBJS += $(libfdt_objs) >> endif >> >> targets := vmlinux vmlinux.lds \ >> diff --git a/arch/arm/boot/compressed/efi-header.S b/arch/arm/boot/compressed/efi-header.S >> new file mode 100644 >> index 0000000..6965e0f >> --- /dev/null >> +++ b/arch/arm/boot/compressed/efi-header.S >> @@ -0,0 +1,111 @@ >> +@ Copyright (C) 2013 Linaro Ltd; >> +@ >> +@ This file contains the PE/COFF header that is part of the >> +@ EFI stub. >> +@ >> + >> + .org 0x3c >> + @ >> + @ The PE header can be anywhere in the file, but for >> + @ simplicity we keep it together with the MSDOS header >> + @ The offset to the PE/COFF header needs to be at offset >> + @ 0x3C in the MSDOS header. >> + @ The only 2 fields of the MSDOS header that are used are this >> + @ PE/COFF offset, and the "MZ" bytes at offset 0x0. >> + @ >> + .long pe_header @ Offset to the PE header. >> + >> + .align 3 >> +pe_header: >> + .ascii "PE" >> + .short 0 >> + >> +coff_header: >> + .short 0x01c2 @ ARM or Thumb >> + .short 2 @ nr_sections >> + .long 0 @ TimeDateStamp >> + .long 0 @ PointerToSymbolTable >> + .long 1 @ NumberOfSymbols >> + .short section_table - optional_header @ SizeOfOptionalHeader >> + .short 0x306 @ Characteristics. >> + @ IMAGE_FILE_32BIT_MACHINE | >> + @ IMAGE_FILE_DEBUG_STRIPPED | >> + @ IMAGE_FILE_EXECUTABLE_IMAGE | >> + @ IMAGE_FILE_LINE_NUMS_STRIPPED >> + >> +optional_header: >> + .short 0x10b @ PE32 format >> + .byte 0x02 @ MajorLinkerVersion >> + .byte 0x14 @ MinorLinkerVersion >> + >> + .long _edata - efi_stub_entry @ SizeOfCode >> + >> + .long 0 @ SizeOfInitializedData >> + .long 0 @ SizeOfUninitializedData >> + >> + .long efi_stub_entry @ AddressOfEntryPoint >> + .long efi_stub_entry @ BaseOfCode >> + .long 0 @ data >> + >> +extra_header_fields: >> + .long 0 @ ImageBase >> + .long 0x20 @ SectionAlignment >> + .long 0x20 @ FileAlignment >> + .short 0 @ MajorOperatingSystemVersion >> + .short 0 @ MinorOperatingSystemVersion >> + .short 0 @ MajorImageVersion >> + .short 0 @ MinorImageVersion >> + .short 0 @ MajorSubsystemVersion >> + .short 0 @ MinorSubsystemVersion >> + .long 0 @ Win32VersionValue >> + >> + .long _edata @ SizeOfImage >> + >> + @ Everything before the entry point is considered part of the header >> + .long efi_stub_entry @ SizeOfHeaders >> + .long 0 @ CheckSum >> + .short 0xa @ Subsystem (EFI application) >> + .short 0 @ DllCharacteristics >> + .long 0 @ SizeOfStackReserve >> + .long 0 @ SizeOfStackCommit >> + .long 0 @ SizeOfHeapReserve >> + .long 0 @ SizeOfHeapCommit >> + .long 0 @ LoaderFlags >> + .long 0x0 @ NumberOfRvaAndSizes >> + >> + # Section table >> +section_table: >> + >> + # >> + # The EFI application loader requires a relocation section >> + # because EFI applications must be relocatable. This is a >> + # dummy section as far as we are concerned. >> + # >> + .ascii ".reloc" >> + .byte 0 >> + .byte 0 @ end of 0 padding of section name >> + .long 0 >> + .long 0 >> + .long 0 @ SizeOfRawData >> + .long 0 @ PointerToRawData >> + .long 0 @ PointerToRelocations >> + .long 0 @ PointerToLineNumbers >> + .short 0 @ NumberOfRelocations >> + .short 0 @ NumberOfLineNumbers >> + .long 0x42100040 @ Characteristics (section flags) >> + >> + >> + .ascii ".text" >> + .byte 0 >> + .byte 0 >> + .byte 0 @ end of 0 padding of section name >> + .long _edata - efi_stub_entry @ VirtualSize >> + .long efi_stub_entry @ VirtualAddress >> + .long _edata - efi_stub_entry @ SizeOfRawData >> + .long efi_stub_entry @ PointerToRawData >> + >> + .long 0 @ PointerToRelocations (0 for executables) >> + .long 0 @ PointerToLineNumbers (0 for executables) >> + .short 0 @ NumberOfRelocations (0 for executables) >> + .short 0 @ NumberOfLineNumbers (0 for executables) >> + .long 0xe0500020 @ Characteristics (section flags) >> diff --git a/arch/arm/boot/compressed/efi-stub.c b/arch/arm/boot/compressed/efi-stub.c >> new file mode 100644 >> index 0000000..4fce68b >> --- /dev/null >> +++ b/arch/arm/boot/compressed/efi-stub.c >> @@ -0,0 +1,448 @@ >> +/* >> + * linux/arch/arm/boot/compressed/efi-stub.c >> + * >> + * Copyright (C) 2013 Linaro Ltd; >> + * >> + * This file implements the EFI boot stub for the ARM kernel >> + * >> + * This program is free software; you can redistribute it and/or modify >> + * it under the terms of the GNU General Public License version 2 as >> + * published by the Free Software Foundation. >> + * >> + */ >> +#include >> +#include >> +#include "efi-stub.h" >> + >> +/* EFI function call wrappers. These are not required for >> + * ARM, but wrappers are required for X86 to convert between >> + * ABIs. These wrappers are provided to allow code sharing >> + * between X86 and ARM. Since these wrappers directly invoke the >> + * EFI function pointer, the function pointer type must be properly >> + * defined, which is not the case for X86 One advantage of this is >> + * it allows for type checking of arguments, which is not >> + * possible with the X86 wrappers. >> + */ >> +#define efi_call_phys0(f) f() >> +#define efi_call_phys1(f, a1) f(a1) >> +#define efi_call_phys2(f, a1, a2) f(a1, a2) >> +#define efi_call_phys3(f, a1, a2, a3) f(a1, a2, a3) >> +#define efi_call_phys4(f, a1, a2, a3, a4) f(a1, a2, a3, a4) >> +#define efi_call_phys5(f, a1, a2, a3, a4, a5) f(a1, a2, a3, a4, a5) >> + >> +/* The maximum uncompressed kernel size is 32 MBytes, so we will reserve >> + * that for the decompressed kernel. We have no easy way to tell what >> + * the actuall size of code + data the uncompressed kernel will use. >> + */ >> +#define MAX_UNCOMP_KERNEL_SIZE 0x02000000 >> + >> +/* The kernel zImage should be located between 32 Mbytes >> + * and 128 MBytes from the base of DRAM. The min >> + * address leaves space for a maximal size uncompressed image, >> + * and the max address is due to how the zImage decompressor >> + * picks a destination address. >> + */ >> +#define ZIMAGE_OFFSET_LIMIT 0x08000000 >> +#define MIN_ZIMAGE_OFFSET MAX_UNCOMP_KERNEL_SIZE >> + >> +#define PRINTK_PREFIX "EFIstub: " >> + >> +struct fdt_region { >> + u64 base; >> + u64 size; >> +}; >> + >> + >> +/* Include shared EFI stub code */ >> +#include "../../../../drivers/firmware/efi/efi-stub-helper.c" >> + >> +static int relocate_kernel(efi_system_table_t *sys_table, >> + unsigned long *zimage_addr, >> + unsigned long zimage_size, >> + unsigned long min_addr, unsigned long max_addr) >> +{ >> + /* Get current address of kernel. */ >> + unsigned long cur_zimage_addr = *zimage_addr; >> + unsigned long new_addr = 0; >> + >> + efi_status_t status; >> + >> + if (!zimage_addr || !zimage_size) >> + return EFI_INVALID_PARAMETER; >> + >> + if (cur_zimage_addr > min_addr >> + && (cur_zimage_addr + zimage_size) < max_addr) { >> + /* We don't need to do anything, as kernel is at an >> + * acceptable address already. >> + */ >> + return EFI_SUCCESS; >> + } >> + /* >> + * The EFI firmware loader could have placed the kernel image >> + * anywhere in memory, but the kernel has restrictions on the >> + * min and max physical address it can run at. >> + */ >> + status = efi_low_alloc(sys_table, zimage_size, 0, >> + &new_addr, min_addr); >> + if (status != EFI_SUCCESS) { >> + efi_printk(sys_table, PRINTK_PREFIX"ERROR: Failed to allocate usable memory for kernel.\n"); >> + return status; >> + } >> + >> + if (new_addr > (max_addr - zimage_size)) { >> + efi_free(sys_table, zimage_size, new_addr); >> + efi_printk(sys_table, PRINTK_PREFIX"ERROR: Failed to allocate usable memory for kernel.\n"); >> + return EFI_INVALID_PARAMETER; >> + } >> + >> + /* We know source/dest won't overlap since both memory ranges >> + * have been allocated by UEFI, so we can safely use memcpy. >> + */ >> + memcpy((void *)new_addr, (void *)(unsigned long)cur_zimage_addr, >> + zimage_size); >> + >> + /* Return the load address */ >> + *zimage_addr = new_addr; >> + >> + return status; >> +} > > This function should be sharable with x86. ARM has more restrictions > that x86, but I don't see any reason for them to be separate. ARM64 is > certainly going to want to use some form of this too. I'll look at sharing this with x86. x86 has the additional concept of a preferred address but other than that looks similar. > >> + >> + >> +/* Convert the unicode UEFI command line to ASCII to pass to kernel. >> + * Size of memory allocated return in *cmd_line_len. >> + * Returns NULL on error. >> + */ >> +static char *convert_cmdline_to_ascii(efi_system_table_t *sys_table, >> + efi_loaded_image_t *image, >> + unsigned long *cmd_line_len, >> + u32 max_addr) > > x86 has equivalent code. This function should be factored out and used > by both. I'll update x86 to use this. > >> +{ >> + u16 *s2; >> + u8 *s1 = NULL; >> + unsigned long cmdline_addr = 0; >> + int load_options_size = image->load_options_size / 2; /* ASCII */ >> + void *options = (u16 *)image->load_options; > > load_options is already a void*. What is the cast here for? Removed. > >> + int options_size = 0; >> + int status; >> + int i; >> + u16 zero = 0; >> + >> + if (options) { >> + s2 = options; >> + while (*s2 && *s2 != '\n' && options_size < load_options_size) { >> + s2++; >> + options_size++; >> + } >> + } >> + >> + if (options_size == 0) { >> + /* No command line options, so return empty string*/ >> + options_size = 1; >> + options = &zero; >> + } >> + >> + options_size++; /* NUL termination */ >> + >> + status = efi_high_alloc(sys_table, options_size, 0, >> + &cmdline_addr, max_addr); >> + if (status != EFI_SUCCESS) >> + return NULL; >> + >> + s1 = (u8 *)(unsigned long)cmdline_addr; >> + s2 = (u16 *)options; >> + >> + for (i = 0; i < options_size - 1; i++) >> + *s1++ = *s2++; >> + >> + *s1 = '\0'; >> + >> + *cmd_line_len = options_size; >> + return (char *)(unsigned long)cmdline_addr; > > Double casting? That looks wrong. Why does the (unsigned long) bit need > to be there? Removed. likely a holdover from a version where that was a 64 bit type. (see more on this below...) > >> +} >> + >> + >> +static u32 update_fdt(efi_system_table_t *sys_table, void *orig_fdt, void *fdt, >> + int new_fdt_size, char *cmdline_ptr, u64 initrd_addr, >> + u64 initrd_size, efi_memory_desc_t *memory_map, >> + int map_size, int desc_size) > > ARM64 will want access to this function. I will move this to be shared. I have discussed with Leif making the EFI FDT entries used by EFI runtime services 64 bit for both ARM32/ARM64, so this function should be identical in both cases. > >> +{ >> + int node; >> + int status; >> + unsigned long fdt_val; >> + >> + status = fdt_open_into(orig_fdt, fdt, new_fdt_size); >> + if (status != 0) >> + goto fdt_set_fail; >> + >> + node = fdt_subnode_offset(fdt, 0, "chosen"); >> + if (node < 0) { >> + node = fdt_add_subnode(fdt, 0, "chosen"); >> + if (node < 0) { >> + status = node; /* node is error code when negative */ >> + goto fdt_set_fail; >> + } >> + } >> + >> + if ((cmdline_ptr != NULL) && (strlen(cmdline_ptr) > 0)) { >> + status = fdt_setprop(fdt, node, "bootargs", cmdline_ptr, >> + strlen(cmdline_ptr) + 1); >> + if (status) >> + goto fdt_set_fail; >> + } > > Some comments in this function would help poor readers like me. I'll review and improve the commenting for this this function > >> + >> + /* Set intird address/end in device tree, if present */ >> + if (initrd_size != 0) { >> + u64 initrd_image_end; >> + u64 initrd_image_start = cpu_to_fdt64(initrd_addr); >> + status = fdt_setprop(fdt, node, "linux,initrd-start", >> + &initrd_image_start, sizeof(u64)); >> + if (status) >> + goto fdt_set_fail; >> + initrd_image_end = cpu_to_fdt64(initrd_addr + initrd_size); >> + status = fdt_setprop(fdt, node, "linux,initrd-end", >> + &initrd_image_end, sizeof(u64)); >> + if (status) >> + goto fdt_set_fail; >> + } >> + >> + /* Add FDT entries for EFI runtime services in chosen node. */ >> + node = fdt_subnode_offset(fdt, 0, "chosen"); >> + fdt_val = cpu_to_fdt32((unsigned long)sys_table); >> + status = fdt_setprop(fdt, node, "efi-system-table", >> + &fdt_val, sizeof(fdt_val)); >> + if (status) >> + goto fdt_set_fail; >> + >> + fdt_val = cpu_to_fdt32(desc_size); >> + status = fdt_setprop(fdt, node, "efi-mmap-desc-size", >> + &fdt_val, sizeof(fdt_val)); >> + if (status) >> + goto fdt_set_fail; >> + >> + fdt_val = cpu_to_fdt32(map_size); >> + status = fdt_setprop(fdt, node, "efi-runtime-mmap-size", >> + &fdt_val, sizeof(fdt_val)); >> + if (status) >> + goto fdt_set_fail; >> + >> + fdt_val = cpu_to_fdt32((unsigned long)memory_map); >> + status = fdt_setprop(fdt, node, "efi-runtime-mmap", >> + &fdt_val, sizeof(fdt_val)); >> + if (status) >> + goto fdt_set_fail; >> + >> + return EFI_SUCCESS; >> + >> +fdt_set_fail: >> + if (status == -FDT_ERR_NOSPACE) >> + return EFI_BUFFER_TOO_SMALL; >> + >> + return EFI_LOAD_ERROR; >> +} >> + >> + >> + >> +int efi_entry(void *handle, efi_system_table_t *sys_table, >> + unsigned long *zimage_addr) >> +{ >> + efi_loaded_image_t *image; >> + int status; >> + unsigned long nr_pages; >> + const struct fdt_region *region; >> + >> + void *fdt; >> + int err; >> + int node; >> + unsigned long zimage_size = 0; >> + unsigned long dram_base; >> + /* addr/point and size pairs for memory management*/ >> + u64 initrd_addr; >> + u64 initrd_size = 0; >> + u64 fdt_addr; >> + u64 fdt_size = 0; >> + u64 kernel_reserve_addr; >> + u64 kernel_reserve_size = 0; >> + char *cmdline_ptr; >> + unsigned long cmdline_size = 0; >> + >> + unsigned long map_size, desc_size; >> + unsigned long mmap_key; >> + efi_memory_desc_t *memory_map; >> + >> + unsigned long new_fdt_size; >> + unsigned long new_fdt_addr; >> + >> + efi_guid_t proto = LOADED_IMAGE_PROTOCOL_GUID; >> + >> + /* Check if we were booted by the EFI firmware */ >> + if (sys_table->hdr.signature != EFI_SYSTEM_TABLE_SIGNATURE) >> + goto fail; >> + >> + efi_printk(sys_table, PRINTK_PREFIX"Booting Linux using EFI stub.\n"); >> + >> + >> + /* get the command line from EFI, using the LOADED_IMAGE protocol */ >> + status = efi_call_phys3(sys_table->boottime->handle_protocol, >> + handle, &proto, (void *)&image); >> + if (status != EFI_SUCCESS) { >> + efi_printk(sys_table, PRINTK_PREFIX"ERROR: Failed to get handle for LOADED_IMAGE_PROTOCOL\n"); >> + goto fail; >> + } >> + >> + /* We are going to copy this into device tree, so we don't care where in >> + * memory it is. >> + */ >> + cmdline_ptr = convert_cmdline_to_ascii(sys_table, image, >> + &cmdline_size, 0xFFFFFFFF); >> + if (!cmdline_ptr) { >> + efi_printk(sys_table, PRINTK_PREFIX"ERROR: converting command line to ascii failed.\n"); >> + goto fail; >> + } >> + >> + /* We first load the device tree, as we need to get the base address of >> + * DRAM from the device tree. The zImage, device tree, and initrd >> + * have address restrictions that are relative to the base of DRAM. >> + */ >> + status = handle_cmdline_files(sys_table, image, cmdline_ptr, "dtb=", >> + 0xffffffff, &fdt_addr, &fdt_size); >> + if (status != EFI_SUCCESS) { >> + efi_printk(sys_table, PRINTK_PREFIX"ERROR: Unable to load device tree blob.\n"); >> + goto fail_free_cmdline; >> + } >> + >> + err = fdt_check_header((void *)(unsigned long)fdt_addr); >> + if (err != 0) { >> + efi_printk(sys_table, PRINTK_PREFIX"ERROR: device tree header not valid\n"); >> + goto fail_free_fdt; >> + } >> + if (fdt_totalsize((void *)(unsigned long)fdt_addr) > fdt_size) { >> + efi_printk(sys_table, PRINTK_PREFIX"ERROR: Incomplete device tree.\n"); >> + goto fail_free_fdt; >> + >> + } >> + >> + >> + /* Look up the base of DRAM from the device tree.*/ >> + fdt = (void *)(u32)fdt_addr; > > More double casting. This one might be legitimate, but you need to > describe why. Is there any possibility that the FDT would appear above > 4G? It can't be above 4GiB. This highlights some inconsistencies in the shared APIs that I need to clean up - some that I added use u64 for passing addresses, which lead to casts like this. I will change the new functions to match the existing alloc/free to use native sizes, as this works for the arm32 and x86 32 bit architectures as well as the 64 bit ones. (The alternative is to move all the APIs to u64/efi_phys_addr_t but this would be more intrusive to the existing x86 code.) > >> + node = fdt_subnode_offset(fdt, 0, "memory"); >> + region = fdt_getprop(fdt, node, "reg", NULL); >> + if (region) { >> + dram_base = fdt64_to_cpu(region->base); >> + } else { >> + efi_printk(sys_table, PRINTK_PREFIX"ERROR: no 'memory' node in device tree.\n"); >> + goto fail_free_fdt; > > Shouldn't fail here. If there is no memory node then create one. It would be good to do that, however I don't know of a way to get this information using boot services. This is available from within the UEFI firmware, but not exposed to EFI applications as far as I can tell. I think that the way to fix this is to have the UEFI firmware 'own' the device tree and make it available to the stub, after filling in platform specific stuff like this. I think this goes along the lines of you have suggested regarding adding a FDT protocol to UEFI. > >> + } >> + >> + /* Reserve memory for the uncompressed kernel image. */ >> + kernel_reserve_addr = dram_base; >> + kernel_reserve_size = MAX_UNCOMP_KERNEL_SIZE; >> + nr_pages = round_up(kernel_reserve_size, EFI_PAGE_SIZE) / EFI_PAGE_SIZE; >> + status = efi_call_phys4(sys_table->boottime->allocate_pages, >> + EFI_ALLOCATE_ADDRESS, EFI_LOADER_DATA, >> + nr_pages, &kernel_reserve_addr); >> + if (status != EFI_SUCCESS) { >> + efi_printk(sys_table, PRINTK_PREFIX"ERROR: unable to allocate memory for uncompressed kernel.\n"); >> + goto fail_free_fdt; >> + } >> + >> + /* Relocate the zImage, if required. */ >> + zimage_size = image->image_size; >> + status = relocate_kernel(sys_table, zimage_addr, zimage_size, >> + dram_base + MIN_ZIMAGE_OFFSET, >> + dram_base + ZIMAGE_OFFSET_LIMIT); >> + if (status != EFI_SUCCESS) { >> + efi_printk(sys_table, PRINTK_PREFIX"ERROR: Failed to relocate kernel\n"); >> + goto fail_free_kernel_reserve; >> + } >> + >> + status = handle_cmdline_files(sys_table, image, cmdline_ptr, "initrd=", >> + dram_base + ZIMAGE_OFFSET_LIMIT, >> + &initrd_addr, &initrd_size); >> + if (status != EFI_SUCCESS) { >> + efi_printk(sys_table, PRINTK_PREFIX"ERROR: Unable to load initrd\n"); >> + goto fail_free_zimage; >> + } >> + >> + /* Estimate size of new FDT, and allocate memory for it. We >> + * will allocate a bigger buffer if this ends up being too >> + * small, so a rough guess is OK here.*/ >> + new_fdt_size = fdt_size + cmdline_size + 0x200; >> + >> +fdt_alloc_retry: >> + status = efi_high_alloc(sys_table, new_fdt_size, 0, &new_fdt_addr, >> + dram_base + ZIMAGE_OFFSET_LIMIT); >> + if (status != EFI_SUCCESS) { >> + efi_printk(sys_table, PRINTK_PREFIX"ERROR: Unable to allocate memory for new device tree.\n"); >> + goto fail_free_initrd; >> + } >> + >> + /* Now that we have done our final memory allocation (and free) >> + * we can get the memory map key needed >> + * forexit_boot_services.*/ >> + status = efi_get_memory_map(sys_table, &memory_map, &map_size, >> + &desc_size, &mmap_key); >> + if (status != EFI_SUCCESS) >> + goto fail_free_new_fdt; >> + >> + status = update_fdt(sys_table, >> + fdt, (void *)new_fdt_addr, new_fdt_size, >> + cmdline_ptr, >> + initrd_addr, initrd_size, >> + memory_map, map_size, desc_size); >> + >> + if (status != EFI_SUCCESS) { >> + if (status == EFI_BUFFER_TOO_SMALL) { >> + /* We need to allocate more space for the new >> + * device tree, so free existing buffer that is >> + * too small. Also free memory map, as we will need >> + * to get new one that reflects the free/alloc we do >> + * on the device tree buffer. */ >> + efi_free(sys_table, new_fdt_size, new_fdt_addr); >> + efi_call_phys1(sys_table->boottime->free_pool, >> + memory_map); >> + new_fdt_size += new_fdt_size/4; >> + goto fdt_alloc_retry; >> + } >> + efi_printk(sys_table, PRINTK_PREFIX"ERROR: Unable to constuct new device tree.\n"); >> + goto fail_free_initrd; >> + } >> + >> + /* Now we are ready to exit_boot_services.*/ >> + status = efi_call_phys2(sys_table->boottime->exit_boot_services, >> + handle, mmap_key); >> + >> + if (status != EFI_SUCCESS) { >> + efi_printk(sys_table, PRINTK_PREFIX"ERROR: exit boot services failed.\n"); >> + goto fail_free_mmap; >> + } >> + >> + >> + /* Now we need to return the FDT address to the calling >> + * assembly to this can be used as part of normal boot. >> + */ >> + return new_fdt_addr; >> + >> +fail_free_mmap: >> + efi_call_phys1(sys_table->boottime->free_pool, memory_map); >> + >> +fail_free_new_fdt: >> + efi_free(sys_table, new_fdt_size, new_fdt_addr); >> + >> +fail_free_initrd: >> + efi_free(sys_table, initrd_size, initrd_addr); >> + >> +fail_free_zimage: >> + efi_free(sys_table, zimage_size, *zimage_addr); >> + >> +fail_free_kernel_reserve: >> + efi_free(sys_table, kernel_reserve_addr, kernel_reserve_size); >> + >> +fail_free_fdt: >> + efi_free(sys_table, fdt_size, fdt_addr); >> + >> +fail_free_cmdline: >> + efi_free(sys_table, cmdline_size, (u32)cmdline_ptr); >> + >> +fail: >> + return EFI_STUB_ERROR; >> +} >> diff --git a/arch/arm/boot/compressed/efi-stub.h b/arch/arm/boot/compressed/efi-stub.h >> new file mode 100644 >> index 0000000..0fe9376 >> --- /dev/null >> +++ b/arch/arm/boot/compressed/efi-stub.h >> @@ -0,0 +1,5 @@ >> +#ifndef _ARM_EFI_STUB_H >> +#define _ARM_EFI_STUB_H >> +/* Error code returned to ASM code instead of valid FDT address. */ >> +#define EFI_STUB_ERROR (~0) >> +#endif >> diff --git a/arch/arm/boot/compressed/head.S b/arch/arm/boot/compressed/head.S >> index 75189f1..5401a3a 100644 >> --- a/arch/arm/boot/compressed/head.S >> +++ b/arch/arm/boot/compressed/head.S >> @@ -10,6 +10,7 @@ >> */ >> #include >> #include >> +#include "efi-stub.h" >> >> .arch armv7-a >> /* >> @@ -120,21 +121,99 @@ >> */ >> .align >> .arm @ Always enter in ARM state >> + .text >> start: >> .type start,#function >> - .rept 7 >> +#ifdef CONFIG_EFI_STUB >> + @ Magic MSDOS signature for PE/COFF + ADD opcode >> + .word 0x62805a4d >> +#else >> + mov r0, r0 >> +#endif >> + .rept 5 >> mov r0, r0 >> .endr >> - ARM( mov r0, r0 ) >> - ARM( b 1f ) >> - THUMB( adr r12, BSYM(1f) ) >> - THUMB( bx r12 ) >> + >> + adrl r12, BSYM(zimage_continue) >> + ARM( mov pc, r12 ) >> + THUMB( bx r12 ) >> + @ zimage_continue will be in ARM or thumb mode as configured >> >> .word 0x016f2818 @ Magic numbers to help the loader >> .word start @ absolute load/run zImage address >> .word _edata @ zImage end address >> + >> +#ifdef CONFIG_EFI_STUB >> + @ Portions of the MSDOS file header must be at offset >> + @ 0x3c from the start of the file. All PE/COFF headers >> + @ are kept contiguous for simplicity. >> +#include "efi-header.S" >> + >> +efi_stub_entry: >> + @ The EFI stub entry point is not at a fixed address, however >> + @ this address must be set in the PE/COFF header. >> + @ EFI entry point is in A32 mode, switch to T32 if configured. >> + THUMB( adr r12, BSYM(1f) ) >> + THUMB( bx r12 ) >> THUMB( .thumb ) >> 1: >> + @ Save lr on stack for possible return to EFI firmware. >> + @ Don't care about fp, but need 64 bit alignment.... >> + stmfd sp!, {fp, lr} >> + >> + @ allocate space on stack for return of new entry point of >> + @ zImage, as EFI stub may copy the kernel. Pass address >> + @ of space in r2 - EFI stub will fill in the pointer. >> + >> + sub sp, sp, #8 @ we only need 4 bytes, >> + @ but keep stack 8 byte aligned. >> + mov r2, sp >> + @ Pass our actual runtime start address in pointer data >> + adr r11, LC0 @ address of LC0 at run time >> + ldr r12, [r11, #0] @ address of LC0 at link time >> + >> + sub r3, r11, r12 @ calculate the delta offset >> + str r3, [r2, #0] >> + bl efi_entry >> + >> + @ get new zImage entry address from stack, put into r3 >> + ldr r3, [sp, #0] >> + add sp, sp, #8 @ restore stack >> + >> + @ Check for error return from EFI stub >> + mov r1, #EFI_STUB_ERROR >> + cmp r0, r1 >> + beq efi_load_fail >> + >> + >> + @ Save return values of efi_entry >> + stmfd sp!, {r0, r3} >> + bl cache_clean_flush >> + bl cache_off >> + ldmfd sp!, {r0, r3} >> + >> + @ Set parameters for booting zImage according to boot protocol >> + @ put FDT address in r2, it was returned by efi_entry() >> + @ r1 is FDT machine type, and r0 needs to be 0 >> + mov r2, r0 >> + mov r1, #0xFFFFFFFF >> + mov r0, #0 >> + >> + @ Branch to (possibly) relocated zImage that is in r3 >> + @ Make sure we are in A32 mode, as zImage requires >> + THUMB( bx r3 ) >> + ARM( mov pc, r3 ) >> + >> +efi_load_fail: >> + @ Return EFI_LOAD_ERROR to EFI firmware on error. >> + @ Switch back to ARM mode for EFI is done based on >> + @ return address on stack >> + ldr r0, =0x80000001 >> + ldmfd sp!, {fp, pc} >> +#endif >> + >> + THUMB( .thumb ) >> +zimage_continue: >> mrs r9, cpsr >> #ifdef CONFIG_ARM_VIRT_EXT >> bl __hyp_stub_install @ get into SVC mode, reversibly >> @@ -167,7 +246,6 @@ not_angel: >> * by the linker here, but it should preserve r7, r8, and r9. >> */ >> >> - .text >> >> #ifdef CONFIG_AUTO_ZRELADDR >> @ determine final kernel image address >> -- >> 1.7.10.4 >> >> -- >> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in >> the body of a message to majordomo@vger.kernel.org >> More majordomo info at http://vger.kernel.org/majordomo-info.html >> Please read the FAQ at http://www.tux.org/lkml/ > -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/