upstream u-boot with additional patches for our devices/boards: https://lists.denx.de/pipermail/u-boot/2017-March/282789.html (AXP crashes) ; Gbit ethernet patch for some LIME2 revisions ; with SPI flash support
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
u-boot/tools/rkcommon.c

379 lines
9.7 KiB

// SPDX-License-Identifier: GPL-2.0+
/*
* (C) Copyright 2015 Google, Inc
* Written by Simon Glass <sjg@chromium.org>
*
* (C) 2017 Theobroma Systems Design und Consulting GmbH
*
* Helper functions for Rockchip images
*/
#include "imagetool.h"
#include <image.h>
#include <rc4.h>
#include "mkimage.h"
#include "rkcommon.h"
#define DIV_ROUND_UP(n, d) (((n) + (d) - 1) / (d))
enum {
RK_SIGNATURE = 0x0ff0aa55,
};
/**
* struct header0_info - header block for boot ROM
*
* This is stored at SD card block 64 (where each block is 512 bytes, or at
* the start of SPI flash. It is encoded with RC4.
*
* @signature: Signature (must be RKSD_SIGNATURE)
* @disable_rc4: 0 to use rc4 for boot image, 1 to use plain binary
* @init_offset: Offset in blocks of the SPL code from this header
* block. E.g. 4 means 2KB after the start of this header.
* Other fields are not used by U-Boot
*/
struct header0_info {
uint32_t signature;
uint8_t reserved[4];
uint32_t disable_rc4;
uint16_t init_offset;
uint8_t reserved1[492];
uint16_t init_size;
uint16_t init_boot_size;
uint8_t reserved2[2];
};
/**
rockchip: mkimage: pad the header to 8-bytes (using a 'nop') for RK3399 The RK3399 boot code (running as AArch64) poses a bit of a challenge for SPL image generation: * The BootROM will start execution right after the 4-byte header (at the odd instruction word loaded into SRAM at 0xff8c2004, with the 'RK33' boot magic residing at 0xff8c2000). * The default padding (during ELF generation) for AArch64 is 0x0, which is an illegal instruction and the .text section needs to be naturally aligned (someone might locate a 64bit constant relative to the section start and unaligned loads trigger a fault for all privileged modes of an ARMv8)... so we can't simply define the CONFIG_SPL_TEXT_BASE option to the odd address (0xff8c2004). * Finally, we don't want to change the values used for padding of the SPL .text section for all ARMv8 targets to the instruction word encoding 'nop', as this would affect all padding in this section and might hide errors that would otherwise quickly trigger an illegal insn exception. To deal with this situation, we modify the rkimage generation to - understand the fact that the RK3399 needs to pad the header to an 8 byte boundary using an AArch64 'nop' - the necessary logic to adjust the header_size (which controls the location where the payload is copied into the image) and to insert this padding (AArch64 insn words are always little-endian) into the image following the 4-byte header magic. X-AffectedPlatforms: RK3399-Q7 Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com> Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
7 years ago
* struct header1_info
*/
struct header1_info {
uint32_t magic;
};
/**
* struct spl_info - spl info for each chip
*
* @imagename: Image name(passed by "mkimage -n")
* @spl_hdr: Boot ROM requires a 4-bytes spl header
* @spl_size: Spl size(include extra 4-bytes spl header)
* @spl_rc4: RC4 encode the SPL binary (same key as header)
*/
rockchip: mkimage: pad the header to 8-bytes (using a 'nop') for RK3399 The RK3399 boot code (running as AArch64) poses a bit of a challenge for SPL image generation: * The BootROM will start execution right after the 4-byte header (at the odd instruction word loaded into SRAM at 0xff8c2004, with the 'RK33' boot magic residing at 0xff8c2000). * The default padding (during ELF generation) for AArch64 is 0x0, which is an illegal instruction and the .text section needs to be naturally aligned (someone might locate a 64bit constant relative to the section start and unaligned loads trigger a fault for all privileged modes of an ARMv8)... so we can't simply define the CONFIG_SPL_TEXT_BASE option to the odd address (0xff8c2004). * Finally, we don't want to change the values used for padding of the SPL .text section for all ARMv8 targets to the instruction word encoding 'nop', as this would affect all padding in this section and might hide errors that would otherwise quickly trigger an illegal insn exception. To deal with this situation, we modify the rkimage generation to - understand the fact that the RK3399 needs to pad the header to an 8 byte boundary using an AArch64 'nop' - the necessary logic to adjust the header_size (which controls the location where the payload is copied into the image) and to insert this padding (AArch64 insn words are always little-endian) into the image following the 4-byte header magic. X-AffectedPlatforms: RK3399-Q7 Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com> Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
7 years ago
struct spl_info {
const char *imagename;
const char *spl_hdr;
const uint32_t spl_size;
const bool spl_rc4;
};
static struct spl_info spl_infos[] = {
{ "rk3036", "RK30", 0x1000, false },
{ "rk3128", "RK31", 0x1800, false },
{ "rk3188", "RK31", 0x8000 - 0x800, true },
{ "rk322x", "RK32", 0x8000 - 0x1000, false },
{ "rk3288", "RK32", 0x8000, false },
{ "rk3328", "RK32", 0x8000 - 0x1000, false },
{ "rk3368", "RK33", 0x8000 - 0x1000, false },
{ "rk3399", "RK33", 0x30000 - 0x2000, false },
{ "rv1108", "RK11", 0x1800, false },
};
static unsigned char rc4_key[16] = {
124, 78, 3, 4, 85, 5, 9, 7,
45, 44, 123, 56, 23, 13, 23, 17
};
static struct spl_info *rkcommon_get_spl_info(char *imagename)
{
int i;
if (!imagename)
return NULL;
for (i = 0; i < ARRAY_SIZE(spl_infos); i++)
if (!strncmp(imagename, spl_infos[i].imagename, 6))
return spl_infos + i;
return NULL;
}
int rkcommon_check_params(struct image_tool_params *params)
{
int i;
if (rkcommon_get_spl_info(params->imagename) != NULL)
return EXIT_SUCCESS;
/*
* If this is a operation (list or extract), the don't require
* imagename to be set.
*/
if (params->lflag || params->iflag)
return EXIT_SUCCESS;
fprintf(stderr, "ERROR: imagename (%s) is not supported!\n",
params->imagename ? params->imagename : "NULL");
fprintf(stderr, "Available imagename:");
for (i = 0; i < ARRAY_SIZE(spl_infos); i++)
fprintf(stderr, "\t%s", spl_infos[i].imagename);
fprintf(stderr, "\n");
return EXIT_FAILURE;
}
const char *rkcommon_get_spl_hdr(struct image_tool_params *params)
{
struct spl_info *info = rkcommon_get_spl_info(params->imagename);
/*
* info would not be NULL, because of we checked params before.
*/
return info->spl_hdr;
}
int rkcommon_get_spl_size(struct image_tool_params *params)
rockchip: mkimage: pad the header to 8-bytes (using a 'nop') for RK3399 The RK3399 boot code (running as AArch64) poses a bit of a challenge for SPL image generation: * The BootROM will start execution right after the 4-byte header (at the odd instruction word loaded into SRAM at 0xff8c2004, with the 'RK33' boot magic residing at 0xff8c2000). * The default padding (during ELF generation) for AArch64 is 0x0, which is an illegal instruction and the .text section needs to be naturally aligned (someone might locate a 64bit constant relative to the section start and unaligned loads trigger a fault for all privileged modes of an ARMv8)... so we can't simply define the CONFIG_SPL_TEXT_BASE option to the odd address (0xff8c2004). * Finally, we don't want to change the values used for padding of the SPL .text section for all ARMv8 targets to the instruction word encoding 'nop', as this would affect all padding in this section and might hide errors that would otherwise quickly trigger an illegal insn exception. To deal with this situation, we modify the rkimage generation to - understand the fact that the RK3399 needs to pad the header to an 8 byte boundary using an AArch64 'nop' - the necessary logic to adjust the header_size (which controls the location where the payload is copied into the image) and to insert this padding (AArch64 insn words are always little-endian) into the image following the 4-byte header magic. X-AffectedPlatforms: RK3399-Q7 Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com> Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
7 years ago
{
struct spl_info *info = rkcommon_get_spl_info(params->imagename);
/*
* info would not be NULL, because of we checked params before.
*/
return info->spl_size;
rockchip: mkimage: pad the header to 8-bytes (using a 'nop') for RK3399 The RK3399 boot code (running as AArch64) poses a bit of a challenge for SPL image generation: * The BootROM will start execution right after the 4-byte header (at the odd instruction word loaded into SRAM at 0xff8c2004, with the 'RK33' boot magic residing at 0xff8c2000). * The default padding (during ELF generation) for AArch64 is 0x0, which is an illegal instruction and the .text section needs to be naturally aligned (someone might locate a 64bit constant relative to the section start and unaligned loads trigger a fault for all privileged modes of an ARMv8)... so we can't simply define the CONFIG_SPL_TEXT_BASE option to the odd address (0xff8c2004). * Finally, we don't want to change the values used for padding of the SPL .text section for all ARMv8 targets to the instruction word encoding 'nop', as this would affect all padding in this section and might hide errors that would otherwise quickly trigger an illegal insn exception. To deal with this situation, we modify the rkimage generation to - understand the fact that the RK3399 needs to pad the header to an 8 byte boundary using an AArch64 'nop' - the necessary logic to adjust the header_size (which controls the location where the payload is copied into the image) and to insert this padding (AArch64 insn words are always little-endian) into the image following the 4-byte header magic. X-AffectedPlatforms: RK3399-Q7 Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com> Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
7 years ago
}
bool rkcommon_need_rc4_spl(struct image_tool_params *params)
{
struct spl_info *info = rkcommon_get_spl_info(params->imagename);
/*
* info would not be NULL, because of we checked params before.
*/
return info->spl_rc4;
}
rockchip: mkimage: pad the header to 8-bytes (using a 'nop') for RK3399 The RK3399 boot code (running as AArch64) poses a bit of a challenge for SPL image generation: * The BootROM will start execution right after the 4-byte header (at the odd instruction word loaded into SRAM at 0xff8c2004, with the 'RK33' boot magic residing at 0xff8c2000). * The default padding (during ELF generation) for AArch64 is 0x0, which is an illegal instruction and the .text section needs to be naturally aligned (someone might locate a 64bit constant relative to the section start and unaligned loads trigger a fault for all privileged modes of an ARMv8)... so we can't simply define the CONFIG_SPL_TEXT_BASE option to the odd address (0xff8c2004). * Finally, we don't want to change the values used for padding of the SPL .text section for all ARMv8 targets to the instruction word encoding 'nop', as this would affect all padding in this section and might hide errors that would otherwise quickly trigger an illegal insn exception. To deal with this situation, we modify the rkimage generation to - understand the fact that the RK3399 needs to pad the header to an 8 byte boundary using an AArch64 'nop' - the necessary logic to adjust the header_size (which controls the location where the payload is copied into the image) and to insert this padding (AArch64 insn words are always little-endian) into the image following the 4-byte header magic. X-AffectedPlatforms: RK3399-Q7 Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com> Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
7 years ago
static void rkcommon_set_header0(void *buf, uint file_size,
struct image_tool_params *params)
{
rockchip: mkimage: pad the header to 8-bytes (using a 'nop') for RK3399 The RK3399 boot code (running as AArch64) poses a bit of a challenge for SPL image generation: * The BootROM will start execution right after the 4-byte header (at the odd instruction word loaded into SRAM at 0xff8c2004, with the 'RK33' boot magic residing at 0xff8c2000). * The default padding (during ELF generation) for AArch64 is 0x0, which is an illegal instruction and the .text section needs to be naturally aligned (someone might locate a 64bit constant relative to the section start and unaligned loads trigger a fault for all privileged modes of an ARMv8)... so we can't simply define the CONFIG_SPL_TEXT_BASE option to the odd address (0xff8c2004). * Finally, we don't want to change the values used for padding of the SPL .text section for all ARMv8 targets to the instruction word encoding 'nop', as this would affect all padding in this section and might hide errors that would otherwise quickly trigger an illegal insn exception. To deal with this situation, we modify the rkimage generation to - understand the fact that the RK3399 needs to pad the header to an 8 byte boundary using an AArch64 'nop' - the necessary logic to adjust the header_size (which controls the location where the payload is copied into the image) and to insert this padding (AArch64 insn words are always little-endian) into the image following the 4-byte header magic. X-AffectedPlatforms: RK3399-Q7 Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com> Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
7 years ago
struct header0_info *hdr = buf;
rockchip: mkimage: pad the header to 8-bytes (using a 'nop') for RK3399 The RK3399 boot code (running as AArch64) poses a bit of a challenge for SPL image generation: * The BootROM will start execution right after the 4-byte header (at the odd instruction word loaded into SRAM at 0xff8c2004, with the 'RK33' boot magic residing at 0xff8c2000). * The default padding (during ELF generation) for AArch64 is 0x0, which is an illegal instruction and the .text section needs to be naturally aligned (someone might locate a 64bit constant relative to the section start and unaligned loads trigger a fault for all privileged modes of an ARMv8)... so we can't simply define the CONFIG_SPL_TEXT_BASE option to the odd address (0xff8c2004). * Finally, we don't want to change the values used for padding of the SPL .text section for all ARMv8 targets to the instruction word encoding 'nop', as this would affect all padding in this section and might hide errors that would otherwise quickly trigger an illegal insn exception. To deal with this situation, we modify the rkimage generation to - understand the fact that the RK3399 needs to pad the header to an 8 byte boundary using an AArch64 'nop' - the necessary logic to adjust the header_size (which controls the location where the payload is copied into the image) and to insert this padding (AArch64 insn words are always little-endian) into the image following the 4-byte header magic. X-AffectedPlatforms: RK3399-Q7 Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com> Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
7 years ago
memset(buf, '\0', RK_INIT_OFFSET * RK_BLK_SIZE);
hdr->signature = RK_SIGNATURE;
hdr->disable_rc4 = !rkcommon_need_rc4_spl(params);
hdr->init_offset = RK_INIT_OFFSET;
hdr->init_size = DIV_ROUND_UP(file_size, RK_BLK_SIZE);
/*
* The init_size has to be a multiple of 4 blocks (i.e. of 2K)
* or the BootROM will not boot the image.
*
* Note: To verify that this is not a legacy constraint, we
* rechecked this against the RK3399 BootROM.
*/
hdr->init_size = ROUND(hdr->init_size, 4);
/*
* init_boot_size needs to be set, as it is read by the BootROM
* to determine the size of the next-stage bootloader (e.g. U-Boot
* proper), when used with the back-to-bootrom functionality.
*
* see https://lists.denx.de/pipermail/u-boot/2017-May/293267.html
* for a more detailed explanation by Andy Yan
*/
hdr->init_boot_size = hdr->init_size + RK_MAX_BOOT_SIZE / RK_BLK_SIZE;
rc4_encode(buf, RK_BLK_SIZE, rc4_key);
rockchip: mkimage: pad the header to 8-bytes (using a 'nop') for RK3399 The RK3399 boot code (running as AArch64) poses a bit of a challenge for SPL image generation: * The BootROM will start execution right after the 4-byte header (at the odd instruction word loaded into SRAM at 0xff8c2004, with the 'RK33' boot magic residing at 0xff8c2000). * The default padding (during ELF generation) for AArch64 is 0x0, which is an illegal instruction and the .text section needs to be naturally aligned (someone might locate a 64bit constant relative to the section start and unaligned loads trigger a fault for all privileged modes of an ARMv8)... so we can't simply define the CONFIG_SPL_TEXT_BASE option to the odd address (0xff8c2004). * Finally, we don't want to change the values used for padding of the SPL .text section for all ARMv8 targets to the instruction word encoding 'nop', as this would affect all padding in this section and might hide errors that would otherwise quickly trigger an illegal insn exception. To deal with this situation, we modify the rkimage generation to - understand the fact that the RK3399 needs to pad the header to an 8 byte boundary using an AArch64 'nop' - the necessary logic to adjust the header_size (which controls the location where the payload is copied into the image) and to insert this padding (AArch64 insn words are always little-endian) into the image following the 4-byte header magic. X-AffectedPlatforms: RK3399-Q7 Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com> Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
7 years ago
}
int rkcommon_set_header(void *buf, uint file_size,
struct image_tool_params *params)
{
struct header1_info *hdr = buf + RK_SPL_HDR_START;
if (file_size > rkcommon_get_spl_size(params))
return -ENOSPC;
rkcommon_set_header0(buf, file_size, params);
/* Set up the SPL name (i.e. copy spl_hdr over) */
rockchip: mkimage: pad the header to 8-bytes (using a 'nop') for RK3399 The RK3399 boot code (running as AArch64) poses a bit of a challenge for SPL image generation: * The BootROM will start execution right after the 4-byte header (at the odd instruction word loaded into SRAM at 0xff8c2004, with the 'RK33' boot magic residing at 0xff8c2000). * The default padding (during ELF generation) for AArch64 is 0x0, which is an illegal instruction and the .text section needs to be naturally aligned (someone might locate a 64bit constant relative to the section start and unaligned loads trigger a fault for all privileged modes of an ARMv8)... so we can't simply define the CONFIG_SPL_TEXT_BASE option to the odd address (0xff8c2004). * Finally, we don't want to change the values used for padding of the SPL .text section for all ARMv8 targets to the instruction word encoding 'nop', as this would affect all padding in this section and might hide errors that would otherwise quickly trigger an illegal insn exception. To deal with this situation, we modify the rkimage generation to - understand the fact that the RK3399 needs to pad the header to an 8 byte boundary using an AArch64 'nop' - the necessary logic to adjust the header_size (which controls the location where the payload is copied into the image) and to insert this padding (AArch64 insn words are always little-endian) into the image following the 4-byte header magic. X-AffectedPlatforms: RK3399-Q7 Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com> Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
7 years ago
memcpy(&hdr->magic, rkcommon_get_spl_hdr(params), RK_SPL_HDR_SIZE);
if (rkcommon_need_rc4_spl(params))
rkcommon_rc4_encode_spl(buf, RK_SPL_HDR_START,
params->file_size - RK_SPL_HDR_START);
return 0;
}
static inline unsigned rkcommon_offset_to_spi(unsigned offset)
{
/*
* While SD/MMC images use a flat addressing, SPI images are padded
* to use the first 2K of every 4K sector only.
*/
return ((offset & ~0x7ff) << 1) + (offset & 0x7ff);
}
static int rkcommon_parse_header(const void *buf, struct header0_info *header0,
struct spl_info **spl_info)
{
unsigned hdr1_offset;
struct header1_info *hdr1_sdmmc, *hdr1_spi;
int i;
if (spl_info)
*spl_info = NULL;
/*
* The first header (hdr0) is always RC4 encoded, so try to decrypt
* with the well-known key.
*/
memcpy((void *)header0, buf, sizeof(struct header0_info));
rc4_encode((void *)header0, sizeof(struct header0_info), rc4_key);
if (header0->signature != RK_SIGNATURE)
return -EPROTO;
/* We don't support RC4 encoded image payloads here, yet... */
if (header0->disable_rc4 == 0)
return -ENOSYS;
hdr1_offset = header0->init_offset * RK_BLK_SIZE;
hdr1_sdmmc = (struct header1_info *)(buf + hdr1_offset);
hdr1_spi = (struct header1_info *)(buf +
rkcommon_offset_to_spi(hdr1_offset));
for (i = 0; i < ARRAY_SIZE(spl_infos); i++) {
if (!memcmp(&hdr1_sdmmc->magic, spl_infos[i].spl_hdr, 4)) {
if (spl_info)
*spl_info = &spl_infos[i];
return IH_TYPE_RKSD;
} else if (!memcmp(&hdr1_spi->magic, spl_infos[i].spl_hdr, 4)) {
if (spl_info)
*spl_info = &spl_infos[i];
return IH_TYPE_RKSPI;
}
}
return -1;
}
int rkcommon_verify_header(unsigned char *buf, int size,
struct image_tool_params *params)
{
struct header0_info header0;
struct spl_info *img_spl_info, *spl_info;
int ret;
ret = rkcommon_parse_header(buf, &header0, &img_spl_info);
/* If this is the (unimplemented) RC4 case, then rewrite the result */
if (ret == -ENOSYS)
return 0;
if (ret < 0)
return ret;
/*
* If no 'imagename' is specified via the commandline (e.g. if this is
* 'dumpimage -l' w/o any further constraints), we accept any spl_info.
*/
if (params->imagename == NULL)
return 0;
/* Match the 'imagename' against the 'spl_hdr' found */
spl_info = rkcommon_get_spl_info(params->imagename);
if (spl_info && img_spl_info)
return strcmp(spl_info->spl_hdr, img_spl_info->spl_hdr);
return -ENOENT;
}
void rkcommon_print_header(const void *buf)
{
struct header0_info header0;
struct spl_info *spl_info;
uint8_t image_type;
int ret;
ret = rkcommon_parse_header(buf, &header0, &spl_info);
/* If this is the (unimplemented) RC4 case, then fail silently */
if (ret == -ENOSYS)
return;
if (ret < 0) {
fprintf(stderr, "Error: image verification failed\n");
return;
}
image_type = ret;
printf("Image Type: Rockchip %s (%s) boot image\n",
spl_info->spl_hdr,
(image_type == IH_TYPE_RKSD) ? "SD/MMC" : "SPI");
printf("Data Size: %d bytes\n", header0.init_size * RK_BLK_SIZE);
}
void rkcommon_rc4_encode_spl(void *buf, unsigned int offset, unsigned int size)
{
unsigned int remaining = size;
while (remaining > 0) {
int step = (remaining > RK_BLK_SIZE) ? RK_BLK_SIZE : remaining;
rc4_encode(buf + offset, step, rc4_key);
offset += RK_BLK_SIZE;
remaining -= step;
}
}
rockchip: mkimage: pad the header to 8-bytes (using a 'nop') for RK3399 The RK3399 boot code (running as AArch64) poses a bit of a challenge for SPL image generation: * The BootROM will start execution right after the 4-byte header (at the odd instruction word loaded into SRAM at 0xff8c2004, with the 'RK33' boot magic residing at 0xff8c2000). * The default padding (during ELF generation) for AArch64 is 0x0, which is an illegal instruction and the .text section needs to be naturally aligned (someone might locate a 64bit constant relative to the section start and unaligned loads trigger a fault for all privileged modes of an ARMv8)... so we can't simply define the CONFIG_SPL_TEXT_BASE option to the odd address (0xff8c2004). * Finally, we don't want to change the values used for padding of the SPL .text section for all ARMv8 targets to the instruction word encoding 'nop', as this would affect all padding in this section and might hide errors that would otherwise quickly trigger an illegal insn exception. To deal with this situation, we modify the rkimage generation to - understand the fact that the RK3399 needs to pad the header to an 8 byte boundary using an AArch64 'nop' - the necessary logic to adjust the header_size (which controls the location where the payload is copied into the image) and to insert this padding (AArch64 insn words are always little-endian) into the image following the 4-byte header magic. X-AffectedPlatforms: RK3399-Q7 Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com> Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
7 years ago
int rkcommon_vrec_header(struct image_tool_params *params,
struct image_type_params *tparams,
unsigned int alignment)
rockchip: mkimage: pad the header to 8-bytes (using a 'nop') for RK3399 The RK3399 boot code (running as AArch64) poses a bit of a challenge for SPL image generation: * The BootROM will start execution right after the 4-byte header (at the odd instruction word loaded into SRAM at 0xff8c2004, with the 'RK33' boot magic residing at 0xff8c2000). * The default padding (during ELF generation) for AArch64 is 0x0, which is an illegal instruction and the .text section needs to be naturally aligned (someone might locate a 64bit constant relative to the section start and unaligned loads trigger a fault for all privileged modes of an ARMv8)... so we can't simply define the CONFIG_SPL_TEXT_BASE option to the odd address (0xff8c2004). * Finally, we don't want to change the values used for padding of the SPL .text section for all ARMv8 targets to the instruction word encoding 'nop', as this would affect all padding in this section and might hide errors that would otherwise quickly trigger an illegal insn exception. To deal with this situation, we modify the rkimage generation to - understand the fact that the RK3399 needs to pad the header to an 8 byte boundary using an AArch64 'nop' - the necessary logic to adjust the header_size (which controls the location where the payload is copied into the image) and to insert this padding (AArch64 insn words are always little-endian) into the image following the 4-byte header magic. X-AffectedPlatforms: RK3399-Q7 Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com> Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
7 years ago
{
unsigned int unpadded_size;
unsigned int padded_size;
rockchip: mkimage: pad the header to 8-bytes (using a 'nop') for RK3399 The RK3399 boot code (running as AArch64) poses a bit of a challenge for SPL image generation: * The BootROM will start execution right after the 4-byte header (at the odd instruction word loaded into SRAM at 0xff8c2004, with the 'RK33' boot magic residing at 0xff8c2000). * The default padding (during ELF generation) for AArch64 is 0x0, which is an illegal instruction and the .text section needs to be naturally aligned (someone might locate a 64bit constant relative to the section start and unaligned loads trigger a fault for all privileged modes of an ARMv8)... so we can't simply define the CONFIG_SPL_TEXT_BASE option to the odd address (0xff8c2004). * Finally, we don't want to change the values used for padding of the SPL .text section for all ARMv8 targets to the instruction word encoding 'nop', as this would affect all padding in this section and might hide errors that would otherwise quickly trigger an illegal insn exception. To deal with this situation, we modify the rkimage generation to - understand the fact that the RK3399 needs to pad the header to an 8 byte boundary using an AArch64 'nop' - the necessary logic to adjust the header_size (which controls the location where the payload is copied into the image) and to insert this padding (AArch64 insn words are always little-endian) into the image following the 4-byte header magic. X-AffectedPlatforms: RK3399-Q7 Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com> Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
7 years ago
/*
* The SPL image looks as follows:
*
* 0x0 header0 (see rkcommon.c)
* 0x800 spl_name ('RK30', ..., 'RK33')
* (start of the payload for AArch64 payloads: we expect the
* first 4 bytes to be available for overwriting with our
* spl_name)
rockchip: mkimage: pad the header to 8-bytes (using a 'nop') for RK3399 The RK3399 boot code (running as AArch64) poses a bit of a challenge for SPL image generation: * The BootROM will start execution right after the 4-byte header (at the odd instruction word loaded into SRAM at 0xff8c2004, with the 'RK33' boot magic residing at 0xff8c2000). * The default padding (during ELF generation) for AArch64 is 0x0, which is an illegal instruction and the .text section needs to be naturally aligned (someone might locate a 64bit constant relative to the section start and unaligned loads trigger a fault for all privileged modes of an ARMv8)... so we can't simply define the CONFIG_SPL_TEXT_BASE option to the odd address (0xff8c2004). * Finally, we don't want to change the values used for padding of the SPL .text section for all ARMv8 targets to the instruction word encoding 'nop', as this would affect all padding in this section and might hide errors that would otherwise quickly trigger an illegal insn exception. To deal with this situation, we modify the rkimage generation to - understand the fact that the RK3399 needs to pad the header to an 8 byte boundary using an AArch64 'nop' - the necessary logic to adjust the header_size (which controls the location where the payload is copied into the image) and to insert this padding (AArch64 insn words are always little-endian) into the image following the 4-byte header magic. X-AffectedPlatforms: RK3399-Q7 Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com> Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
7 years ago
* 0x804 first instruction to be executed
* (start of the image/payload for 32bit payloads)
rockchip: mkimage: pad the header to 8-bytes (using a 'nop') for RK3399 The RK3399 boot code (running as AArch64) poses a bit of a challenge for SPL image generation: * The BootROM will start execution right after the 4-byte header (at the odd instruction word loaded into SRAM at 0xff8c2004, with the 'RK33' boot magic residing at 0xff8c2000). * The default padding (during ELF generation) for AArch64 is 0x0, which is an illegal instruction and the .text section needs to be naturally aligned (someone might locate a 64bit constant relative to the section start and unaligned loads trigger a fault for all privileged modes of an ARMv8)... so we can't simply define the CONFIG_SPL_TEXT_BASE option to the odd address (0xff8c2004). * Finally, we don't want to change the values used for padding of the SPL .text section for all ARMv8 targets to the instruction word encoding 'nop', as this would affect all padding in this section and might hide errors that would otherwise quickly trigger an illegal insn exception. To deal with this situation, we modify the rkimage generation to - understand the fact that the RK3399 needs to pad the header to an 8 byte boundary using an AArch64 'nop' - the necessary logic to adjust the header_size (which controls the location where the payload is copied into the image) and to insert this padding (AArch64 insn words are always little-endian) into the image following the 4-byte header magic. X-AffectedPlatforms: RK3399-Q7 Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com> Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
7 years ago
*
* For AArch64 (ARMv8) payloads, natural alignment (8-bytes) is
* required for its sections (so the image we receive needs to
* have the first 4 bytes reserved for the spl_name). Reserving
* these 4 bytes is done using the BOOT0_HOOK infrastructure.
rockchip: mkimage: pad the header to 8-bytes (using a 'nop') for RK3399 The RK3399 boot code (running as AArch64) poses a bit of a challenge for SPL image generation: * The BootROM will start execution right after the 4-byte header (at the odd instruction word loaded into SRAM at 0xff8c2004, with the 'RK33' boot magic residing at 0xff8c2000). * The default padding (during ELF generation) for AArch64 is 0x0, which is an illegal instruction and the .text section needs to be naturally aligned (someone might locate a 64bit constant relative to the section start and unaligned loads trigger a fault for all privileged modes of an ARMv8)... so we can't simply define the CONFIG_SPL_TEXT_BASE option to the odd address (0xff8c2004). * Finally, we don't want to change the values used for padding of the SPL .text section for all ARMv8 targets to the instruction word encoding 'nop', as this would affect all padding in this section and might hide errors that would otherwise quickly trigger an illegal insn exception. To deal with this situation, we modify the rkimage generation to - understand the fact that the RK3399 needs to pad the header to an 8 byte boundary using an AArch64 'nop' - the necessary logic to adjust the header_size (which controls the location where the payload is copied into the image) and to insert this padding (AArch64 insn words are always little-endian) into the image following the 4-byte header magic. X-AffectedPlatforms: RK3399-Q7 Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com> Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
7 years ago
*
* The header is always at 0x800 (as we now use a payload
* prepadded using the boot0 hook for all targets): the first
* 4 bytes of these images can safely be overwritten using the
* boot magic.
rockchip: mkimage: pad the header to 8-bytes (using a 'nop') for RK3399 The RK3399 boot code (running as AArch64) poses a bit of a challenge for SPL image generation: * The BootROM will start execution right after the 4-byte header (at the odd instruction word loaded into SRAM at 0xff8c2004, with the 'RK33' boot magic residing at 0xff8c2000). * The default padding (during ELF generation) for AArch64 is 0x0, which is an illegal instruction and the .text section needs to be naturally aligned (someone might locate a 64bit constant relative to the section start and unaligned loads trigger a fault for all privileged modes of an ARMv8)... so we can't simply define the CONFIG_SPL_TEXT_BASE option to the odd address (0xff8c2004). * Finally, we don't want to change the values used for padding of the SPL .text section for all ARMv8 targets to the instruction word encoding 'nop', as this would affect all padding in this section and might hide errors that would otherwise quickly trigger an illegal insn exception. To deal with this situation, we modify the rkimage generation to - understand the fact that the RK3399 needs to pad the header to an 8 byte boundary using an AArch64 'nop' - the necessary logic to adjust the header_size (which controls the location where the payload is copied into the image) and to insert this padding (AArch64 insn words are always little-endian) into the image following the 4-byte header magic. X-AffectedPlatforms: RK3399-Q7 Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com> Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
7 years ago
*/
tparams->header_size = RK_SPL_HDR_START;
rockchip: mkimage: pad the header to 8-bytes (using a 'nop') for RK3399 The RK3399 boot code (running as AArch64) poses a bit of a challenge for SPL image generation: * The BootROM will start execution right after the 4-byte header (at the odd instruction word loaded into SRAM at 0xff8c2004, with the 'RK33' boot magic residing at 0xff8c2000). * The default padding (during ELF generation) for AArch64 is 0x0, which is an illegal instruction and the .text section needs to be naturally aligned (someone might locate a 64bit constant relative to the section start and unaligned loads trigger a fault for all privileged modes of an ARMv8)... so we can't simply define the CONFIG_SPL_TEXT_BASE option to the odd address (0xff8c2004). * Finally, we don't want to change the values used for padding of the SPL .text section for all ARMv8 targets to the instruction word encoding 'nop', as this would affect all padding in this section and might hide errors that would otherwise quickly trigger an illegal insn exception. To deal with this situation, we modify the rkimage generation to - understand the fact that the RK3399 needs to pad the header to an 8 byte boundary using an AArch64 'nop' - the necessary logic to adjust the header_size (which controls the location where the payload is copied into the image) and to insert this padding (AArch64 insn words are always little-endian) into the image following the 4-byte header magic. X-AffectedPlatforms: RK3399-Q7 Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com> Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
7 years ago
/* Allocate, clear and install the header */
tparams->hdr = malloc(tparams->header_size);
if (!tparams->hdr)
return -ENOMEM;
rockchip: mkimage: pad the header to 8-bytes (using a 'nop') for RK3399 The RK3399 boot code (running as AArch64) poses a bit of a challenge for SPL image generation: * The BootROM will start execution right after the 4-byte header (at the odd instruction word loaded into SRAM at 0xff8c2004, with the 'RK33' boot magic residing at 0xff8c2000). * The default padding (during ELF generation) for AArch64 is 0x0, which is an illegal instruction and the .text section needs to be naturally aligned (someone might locate a 64bit constant relative to the section start and unaligned loads trigger a fault for all privileged modes of an ARMv8)... so we can't simply define the CONFIG_SPL_TEXT_BASE option to the odd address (0xff8c2004). * Finally, we don't want to change the values used for padding of the SPL .text section for all ARMv8 targets to the instruction word encoding 'nop', as this would affect all padding in this section and might hide errors that would otherwise quickly trigger an illegal insn exception. To deal with this situation, we modify the rkimage generation to - understand the fact that the RK3399 needs to pad the header to an 8 byte boundary using an AArch64 'nop' - the necessary logic to adjust the header_size (which controls the location where the payload is copied into the image) and to insert this padding (AArch64 insn words are always little-endian) into the image following the 4-byte header magic. X-AffectedPlatforms: RK3399-Q7 Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com> Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
7 years ago
memset(tparams->hdr, 0, tparams->header_size);
/*
* If someone passed in 0 for the alignment, we'd better handle
* it correctly...
*/
if (!alignment)
alignment = 1;
unpadded_size = tparams->header_size + params->file_size;
padded_size = ROUND(unpadded_size, alignment);
return padded_size - unpadded_size;
rockchip: mkimage: pad the header to 8-bytes (using a 'nop') for RK3399 The RK3399 boot code (running as AArch64) poses a bit of a challenge for SPL image generation: * The BootROM will start execution right after the 4-byte header (at the odd instruction word loaded into SRAM at 0xff8c2004, with the 'RK33' boot magic residing at 0xff8c2000). * The default padding (during ELF generation) for AArch64 is 0x0, which is an illegal instruction and the .text section needs to be naturally aligned (someone might locate a 64bit constant relative to the section start and unaligned loads trigger a fault for all privileged modes of an ARMv8)... so we can't simply define the CONFIG_SPL_TEXT_BASE option to the odd address (0xff8c2004). * Finally, we don't want to change the values used for padding of the SPL .text section for all ARMv8 targets to the instruction word encoding 'nop', as this would affect all padding in this section and might hide errors that would otherwise quickly trigger an illegal insn exception. To deal with this situation, we modify the rkimage generation to - understand the fact that the RK3399 needs to pad the header to an 8 byte boundary using an AArch64 'nop' - the necessary logic to adjust the header_size (which controls the location where the payload is copied into the image) and to insert this padding (AArch64 insn words are always little-endian) into the image following the 4-byte header magic. X-AffectedPlatforms: RK3399-Q7 Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com> Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
7 years ago
}