mirror of
https://github.com/AsahiLinux/u-boot
synced 2024-12-30 06:53:09 +00:00
6039e0edc8
The current mechanism is unnecessarily complex. Simplify the whole mechanism such that the entire fitImage is signed, IVT is placed at the end, followed by CSF, and this entire bundle is also authenticated. This makes the signing scripting far simpler. Signed-off-by: Marek Vasut <marex@denx.de>
221 lines
9.7 KiB
Text
221 lines
9.7 KiB
Text
+=========================================================+
|
|
+ i.MX8M U-Boot HABv4 Secure Boot guide for SPL targets +
|
|
+=========================================================+
|
|
|
|
1. HABv4 secure boot process
|
|
-----------------------------
|
|
|
|
This document is an addendum of mx6_mx7_spl_secure_boot.txt guide describing
|
|
a step-by-step procedure on how to sign and securely boot an U-Boot image for
|
|
SPL targets on i.MX8M, i.MX8M Mini, i.MX8M Nano, i.MX8M Plus.
|
|
|
|
Details about HAB can be found in the application note AN4581[1] and in the
|
|
introduction_habv4.txt document.
|
|
|
|
1.1 Building a SPL target supporting secure boot
|
|
-------------------------------------------------
|
|
|
|
The U-Boot build for i.MX8M SoC makes use of Second Program Loader (SPL)
|
|
support, fitImage support and custom i.MX8M specific flash.bin container.
|
|
This leads to a generation of multiple intermediate build artifacts, the
|
|
U-Boot SPL, U-Boot binary, DT blob. These later two artifacts are bundled
|
|
with external ATF BL31 blob to form a fitImage. The fitImage is bundled
|
|
with SPL and external DDR and optional HDMI PHY initialization blobs to
|
|
form the final flash.bin container. The HABv4 can be used to authenticate
|
|
all of the input binaries separately.
|
|
|
|
Out of reset the ROM code authenticates the SPL and PHY initialization
|
|
blobs, combination of which is responsible for initializing essential
|
|
features such as DDR, UART, PMIC and clock enablement. Once the DDR is
|
|
available, the SPL code loads the secondary fitImage to its specific
|
|
address and call the HAB APIs to extend the root of trust on its
|
|
components.
|
|
|
|
The U-Boot SPL provides support to secure boot configuration and also
|
|
provide access to the HAB APIs exposed by the ROM vector table, the
|
|
U-Boot provides access to HAB APIs via SMC calls to ATF. The support
|
|
is enabled by selecting the CONFIG_IMX_HAB option.
|
|
|
|
When built with this configuration the U-Boot correctly pads combined
|
|
SPL and PHY initialization blob image, called u-boot-spl-ddr.bin, by
|
|
aligning to the next 0xC00 address, so the CSF signature data generated
|
|
by CST can be concatenated to the image.
|
|
|
|
The U-Boot also reserves space in the fitImage binary (u-boot.itb) between
|
|
the fitImage tree and external blobs included in it, so it can be used to
|
|
inject IVT and CST signatures used by SPL HAB calls to authenticate the
|
|
fitImage components.
|
|
|
|
The diagram below illustrate a signed SPL combined with DDR PHY
|
|
initialization firmware blobs part of flash.bin container layout.
|
|
This part is loaded to memory address ( CONFIG_SPL_TEXT_BASE - 0x40 ) and
|
|
authenticated the BootROM. The reason for the offset is so that the *entry
|
|
would be at memory address CONFIG_SPL_TEXT_BASE when BootROM executes the
|
|
code within it:
|
|
|
|
------- +-----------------------------+ <-- *start
|
|
^ | Image Vector Table |
|
|
| | (0x20 bytes) |
|
|
| +-----------------------------+ <-- *boot_data
|
|
| | Boot Data |
|
|
| +-----------------------------+
|
|
| | Padding |
|
|
Signed | | to 0x40 bytes from *start |
|
|
Data | +-----------------------------+ <-- *entry
|
|
| | |
|
|
| | SPL combined with DDR PHY |
|
|
| | initialization blobs |
|
|
| | (u-boot-spl-ddr.bin) |
|
|
| | |
|
|
| +-----------------------------+
|
|
v | Padding |
|
|
------- +-----------------------------+ <-- *csf
|
|
| |
|
|
| Command Sequence File (CSF) |
|
|
| |
|
|
+-----------------------------+
|
|
| Padding (optional) |
|
|
+-----------------------------+
|
|
|
|
The diagram below illustrate a signed U-Boot binary, DT blob and external
|
|
ATF BL31 blob combined to form fitImage part of flash.bin container layout.
|
|
The *load_address is CONFIG_SPL_LOAD_FIT_ADDRESS, the fitImage is loaded
|
|
including all of its embedded data, authenticated using IVT+CSF concatenated
|
|
at the end of the fitImage at offset aligned to 4 kiB. The fitImage with
|
|
external data is not supported.
|
|
|
|
------- +-----------------------------+ <-- *load_address
|
|
^ | |
|
|
| | fitImage tree |
|
|
| | with embedded data |
|
|
| | (cca. 1 MiB) |
|
|
Signed | | |
|
|
.----- Tree | +-----------------------------+
|
|
| Data | | Padding to next 4k aligned |
|
|
| | | from *load_address |
|
|
| | +-----------------------------+ <-- *ivt
|
|
| | | Image Vector Table |
|
|
| v | (0x20 bytes) |
|
|
| ------- +-----------------------------+ <-- *csf
|
|
| | Command Sequence File (CSF) |
|
|
| | for all signed entries in |
|
|
'---------------->| the fitImage, tree and data |
|
|
| (cca 6-7 kiB) |
|
|
+-----------------------------+
|
|
|
|
The diagram below illustrate a combined flash.bin container layout:
|
|
|
|
+-----------------------------+
|
|
| Signed SPL part |
|
|
+-----------------------------+
|
|
| Signed fitImage part |
|
|
+-----------------------------+
|
|
|
|
1.2 Enabling the secure boot support
|
|
-------------------------------------
|
|
|
|
The first step is to generate an U-Boot image supporting the HAB features
|
|
mentioned above, this can be achieved by adding CONFIG_IMX_HAB to the
|
|
build configuration:
|
|
|
|
- Defconfig:
|
|
|
|
CONFIG_IMX_HAB=y
|
|
|
|
- Kconfig:
|
|
|
|
ARM architecture -> Support i.MX HAB features
|
|
|
|
1.3 Signing the images
|
|
-----------------------
|
|
|
|
The CSF contains all the commands that the HAB executes during the secure
|
|
boot. These commands instruct the HAB code on which memory areas of the image
|
|
to authenticate, which keys to install, use and etc.
|
|
|
|
CSF examples are available under doc/imx/habv4/csf_examples/ directory.
|
|
|
|
CSF "Blocks" line for csf_spl.txt can be generated as follows:
|
|
|
|
```
|
|
spl_block_base=$(printf "0x%x" $(( $(sed -n "/CONFIG_SPL_TEXT_BASE=/ s@.*=@@p" .config) - 0x40)) )
|
|
spl_block_size=$(printf "0x%x" $(stat -tc %s u-boot-spl-ddr.bin))
|
|
sed -i "/Blocks = / s@.*@ Blocks = $spl_block_base 0x0 $spl_block_size \"flash.bin\"@" csf_spl.txt
|
|
```
|
|
|
|
The resulting line looks as follows:
|
|
```
|
|
Blocks = 0x7e0fc0 0x0 0x306f0 "flash.bin"
|
|
```
|
|
|
|
The columns mean:
|
|
- CONFIG_SPL_TEXT_BASE - 0x40 -- Start address of signed data, in DRAM
|
|
- 0x0 -- Start address of signed data, in "flash.bin"
|
|
- 0x306f0 -- Length of signed data, in "flash.bin"
|
|
- Filename -- "flash.bin"
|
|
|
|
To generate signature for the SPL part of flash.bin container, use CST:
|
|
```
|
|
cst -i csf_spl.tmp -o csf_spl.bin
|
|
```
|
|
|
|
The newly generated CST blob has to be patched into existing flash.bin
|
|
container. Conveniently, flash.bin IVT contains physical address of the
|
|
CSF blob. Remember, the SPL part of flash.bin container is loaded by the
|
|
BootROM at CONFIG_SPL_TEXT_BASE - 0x40 , so the offset of CSF blob in
|
|
the fitImage can be calculated and inserted into the flash.bin in the
|
|
correct location as follows:
|
|
```
|
|
# offset = IVT_HEADER[6 = CSF address] - CONFIG_SPL_TEXT_BASE - 0x40
|
|
spl_csf_offset=$(xxd -s 24 -l 4 -e flash.bin | cut -d " " -f 2 | sed "s@^@0x@")
|
|
spl_bin_offset=$(xxd -s 4 -l 4 -e flash.bin | cut -d " " -f 2 | sed "s@^@0x@")
|
|
spl_dd_offset=$((${spl_csf_offset} - ${spl_bin_offset} + 0x40))
|
|
dd if=csf_spl.bin of=flash.bin bs=1 seek=${spl_dd_offset} conv=notrunc
|
|
```
|
|
|
|
CSF "Blocks" line for csf_fit.txt can be generated as follows:
|
|
```
|
|
# fitImage
|
|
fit_block_base=$(printf "0x%x" $(sed -n "/CONFIG_SPL_LOAD_FIT_ADDRESS=/ s@.*=@@p" .config) )
|
|
fit_block_offset=$(printf "0x%s" $(fdtget -t x u-boot.dtb /binman/imx-boot/uboot offset))
|
|
fit_block_size=$(printf "0x%x" $(( ( ( $(stat -tc %s u-boot.itb) + 0x1000 - 0x1 ) & ~(0x1000 - 0x1)) + 0x20 )) )
|
|
sed -i "/Blocks = / s@.*@ Blocks = $fit_block_base $fit_block_offset $fit_block_size \"flash.bin\"@" csf_fit.tmp
|
|
```
|
|
|
|
The fitImage part of flash.bin requires separate IVT. Generate the IVT and
|
|
patch it into the correct aligned location of flash.bin as follows:
|
|
```
|
|
# IVT
|
|
ivt_ptr_base=$(printf "%08x" ${fit_block_base} | sed "s@\(..\)\(..\)\(..\)\(..\)@0x\4\3\2\1@")
|
|
ivt_block_base=$(printf "%08x" $(( ${fit_block_base} + ${fit_block_size} - 0x20 )) | sed "s@\(..\)\(..\)\(..\)\(..\)@0x\4\3\2\1@")
|
|
csf_block_base=$(printf "%08x" $(( ${fit_block_base} + ${fit_block_size} )) | sed "s@\(..\)\(..\)\(..\)\(..\)@0x\4\3\2\1@")
|
|
ivt_block_offset=$((${fit_block_offset} + ${fit_block_size} - 0x20))
|
|
csf_block_offset=$((${ivt_block_offset} + 0x20))
|
|
|
|
echo "0xd1002041 ${ivt_block_base} 0x00000000 0x00000000 0x00000000 ${ivt_block_base} ${csf_block_base} 0x00000000" | xxd -r -p > ivt.bin
|
|
dd if=ivt.bin of=flash.bin bs=1 seek=${ivt_block_offset} conv=notrunc
|
|
```
|
|
|
|
To generate CSF signature for the fitImage part of flash.bin container, use CST:
|
|
```
|
|
cst -i csf_fit.tmp -o csf_fit.bin
|
|
```
|
|
|
|
Finally, patch the CSF signature into the fitImage right past the IVT:
|
|
```
|
|
dd if=csf_fit.bin of=flash.bin bs=1 seek=${csf_block_offset} conv=notrunc
|
|
```
|
|
|
|
The entire script is available in doc/imx/habv4/csf_examples/mx8m/csf.sh
|
|
|
|
1.4 Closing the device
|
|
-----------------------
|
|
|
|
The procedure for closing the device is similar as in Non-SPL targets, for a
|
|
complete procedure please refer to section "1.5 Programming SRK Hash" in
|
|
mx6_mx7_secure_boot.txt document available under doc/imx/habv4/guides/
|
|
directory.
|
|
|
|
References:
|
|
[1] AN4581: "Secure Boot on i.MX 50, i.MX 53, i.MX 6 and i.MX 7 Series using
|
|
HABv4" - Rev 2.
|