04 Dec, 2018
1 commit
-
Add commands to write/read vbmeta public key to/from secure
storage. The vbmeta public key can only be set once.
Comands to set the public key:
fastboot stage
fastboot oem set-public-keyTest: build and boot on imx8qxp_mek.
Change-Id: Id3ad4aa5aacef4fc8443f6a2d6ccb931310970ca
Signed-off-by: Ji Luo
29 Nov, 2018
2 commits
-
now the uboot used by uuu is compiled indepently from the one flashed to
the board, and for imx6 and imx7, defconfig from BSP team is used to
configure the uboot for uuu, so there is no need to unlock the device
when flash the board.Change-Id: Ieefbd67e095dd51eab87e9a32b586cdf88f05d81
Signed-off-by: faqiang.zhu -
to handle situations that it is not allowed to unlock but the board is
already unlocked, unlock logic is modified in this patchChange-Id: I90b22c2dfbc1c64ed9514032986e98b27526c7b9
Signed-off-by: faqiang.zhu
22 Nov, 2018
1 commit
-
Too many macros are used in fsl_avbkey.c and
make it difficult to maintain.
This patch made some refine by:
1. Move all avb/atx operations to fsl_avb.c.
2. Refine the functions logic.
3. Drop some unsupported conditions/functions.Test: build and boot on
imx8qm_mek/imx8mq_evk/imx6qp_sabresd/imx7d_pico/imx8m_aiy.Change-Id: I5c99732acfc47d53cdf188d69223983777e577f4
Signed-off-by: Luo Ji
20 Nov, 2018
1 commit
-
Update the ddrc Qos setting for B1 to align with B0'ssetting.
Correct the initial clock for dram_pll. This setting will be
overwrite before ddr phy training. Although there is no impact
on the dram init, we still need to correct it to eliminate
confusion.Signed-off-by: Bai Ping
Reviewed-by: Ye Li
Tested-by: Robby Cai
(cherry picked from commit 566b798213ab9690966f163de2765acdbfe647a7)
15 Nov, 2018
2 commits
-
Support output/input using `xl console [domid]`.
Signed-off-by: Peng Fan
Reviewed-by: Peng Fan
Reviewed-by: Flynn xu -
Introduce puts hook for dm serial driver.
Signed-off-by: Peng Fan
Reviewed-by: Peng Fan
Reviewed-by: Flynn xu
14 Nov, 2018
1 commit
-
Fix coverity issue CID 43665: Free of address-of expression (BAD_FREE)
incorrect_free: free frees incorrect pointer pp.pp points the port array field of struct ahci_uc_priv, should not free it.
Signed-off-by: Ye Li
Acked-by: Peng Fan
13 Nov, 2018
9 commits
-
Current code uses strlen of string to get the property data
length, which is wrong for unicode string, also the whole
property length also should be corrected(descriptor length
minus head length, 142-10=132), detail data format of single
GUID see below table:Table 4a: Microsoft Extended Properties Feature Descriptor
===================================================================
Value | TYPE | Description
===================================================================
0x8E, 0x00, 0x00, 0x00 | DWORD (LE) | Descriptor length
| | (142 bytes)
--------------------------------------------------------------------
0x00, 0x01 | BCD WORD (LE) | Version ('1.0')
--------------------------------------------------------------------
0x05, 0x00 | WORD (LE) | Extended Property
| | Descriptor index (5)
--------------------------------------------------------------------
0x01, 0x00 | WORD (LE) | Number of sections (1)
--------------------------------------------------------------------
0x84, 0x00, 0x00, 0x00 | DWORD (LE) | Size of the property
| | section (132 bytes)
--------------------------------------------------------------------
0x01, 0x00, 0x00, 0x00 | DWORD (LE) | Property data type
| | (1 = Unicode
| | REG_SZ,
| | see table below)
--------------------------------------------------------------------
0x28, 0x00 | WORD (LE) | Property name
| | length (40 bytes)
--------------------------------------------------------------------
0x44, 0x00, 0x65, 0x00, | NUL-terminated | Property name
(...) | Unicode String | "DeviceInterfaceGUID"
0x74, 0x00, 0x00, 0x00 | (LE) |
--------------------------------------------------------------------
0x4e, 0x00, 0x00, 0x00 | DWORD (LE) | Property data
| | length (78 bytes)
--------------------------------------------------------------------
0x7b, 0x00, 0x46, 0x00, | NUL-terminated | Property name
(...) | Unicode String | "{xxxxxxxx-xxxx-
0x7d, 0x00, 0x00, 0x00 | (LE) | xxxx-xxxx-
| | xxxxxxxxxxxx}\0"
--------------------------------------------------------------------Details of WCID see below link:
https://github.com/pbatard/libwdi/wiki/WCID-DevicesReviewed-by: Ye Li
Signed-off-by: Li Jun -
Fix coverity issue CID 5015449: Dereference before null check (REVERSE_INULL)
check_after_deref: Null-checking mmc suggests that it may be null, but it has
already been dereferenced on all paths leading to the checkSigned-off-by: Ye Li
-
Fix coverity issue CID 1898965: Dereference null return value (NULL_RETURNS)
dereference: Dereferencing a pointer that might be NULL dev_desc when calling
write_backup_gpt_partitionsSigned-off-by: Ye Li
-
Fix coverity issue CID 3826473: Destination buffer too small (STRING_OVERFLOW)
string_overflow: You might overrun the 16-character destination string buffer by
writing 20 characters fromThe size of g_ptable->name is 20, but we use 16 bytes buffer for strcpy. It may
cause buffer overflow.Signed-off-by: Ye Li
-
Fix coverity issue CID 3606685: Structurally dead code (UNREACHABLE)
unreachable: This code cannot be reached: return esdhc_getcd_commonShould not return true directly, otherwise the esdhc_getcd_common is
bypassed.Signed-off-by: Ye Li
-
Fix coverity issue CID 3411367: Missing break in switch (MISSING_BREAK)
unterminated_case: The case for value 1U is not terminated by a 'break' statement.Signed-off-by: Ye Li
-
Fix coverity issue CID 3768406: Dereference null return value (NULL_RETURNS)
dereference: Dereferencing a pointer that might be NULL gis_input when calling strcmp.The gis_input may returns NULL from env_get, need check it before strcmp.
Signed-off-by: Ye Li
-
Fix coverity issue CID 3261683: Wrong operator used (CONSTANT_EXPRESSION_RESULT)
operator_confusion: ({...; __v;}) | 67108864 is always 1/true regardless of the values
of its operand. This occurs as the logical operand of !When DIAG_X is set, the PHY COMINIT signal is detected, so should use '&' to check
whether it is set.Signed-off-by: Ye Li
-
Fix coverity issue CID 3606684: Resource leak (RESOURCE_LEAK)
leaked_storage: Variable uc_priv going out of scope leaks the storage it points toSigned-off-by: Ye Li
12 Nov, 2018
9 commits
-
The rollback index should be updated when avb verify pass
and the slot has been marked as successful, update the
rollback index also for those enabled dual bootloader
feature.
This commit also fix some configs condition issue so
read/write rollback index with trusty will work.Test: rollback index updated successfully on
imx7d_pico and AIY.Change-Id: I2344d6462249d8d88f0622d331cdeffc7e12f885
Signed-off-by: Ji Luo -
SHA256 hash of the entire verified software stack should be calculated
and passed to 'trusty_set_boot_params'. This commit will calculate the
SHA256 hash which represents spl.bin and bootloader.img, and then extend
that hash with the hash of vbmeta image into the final VBH.Bug: 110905171
Test: VBH is calculated and passed ok on AIY.
Change-Id: Id83ad36f3de79bedd435ca8f26035a35cca66b07
Signed-off-by: Ji Luo -
Add support for fastboot variable 'at-vboot-state', it's composed
by 6 sub-variable: 'bootloader-locked', 'bootloader-min-versions',
'avb-perm-attr-set', 'avb-locked', 'avb-unlock-disabled' and
'avb-min-versions'.Test: All 'at-vboot-state' variables are returned
correctly on imx7d_pico and AIY.Change-Id: Ibb855cbcc7c41657af62dafb98a96c4dfb96ef22
Signed-off-by: Ji Luo -
Device will be locked permanently after disabling the unlcok vboot, store
the disable unlock vboot status into fuse. Since the fuse write operation
is irreversible so config 'CONFIG_AVB_FUSE' is disabled by default, user
need to add this config manually and run this command again.Test: Disable unlock vboot bit is set after enabling "CONFIG_AVB_FUSE",
device was locked permanently after running this command. This is
verified on both imx7d_pico and AIY.Change-Id: Iad8991a238763b1d662e33cba65f0b9eb44e97ef
Signed-off-by: Ji Luo -
Supoort "fastboot oem at-lock-vboot" command for Android
Things, this command can only be called after perm-attr
have been fused.Test: build and boot ok on imx7d_pico and AIY.
Change-Id: Ifcfeb2a38d88c5d12b46a1d9ea61b182ae2e7bcb
Signed-off-by: Ji Luo -
Add fastboot commands "fastboot oem at-get-vboot-unlock-challenge"
and "fastboot oem at-unlock-vboot" to support the authenticated
unlock feature for Android Things devices. Use software random
numbers generator to generate the 16 bytes random challenge, it
should be replaced with hardware encrypted random generator when
the TEE part is ready.Test: Generate unlock challenge by:
./avbtool make_atx_unlock_credential
--output=atx_unlock_credential.bin
--intermediate_key_certificate=atx_pik_certificate.bin
--unlock_key_certificate=atx_puk_certificate.bin
--challenge=my_generated_challenge.bin
--unlock_key=testkey_atx_puk.pem
validated the unlock credential successfully on imx7d_pico
and AIY.Change-Id: I4b8cee87c9e96924169479b65020a081136681f6
Signed-off-by: Ji Luo -
Per security requirement, attestation of keymaster
supported by Trusty OS should support both P256
and curve25519 algorithm for Diffie-Hellman.TEST: Works on AIY boards by below commands:
$fastboot getvar at-attest-dhChange-Id: I3244f8d8b677222463b4e8fc75412e54dadeb23e
Signed-off-by: Haoran.Wang -
Flashing a signed vbmeta image adds extra
parameters to the kernel command line, overflowing
the buffer and corrupting the global data struct.
Increase the command line buffer size from 512
to 1024 to accomodate the extra data, and change
the sprintf's to snprintf's to prevent this from
happening if more parameters are added later.This bug was exposed by
Idd55dde79eed793dccdd7319600fbd04e11ca12d.Bug: 112397808
Test: Device boots with images from console and
from local build.Change-Id: If73d04007d54193527e11e11c6ef326110e899b6
-
Change-Id: I5969217e400ab494f9a74662d1d228fcf2e2d465
07 Nov, 2018
1 commit
-
When booting from NAND/SPINOR/WEIMNOR boot devices, the fastboot will
print "unsupported boot device". This warning is used by android fastboot
when setting its "bootcmd". Since android does not support these devices.
so it gives the warning correctly.
But for BSP normal boot, this warning will bring confuse to users. So
change to check the "bootcmd" before giving such warning.Signed-off-by: Ye Li
Reviewed-by: Peng Fan
03 Nov, 2018
5 commits
-
Sometimes, SPL need to pass the trained FSP drate to ATF
if DDR PHY bypass mode is not enabled. So add a fsp_table
to pass these info to ATF. additionally, add more clock
frequency point config to support for code reuse for i.MX8MQ.Signed-off-by: Bai Ping
-
On iMX8 platforms like 8QM/QXP, we must power up the USB PHY resource
before accessing the PHY. However, current init flow access the USB PHY
in ehci_mx6_common_init prior than ehci_usb_phy_mode where the PHY is power
up.Fix the issue by adding ehci_get_usb_phy function to parse the PHY address
from DTB and power up the PHY before ehci_mx6_common_init.Signed-off-by: Ye Li
-
Type "ulong" will be 32 bits long on imx6/7 but we may have
userdate partition more than 4G on 32 bits platforms. Use
"uint64_t" instead of "ulong" here to fix this issue.Test: Get correct userdata partition size on both imx6dl
and imx8qxp.Change-Id: Ia6b242f3998a65b157737e83da0ad3126b689713
Signed-off-by: Ji Luo -
When building mxc_spi driver on ARM64 platforms like iMX8MM, get
such build warnings below. Fix it in this patch.In file included from include/common.h:48:0,
from drivers/spi/mxc_spi.c:9:
drivers/spi/mxc_spi.c: In function ‘spi_xchg_single’:
drivers/spi/mxc_spi.c:232:21: warning: cast from pointer to integer of
different size [-Wpointer-to-int-cast]
_func_, bitlen, (u32)dout, (u32)din);
^
include/log.h:135:26: note: in definition of macro ‘debug_cond’
printf(pr_fmt(fmt), ##args); \
^~~~
drivers/spi/mxc_spi.c:231:2: note: in expansion of macro ‘debug’
debug("%s: bitlen %d dout 0x%x din 0x%x\n",
^~~~~
drivers/spi/mxc_spi.c:232:32: warning: cast from pointer to integer of
different size [-Wpointer-to-int-cast]
_func_, bitlen, (u32)dout, (u32)din);
^
include/log.h:135:26: note: in definition of macro ‘debug_cond’
printf(pr_fmt(fmt), ##args); \
^~~~
drivers/spi/mxc_spi.c:231:2: note: in expansion of macro ‘debug’
debug("%s: bitlen %d dout 0x%x din 0x%x\n",
^~~~~Signed-off-by: Ye Li
Reviewed-by: Peng Fan -
iMX8MM DDR3L validation board uses GD25LQ16, but its id is not in
u-boot flash ids table. Add the new id and parameters into the table.Signed-off-by: Ye Li
29 Oct, 2018
1 commit
-
In flexspi driver, each sf command will disable the module in release bus
function. So reading from flexspi memory-map address using "md" command
can't work. When iMX8MM kicks M4 image to run flexspi NOR XIP,
this causes problem.Signed-off-by: Ye Li
26 Oct, 2018
1 commit
-
Currently the clocks and power of USB controller and USB PHY are both
controlled by ehci-mx6 driver in device probe. However, the function
"ehci_usb_ofdata_to_platdata" calls "ehci_usb_phy_mode"
to access PHY registers when "dr_mode" is set to OTG, both "dr_mode" and
"extcon" properties are not set in DTB. This may cause hang at accessing
USB PHY registers if the power and clocks are not enabled.Change the usb type logic to more clear way:
1. plat->init_type: The requested USB mode type from uplayers
2. priv->init_type: The USB mode type specified by DTB or by the USB ID pin or
by external controller like tcpc or GPIO.
3. If two init_type are not same, return failure. Align with non-DM driver.
4. USB PHY access is moved after power and clock enabled.Signed-off-by: Ye Li
24 Oct, 2018
1 commit
-
Found the imx8mq Rev A chip (B0 and B1 chips are ok) boot hang at CAAM RNG init.
The jobring 0 can't complete instantiation descriptor and spins on checking ORSFR_JR0.In current implementation, the descriptor and jobring input and output base address locate
on TCM, because the driver uses raw_data array in jr_data_st structure as the buffer.
This seems cause the issue. If switched from TCM to OCRAM, the issue will go.Since accessing TCM by CAAM is not very reliable. Add this patch to use OCRAM for SPL case.
The early malloc is ready on SPL before calling board_init_f. So we can use malloc to allocate
memory instead of the raw_data array.Signed-off-by: Ye Li
Reviewed-by: Peng Fan
12 Oct, 2018
2 commits
-
Update the refresh_mode setting. Clear the RFSHCTL3.refresh_mode bit
to set it to normal_mode.Signed-off-by: Bai Ping
Reviewed-by: Ye Li -
Add fastboot command "fastboot oem set-rpmb-key" to program the rpmb
key which should be staged first.
Usage:
1. fastboot stage my-rpmb-key.bin
2. fastboot oem set-rpmb-keyTest: rpmb key programed successfully on imx8qxp.
Change-Id: I95474a6367eb8ef0db16bb38680975b8c45b84f1
Signed-off-by: Ji Luo
09 Oct, 2018
1 commit
-
The current keep-alive check mechanism uses a static variable that is
initialized to 0. When the function is first called, it may happen to
catch the 8-bit keep-alive counter right when it overflows, hence
returning BUSY.This patch will keep checking the counter for 10us, every 1us, but it
will immediately return if the keep-alive counter changed.Signed-off-by: Laurentiu Palcu
01 Oct, 2018
2 commits
-
Optimize the DDR4 init flow. Split the common flow
with the DDR specific timing config. So the common
flow can be reused.Signed-off-by: Bai Ping
-
For LPDDR4 or DDR4, the ddr phy train flow is the same.
So rename the 'lpddr4_ddrphy_train.c' to 'ddrphy_train.c'.
make it more common for reuse and move it to driver/ddr/imx8m/.Signed-off-by: Bai Ping