October 2024 - U-Boot

[PATCH v3 0/3] This series adds a few patches to make it easy to use the system dtc in
by Simon Glass 15 Oct '24

15 Oct '24

buildman. It updates CI to do this for world builds, to save time. It also includes an old patch which is now needed again since: e13fcae3fce Revert "buildman: Always use the full path in... Changes in v3: - Drop patches previously applied - Add dtc patches Changes in v2: - Fix 'envronment' typo Simon Glass (3): buildman: Support building within a Python venv buildman: Allow skipping the dtc build RFC: CI: Skip building dtc for world builds .azure-pipelines.yml | 2 +- .gitlab-ci.yml | 8 +-- tools/buildman/bsettings.py | 3 + tools/buildman/builder.py | 27 +++++++- tools/buildman/builderthread.py | 4 +- tools/buildman/buildman.rst | 3 + tools/buildman/cmdline.py | 2 + tools/buildman/control.py | 3 +- tools/buildman/test.py | 115 ++++++++++++++++++++++++++++++++ tools/buildman/toolchain.py | 31 +++++++-- 10 files changed, 185 insertions(+), 13 deletions(-) -- 2.34.1

2 26

[PATCH 1/2] qconfig: Add an way to select defconfigs without stdin
by Simon Glass 14 Oct '24

14 Oct '24

Sometimes it is useful to process just one or two defconfigs and it is convenient to do this just by listing them in the arguments. Add a -D option for this. Update the docs to avoid mentioning boards which have been removed. Signed-off-by: Simon Glass <sjg(a)chromium.org> --- doc/develop/qconfig.rst | 7 +++++-- tools/qconfig.py | 23 +++++++++++++++-------- 2 files changed, 20 insertions(+), 10 deletions(-) diff --git a/doc/develop/qconfig.rst b/doc/develop/qconfig.rst index a18f32470ca..3b995355967 100644 --- a/doc/develop/qconfig.rst +++ b/doc/develop/qconfig.rst @@ -62,10 +62,13 @@ or:: grep -l X86 configs/* | ./tools/qconfig.py -s -d - +or:: + + ./tools/qconfig.py -s -D $(grep -l X86 configs/*) + To process CONFIG_CMD_FPGAD only for a subset of configs based on path match:: - ls configs/{hrcon*,iocon*,strider*} | \ - ./tools/qconfig.py -C CONFIG_CMD_FPGAD -d - + ./tools/qconfig.py -C CONFIG_CMD_FPGAD -D configs/{amd_versal2,am68_sk}* Finding boards with particular CONFIG combinations diff --git a/tools/qconfig.py b/tools/qconfig.py index 058d72cf4bc..c5a33412c8c 100755 --- a/tools/qconfig.py +++ b/tools/qconfig.py @@ -125,23 +125,26 @@ def get_matched_defconfig(line): pattern = os.path.join('configs', line) return glob.glob(pattern) + glob.glob(pattern + '_defconfig') -def get_matched_defconfigs(defconfigs_file): - """Get all the defconfig files that match the patterns in a file. +def get_matched_defconfigs(defconfigs_in): + """Get all the defconfig files that match the patterns given. Args: - defconfigs_file (str): File containing a list of defconfigs to process, - or '-' to read the list from stdin + defconfigs_file (str or list of str): File containing a list of + defconfigs to process, or '-' to read the list from stdin, or a + list of defconfig names Returns: list of str: A list of paths to defconfig files, with no duplicates """ defconfigs = [] with ExitStack() as stack: - if defconfigs_file == '-': + if isinstance(defconfigs_in, list): + inf = defconfigs_in + elif defconfigs_in == '-': inf = sys.stdin - defconfigs_file = 'stdin' + defconfigs_in = 'stdin' else: - inf = stack.enter_context(open(defconfigs_file, encoding='utf-8')) + inf = stack.enter_context(open(defconfigs_in, encoding='utf-8')) for i, line in enumerate(inf): line = line.strip() if not line: @@ -150,7 +153,7 @@ def get_matched_defconfigs(defconfigs_file): line = line.split(' ')[0] # handle 'git log' input matched = get_matched_defconfig(line) if not matched: - print(f"warning: {defconfigs_file}:{i + 1}: no defconfig matched '{line}'", + print(f"warning: {defconfigs_in}:{i + 1}: no defconfig matched '{line}'", file=sys.stderr) defconfigs += matched @@ -738,6 +741,8 @@ def move_config(args): if args.defconfigs: defconfigs = get_matched_defconfigs(args.defconfigs) + elif args.defconfiglist: + defconfigs = get_matched_defconfigs(args.defconfiglist) else: defconfigs = get_all_defconfigs() @@ -1532,6 +1537,8 @@ doc/develop/moveconfig.rst for documentation.''' help='a file containing a list of defconfigs to move, ' "one per line (for example 'snow_defconfig') " "or '-' to read from stdin") + parser.add_argument('-D', '--defconfiglist', type=str, nargs='*', + help='list of defconfigs to move') parser.add_argument('-e', '--exit-on-error', action='store_true', default=False, help='exit immediately on any error') -- 2.43.0

2 8

[PATCH v3] rockchip: board: Increase rng-seed size to make it sufficient for modern Linux
by Alex Shumsky 14 Oct '24

14 Oct '24

Increase rng-seed size to make Linux happy and initialize rng pool instantly. Linux 5.19+ requires 32 bytes of entropy to initialize random pool, but u-boot currently provides only 8 bytes. Linux 5.18 and probably some vesrions before it used to require 64 bytes. Bump min value to 64 bytes to be on a safe side. Boot with 8 byte rng-seed (Linux 6.11): # dmesg | grep crng [ 12.089286] random: crng init done Boot with 32 byte rng-seed (Linux 6.11): # dmesg | grep crng [ 0.000000] random: crng init done Linux source references: https://github.com/torvalds/linux/blob/7234e2ea0edd00bfb6bb2159e55878c19885… https://elixir.bootlin.com/linux/v5.18/source/drivers/char/random.c#L236 https://elixir.bootlin.com/linux/v5.19/source/drivers/char/random.c#L551 Signed-off-by: Alex Shumsky <alexthreed(a)gmail.com> Fixes: d2048bafae40 ("rockchip: board: Add board_rng_seed() for all Rockchip devices") --- Changes in v3: - reword warning - increase default rng_seed_size to support older linux as well - reword commit message Changes in v2: - add env config knob rng_seed_size - add warning for small rng_seed_size - 12-character commit SHA in Fixes arch/arm/mach-rockchip/board.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) diff --git a/arch/arm/mach-rockchip/board.c b/arch/arm/mach-rockchip/board.c index 3fadf7e412..9ff1567fb1 100644 --- a/arch/arm/mach-rockchip/board.c +++ b/arch/arm/mach-rockchip/board.c @@ -472,9 +472,16 @@ __weak int misc_init_r(void) __weak int board_rng_seed(struct abuf *buf) { struct udevice *dev; - size_t len = 0x8; + ulong len = env_get_ulong("rng_seed_size", 10, 64); u64 *data; + if (len < 64) { + // rng_seed_size should be at least 32 bytes for Linux 5.19+, or 64 for + // older Linux'es + log_warning("Value for rng_seed_size too low (%lu) and likely insufficient for the Linux RNG initialization\n", + len); + } + data = malloc(len); if (!data) { printf("Out of memory\n"); -- 2.34.1

3 2

[PATCH v2 0/7] Allow showing the memory map
by Simon Glass 14 Oct '24

14 Oct '24

This little series adds a new 'memmap' command, intended to show the layout of memory within U-Boot and how much memory is available for loading images. Changes in v2: - Split into its own patch - Modify the existing 'meminfo' command instead Simon Glass (7): common: Fix up malloc() comment in reserve_noncached() common: Tidy up how malloc() is inited am65x: Use map_to_sysmem() to convert from pointer global_data: Add some more accessors bootstage: Allow counting memory without strings cmd: Move meminfo command into its own file cmd: Update the meminfo command to show the memory map arch/arm/mach-k3/am65x/am654_init.c | 11 +-- cmd/Kconfig | 1 + cmd/Makefile | 1 + cmd/mem.c | 19 ----- cmd/meminfo.c | 72 ++++++++++++++++ common/board_f.c | 8 +- common/board_r.c | 3 +- common/bootstage.c | 16 ++-- common/dlmalloc.c | 8 +- common/spl/spl.c | 4 +- doc/usage/cmd/meminfo.rst | 126 ++++++++++++++++++++++++++++ doc/usage/index.rst | 1 + include/asm-generic/global_data.h | 30 +++++++ include/bootstage.h | 5 +- include/malloc.h | 8 ++ test/cmd/Makefile | 3 +- test/cmd/meminfo.c | 38 +++++++++ 17 files changed, 308 insertions(+), 46 deletions(-) create mode 100644 cmd/meminfo.c create mode 100644 doc/usage/cmd/meminfo.rst create mode 100644 test/cmd/meminfo.c -- 2.34.1

4 18

[PATCH v2] rockchip: board: Increase rng-seed size to make it sufficient for modern Linux
by Alex Shumsky 14 Oct '24

14 Oct '24

Modern Linux requires 32 byte seed to initialize random pool, but u-boot currently provides only 8 bytes. Increase rng-seed size to make Linux happy and initialize rng pool instantly. Boot with 8 byte rng-seed: # dmesg | grep crng [ 12.089286] random: crng init done Boot with 32 byte rng-seed: # dmesg | grep crng [ 0.000000] random: crng init done https://github.com/torvalds/linux/blob/7234e2ea0edd00bfb6bb2159e55878c19885… Signed-off-by: Alex Shumsky <alexthreed(a)gmail.com> Fixes: d2048bafae40 ("rockchip: board: Add board_rng_seed() for all Rockchip devices") --- Changes in v2: - add env config knob rng_seed_size - 12-character commit SHA in Fixes arch/arm/mach-rockchip/board.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/arch/arm/mach-rockchip/board.c b/arch/arm/mach-rockchip/board.c index 3fadf7e412..f9f0d7214c 100644 --- a/arch/arm/mach-rockchip/board.c +++ b/arch/arm/mach-rockchip/board.c @@ -472,9 +472,15 @@ __weak int misc_init_r(void) __weak int board_rng_seed(struct abuf *buf) { struct udevice *dev; - size_t len = 0x8; + ulong len = env_get_ulong("rng_seed_size", 10, 32); u64 *data; + if (len < 32) { + // rng_seed_size should be 32 bytes for Linux 5.19+, or 64 for older Linux'es + log_warning("Too small rng_seed_size (%lu). It is likely insufficient to init linux crng\n", + len); + } + data = malloc(len); if (!data) { printf("Out of memory\n"); -- 2.34.1

4 6

Mainline kernel bad magic for Pine64-A64-LTS
by Louis Holbrook 14 Oct '24

14 Oct '24

(Also posted in pine64 forum https://forum.pine64.org/showthread.php?tid=19460) In an attempt to troubleshoot failing to get generic alpine linux to run on PA64-2G-LTS, I've tried to reduce the problem to the smallest steps that are legible to me. When booting the kernel, I get "Bad ARM64 Image Magic" I can't figure out what this means, and why it is happening. Sources I have searched seem to be board specific, and found nothing on this board. I've arrived at a point where I cannot find a way to reason further about it, so thus I am reaching out. I am currently loading the kernel and dtb manually at the uboot prompt: setenv bootargs loglevel=7 earlyprintk fatload mmc 0:1 0x42000000 boot/vmlinux fatload mmc 0:1 0x4a000000 boot/u-boot.dtb booti 0x42000000 - 0x4a000000 My current configuration is: - linux 5.14 from mainline (git:7d2a07b769330c34b4deabeed939325c77a7ec2f) - u-boot v2024.10 from https://source.denx.de/u-boot/u-boot (git:f919c3a889f0ec7d63a48b5d0ed064386b0980bd) - dtb from the u-boot build - bl31.bin from PLAT=sun50i_a64 https://git.trustedfirmware.org/TF-A/tru...ware-a.git lts-v2.10.6 (git:f2aebf361050d69176d60016170291efa9e08a68) - scp.bin from https://github.com/crust-firmware/crust 0.6 (git:ffe9f1ac9c675e6e67db9084bd19fbdeffd8e162) - u-boot-sunxi-with-spl.bin written to offset 8k - tried sd and emmc vfat single partition (whole disk) with * boot/vmlinux * boot/u-boot.dtb Files: - uboot defconfig: https://g33k.holbrook.no/c36cab0e6a15127...7b25f71bdd - uboot environment (saved from soc): https://g33k.holbrook.no/1c4d93e6510eb49...2329468b73 - linux defconfig: https://g33k.holbrook.no/5bde4487e333f5a...19beb3382f

3 2

Re: FIT signature security flay
by Sean Anderson 14 Oct '24

14 Oct '24

Hi Lev, On 10/14/24 04:42, Lev R. Oshvang wrote: > Hi Sean, > > I am looking for help with Uboot FIT signatures problem > > > > I started to work with FIT image (u-boot 2024) and managed to sign > kernel and load this image with Uboot using 'required' property in > signature as : > > signature-1 { > > algo = "sha1,rsa2048"; > > key-name-hint = "dev_key"; > > sign-images="kernel"; > > required="conf"; > > { > > Iminfo reports" > > ## Checking hash(es) for FIT Image at 01000000 ... > > Hash(es) for Image 0 (kernel-1): sha256+ sha256,rsa2048:dev_key- > > > > To test the procedure, I generated another private key and signed > another kernel with this new key on another Linux host. > > > > > > I expected bootm to fail, but it just happily loads this image!!! > > Even an image without a signature but with a valid hash is not > rejected against my expectations. > > In this case iminfo report only hash is OK Did you embed the public key into your U-Boot devicetree with `mkimage -K` ? --Sean

2 1

ll_entry_count returns 0 even though the linker list has 5 entries
by Osman 14 Oct '24

14 Oct '24

Hello, After enabling CONFIG_FIT in defconfig the u-boot-spl becomes unable to detect the root_driver in function lists_driver_lookup_name(). The spl executes without issues when manually changing the n_ents variable to 5 while debugging so i am guessing the ll_entry_count() function somehow returns the wrong value but i am not sure why. For reference this is the section in u-boot-spl.map file that contains the "root_driver": __u_boot_list_2_driver_1 0x0000000080009b78 0x0 drivers/core/root.o __u_boot_list_2_driver_2_root_driver 0x0000000080009c68 0x78 drivers/core/root.o 0x0000000080009c68 _u_boot_list_2_driver_2_root_driver __u_boot_list_2_driver_3 0x0000000080009dd0 0x0 drivers/core/root.o (gdb) x /8xw 0x0000000080009c68 0x80009c68 <_u_boot_list_2_driver_2_root_driver>: 0x800091a0 0x00000000 0x00000000 0x00000000 By manually calculating (end-start)/size from the symbol addresses of the .map file i found the correct value to be 5 but ll_entry_count returns 0 while the function lists_driver_lookup_name executes: (gdb) s 27 const int n_ents = ll_entry_count(struct driver, driver); (gdb) s 30 for (entry = drv; entry != drv + n_ents; entry++) { (gdb) info locals drv = 0x80009b78 <_u_boot_list_2_driver_2_cpu_bus> n_ents = 0 entry = 0x80009b78 <_u_boot_list_2_driver_2_cpu_bus>(gdb) frame#0 lists_driver_lookup_name (name=0x800091a0 "root_driver") I am not sure whats causing this but disabling CONFIG_FIT in my defconfig causes the program to execute without issues again. Could someone point out what i am missing here? Regards, Osman

2 1

[PATCH] rockchip: board: Increase rng-seed size to make it sufficient for modern Linux
by Alex Shumsky 14 Oct '24

14 Oct '24

Modern Linux requires 32 byte seed to initialize random pool, but u-boot currently provides only 8 bytes. Increase rng-seed size to make Linux happy and initialize rng pool instantly. Boot with 8 byte rng-seed: # dmesg | grep crng [ 12.089286] random: crng init done Boot with 32 byte rng-seed: # dmesg | grep crng [ 0.000000] random: crng init done https://github.com/torvalds/linux/blob/7234e2ea0edd00bfb6bb2159e55878c19885… Signed-off-by: Alex Shumsky <alexthreed(a)gmail.com> Fixes: d2048ba ("rockchip: board: Add board_rng_seed() for all Rockchip devices") --- arch/arm/mach-rockchip/board.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/arm/mach-rockchip/board.c b/arch/arm/mach-rockchip/board.c index 3fadf7e412..eee75f7bcb 100644 --- a/arch/arm/mach-rockchip/board.c +++ b/arch/arm/mach-rockchip/board.c @@ -472,7 +472,7 @@ __weak int misc_init_r(void) __weak int board_rng_seed(struct abuf *buf) { struct udevice *dev; - size_t len = 0x8; + size_t len = 32; u64 *data; data = malloc(len); -- 2.34.1

4 12

[GIT PULL] Please pull u-boot-ufs-next-20241014
by Neil Armstrong 14 Oct '24

14 Oct '24

Hi Tom, This is my first UFS PR with: - Set of fixes/updates on UFS core for dcache, quirks, memory barriers and startup logic - Add myself to UFS maintainers - Add UFS DesignWare Controller for AMD versal2 platforms - Add UFS Qualcomm controller driver Thanks, Neil The following changes since commit 580fb5773634c4a01f907a164b2a05961840b4c0: Merge tag 'u-boot-imx-master-20241013' of https://gitlab.denx.de/u-boot/custodians/u-boot-imx (2024-10-13 16:25:01 -0600) are available in the Git repository at: https://source.denx.de/u-boot/custodians/u-boot-ufs.git tags/u-boot-ufs-next-20241014 for you to fetch changes up to 0c9c501a87e0070b944ffcd8f0d9942353048f41: ufs: Add Support for Qualcomm UFS HC driver (2024-10-14 08:56:03 +0200) ---------------------------------------------------------------- - Set of fixes/updated on UFS core - Fix dcache/flush - Update & Sync Quirks - Backport fixes from Linux - Add missing memory barriers - Remove link_startup_again logic - Add Neil to UFS maintainers - Add UFS DesignWare Controller for AMD vers2 platforms - Add UFS Qualcomm controller driver ---------------------------------------------------------------- Bhupesh Sharma (6): ufs/ufs.h: Add definition of 'ufshcd_rmwl()' ufs: Clear UECPA once due to LINERESET has happened during LINK_STARTUP ufs: Sync possible UFS Quirks with Linux UFS driver ufs: Add missing memory barriers ufs: Fix debug message in 'ufs_start' ufs: Add Support for Qualcomm UFS HC driver Marek Vasut (2): ufs: Add UFSHCD_QUIRK_BROKEN_64BIT_ADDRESS ufs: Add UFSHCD_QUIRK_HIBERN_FASTAUTO Neil Armstrong (9): ufs: allocate descriptors with size aligned with DMA_MINALIGN ufs: fix dcache flush and invalidate range calculation ufs: split flush and invalidate to only invalidate when required ufs: use dcache helpers for scsi_cmd data and only invalidate if necessary ufs: core: remove link_startup_again logic MAINTAINERS: Add myself to the list of UFS maintainers ufs: add device_reset callback ufs: add get_max_pwr_mode callback ufs: allow UFSHCI version 4.0 Venkatesh Yadav Abbarapu (3): ufs: add support for DesignWare Controller ufs: ufs-amd-versal2: Add support for AMD UFS controller configs: versal2: Add support for AMD UFS platform driver MAINTAINERS | 1 + configs/amd_versal2_virt_defconfig | 2 +- drivers/ufs/Kconfig | 15 + drivers/ufs/Makefile | 2 + drivers/ufs/ufs-amd-versal2.c | 501 +++++++++++++++++++++++++++ drivers/ufs/ufs-qcom.c | 670 +++++++++++++++++++++++++++++++++++++ drivers/ufs/ufs-qcom.h | 147 ++++++++ drivers/ufs/ufs.c | 111 +++--- drivers/ufs/ufs.h | 224 +++++++++++-- drivers/ufs/ufshcd-dwc.c | 133 ++++++++ drivers/ufs/ufshcd-dwc.h | 69 ++++ drivers/ufs/ufshci-dwc.h | 32 ++ drivers/ufs/unipro.h | 7 + 13 files changed, 1850 insertions(+), 64 deletions(-) create mode 100644 drivers/ufs/ufs-amd-versal2.c create mode 100644 drivers/ufs/ufs-qcom.c create mode 100644 drivers/ufs/ufs-qcom.h create mode 100644 drivers/ufs/ufshcd-dwc.c create mode 100644 drivers/ufs/ufshcd-dwc.h create mode 100644 drivers/ufs/ufshci-dwc.h

2 1