March 2019 - U-Boot

[U-Boot] [RFC PATCH v2 00/12] Add support for applications of overlays in SPL
by Jean-Jacques Hiblot 18 Apr '19

18 Apr '19

The purpose of this series is to provide the SPL with ability to apply overlays for u-boot. this is only a RFC so far, to get a feedback on the approach. Our use-case is the support of the daughter boards of the AM65x EVM. In Linux, each board is supported by a unique overlay. The presence of the boards is detected at runtime, and some useful features (like USB) are implemented on those daughter boards. Instead of providing multiple dtbs and fall in a combinatorial pit, we propose to use DT overlays. Patch #1 "spl: fit: Add support for applying DT overlay" has been posted a few weeks ago by Michal Simek. Patch #2 and #3 amend Michal's patch. Patch #4 and #5 are simple fixes for the Makefile Patch #6 is not required but relates to this series and will be required later by the AM6x platform Patch #7 may be a bit controversial. It basically replaces u-boot.img with a symlink to u-boot.itb in case we use a "complex" FIT (ie: if SPL_FIT_SOURCE or SPL_FIT_GENERATOR are set). This breaks buildman for several platforms because not all the binaries embedded in the FIT are available. Patch #9 adds a way to dynamically select the DT overlays. That is were we would use HW detection to select the required overlays. In that case, the board-level code tells what overlay it needs (it gives the name of the node). Patch #10 disables advanced SPL FIT features to keep the size of the SPL of the am335x_evm in check Patch #11 is not required, but demonstrates on a DRA76-evm how this series can be used. On arm, if overlay are supported, this series increases the size of the SPL by 3-4 kB. Changes in v2: - reworked board_fit_get_additionnal_images() and how it used in spl_fit.c - removed dtbo generation from dtso files and use .dts extension for the overlays - add dynamic allocation usage in a separate patch - defconfig change for the am335x_evm Jean-Jacques Hiblot (11): spl: fit: Make room in the FDT before applying overlays spl: fit: allocate a temporary buffer to load the overlays Makefile.lib: include /__symbols__ in dtb if OF_LIBFDT_OVERLAY is enabled Makefile: Fix tests for CONFIG_SPL_LOAD_FIT and CONFIG_SPL_FIT_GENERATOR Makefile: Fix u-boot.itb generation when building outside the source tree Makefile: Pass the board name to the FIT generator scripts Makefile: use custom ITS to build u-boot.img if SPL_FIT_SOURCE or SPL_FIT_GENERATOR are set spl: fit: constify the output parameter of spl_fit_get_image_name() spl: fit: Allow the board to tell if more images must be loaded from FIT configs: am335x_evm: enable SPL_FIT_IMAGE_TINY !!! TEMP !!! For demonstration only !!! DRA76: Usage of overlays in SPL Michal Simek (1): spl: fit: Add support for applying DT overlay Makefile | 27 +++++-- arch/arm/dts/Makefile | 2 +- arch/arm/dts/dra76-evm-dummy.dts | 14 ++++ arch/arm/dts/dra76-evm-dummy2.dts | 15 ++++ arch/arm/mach-imx/mkimage_fit_atf.sh | 3 +- arch/arm/mach-rockchip/make_fit_atf.py | 5 +- board/sunxi/mksunxi_fit_atf.sh | 2 + .../lion_rk3368/fit_spl_atf.its | 6 +- .../puma_rk3399/fit_spl_atf.its | 8 +- board/ti/dra7xx/evm.c | 30 +++++++ board/ti/dra7xx/evm.its | 48 +++++++++++ common/spl/spl_fit.c | 81 +++++++++++++++++-- configs/am335x_evm_defconfig | 1 + configs/dra7xx_evm_defconfig | 2 + include/spl.h | 16 ++++ scripts/Makefile.lib | 4 + 16 files changed, 239 insertions(+), 25 deletions(-) create mode 100644 arch/arm/dts/dra76-evm-dummy.dts create mode 100644 arch/arm/dts/dra76-evm-dummy2.dts create mode 100644 board/ti/dra7xx/evm.its -- 2.17.1

3 19

[U-Boot] [PATCH 0/8] video/console: Fix various DM_VIDEO console issues
by Andre Przywara 17 Apr '19

17 Apr '19

The graphical console (vidconsole-uclass.c) for DM_VIDEO based display drivers has several issues: - Many ANSI sequences are not handled properly. - The character set used is assumed to be the original IBM PC code page 437, even though the UEFI code expect the terminal to display UTF-8 encoded characters. Truetype fonts expect ISO8859-1. - The USB keyboard does not handle arrow keys correctly. - There is no visible cursor displayed. - Truetype fonts only work in 16-bit screen modes. - The maximum font width supported is 8 pixels. - The "bmp" command to display bitmaps only works in certain video modes. This leads to a very poor user experience, up to a point where applications become unusable (Grub/EFI and U-Boot's bootmenu come to mind). This patch set aims to fix the most important of those issues (the first three). I have further patches to address the rest, but they are not so nice or have issues, so I will send them later. Patch 1/8 is a bug fix (for ASCII characters > 127). Patch 2-4 extend the ANSI sequence handling. Patch 5 and 6 fix the character encoding for the bitmap fonts. The USB keyboard learns about proper arrow key handling in patch 7, while patch 8 fixes a minor Kconfig omission for sunxi. After this series I can use both the bootmenu and Grub/EFI properly. Tested on Sandbox and an Pine64-LTS board. Cheers, Andre. Andre Przywara (8): video/console: Fix DM_VIDEO font glyph array indexing video/console: Implement reverse video ANSI sequence for DM_VIDEO video/console: Implement relative cursor movement ANSI handling video/console: Implement ANSI clear line command video/console: Factor out actual character output video/console: Convert UTF-8 codes to CP437 code points usb: kbd: Properly translate up/down arrow keys sunxi: allow boards to de-select SYS_WHITE_ON_BLACK font scheme common/usb_kbd.c | 24 +++++- drivers/video/Kconfig | 2 +- drivers/video/Makefile | 1 + drivers/video/console_normal.c | 3 +- drivers/video/console_rotate.c | 7 +- drivers/video/utf8_cp437.c | 170 ++++++++++++++++++++++++++++++++++++++ drivers/video/vidconsole-uclass.c | 113 +++++++++++++++++++++---- drivers/video/video-uclass.c | 1 + include/configs/sunxi-common.h | 1 - include/video.h | 2 + include/video_console.h | 9 ++ 11 files changed, 310 insertions(+), 23 deletions(-) create mode 100644 drivers/video/utf8_cp437.c -- 2.14.5

5 33

[U-Boot] [RFC PATCH] common: fdt_support: Remove fdt_add_mem_rsv in fdt_shrink_to_minimum
by Keerthy 16 Apr '19

16 Apr '19

With introduction of commit: a156c47e39ad: tftp: prevent overwriting reserved memory tftp with loadaddr is failing for Images of the size bigger than SPL. The issue is that SPL image is located at bootaddr and dtb is allocated at bootaddr + spl_size. The region where dtb is present is marked reserved. So when the tftp tries to load the Image it tries to find a location from loadaddr to the first reserved memory region which is the dtb base address. The max_size obtained is pretty less and hence tftp fails with the below error: TFTP error: trying to overwrite reserved memory... Signed-off-by: Keerthy <j-keerthy(a)ti.com> --- I am sure there are better solutions so just wanted to know if there is one. common/fdt_support.c | 5 ----- 1 file changed, 5 deletions(-) diff --git a/common/fdt_support.c b/common/fdt_support.c index 42583e3ed8..2bb101e56e 100644 --- a/common/fdt_support.c +++ b/common/fdt_support.c @@ -633,11 +633,6 @@ int fdt_shrink_to_minimum(void *blob, uint extrasize) /* Change the fdt header to reflect the correct size */ fdt_set_totalsize(blob, actualsize); - /* Add the new reservation */ - ret = fdt_add_mem_rsv(blob, map_to_sysmem(blob), actualsize); - if (ret < 0) - return ret; - return actualsize; } -- 2.17.1

4 4

[U-Boot] [RFC v2 00/11] efi_loader: rework bootefi/bootmgr
by AKASHI Takahiro 16 Apr '19

16 Apr '19

There are several reasons that I want to rework/refactor bootefi command as well as bootmgr: * Some previous commits on bootefi.c have made the code complicated and a bit hard to understand. * do_bootefi_exec() would better be implemented using load_image() along with start_image() to be aligned with UEFI interfaces. * Contrary to the other part, efi_selftest part of the code is unusal in terms of loading/execution path in do_bootefi(). * When we will support "secure boot" in the future, EFI Boot Manager is expected to be invoked as a standalone command without any arguments to mitigate security surfaces. In this patch set, Patch#1 is a bug fix. Patch#2 to #9 are preparatory patches for patch#10. Patch#10 is a core part of reworking. Patch#11 is for standalone boot manager. # Please note that some patches, say patch#4 and #5, can be combined into one # but I intentionally keep them separated to clearify my intentions of changes. Prerequsite patch: * "efi_loader: bootmgr: support BootNext and BootCurrent variable behavior" * "efi_loader: release file buffer after loading image" Issues: * load_image() will take an argument of "parent_handle," but obviously we don't have any parent when invoking an application from command line. (See FIXME in patch#10.) * Starting EDK2's Shell.efi from boot manager fails. (I need to find out a fix.) * The semantics of efi_dp_from_name() should be changed. (See FIXME in patch#10.) -Takahiro Akashi Changes in RFC v2 (Mar 27, 2019) * rebased on v2019.04-rc4 * use load_image API in do_bootmgr_load() * merge efi_install_fdt() and efi_process_fdt() * add EFI_LOADED_IMAGE_DEVICE_PATH_PROTOCOL to image (patch#1) * lots of minor changes AKASHI Takahiro (11): efi_loader: boottime: add loaded image device path protocol to image handle efi_loader: boottime: export efi_[un]load_image() efi_loader: device_path: handle special case of loading cmd: bootefi: carve out fdt handling from do_bootefi() cmd: bootefi: merge efi_install_fdt() and efi_process_fdt() cmd: bootefi: carve out efi_selftest code from do_bootefi() cmd: bootefi: move do_bootefi_bootmgr_exec() forward cmd: bootefi: carve out bootmgr code from do_bootefi() cmd: bootefi: carve out do_boot_efi() from do_bootefi() efi_loader: rework bootmgr/bootefi using load_image API cmd: add efibootmgr command cmd/Kconfig | 8 + cmd/bootefi.c | 521 +++++++++++++++++++----------- include/efi_api.h | 4 + include/efi_loader.h | 15 +- lib/efi_loader/efi_bootmgr.c | 43 +-- lib/efi_loader/efi_boottime.c | 34 +- lib/efi_loader/efi_device_path.c | 8 + lib/efi_loader/efi_image_loader.c | 2 + 8 files changed, 411 insertions(+), 224 deletions(-) -- 2.20.1

3 42

[U-Boot] [PATCH v4] cmd: clk: Handle ENODEV from clk_get_rate
by Ismael Luceno Cortes 15 Apr '19

15 Apr '19

clk_get_rate may return -ENODEV if the clock isn't valid. Also, make the error cases go through a single path. Fixes: ff8eee0330a6 ("cmd: clk: Add trivial implementation of clock dump for DM") Signed-off-by: Ismael Luceno <ismael.luceno(a)silicon-gears.com> Reviewed-by: Matthias Brugger <mbrugger(a)suse.com> Reviewed-by: Marek Vasut <marek.vasut(a)gmail.com> --- Notes: Changes since v3: - Moved the "rate" variable declaration to the begining of the function. Changes since v2: - Further improved commit message. Changes since v1: - Added further explanation to the commit message. cmd/clk.c | 24 ++++++++++++++---------- 1 file changed, 14 insertions(+), 10 deletions(-) diff --git a/cmd/clk.c b/cmd/clk.c index fd4231589c..5402c87de7 100644 --- a/cmd/clk.c +++ b/cmd/clk.c @@ -17,6 +17,7 @@ int __weak soc_clk_dump(void) struct uclass *uc; struct clk clk; int ret; + ulong rate; /* Device addresses start at 1 */ ret = uclass_get(UCLASS_CLK, &uc); @@ -26,20 +27,23 @@ int __weak soc_clk_dump(void) uclass_foreach_dev(dev, uc) { memset(&clk, 0, sizeof(clk)); ret = device_probe(dev); - if (ret) { - printf("%-30.30s : ? Hz\n", dev->name); - continue; - } + if (ret) + goto noclk; ret = clk_request(dev, &clk); - if (ret) { - printf("%-30.30s : ? Hz\n", dev->name); - continue; - } - - printf("%-30.30s : %lu Hz\n", dev->name, clk_get_rate(&clk)); + if (ret) + goto noclk; + rate = clk_get_rate(&clk); clk_free(&clk); + + if (rate == -ENODEV) + goto noclk; + + printf("%-30.30s : %lu Hz\n", dev->name, rate); + continue; + noclk: + printf("%-30.30s : ? Hz\n", dev->name); } return 0; base-commit: 9ba5e5bc261a16f51662490da0cf620dc7f29013 -- 2.19.1

2 2

[U-Boot] [PATCH v3 00/13] ARM: tegra: Add support for framebuffer carveouts
by Thierry Reding 15 Apr '19

15 Apr '19

From: Thierry Reding <treding(a)nvidia.com> The series adds a couple of helpers that allow passing framebuffer carveouts to the kernel on boot. This is required in order to reserve the memory region that the framebuffer resides in and that the display controller is scanning out from, so that it is not reused by the kernel for the buddy allocator. At the same time, this information can be used to set up a 1:1 mapping for the IOMMU during boot in order to allow the display controller to keep scanning out from the same address without causing IOMMU faults. Note that the last two patches depend on this series: http://patchwork.ozlabs.org/project/uboot/list/?series=98501 I've included them here to give an example for how they are used. Tom would need to apply these on top of that series once the remainder of the patches here have been merged. Given the build time dependencies it might be best for Tom to pick up the fdtdec pieces as well, with Simon's ack. Or alternatively Simon could merge these through his tree and Tom can apply once the dependencies have gone into master. Tom, Simon, any preferences? Thierry Thierry Reding (13): libfdt: Add phandle generation helper fdtdec: Add cpu_to_fdt_{addr,size}() macros fdtdec: Add fdt_{addr,size}_unpack() helpers fdtdec: Implement fdtdec_set_phandle() fdtdec: Implement fdtdec_add_reserved_memory() fdtdec: Implement carveout support functions fdtdec: Add Kconfig symbol for tests fdtdec: test: Fix build warning fdtdec: test: Use compound statement macros fdtdec: test: Add carveout tests sandbox: Enable fdtdec tests p2371-2180: Add support for framebuffer carveouts p2771-0000: Add support for framebuffer carveouts .../reserved-memory/reserved-memory.txt | 136 +++++++++++ board/nvidia/p2371-2180/p2371-2180.c | 47 ++++ board/nvidia/p2771-0000/p2771-0000.c | 66 ++++- configs/sandbox64_defconfig | 1 + configs/sandbox_defconfig | 1 + include/fdtdec.h | 169 +++++++++++++ lib/Kconfig | 4 + lib/fdtdec.c | 225 ++++++++++++++++++ lib/fdtdec_test.c | 216 +++++++++++++---- lib/libfdt/fdt_ro.c | 31 +++ scripts/dtc/libfdt/fdt_ro.c | 31 +++ scripts/dtc/libfdt/libfdt.h | 19 ++ scripts/dtc/libfdt/libfdt_env.h | 1 + 13 files changed, 902 insertions(+), 45 deletions(-) create mode 100644 Documentation/devicetree/bindings/reserved-memory/reserved-memory.txt -- 2.21.0

3 44

[U-Boot] [PATCH v3 00/13] Add Support for eMMC in AM65x-evm
by Faiz Abbas 15 Apr '19

15 Apr '19

Add Support for eMMC in TI's AM65x-evm. The series starts by syncing the sdhci0 node from the kernel. Then it adds APIs and changes to the driver required for handling the driver's integrated phy. The current maximum supported speed is DDR52. Higher speeds and tuning support will be added in a subsequent series. Support for booting from eMMC will also be added in subsequent patches. Tested on Lokesh's branch[1] as sysfw loading for AM65x is not yet upstream. [1] https://github.com/lokeshvutla/u-boot/tree/devel/am65x-evm-boot Changes in v3: Patch 9: Got rid of variable ret Changes in v2: Patch 9: Fixed return value Patch 12: Fixed spacing Faiz Abbas (13): arm64: dts: k3: Sync sdhci0 node from kernel mmc: am654_mmc: Change driver name mmc: am654_sdhci: Remove quirks regmap: Add API regmap_init_mem_index() regmap: Add support for polling on a register mmc: sdhci: Add support for sdhci-caps-mask mmc: sdhci: Make sdhci_set_clock() non static arm: dts: k3: Add phy specific properties to SD card node mmc: sdhci: Make set_ios_post() return int mmc: am654_sdhci: Add Support for PHY configs: am65x_evm: Enable CONFIG_REGMAP mmc: sdhci: Add support for HOST_CONTROL2 and setting UHS timings mmc: am654_sdhci: Add a platform specific set_control_reg() callback arch/arm/dts/k3-am65-main.dtsi | 22 ++ arch/arm/dts/k3-am654-base-board-u-boot.dtsi | 40 +-- arch/arm/dts/k3-am654-base-board.dts | 28 ++ configs/am65x_evm_a53_defconfig | 4 +- configs/am65x_evm_r5_defconfig | 4 +- drivers/core/regmap.c | 42 +++ drivers/mmc/Kconfig | 9 +- drivers/mmc/Makefile | 2 +- drivers/mmc/am654_sdhci.c | 340 +++++++++++++++++++ drivers/mmc/k3_arsan_sdhci.c | 109 ------ drivers/mmc/sdhci.c | 47 ++- drivers/mmc/xenon_sdhci.c | 4 +- drivers/mmc/zynq_sdhci.c | 31 +- include/regmap.h | 36 ++ include/sdhci.h | 22 +- 15 files changed, 549 insertions(+), 191 deletions(-) create mode 100644 drivers/mmc/am654_sdhci.c delete mode 100644 drivers/mmc/k3_arsan_sdhci.c -- 2.19.2

4 32

[U-Boot] [PATCH 0/3] remove dm_fdt_pre_reloc, replaced by dm_ofnode_pre_reloc
by Patrick Delaunay 12 Apr '19

12 Apr '19

This Serie replace any call to dm_fdt_pre_reloc by dm_ofnode_pre_reloc and remove this function only used 2 times and avoid duplicated code. Patches created after Simon comment on http://patchwork.ozlabs.org/patch/1052734/ >> bool dm_fdt_pre_reloc(const void *blob, int offset) > This function should really be removed. > We should be able to use the ofnode version always. Patrick Delaunay (3): clk: at91: replace dm_fdt_pre_reloc by dm_ofnode_pre_reloc clk: socfpga: replace dm_fdt_pre_reloc by dm_ofnode_pre_reloc dm: remove unused function dm_fdt_pre_reloc drivers/clk/altera/clk-arria10.c | 3 ++- drivers/clk/at91/pmc.c | 2 +- drivers/core/util.c | 24 ------------------------ include/dm/util.h | 26 -------------------------- 4 files changed, 3 insertions(+), 52 deletions(-) -- 2.7.4

3 8

[U-Boot] [PATCH 1/3] vsprintf: Support phys_addr_t specifier unconditionally
by Thierry Reding 12 Apr '19

12 Apr '19

From: Thierry Reding <treding(a)nvidia.com> When phys_addr_t printf specifier support was first introduced in commit 1eebd14b7902 ("vsprintf: Add modifier for phys_addr_t"), it was enabled only if CONFIG_CMD_NET was selected. Since physical addresses are not unique to networking support it doesn't make sense to conditionally add it in those cases only. Move support for it outside of the CMD_NET guard so that the specifier is always supported. Signed-off-by: Thierry Reding <treding(a)nvidia.com> --- lib/vsprintf.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/lib/vsprintf.c b/lib/vsprintf.c index 1b6c154d8d72..2403825dc98d 100644 --- a/lib/vsprintf.c +++ b/lib/vsprintf.c @@ -457,7 +457,6 @@ static char *pointer(const char *fmt, char *buf, char *end, void *ptr, return device_path_string(buf, end, ptr, field_width, precision, flags); #endif -#ifdef CONFIG_CMD_NET case 'a': flags |= SPECIAL | ZEROPAD; @@ -469,6 +468,7 @@ static char *pointer(const char *fmt, char *buf, char *end, void *ptr, break; } break; +#ifdef CONFIG_CMD_NET case 'm': flags |= SPECIAL; /* Fallthrough */ -- 2.20.1

2 5

[U-Boot] [PATCH] ARM: HYP/non-sec: Don't enable ARMV7_LPAE for old sunxi kernels
by Jonathan Liu 12 Apr '19

12 Apr '19

Old sunxi kernels fail to boot with ARMV7_LPAE enabled. Signed-off-by: Jonathan Liu <net147(a)gmail.com> --- arch/arm/cpu/armv7/Kconfig | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/arm/cpu/armv7/Kconfig b/arch/arm/cpu/armv7/Kconfig index 73d57a2aae..cdb7e402b4 100644 --- a/arch/arm/cpu/armv7/Kconfig +++ b/arch/arm/cpu/armv7/Kconfig @@ -53,7 +53,7 @@ config ARMV7_PSCI_NR_CPUS config ARMV7_LPAE bool "Use LPAE page table format" if EXPERT depends on CPU_V7A - default y if ARMV7_VIRT + default y if ARMV7_VIRT && !OLD_SUNXI_KERNEL_COMPAT ---help--- Say Y here to use the long descriptor page table format. This is required if U-Boot runs in HYP mode. -- 2.21.0

4 5