September 2018 - U-Boot

[U-Boot] [Patch v3] socfpga: stratix10: fix sdram_calculate_size
by Dalon Westergreen 11 Sep '18

11 Sep '18

Incorrect type of size variable results in 0 being returned for sdram sizes greater than or equal to 4GB. Signed-off-by: Dalon Westergreen <dwesterg(a)gmail.com> --- v3: Fix commit message v2: Move function type to phys_size_t to match gd->ram_size --- arch/arm/mach-socfpga/include/mach/sdram_s10.h | 2 +- drivers/ddr/altera/sdram_s10.c | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/arch/arm/mach-socfpga/include/mach/sdram_s10.h b/arch/arm/mach-socfpga/include/mach/sdram_s10.h index 91bfc0e5ec..ca68594445 100644 --- a/arch/arm/mach-socfpga/include/mach/sdram_s10.h +++ b/arch/arm/mach-socfpga/include/mach/sdram_s10.h @@ -7,7 +7,7 @@ #ifndef _SDRAM_S10_H_ #define _SDRAM_S10_H_ -unsigned long sdram_calculate_size(void); +phys_size_t sdram_calculate_size(void); int sdram_mmr_init_full(unsigned int sdr_phy_reg); int sdram_calibration_full(void); diff --git a/drivers/ddr/altera/sdram_s10.c b/drivers/ddr/altera/sdram_s10.c index 48f4f47b14..a48567c109 100644 --- a/drivers/ddr/altera/sdram_s10.c +++ b/drivers/ddr/altera/sdram_s10.c @@ -371,11 +371,11 @@ int sdram_mmr_init_full(unsigned int unused) * Calculate SDRAM device size based on SDRAM controller parameters. * Size is specified in bytes. */ -unsigned long sdram_calculate_size(void) +phys_size_t sdram_calculate_size(void) { u32 dramaddrw = hmc_readl(DRAMADDRW); - u32 size = 1 << (DRAMADDRW_CFG_CS_ADDR_WIDTH(dramaddrw) + + phys_size_t size = 1 << (DRAMADDRW_CFG_CS_ADDR_WIDTH(dramaddrw) + DRAMADDRW_CFG_BANK_GRP_ADDR_WIDTH(dramaddrw) + DRAMADDRW_CFG_BANK_ADDR_WIDTH(dramaddrw) + DRAMADDRW_CFG_ROW_ADDR_WIDTH(dramaddrw) + -- 2.17.1

2 1

[U-Boot] [PATCH v1 0/4] MIPS: optimise cache initialization
by Daniel Schwierzeck 11 Sep '18

11 Sep '18

Do some minor cleanups. Improve the code for changing the k0 CCA. Make the index base address for cache initialisation configurable. Some changes are insprired by the book "MD0901 Boot-MIPS" [1]. Also this is partly a preparation for the upcoming support for Mediatek MT76xx platform. [1] https://training.mips.com/cps_mips/Examples/MD00901_Boot-MIPS_Example_Boot_… Daniel Schwierzeck (4): MIPS: cache: reimplement dcache_[status,enable,disable] MIPS: cache: optimise changing of k0 CCA mode MIPS: cache: make index base address configurable MIPS: cache: remove config option CONFIG_SYS_MIPS_CACHE_MODE README | 14 ---- arch/mips/Kconfig | 12 ++++ arch/mips/lib/cache.c | 20 ++++++ arch/mips/lib/cache_init.S | 118 ++++++++++--------------------- include/configs/imgtec_xilfpga.h | 3 - include/configs/pic32mzdask.h | 3 - scripts/config_whitelist.txt | 1 - 7 files changed, 70 insertions(+), 101 deletions(-) -- 2.18.0

2 8

[U-Boot] [PATCH] socfpga: stratix10: fix sdram_calculate_size
by Dalon Westergreen 11 Sep '18

11 Sep '18

Incorrect type of size variable results in 0 being returned for sdram sizes greater than or equal to 4GB. Signed-off-by: Dalon Westergreen <dwesterg(a)gmail.com> --- drivers/ddr/altera/sdram_s10.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/ddr/altera/sdram_s10.c b/drivers/ddr/altera/sdram_s10.c index 48f4f47b14..dad0147b64 100644 --- a/drivers/ddr/altera/sdram_s10.c +++ b/drivers/ddr/altera/sdram_s10.c @@ -375,7 +375,7 @@ unsigned long sdram_calculate_size(void) { u32 dramaddrw = hmc_readl(DRAMADDRW); - u32 size = 1 << (DRAMADDRW_CFG_CS_ADDR_WIDTH(dramaddrw) + + unsigned long size = 1 << (DRAMADDRW_CFG_CS_ADDR_WIDTH(dramaddrw) + DRAMADDRW_CFG_BANK_GRP_ADDR_WIDTH(dramaddrw) + DRAMADDRW_CFG_BANK_ADDR_WIDTH(dramaddrw) + DRAMADDRW_CFG_ROW_ADDR_WIDTH(dramaddrw) + -- 2.17.1

3 5

[U-Boot] [Patch v2] socfpga: stratix10: fix sdram_calculate_size
by Dalon Westergreen 11 Sep '18

11 Sep '18

Incorrect type of size variable results in 0 being returned for sdram sizes greater than or equal to 4GB. v2: -> Move function type to phys_size_t to match gd->ram_size Signed-off-by: Dalon Westergreen <dwesterg(a)gmail.com> --- arch/arm/mach-socfpga/include/mach/sdram_s10.h | 2 +- drivers/ddr/altera/sdram_s10.c | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/arch/arm/mach-socfpga/include/mach/sdram_s10.h b/arch/arm/mach-socfpga/include/mach/sdram_s10.h index 91bfc0e5ec..ca68594445 100644 --- a/arch/arm/mach-socfpga/include/mach/sdram_s10.h +++ b/arch/arm/mach-socfpga/include/mach/sdram_s10.h @@ -7,7 +7,7 @@ #ifndef _SDRAM_S10_H_ #define _SDRAM_S10_H_ -unsigned long sdram_calculate_size(void); +phys_size_t sdram_calculate_size(void); int sdram_mmr_init_full(unsigned int sdr_phy_reg); int sdram_calibration_full(void); diff --git a/drivers/ddr/altera/sdram_s10.c b/drivers/ddr/altera/sdram_s10.c index 48f4f47b14..a48567c109 100644 --- a/drivers/ddr/altera/sdram_s10.c +++ b/drivers/ddr/altera/sdram_s10.c @@ -371,11 +371,11 @@ int sdram_mmr_init_full(unsigned int unused) * Calculate SDRAM device size based on SDRAM controller parameters. * Size is specified in bytes. */ -unsigned long sdram_calculate_size(void) +phys_size_t sdram_calculate_size(void) { u32 dramaddrw = hmc_readl(DRAMADDRW); - u32 size = 1 << (DRAMADDRW_CFG_CS_ADDR_WIDTH(dramaddrw) + + phys_size_t size = 1 << (DRAMADDRW_CFG_CS_ADDR_WIDTH(dramaddrw) + DRAMADDRW_CFG_BANK_GRP_ADDR_WIDTH(dramaddrw) + DRAMADDRW_CFG_BANK_ADDR_WIDTH(dramaddrw) + DRAMADDRW_CFG_ROW_ADDR_WIDTH(dramaddrw) + -- 2.17.1

2 1

[U-Boot] [PATCH] cmd: part: use MAX_SEARCH_PARTITIONS for part search
by Kever Yang 11 Sep '18

11 Sep '18

Use Macro instead of hard code. Signed-off-by: Kever Yang <kever.yang(a)rock-chips.com> --- cmd/part.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/cmd/part.c b/cmd/part.c index bee204fff0..bfb6488b0f 100644 --- a/cmd/part.c +++ b/cmd/part.c @@ -90,7 +90,7 @@ static int do_part_list(int argc, char * const argv[]) char str[512] = { '\0', }; disk_partition_t info; - for (p = 1; p < 128; p++) { + for (p = 1; p < MAX_SEARCH_PARTITIONS; p++) { char t[5]; int r = part_get_info(desc, p, &info); -- 2.18.0

3 2

[U-Boot] [PATCH] arm: armv7m: clean up armv7m unified code compilation
by Vikas Manocha 11 Sep '18

11 Sep '18

unified syntax should be selected by config ARM_ASM_UNIFIED Signed-off-by: Vikas Manocha <vikas.manocha(a)st.com> --- arch/arm/include/asm/armv7m.h | 5 ----- arch/arm/lib/crt0.S | 4 +--- arch/arm/lib/relocate.S | 1 + arch/arm/lib/vectors_m.S | 2 +- 4 files changed, 3 insertions(+), 9 deletions(-) diff --git a/arch/arm/include/asm/armv7m.h b/arch/arm/include/asm/armv7m.h index 278f302..ad67b4f 100644 --- a/arch/arm/include/asm/armv7m.h +++ b/arch/arm/include/asm/armv7m.h @@ -10,11 +10,6 @@ #ifndef ARMV7M_H #define ARMV7M_H -#if defined(__ASSEMBLY__) -.syntax unified -.thumb -#endif - /* armv7m fixed base addresses */ #define V7M_SCS_BASE 0xE000E000 #define V7M_NVIC_BASE (V7M_SCS_BASE + 0x0100) diff --git a/arch/arm/lib/crt0.S b/arch/arm/lib/crt0.S index d7ff9f0..fe312db 100644 --- a/arch/arm/lib/crt0.S +++ b/arch/arm/lib/crt0.S @@ -8,9 +8,7 @@ #include <config.h> #include <asm-offsets.h> #include <linux/linkage.h> -#ifdef CONFIG_CPU_V7M -#include <asm/armv7m.h> -#endif +#include <asm/assembler.h> /* * This file handles the target-independent stages of the U-Boot diff --git a/arch/arm/lib/relocate.S b/arch/arm/lib/relocate.S index c5b135d..e5f7267 100644 --- a/arch/arm/lib/relocate.S +++ b/arch/arm/lib/relocate.S @@ -6,6 +6,7 @@ */ #include <asm-offsets.h> +#include <asm/assembler.h> #include <config.h> #include <elf.h> #include <linux/linkage.h> diff --git a/arch/arm/lib/vectors_m.S b/arch/arm/lib/vectors_m.S index d75e477..7d2d55c 100644 --- a/arch/arm/lib/vectors_m.S +++ b/arch/arm/lib/vectors_m.S @@ -5,7 +5,7 @@ */ #include <config.h> -#include <asm/armv7m.h> +#include <asm/assembler.h> #include <linux/linkage.h> .type __hard_fault_entry, %function -- 2.7.4

2 1

[U-Boot] [PATCH] arm: armv7m: remove un-necessary If then instruction
by Vikas Manocha 11 Sep '18

11 Sep '18

With gas option -mimplicit-it=always, IT block is inserted by the assembler for thumb2. Signed-off-by: Vikas Manocha <vikas.manocha(a)st.com> --- arch/arm/lib/crt0.S | 3 --- 1 file changed, 3 deletions(-) diff --git a/arch/arm/lib/crt0.S b/arch/arm/lib/crt0.S index 0decce2..d7ff9f0 100644 --- a/arch/arm/lib/crt0.S +++ b/arch/arm/lib/crt0.S @@ -139,9 +139,6 @@ here: mov r2, #0x00000000 /* prepare zero to clear BSS */ clbss_l:cmp r0, r1 /* while not at end of BSS */ -#if defined(CONFIG_CPU_V7M) - itt lo -#endif strlo r2, [r0] /* clear 32-bit BSS word */ addlo r0, r0, #4 /* move to next */ blo clbss_l -- 2.7.4

2 1

[U-Boot] [PATCH] arm: stm32: Remove redundant thumb build selection
by Vikas Manocha 11 Sep '18

11 Sep '18

All armv7m arch builds are thumb & SYS_THUMB_BUILD is already selected by CPU_ARMV7M. Signed-off-by: Vikas Manocha <vikas.manocha(a)st.com> --- arch/arm/Kconfig | 1 - 1 file changed, 1 deletion(-) diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig index 8a23c76..b711605 100644 --- a/arch/arm/Kconfig +++ b/arch/arm/Kconfig @@ -1266,7 +1266,6 @@ config STM32 select CPU_V7M select DM select DM_SERIAL - select SYS_THUMB_BUILD imply CMD_DM config ARCH_STI -- 2.7.4

2 1

[U-Boot] [PATCH] arm: spear: enable GPIO3 and 4 clocks when the GPIO controller driver is built
by Quentin Schulz 11 Sep '18

11 Sep '18

There are multiple GPIOs that can be used with the same driver depending on the CONFIG_GPIO_BASE defined in the controller driver. GPIO3 and 4 require a clock to be enabled so let's enable them when the driver is built. Signed-off-by: Quentin Schulz <quentin.schulz(a)bootlin.com> --- arch/arm/cpu/arm926ejs/spear/cpu.c | 3 +++ arch/arm/include/asm/arch-spear/spr_misc.h | 1 + 2 files changed, 4 insertions(+) diff --git a/arch/arm/cpu/arm926ejs/spear/cpu.c b/arch/arm/cpu/arm926ejs/spear/cpu.c index 0970aca2ef..6fc29e77ff 100644 --- a/arch/arm/cpu/arm926ejs/spear/cpu.c +++ b/arch/arm/cpu/arm926ejs/spear/cpu.c @@ -49,6 +49,9 @@ int arch_cpu_init(void) #if defined(CONFIG_USB_EHCI_SPEAR) periph1_clken |= PERIPH_USBH1 | PERIPH_USBH2; #endif +#if defined(CONFIG_SPEAR_GPIO) + periph1_clken |= MISC_GPIO3ENB | MISC_GPIO4ENB; +#endif writel(periph1_clken, &misc_p->periph1_clken); diff --git a/arch/arm/include/asm/arch-spear/spr_misc.h b/arch/arm/include/asm/arch-spear/spr_misc.h index c906730699..65063fca51 100644 --- a/arch/arm/include/asm/arch-spear/spr_misc.h +++ b/arch/arm/include/asm/arch-spear/spr_misc.h @@ -144,6 +144,7 @@ struct misc_regs { #define MISC_USBDENB 0x01000000 #define MISC_ETHENB 0x00800000 #define MISC_SMIENB 0x00200000 +#define MISC_GPIO3ENB 0x00040000 #define MISC_GPT3ENB 0x00010000 #define MISC_GPIO4ENB 0x00002000 #define MISC_GPT2ENB 0x00000800 -- 2.17.1

2 1

[U-Boot] mkimage (kwbimage) unmapping more bytes than it maps
by Chris Packham 11 Sep '18

11 Sep '18

Hi, Mark and I have been looking at a problem we've been experiencing on our build system. We're running a number of similarly configured makers and seemingly randomly one will fail to build one of our kirkwood based targets (specifically the SBx81LIFXCAT that was recently added) with a segfault in mkimage. When I tried to run the same build it would work fine on my PC (in theory the same base install as the makers). I'd been blaming specific makers as having bad RAM (these things get hammered so we've had them die before) but Mark managed to track it down to mmap and munmap disagreeing on the size of the area mapped. You can see this by running strace on mkimage strace -y ./tools/mkimage -n ./arch/arm/mach-mvebu/kwbimage.cfg -T kwbimage -a 0x00800000 -e 0x00800000 -d u-boot.img u-boot-spl.kwb ... mmap(NULL, 641024, PROT_READ|PROT_WRITE, MAP_SHARED, 3, 0) = 0x7f1b5ce3e000 ... munmap(0x7f1b5ce3e000, 641028) = 0 Note that the size has been incremented by 4. Most of the time this is harmless but on our build system the size of the image worked out to an exact multiple of the page size so the munmap ended up spilling over and unmapping another page which depending on the in-memory layout of the process could result in a segfault. This extra 4 bytes comes from kwbimage.c. Towards the end of kwbimage_set_header() we have checksum = cpu_to_le32(image_checksum32((uint32_t *)ptr, sbuf->st_size)); size = write(ifd, &checksum, sizeof(uint32_t)); ... sbuf->st_size += sizeof(uint32_t); The st_size increment seems like a leftover from commit 4acd2d24b651 ("tools: kwbimage: Add image version 1 support for Armada XP / 370") as aside from the munmap in mkimage nothing else uses it. This is where Mark and I disagree. A minimal change would simply to remove the increment of st_size from kwbimage.c. But a more robust fix would be to have mkimage not trust sbuf after it's been passed to tparams->set_header. Is there are preference for how to handle this? I'll follow up this mail with Marks proposed fix.

2 2