February 2017 - U-Boot

[U-Boot] [RFC PATCH] armv8: cache: Switch MMU table without turning off MMU
by York Sun 10 Feb '17

10 Feb '17

We don't have to completely turn off MMU and cache to switch to another MMU table as far as the data is coherent before and after the switching. This patch relaxes the procedure. Signed-off-by: York Sun <york.sun(a)nxp.com> CC: Alexander Graf <agraf(a)suse.de> --- I found this issue while trying to change MMU table for a SPL boot. The RAM version U-Boot was copied to memory when d-cache was enabled. SPL code never flushed the cache (I will send another patch to fix that). With existing code, U-Boot stops running as soon as the MMU is diabled. With below propsed change, U-Boot continues to run. I have been in contact with ARM support and got very useful information. However, this switching TTBR method is "behavious that should work" but not well verified by ARM. During my debugging, I found other minor issue which convinced me this code wasn't exercised. I don't intend to use this method in long term, but figure it may help others by fixing it. arch/arm/cpu/armv8/cache.S | 44 +++++++++----------------------------------- 1 file changed, 9 insertions(+), 35 deletions(-) diff --git a/arch/arm/cpu/armv8/cache.S b/arch/arm/cpu/armv8/cache.S index f1deaa7..63fb112 100644 --- a/arch/arm/cpu/armv8/cache.S +++ b/arch/arm/cpu/armv8/cache.S @@ -171,35 +171,10 @@ ENDPROC(__asm_invalidate_l3_icache) /* * void __asm_switch_ttbr(ulong new_ttbr) * - * Safely switches to a new page table. + * Switches to a new page table. Cache coherency must be maintained + * before calling this function. */ ENTRY(__asm_switch_ttbr) - /* x2 = SCTLR (alive throghout the function) */ - switch_el x4, 3f, 2f, 1f -3: mrs x2, sctlr_el3 - b 0f -2: mrs x2, sctlr_el2 - b 0f -1: mrs x2, sctlr_el1 -0: - - /* Unset CR_M | CR_C | CR_I from SCTLR to disable all caches */ - movn x1, #(CR_M | CR_C | CR_I) - and x1, x2, x1 - switch_el x4, 3f, 2f, 1f -3: msr sctlr_el3, x1 - b 0f -2: msr sctlr_el2, x1 - b 0f -1: msr sctlr_el1, x1 -0: isb - - /* This call only clobbers x30 (lr) and x9 (unused) */ - mov x3, x30 - bl __asm_invalidate_tlb_all - - /* From here on we're running safely with caches disabled */ - /* Set TTBR to our first argument */ switch_el x4, 3f, 2f, 1f 3: msr ttbr0_el3, x0 @@ -209,14 +184,13 @@ ENTRY(__asm_switch_ttbr) 1: msr ttbr0_el1, x0 0: isb - /* Restore original SCTLR and thus enable caches again */ - switch_el x4, 3f, 2f, 1f -3: msr sctlr_el3, x2 - b 0f -2: msr sctlr_el2, x2 - b 0f -1: msr sctlr_el1, x2 -0: isb + /* invalidate i-cache */ + ic ialluis + isb sy + + /* This call only clobbers x30 (lr) and x9 (unused) */ + mov x3, x30 + bl __asm_invalidate_tlb_all ret x3 ENDPROC(__asm_switch_ttbr) -- 2.7.4

4 3

[U-Boot] [PATCH v2 0/4] Allow disabling non-FIT image loading from SPL
by Andrew F. Davis 10 Feb '17

10 Feb '17

Hello all, To address a needed feature brought up by Andreas[0], we need a way to disable SPL from loading non-FIT images. The function spl_parse_image_header is common to all SPL loading paths (common/spl/spl_(nand|net|nor|etc..)) so we add the check here, much like the existing CONFIG_SPL_ABORT_ON_RAW_IMAGE. My original attempt was to add CONFIG_SPL_PANIC_ON_MKIMAGE, but then if other formats are added, flaws in restricting image types may be introduced, so we would like a single option to restrict all non-FIT types vs disabling types individually. Thanks, Andrew [0] https://www.mail-archive.com/u-boot@lists.denx.de/msg219253.html Changes from v1: - Changed to abort on non-FIT so other boot methods can be attempted. - Made this a default if SPL_FIT_SIGNATURE is selected. Andrew F. Davis (4): Introduce CONFIG_SPL_ABORT_ON_NON_FIT_IMAGE ARM: AM57xx: Disable non-FIT based image loading for HS devices ARM: AM437x: Disable non-FIT based image loading for HS devices ARM: DRA7xx: Disable non-FIT based image loading for HS devices Kconfig | 9 +++++++++ common/spl/spl.c | 5 +++++ configs/am43xx_hs_evm_defconfig | 1 + configs/am57xx_hs_evm_defconfig | 1 + configs/dra7xx_hs_evm_defconfig | 1 + 5 files changed, 17 insertions(+) -- 2.10.1

2 12

[U-Boot] [PATCH] armv8: Fix MMU table page descriptor
by York Sun 10 Feb '17

10 Feb '17

Page descriptor uses different format than block descriptor. Both bit 1 and 0 must be set to indicate a valid page. Signed-off-by: York Sun <york.sun(a)nxp.com> CC: Alexander Graf <agraf(a)suse.de> --- This issue didn't reveal itself unless we use MMU tables down to page size. arch/arm/cpu/armv8/cache_v8.c | 2 +- arch/arm/include/asm/armv8/mmu.h | 1 + 2 files changed, 2 insertions(+), 1 deletion(-) diff --git a/arch/arm/cpu/armv8/cache_v8.c b/arch/arm/cpu/armv8/cache_v8.c index 6c5630c..afa76c1 100644 --- a/arch/arm/cpu/armv8/cache_v8.c +++ b/arch/arm/cpu/armv8/cache_v8.c @@ -230,7 +230,7 @@ static void add_map(struct mm_region *map) /* Page fits, create block PTE */ debug("Setting PTE %p to block virt=%llx\n", pte, virt); - *pte = phys | attrs; + *pte = phys | attrs | (level == 3 ? PTE_TYPE_PAGE : 0); virt += blocksize; phys += blocksize; size -= blocksize; diff --git a/arch/arm/include/asm/armv8/mmu.h b/arch/arm/include/asm/armv8/mmu.h index aa0f3c4..58aecb9 100644 --- a/arch/arm/include/asm/armv8/mmu.h +++ b/arch/arm/include/asm/armv8/mmu.h @@ -60,6 +60,7 @@ #define PTE_TYPE_MASK (3 << 0) #define PTE_TYPE_FAULT (0 << 0) #define PTE_TYPE_TABLE (3 << 0) +#define PTE_TYPE_PAGE (3 << 0) #define PTE_TYPE_BLOCK (1 << 0) #define PTE_TABLE_PXN (1UL << 59) -- 2.7.4

2 1

[U-Boot] [RFC PATCH 1/2] video: add a function to acquire the GraphicDevice of cfb_console
by Icenowy Zheng 10 Feb '17

10 Feb '17

Sometimes the GraphicDevice of cfb_console is useful at other place. Add a function to acquire it, so that the graphics hardware (especially the framebuffer) can be reused. Signed-off-by: Icenowy Zheng <icenowy(a)aosc.xyz> --- drivers/video/cfb_console.c | 5 +++++ include/video_fb.h | 2 ++ 2 files changed, 7 insertions(+) diff --git a/drivers/video/cfb_console.c b/drivers/video/cfb_console.c index c0b1b8dc17..c80ed1646f 100644 --- a/drivers/video/cfb_console.c +++ b/drivers/video/cfb_console.c @@ -2199,3 +2199,8 @@ int video_get_screen_columns(void) { return CONSOLE_COLS; } + +GraphicDevice *video_get_graphic_device(void) +{ + return pGD; +}; diff --git a/include/video_fb.h b/include/video_fb.h index b008853f30..29f2b24648 100644 --- a/include/video_fb.h +++ b/include/video_fb.h @@ -89,4 +89,6 @@ void video_set_lut ( unsigned char b /* blue */ ); +GraphicDevice *video_get_graphic_device(void); + #endif /*_VIDEO_FB_H_ */ -- 2.11.0

3 3

[U-Boot] [RFC PATCH 0/4] arm: sunxi: Allwinner SPI driver
by S.J.R. van Schaik 10 Feb '17

10 Feb '17

Hi, The following patches implement a working SPI driver based on driver model for Allwinner (sunxi) devices. We have tested the driver on the A20 OLinuXino LIME 2 (sun7i) with a Winbond w25q128bv chip loading and booting a FIT image from the chip. The sun6i/sun8i/sun9i/sun50i code also builds properly, but still has to be tested on actual devices. The additional SPI controllers (SPI1, SPI2, SPI3) are not yet supported by this driver because the clock gating code only sets up the clock for SPI0. We are going to implement this in the near future. It might make sense to migrate the clock code to a separate driver or at least to mach-sunxi. We would like to hear some feedback to further improve these patches for mainline integration. Yours sincerely, Stephan. S.J.R. van Schaik (4): arch-sunxi: added missing AHB_GATE_OFFSET_SPIx defines for sun6i/sun9i arch-sunxi: added SPI register definitions for sun4i/sun7i arch-sunxi: added SPI register definitions for sun6i/sun8i/sun9i/sun50i sunxi-spi: added SPI driver for Allwinner devices (sunxi) arch/arm/include/asm/arch-sunxi/clock_sun6i.h | 2 + arch/arm/include/asm/arch-sunxi/clock_sun9i.h | 4 + arch/arm/include/asm/arch-sunxi/spi.h | 29 +++ arch/arm/include/asm/arch-sunxi/spi_sun4i.h | 53 +++++ arch/arm/include/asm/arch-sunxi/spi_sun6i.h | 56 +++++ drivers/spi/Kconfig | 5 + drivers/spi/Makefile | 1 + drivers/spi/sunxi_spi.c | 329 ++++++++++++++++++++++++++ 8 files changed, 479 insertions(+) create mode 100644 arch/arm/include/asm/arch-sunxi/spi.h create mode 100644 arch/arm/include/asm/arch-sunxi/spi_sun4i.h create mode 100644 arch/arm/include/asm/arch-sunxi/spi_sun6i.h create mode 100644 drivers/spi/sunxi_spi.c -- 2.11.0

1 4

[U-Boot] [PATCH] bcm2835: wdt: support for the bcm2835/2836 wdt
by Paolo Pisati 10 Feb '17

10 Feb '17

The following patch add support for the watchdog built into the bcm2835/2836 chips, using the max timeout (~15s). By default the watchdog is compiled in, but disabled. To enable it, apply the following patch: --- a/include/configs/rpi.h +++ b/include/configs/rpi.h @@ -109,6 +109,7 @@ /* Watchdog support */ #define CONFIG_BCM2835_WDT +#define CONFIG_HW_WATCHDOG /* Console configuration */ #define CONFIG_SYS_CBSIZE 1024 -- 2.7.4 Tested on my RaspberryPi2. Paolo Pisati (1): bcm2835_wdt: support for the BCM2835/2836 watchdog arch/arm/mach-bcm283x/reset.c | 21 ++++++++++++++++++--- board/raspberrypi/rpi/rpi.c | 4 ++++ drivers/watchdog/Makefile | 1 + drivers/watchdog/bcm2835_wdt.c | 34 ++++++++++++++++++++++++++++++++++ include/configs/rpi.h | 3 +++ scripts/config_whitelist.txt | 1 + 6 files changed, 61 insertions(+), 3 deletions(-) create mode 100644 drivers/watchdog/bcm2835_wdt.c -- 2.7.4

2 3

[U-Boot] [PATCH v2 0/5] ARM: DRA7: AM57xx: Enable SPL_DM
by Lokesh Vutla 10 Feb '17

10 Feb '17

- Enable SPL_DM on all DRA7 and AM57xx based platforms. - SPL uses a generic dts on all platforms with same defconfig. - After this series SPL size is increased by ~8KB. Verified MMC, eMMC boot on DRA74-evm, DRA72-evm. MMC boot on AM57xx-evm. Changes since v1: - Instead of creating a new dts file for each platform, created a omap5 specific common dtsi that will be included automatically. Lokesh Vutla (5): tools: omapimage: Fix size in header ARM: OMAP2+: define _image_binary_end to fix SPL_OF_CONTROL ARM: dts: OMAP5+: Add u-boot specific dtsi configs: am57xx_evm: Enable SPL_DM configs: dra7xx_evm: Enable SPL_DM arch/arm/dts/omap5-u-boot.dtsi | 46 ++++++++++++++++++++++++++++++++++++++ arch/arm/mach-omap2/u-boot-spl.lds | 2 ++ configs/am57xx_evm_defconfig | 7 ++++-- configs/am57xx_hs_evm_defconfig | 5 ++++- configs/dra7xx_evm_defconfig | 5 ++++- configs/dra7xx_hs_evm_defconfig | 5 ++++- include/configs/am57xx_evm.h | 5 ----- include/configs/dra7xx_evm.h | 5 ----- include/configs/ti_omap5_common.h | 8 +------ tools/gpimage.c | 2 +- tools/omapimage.c | 2 +- 11 files changed, 68 insertions(+), 24 deletions(-) create mode 100644 arch/arm/dts/omap5-u-boot.dtsi -- 2.11.0

2 16

[U-Boot] [PATCH v4 0/5] Add FreeBSD kconfig options
by Emmanuel Vadot 10 Feb '17

10 Feb '17

This series of patches add the needed bits for booting the FreeBSD loader and kernel. FreeBSD loader and kernel needs the U-Boot API and dcache/icache flushed for it to run so add this operation in go/bootelf command and when closing the API. Also add some some boot command that locate and run the FreeBSD loader if found. Changes in v4: Convert boards to CONFIG_API Change commit log for patch 3 to be not FreeBSD specific and add Reviewed-By Change patch order. Changes in v3: Always flush icache/dcache if enabled for bootelf/go commands Correct some typos Point to api/README in the help of Kconfig API Changes in v2: Remove SYS_DCACHE_OFF option Flush dcache/icache in go/boot command Flush dcache/icache in API closedev Emmanuel Vadot (4): kconfig: Add API kconfig file api: Convert to Kconfig distro_bootcmd: Add command to run FreeBSD kconfig: Add a DISTRO_FREEBSD option Warner Losh (1): api: bootelf: go: flush cache before starting Kconfig | 2 ++ api/Kconfig | 9 +++++++++ api/api.c | 5 +++++ cmd/boot.c | 4 ++++ cmd/elf.c | 5 +++++ common/Kconfig | 11 +++++++++++ configs/PMC440_defconfig | 1 + configs/lsxhl_defconfig | 1 + include/config_distro_bootcmd.h | 32 ++++++++++++++++++++++++++++++++ include/configs/PMC440.h | 2 -- include/configs/lsxl.h | 5 ----- scripts/config_whitelist.txt | 1 - 12 files changed, 70 insertions(+), 8 deletions(-) create mode 100644 api/Kconfig -- 2.11.0

4 19

[U-Boot] [BUG] x86: invalid size calculations in interrupts.c with newer GCC
by J. Tang 10 Feb '17

10 Feb '17

On latest master (commit 0ff27d), when using the stock qemu-x86_defconfig, it is possible to build an invalid U-Boot. I traced the issue to arch/x86/cpu/interrupts.c. The problem is that cpu_init_interrupts() assumes that each IRQ entry consumes the same number of bytes. Depending upon the compiler used, this assumption is incorrect. Commit 564a9984 added this hunk: +void irq_0(void); +void irq_1(void); int cpu_init_interrupts(void) { int i; + int irq_entry_size = irq_1 - irq_0; + void *irq_entry = (void *)irq_0; When I used a Buildroot derived compiler, I got this assembly code for arch/x86/cpu/interrupts.o: $ /home/tang/buildroot-x86/output/host/usr/bin/x86_64-linux-gcc --version x86_64-linux-gcc.br_real (Buildroot 2016.11.1) 5.4.0 $ objdump -d interrupts.o <snip> 00000207 <irq_18>: 207: 6a 12 push $0x12 209: eb 85 jmp 190 <irq_common_entry> 0000020b <irq_19>: 20b: 6a 13 push $0x13 20d: eb 81 jmp 190 <irq_common_entry> 0000020f <irq_20>: 20f: 6a 14 push $0x14 211: e9 7a ff ff ff jmp 190 <irq_common_entry> 00000216 <irq_21>: 216: 6a 15 push $0x15 218: e9 73 ff ff ff jmp 190 <irq_common_entry> Based upon the offset difference, the newer GCC optimizer will use a shorter jmp opcode instruction. This results in varying size IRQ entries. As a comparison, a stock Debian compiler produced this assembly: $ gcc --version gcc (Debian 4.9.2-10) 4.9.2 $ objdump -d interrupts.o <snip> 00000240 <irq_18>: 240: 6a 12 push $0x12 242: e9 fc ff ff ff jmp 243 <irq_18+0x3> 00000247 <irq_19>: 247: 6a 13 push $0x13 249: e9 fc ff ff ff jmp 24a <irq_19+0x3> 0000024e <irq_20>: 24e: 6a 14 push $0x14 250: e9 fc ff ff ff jmp 251 <irq_20+0x3> 00000255 <irq_21>: 255: 6a 15 push $0x15 257: e9 fc ff ff ff jmp 258 <irq_21+0x3> I got around this temporarily by having directly referencing all 256 IRQ symbols, rather than compute the next IRQ address relative to the previous. Hopefully someone can come up with a more elegant solution. -- Jason Tang / tang(a)jtang.org

2 3

[U-Boot] unused-const-variable warnings in FSL I2C driver
by Thomas Schaefer 09 Feb '17

09 Feb '17

Hi all, When compiling latest u-boot with gcc 6.3 for a T1024RDB system, I get a 'unused-const-variable' warning in the FSL I2C driver. Affected variable is fsl_i2c_speed_map. I guess this warning can be fixed with the patch applied, but I'm not sure. What do you think? Best regards, Thomas

2 1