April 2013 - coreboot-gerrit - mail.coreboot.org

Patch merged into coreboot/master: aee444f exynos5250: ungate the product ID register
by gerrit＠coreboot.org April 23, 2013

April 23, 2013

the following patch was just integrated into master: commit aee444f453f38e53f3e2ac54b560707616767869 Author: David Hendricks <dhendrix(a)chromium.org> Date: Mon Apr 22 16:03:11 2013 -0700 exynos5250: ungate the product ID register This makes sure that the product ID (PRO_ID) register can be read when the OS kernel is figuring out what kind of CPU it's running on. For historical reference, the original U-Boot code seems to have worked basically by accident here. The hardware has a quirk where by reading the value before gating the IP block keeps the value persistent. U-Boot reads the chip ID early on to distinguish between chip family, but we do not mix code the same way so we do not read the chip ID. Since the value has been read before the clock gating happens, the value remains available for the kernel to use during the decompression stage. We don't want to rely on that behavior when using coreboot. Instead the kernel should gate unused IPs. (credit to Gabe for finding symptom in the kernel) Change-Id: Iaa21e6e718b9000b5558f568020f393779fd208e Signed-off-by: Gabe Black <gabeblack(a)chromium.org> Signed-off-by: David Hendricks <dhendrix(a)chromium.org> Reviewed-on: http://review.coreboot.org/3121 Reviewed-by: Ronald G. Minnich <rminnich(a)gmail.com> Reviewed-by: Stefan Reinauer <stefan.reinauer(a)coreboot.org> Tested-by: build bot (Jenkins) Build-Tested: build bot (Jenkins) at Tue Apr 23 03:06:56 2013, giving +1 Reviewed-By: Stefan Reinauer <stefan.reinauer(a)coreboot.org> at Tue Apr 23 02:05:50 2013, giving +2 See http://review.coreboot.org/3121 for details. -gerrit

1 0

Patch set updated for coreboot: 4882503 GOOGLE/SNOW: get graphics working
by David Hendricks April 23, 2013

April 23, 2013

David Hendricks (dhendrix(a)chromium.org) just uploaded a new patch set to gerrit, which you can find at http://review.coreboot.org/3116 -gerrit commit 4882503f6a11413a4d8cb42e6e1ce87a36854999 Author: Ronald G. Minnich <rminnich(a)gmail.com> Date: Thu Apr 18 18:09:24 2013 -0700 GOOGLE/SNOW: get graphics working This adds support for display bring-up on Snow. It includes framebuffer initialization and LCD enable functions. Change-Id: I16e711c97e9d02c916824f621e2313297448732b Signed-off-by: Ronald G. Minnich <rminnich(a)gmail.com> Signed-off-by: David Hendricks <dhendrix(a)chromium.org> --- src/cpu/samsung/exynos5-common/exynos-fb.c | 76 ++++++++++++++++++----------- src/cpu/samsung/exynos5-common/s5p-dp-reg.c | 11 ++--- src/cpu/samsung/exynos5250/cpu.c | 25 ++++++++-- src/mainboard/google/snow/Kconfig | 2 + src/mainboard/google/snow/ramstage.c | 36 ++++++++++++++ 5 files changed, 113 insertions(+), 37 deletions(-) diff --git a/src/cpu/samsung/exynos5-common/exynos-fb.c b/src/cpu/samsung/exynos5-common/exynos-fb.c index 54a2d7a..00ff19e 100644 --- a/src/cpu/samsung/exynos5-common/exynos-fb.c +++ b/src/cpu/samsung/exynos5-common/exynos-fb.c @@ -27,6 +27,7 @@ #include <arch/io.h> #include <stdlib.h> #include <string.h> +#include <time.h> #include <console/console.h> #include <cpu/samsung/exynos5250/cpu.h> #include <cpu/samsung/exynos5250/power.h> @@ -41,6 +42,28 @@ #include "cpu/samsung/exynos5250/s5p-dp.h" #include "s5p-dp-core.h" +/* + * Here is the rough outline of how we bring up the display: + * 1. Upon power-on Sink generates a hot plug detection pulse thru HPD + * 2. Source determines video mode by reading DPCD receiver capability field + * (DPCD 00000h to 0000Dh) including eDP CP capability register (DPCD + * 0000Dh). + * 3. Sink replies DPCD receiver capability field. + * 4. Source starts EDID read thru I2C-over-AUX. + * 5. Sink replies EDID thru I2C-over-AUX. + * 6. Source determines link configuration, such as MAX_LINK_RATE and + * MAX_LANE_COUNT. Source also determines which type of eDP Authentication + * method to use and writes DPCD link configuration field (DPCD 00100h to + * 0010Ah) including eDP configuration set (DPCD 0010Ah). + * 7. Source starts link training. Sink does clock recovery and equalization. + * 8. Source reads DPCD link status field (DPCD 00200h to 0020Bh). + * 9. Sink replies DPCD link status field. If main link is not stable, Source + * repeats Step 7. + * 10. Source sends MSA (Main Stream Attribute) data. Sink extracts video + * parameters and recovers stream clock. + * 11. Source sends video data. + */ + /* To help debug any init errors here, define a list of possible errors */ enum { ERR_PLL_NOT_UNLOCKED = 2, @@ -126,10 +149,8 @@ void fb_init(vidinfo_t *panel_info, void *lcdbase, { unsigned int val; u32 fbsize; - struct exynos5_fimd *fimd = - samsung_get_base_fimd(); - struct exynos5_disp_ctrl *disp_ctrl = - samsung_get_base_disp_ctrl(); + struct exynos5_fimd *fimd = samsung_get_base_fimd(); + struct exynos5_disp_ctrl *disp_ctrl = samsung_get_base_disp_ctrl(); writel(pd->ivclk | pd->fixvclk, &disp_ctrl->vidcon1); val = ENVID_ON | ENVID_F_ON | (pd->clkval_f << CLKVAL_F_OFFSET); @@ -191,7 +212,7 @@ static int s5p_dp_config_video(struct s5p_dp_device *dp, struct video_info *video_info) { int timeout = 0; - u32 start; + u32 start, end; struct exynos5_dp *base = dp->base; s5p_dp_config_video_slave_mode(dp, video_info); @@ -206,16 +227,18 @@ static int s5p_dp_config_video(struct s5p_dp_device *dp, return -ERR_PLL_NOT_UNLOCKED; } - start = get_timer(0); + start = timer_us(); + end = start + STREAM_ON_TIMEOUT*1000; do { if (s5p_dp_is_slave_video_stream_clock_on(dp) == 0) { timeout++; break; } - } while (get_timer(start) <= STREAM_ON_TIMEOUT); + } while (timer_us() < end); if (!timeout) { - printk(BIOS_DEBUG, "Video Clock Not ok\n"); + printk(BIOS_ERR, "Video Clock Not ok after %uus.\n", + timer_us() - start); return -ERR_VIDEO_CLOCK_BAD; } @@ -255,7 +278,6 @@ static int s5p_dp_enable_rx_to_enhanced_mode(struct s5p_dp_device *dp) printk(BIOS_DEBUG, "DPCD read error\n"); return -ERR_DPCD_READ_ERROR1; } - if (s5p_dp_write_byte_to_dpcd(dp, DPCD_ADDR_LANE_COUNT_SET, DPCD_ENHANCED_FRAME_EN | (data & DPCD_LANE_COUNT_SET_MASK))) { @@ -408,6 +430,7 @@ static int s5p_dp_hw_link_training(struct s5p_dp_device *dp, unsigned int max_lane, unsigned int max_rate) { + int pll_is_locked = 0; u32 data; u32 start; int lane; @@ -417,14 +440,15 @@ static int s5p_dp_hw_link_training(struct s5p_dp_device *dp, clrbits_le32(&base->video_ctl_1, VIDEO_EN); start = get_timer(0); - while (s5p_dp_get_pll_lock_status(dp) == PLL_UNLOCKED) { + while ((pll_is_locked = s5p_dp_get_pll_lock_status(dp)) == PLL_UNLOCKED) { if (get_timer(start) > PLL_LOCK_TIMEOUT) { /* Ignore this error, and try to continue */ printk(BIOS_ERR, "PLL is not locked yet.\n"); break; } } - + printk(BIOS_SPEW, "PLL is %slocked\n", + pll_is_locked == PLL_LOCKED ? "": "not "); /* Reset Macro */ setbits_le32(&base->dp_phy_test, MACRO_RST); @@ -448,6 +472,9 @@ static int s5p_dp_hw_link_training(struct s5p_dp_device *dp, s5p_dp_get_max_rx_bandwidth(dp, &dp->link_train.link_rate); s5p_dp_get_max_rx_lane_count(dp, &dp->link_train.lane_count); + printk(BIOS_SPEW, "%s: rate 0x%x, lane_count %d\n", __func__, + dp->link_train.link_rate, dp->link_train.lane_count); + if ((dp->link_train.link_rate != LINK_RATE_1_62GBPS) && (dp->link_train.link_rate != LINK_RATE_2_70GBPS)) { printk(BIOS_DEBUG, "Rx Max Link Rate is abnormal :%x !\n", @@ -480,13 +507,14 @@ static int s5p_dp_hw_link_training(struct s5p_dp_device *dp, /* Start HW link training */ writel(HW_TRAINING_EN, &base->dp_hw_link_training); - /* Wait unitl HW link training done */ + /* Wait until HW link training done */ s5p_dp_wait_hw_link_training_done(dp); /* Get hardware link training status */ data = readl(&base->dp_hw_link_training); + printk(BIOS_SPEW, "hardware link training status: 0x%08x\n", data); if (data != 0) { - printk(BIOS_DEBUG, " H/W link training failure: 0x%x\n", data); + printk(BIOS_ERR, " H/W link training failure: 0x%x\n", data); return -ERR_LINK_TRAINING_FAILURE; } @@ -497,6 +525,8 @@ static int s5p_dp_hw_link_training(struct s5p_dp_device *dp, data = readl(&base->lane_count_set); dp->link_train.lane_count = data; + printk(BIOS_SPEW, "Done training: Link bandwidth: 0x%x, lane_count: %d\n", + dp->link_train.link_rate, data); return 0; } @@ -510,24 +540,19 @@ int dp_controller_init(struct s5p_dp_device *dp_device) struct s5p_dp_device *dp = dp_device; struct exynos5_dp *base; - //dp->base = (struct exynos5_dp *)addr; - /* yes. we're a snow. Yet somehow our config is from a development kit? - * This Must Change */ - //dp->video_info = &smdk5250_dp_config; - clock_init_dp_clock(); power_enable_dp_phy(); ret = s5p_dp_init_dp(dp); if (ret) { - printk(BIOS_DEBUG, "%s: Could not initialize dp\n", __func__); + printk(BIOS_ERR, "%s: Could not initialize dp\n", __func__); return ret; } ret = s5p_dp_hw_link_training(dp, dp->video_info->lane_count, dp->video_info->link_rate); if (ret) { - printk(BIOS_DEBUG, "unable to do link train\n"); + printk(BIOS_ERR, "unable to do link train\n"); return ret; } /* Minimum delay after H/w Link training */ @@ -535,13 +560,13 @@ int dp_controller_init(struct s5p_dp_device *dp_device) ret = s5p_dp_enable_scramble(dp); if (ret) { - printk(BIOS_DEBUG, "unable to set scramble mode\n"); + printk(BIOS_ERR, "unable to set scramble mode\n"); return ret; } ret = s5p_dp_enable_rx_to_enhanced_mode(dp); if (ret) { - printk(BIOS_DEBUG, "unable to set enhanced mode\n"); + printk(BIOS_ERR, "unable to set enhanced mode\n"); return ret; } @@ -556,7 +581,7 @@ int dp_controller_init(struct s5p_dp_device *dp_device) s5p_dp_init_video(dp); ret = s5p_dp_config_video(dp, dp->video_info); if (ret) { - printk(BIOS_DEBUG, "unable to config video\n"); + printk(BIOS_ERR, "unable to config video\n"); return ret; } @@ -576,10 +601,5 @@ int lcd_ctrl_init(vidinfo_t *panel_info, fimd_bypass(); fb_init(panel_info, lcdbase, panel_data); - printk(BIOS_SPEW, - "fb_init(%p, %p, %p) done\n", panel_info, lcdbase, panel_data); - - /* Enable flushing after LCD writes if requested */ - // forget it. lcd_set_flush_dcache(1); return ret; } diff --git a/src/cpu/samsung/exynos5-common/s5p-dp-reg.c b/src/cpu/samsung/exynos5-common/s5p-dp-reg.c index 60e1398..cce8c6d 100644 --- a/src/cpu/samsung/exynos5-common/s5p-dp-reg.c +++ b/src/cpu/samsung/exynos5-common/s5p-dp-reg.c @@ -129,7 +129,8 @@ int s5p_dp_init_analog_func(struct s5p_dp_device *dp) start = get_timer(0); while (s5p_dp_get_pll_lock_status(dp) == PLL_UNLOCKED) { if (get_timer(start) > PLL_LOCK_TIMEOUT) { - printk(BIOS_DEBUG, "%s: PLL is not locked yet\n", __func__); + printk(BIOS_ERR, "%s: PLL is not locked\n", + __func__); return -1; } } @@ -176,9 +177,6 @@ int s5p_dp_start_aux_transaction(struct s5p_dp_device *dp) /* Enable AUX CH operation */ setbits_le32(&base->aux_ch_ctl_2, AUX_EN); - printk(BIOS_DEBUG, "%s: base: 0x%p, &base->aux_ch_ctl_2: 0x%p, aux_ch_ctl_2: 0x%08x\n", - __func__, base, &base->aux_ch_ctl_2, readl(&base->aux_ch_ctl_2)); - /* Is AUX CH command reply received? */ reg = readl(&base->dp_int_sta); while (!(reg & RPLY_RECEIV)) @@ -189,8 +187,9 @@ int s5p_dp_start_aux_transaction(struct s5p_dp_device *dp) /* Clear interrupt source for AUX CH access error */ reg = readl(&base->dp_int_sta); - printk(BIOS_DEBUG, "%s: dp_int_sta: 0x%02x\n", __func__, reg); if (reg & AUX_ERR) { + printk(BIOS_ERR, "%s: AUX_ERR encountered, dp_int_sta: " + "0x%02x\n", __func__, reg); writel(AUX_ERR, &base->dp_int_sta); return -1; } @@ -198,7 +197,7 @@ int s5p_dp_start_aux_transaction(struct s5p_dp_device *dp) /* Check AUX CH error access status */ reg = readl(&base->dp_int_sta); if ((reg & AUX_STATUS_MASK) != 0) { - printk(BIOS_DEBUG, "AUX CH error happens: %d\n\n", + printk(BIOS_ERR, "AUX CH error happens: %d\n\n", reg & AUX_STATUS_MASK); return -1; } diff --git a/src/cpu/samsung/exynos5250/cpu.c b/src/cpu/samsung/exynos5250/cpu.c index e122413..8bf0d49 100644 --- a/src/cpu/samsung/exynos5250/cpu.c +++ b/src/cpu/samsung/exynos5250/cpu.c @@ -5,6 +5,7 @@ #include <console/console.h> #include <device/device.h> #include <cbmem.h> +#include <arch/cache.h> #include <cpu/samsung/exynos5250/fimd.h> #include <cpu/samsung/exynos5-common/s5p-dp-core.h> #include "chip.h" @@ -32,7 +33,6 @@ static void exynos_displayport_init(device_t dev) unsigned long int fb_size; u32 lcdbase; - printk(BIOS_SPEW, "%s: dev 0x%p, conf 0x%p\n", __func__, dev, conf); memset(&vi, 0, sizeof(vi)); memset(&panel, 0, sizeof(panel)); @@ -61,16 +61,35 @@ static void exynos_displayport_init(device_t dev) * The size is a magic number from hardware. Allocate enough for the * frame buffer and color map. */ - fb_size = conf->xres * conf->yres * sizeof(unsigned long); + fb_size = conf->xres * conf->yres * (conf->bpp / 8); lcdbase = (uintptr_t)cbmem_add(CBMEM_ID_CONSOLE, fb_size + 64*KiB); printk(BIOS_SPEW, "lcd colormap base is %p\n", (void *)(lcdbase)); mmio_resource(dev, 0, lcdbase/KiB, 64); vi.cmap = (void *)lcdbase; + /* + * We need to clean and invalidate the framebuffer region and disable + * caching as well. We assume that our dcache <--> memory address + * space is identity-mapped in 1MB chunks, so align accordingly. + * + * Note: We may want to do something clever to ensure the framebuffer + * region is aligned such that we don't change dcache policy for other + * stuff inadvertantly. + * + * FIXME: Is disabling/re-enabling the MMU entirely necessary? + */ + uint32_t lower = ALIGN_DOWN(lcdbase, MiB); + uint32_t upper = ALIGN_UP(lcdbase + fb_size + 64*KiB, MiB); + dcache_clean_invalidate_by_mva(lower, upper - lower); + dcache_mmu_disable(); + mmu_config_range(lower/MiB, (upper - lower)/MiB, DCACHE_OFF); + dcache_mmu_enable(); + lcdbase += 64*KiB; mmio_resource(dev, 1, lcdbase/KiB, (fb_size + KiB - 1)/KiB); printk(BIOS_DEBUG, - "Initializing exynos VGA, base %p\n",(void *)lcdbase); + "Initializing exynos VGA, base %p\n", (void *)lcdbase); + memset((void *)lcdbase, 0, fb_size); /* clear the framebuffer */ ret = lcd_ctrl_init(&vi, &panel, (void *)lcdbase); } diff --git a/src/mainboard/google/snow/Kconfig b/src/mainboard/google/snow/Kconfig index 6d3d7c8..c28a986 100644 --- a/src/mainboard/google/snow/Kconfig +++ b/src/mainboard/google/snow/Kconfig @@ -33,6 +33,8 @@ config BOARD_SPECIFIC_OPTIONS # dummy select EXYNOS_DISPLAYPORT select CHROMEOS select DRIVER_TI_TPS65090 + select MAINBOARD_HAS_NATIVE_VGA_INIT + select MAINBOARD_DO_NATIVE_VGA_INIT config MAINBOARD_DIR string diff --git a/src/mainboard/google/snow/ramstage.c b/src/mainboard/google/snow/ramstage.c index 54d13d3..09b51a7 100644 --- a/src/mainboard/google/snow/ramstage.c +++ b/src/mainboard/google/snow/ramstage.c @@ -23,6 +23,7 @@ #include <drivers/ti/tps65090/tps65090.h> #include <cbmem.h> #include <delay.h> +#include <boot/coreboot_tables.h> #include <arch/cache.h> #include <arch/exception.h> #include <arch/gpio.h> @@ -41,6 +42,41 @@ #define DRAM_SIZE CONFIG_DRAM_SIZE_MB #define DRAM_END (DRAM_START + DRAM_SIZE) /* plus one... */ +int vbe_mode_info_valid(void); +int vbe_mode_info_valid(void) +{ + return 1; +} + +void fill_lb_framebuffer(struct lb_framebuffer *framebuffer); +void fill_lb_framebuffer(struct lb_framebuffer *framebuffer) +{ + /* + * The address returned points at the LCD colormap base. The + * 64KiB offset points at the LCD base. + */ + framebuffer->physical_address = + (u32)cbmem_find(CBMEM_ID_CONSOLE) + 64*KiB; + printk(BIOS_SPEW, "%s: framebuffer->physical address is 0x%llx\n", + __func__, framebuffer->physical_address); + + framebuffer->x_resolution = 1366; + framebuffer->y_resolution = 768; + framebuffer->bits_per_pixel = 16; + framebuffer->bytes_per_line = + (framebuffer->x_resolution * framebuffer->bits_per_pixel) / 8; + + framebuffer->red_mask_pos = 11; + framebuffer->red_mask_size = 5; + framebuffer->green_mask_pos = 6; + framebuffer->green_mask_size = 5; + framebuffer->blue_mask_pos = 0; + framebuffer->blue_mask_size = 5; + framebuffer->reserved_mask_pos = 0; + framebuffer->reserved_mask_size = 0; +} + + void hardwaremain(int boot_complete); void main(void) {

1 0

Patch set updated for coreboot: 9cd5ebe exynos5250: ungate the product ID register
by David Hendricks April 23, 2013

April 23, 2013

David Hendricks (dhendrix(a)chromium.org) just uploaded a new patch set to gerrit, which you can find at http://review.coreboot.org/3121 -gerrit commit 9cd5ebe34dae69febebce240ca0123d4f4f9d2ec Author: David Hendricks <dhendrix(a)chromium.org> Date: Mon Apr 22 16:03:11 2013 -0700 exynos5250: ungate the product ID register This makes sure that the product ID (PRO_ID) register can be read when the OS kernel is figuring out what kind of CPU it's running on. For historical reference, the original U-Boot code seems to have worked basically by accident here. The hardware has a quirk where by reading the value before gating the IP block keeps the value persistent. U-Boot reads the chip ID early on to distinguish between chip family, but we do not mix code the same way so we do not read the chip ID. Since the value has been read before the clock gating happens, the value remains available for the kernel to use during the decompression stage. We don't want to rely on that behavior when using coreboot. Instead the kernel should gate unused IPs. (credit to Gabe for finding symptom in the kernel) Change-Id: Iaa21e6e718b9000b5558f568020f393779fd208e Signed-off-by: Gabe Black <gabeblack(a)chromium.org> Signed-off-by: David Hendricks <dhendrix(a)chromium.org> --- src/cpu/samsung/exynos5250/clock_init.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) diff --git a/src/cpu/samsung/exynos5250/clock_init.c b/src/cpu/samsung/exynos5250/clock_init.c index c94cadf..0989b88 100644 --- a/src/cpu/samsung/exynos5250/clock_init.c +++ b/src/cpu/samsung/exynos5250/clock_init.c @@ -416,7 +416,12 @@ void clock_gate(void) CLK_SPI2_MASK | CLK_SPI0_MASK); - /* CLK_GATE_IP_PERIS */ + /* + * CLK_GATE_IP_PERIS + * Note: Keep CHIPID_APBIF ungated to ensure reading the product ID + * register (PRO_ID) works correctly when the OS kernel determines + * which chip it is running on. + */ clrbits_le32(&clk->gate_ip_peris, CLK_RTC_MASK | CLK_TZPC9_MASK | CLK_TZPC8_MASK | @@ -427,8 +432,7 @@ void clock_gate(void) CLK_TZPC3_MASK | CLK_TZPC2_MASK | CLK_TZPC1_MASK | - CLK_TZPC0_MASK | - CLK_CHIPID_MASK); + CLK_TZPC0_MASK); /* CLK_GATE_BLOCK */ clrbits_le32(&clk->gate_block, CLK_ACP_MASK);

1 0

Patch set updated for coreboot: bca95c7 exynos5250: ungate the product ID register
by David Hendricks April 23, 2013

April 23, 2013

David Hendricks (dhendrix(a)chromium.org) just uploaded a new patch set to gerrit, which you can find at http://review.coreboot.org/3121 -gerrit commit bca95c71bfa5c3fa606f199b4ae69bb42a492873 Author: David Hendricks <dhendrix(a)chromium.org> Date: Mon Apr 22 16:03:11 2013 -0700 exynos5250: ungate the product ID register This makes sure that the product ID (PRO_ID) register can be read when the OS kernel is figuring out what kind of CPU it's running on. For historical reference, the original u-boot code seems to have worked basically by accident here. The hardware has a quirk where by reading the value before gating the IP block keeps the value persistent. That is why the kernel is able to read the product ID during the decompression stage. We don't want to rely on that behavior when using coreboot, instead the kernel should gate unused IPs. (credit to Gabe for finding symptom in the kernel) Change-Id: Iaa21e6e718b9000b5558f568020f393779fd208e Signed-off-by: Gabe Black <gabeblack(a)chromium.org> Signed-off-by: David Hendricks <dhendrix(a)chromium.org> --- src/cpu/samsung/exynos5250/clock_init.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) diff --git a/src/cpu/samsung/exynos5250/clock_init.c b/src/cpu/samsung/exynos5250/clock_init.c index c94cadf..0989b88 100644 --- a/src/cpu/samsung/exynos5250/clock_init.c +++ b/src/cpu/samsung/exynos5250/clock_init.c @@ -416,7 +416,12 @@ void clock_gate(void) CLK_SPI2_MASK | CLK_SPI0_MASK); - /* CLK_GATE_IP_PERIS */ + /* + * CLK_GATE_IP_PERIS + * Note: Keep CHIPID_APBIF ungated to ensure reading the product ID + * register (PRO_ID) works correctly when the OS kernel determines + * which chip it is running on. + */ clrbits_le32(&clk->gate_ip_peris, CLK_RTC_MASK | CLK_TZPC9_MASK | CLK_TZPC8_MASK | @@ -427,8 +432,7 @@ void clock_gate(void) CLK_TZPC3_MASK | CLK_TZPC2_MASK | CLK_TZPC1_MASK | - CLK_TZPC0_MASK | - CLK_CHIPID_MASK); + CLK_TZPC0_MASK); /* CLK_GATE_BLOCK */ clrbits_le32(&clk->gate_block, CLK_ACP_MASK);

1 0

New patch to review for coreboot: 55c70fc exynos5250: read product ID register before gating its clock
by David Hendricks April 23, 2013

April 23, 2013

David Hendricks (dhendrix(a)chromium.org) just uploaded a new patch set to gerrit, which you can find at http://review.coreboot.org/3121 -gerrit commit 55c70fcfccdf8591af4afc05f8b65b7a955b4ef0 Author: David Hendricks <dhendrix(a)chromium.org> Date: Mon Apr 22 16:03:11 2013 -0700 exynos5250: read product ID register before gating its clock The product ID comes from a special-function register (SFR) which has interesting behavior whereby reading its value causes it to persist in memory even after the clock is gated. The chip ID may be read by the OS kernel; for Linux, it is used while decompressing the kernel. So to make this information available after coreboot finishes and save a little bit of power we can force a read to occur before gating its IP block. (credit to Gabe for finding symptom in the kernel) Change-Id: Iaa21e6e718b9000b5558f568020f393779fd208e Signed-off-by: Gabe Black <gabeblack(a)chromium.org> Signed-off-by: David Hendricks <dhendrix(a)chromium.org> --- src/cpu/samsung/exynos5250/clock_init.c | 13 +++++++++++-- 1 file changed, 11 insertions(+), 2 deletions(-) diff --git a/src/cpu/samsung/exynos5250/clock_init.c b/src/cpu/samsung/exynos5250/clock_init.c index c94cadf..2fd9947 100644 --- a/src/cpu/samsung/exynos5250/clock_init.c +++ b/src/cpu/samsung/exynos5250/clock_init.c @@ -295,6 +295,7 @@ void system_clock_init(struct mem_timings *mem, void clock_gate(void) { struct exynos5_clock *clk = (struct exynos5_clock *)EXYNOS5_CLOCK_BASE; + uint32_t chip_id; /* CLK_GATE_IP_SYSRGT */ clrbits_le32(&clk->gate_ip_sysrgt, CLK_C2C_MASK); @@ -427,8 +428,16 @@ void clock_gate(void) CLK_TZPC3_MASK | CLK_TZPC2_MASK | CLK_TZPC1_MASK | - CLK_TZPC0_MASK | - CLK_CHIPID_MASK); + CLK_TZPC0_MASK); + + /* + * Ensure the chip ID is read before gating its IP block. This causes + * the value to remain persistent in memory so when other code + * references the PRO_ID SFR it will identify the chip correctly. + */ + chip_id = (readl((void *)EXYNOS_PRO_ID) >> 20) & 0xf; + if (chip_id == 5) + clrbits_le32(&clk->gate_ip_peris, CLK_CHIPID_MASK); /* CLK_GATE_BLOCK */ clrbits_le32(&clk->gate_block, CLK_ACP_MASK);

1 0

Patch set updated for coreboot: f51ed29 AMD CIMx SB800: Remove unused function `get_sbdn()`
by Paul Menzel April 22, 2013

April 22, 2013

Paul Menzel (paulepanter(a)users.sourceforge.net) just uploaded a new patch set to gerrit, which you can find at http://review.coreboot.org/3119 -gerrit commit f51ed298309f817db1624488291a38e4bace922e Author: Paul Menzel <paulepanter(a)users.sourceforge.net> Date: Fri Apr 19 10:23:03 2013 +0200 AMD CIMx SB800: Remove unused function `get_sbdn()` No SB800 based board uses the function `get_sbdn()`, which was probably copied from SB700 code. So remove it. Just as a test remove some code from AMD model 10xxx. But looking at $ more src/mainboard/asus/m5a88-v/romstage.c […] #include <lib.h> #include <stdint.h> #include <string.h> #include <device/pci_def.h> #include <device/pci_ids.h> #include <arch/io.h> #include <device/pnp_def.h> #include <cpu/x86/lapic.h> #include <console/console.h> #include <cpu/amd/model_10xxx_rev.h> #include "northbridge/amd/amdfam10/raminit.h" #include "northbridge/amd/amdfam10/amdfam10.h" #include "cpu/x86/lapic/boot_cpu.c" #include "northbridge/amd/amdfam10/reset_test.c" #include <console/loglevel.h> #include "cpu/x86/bist.h" #include "superio/ite/it8721f/early_serial.c" #include "cpu/x86/mtrr/earlymtrr.c" #include <cpu/amd/mtrr.h> #include "northbridge/amd/amdfam10/setup_resource_map.c" #include "southbridge/amd/rs780/early_setup.c" #include <sb_cimx.h> #include <SBPLATFORM.h> /* SB OEM constants */ #include <southbridge/amd/cimx/sb800/smbus.h> #include "northbridge/amd/amdfam10/debug.c" static void activate_spd_rom(const struct mem_controller *ctrl) { } static int spd_read_byte(u32 device, u32 address) { return do_smbus_read_byte(SMBUS_IO_BASE, device, address); } #include "northbridge/amd/amdfam10/raminit_sysinfo_in_ram.c" #include "northbridge/amd/amdfam10/pci.c" #include "resourcemap.c" #include "cpu/amd/quadcore/quadcore.c" #include "cpu/amd/car/post_cache_as_ram.c" #include "cpu/amd/microcode/microcode.c" #if CONFIG_UPDATE_CPU_MICROCODE #include "cpu/amd/model_10xxx/update_microcode.c" #endif #include "cpu/amd/model_10xxx/init_cpus.c" #include "northbridge/amd/amdfam10/early_ht.c" #include "spd.h" #include <reset.h> […] this should be left alone probably as cleaning this mess up is going to take a lot of time with no real benefits. Change-Id: I85fe3db0abf2648b6eb7a659f2112d130f856afa Signed-off-by: Paul Menzel <paulepanter(a)users.sourceforge.net> --- src/cpu/amd/model_10xxx/init_cpus.c | 10 ---------- src/southbridge/amd/cimx/sb800/early.c | 22 ---------------------- src/southbridge/amd/cimx/sb800/sb_cimx.h | 8 -------- 3 files changed, 40 deletions(-) diff --git a/src/cpu/amd/model_10xxx/init_cpus.c b/src/cpu/amd/model_10xxx/init_cpus.c index 3d5ddcf..2f0654f 100644 --- a/src/cpu/amd/model_10xxx/init_cpus.c +++ b/src/cpu/amd/model_10xxx/init_cpus.c @@ -905,16 +905,6 @@ static void finalize_node_setup(struct sys_info *sysinfo) u8 i; u8 nodes = get_nodes(); -#if CONFIG_RAMINIT_SYSINFO - u32 reg; - /* read Node0 F0_0x64 bit [8:10] to find out SbLink # */ - reg = pci_read_config32(NODE_HT(0), 0x64); - sysinfo->sblk = (reg >> 8) & 7; - sysinfo->sbbusn = 0; - sysinfo->nodes = nodes; - sysinfo->sbdn = get_sbdn(sysinfo->sbbusn); -#endif - for (i = 0; i < nodes; i++) { cpuSetAMDPCI(i); } diff --git a/src/southbridge/amd/cimx/sb800/early.c b/src/southbridge/amd/cimx/sb800/early.c index 83087f5..dad215f 100644 --- a/src/southbridge/amd/cimx/sb800/early.c +++ b/src/southbridge/amd/cimx/sb800/early.c @@ -26,28 +26,6 @@ #include "cfg.h" /*sb800_cimx_config*/ #include "cbmem.h" -#if CONFIG_RAMINIT_SYSINFO -/** - * @brief Get SouthBridge device number - * @param[in] bus target bus number - * @return southbridge device number - */ -u32 get_sbdn(u32 bus) -{ - device_t dev; - - printk(BIOS_DEBUG, "SB800 - %s - %s - Start.\n", __FILE__, __func__); - //dev = PCI_DEV(bus, 0x14, 0); - dev = pci_locate_device_on_bus( - PCI_ID(PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_SB800_SM), - bus); - - printk(BIOS_DEBUG, "SB800 - %s - %s - End.\n", __FILE__, __func__); - return (dev >> 15) & 0x1f; -} -#endif - - /** * @brief South Bridge CIMx romstage entry, * wrapper of sbPowerOnInit entry point. diff --git a/src/southbridge/amd/cimx/sb800/sb_cimx.h b/src/southbridge/amd/cimx/sb800/sb_cimx.h index 3bec5d8..f07b0e0 100644 --- a/src/southbridge/amd/cimx/sb800/sb_cimx.h +++ b/src/southbridge/amd/cimx/sb800/sb_cimx.h @@ -40,12 +40,4 @@ int acpi_is_wakeup_early(void); */ void sb800_clk_output_48Mhz(void); -#if CONFIG_RAMINIT_SYSINFO -/** - * @brief Get SouthBridge device number, called by finalize_node_setup() - * @param[in] bus target bus number - * @return southbridge device number - */ -u32 get_sbdn(u32 bus); -#endif #endif

1 0

Patch set updated for coreboot: 59455b1 AMD CIMx SB800: Remove unused function `get_sbdn()`
by Paul Menzel April 22, 2013

April 22, 2013

Paul Menzel (paulepanter(a)users.sourceforge.net) just uploaded a new patch set to gerrit, which you can find at http://review.coreboot.org/3119 -gerrit commit 59455b19a2765ed878570be0f5e94d3e132fc391 Author: Paul Menzel <paulepanter(a)users.sourceforge.net> Date: Fri Apr 19 10:23:03 2013 +0200 AMD CIMx SB800: Remove unused function `get_sbdn()` No SB800 based board uses the function `get_sbdn()`, which was probably copied from SB700 code. So remove it. Change-Id: I85fe3db0abf2648b6eb7a659f2112d130f856afa Signed-off-by: Paul Menzel <paulepanter(a)users.sourceforge.net> --- src/southbridge/amd/cimx/sb800/early.c | 22 ---------------------- src/southbridge/amd/cimx/sb800/sb_cimx.h | 8 -------- 2 files changed, 30 deletions(-) diff --git a/src/southbridge/amd/cimx/sb800/early.c b/src/southbridge/amd/cimx/sb800/early.c index 83087f5..dad215f 100644 --- a/src/southbridge/amd/cimx/sb800/early.c +++ b/src/southbridge/amd/cimx/sb800/early.c @@ -26,28 +26,6 @@ #include "cfg.h" /*sb800_cimx_config*/ #include "cbmem.h" -#if CONFIG_RAMINIT_SYSINFO -/** - * @brief Get SouthBridge device number - * @param[in] bus target bus number - * @return southbridge device number - */ -u32 get_sbdn(u32 bus) -{ - device_t dev; - - printk(BIOS_DEBUG, "SB800 - %s - %s - Start.\n", __FILE__, __func__); - //dev = PCI_DEV(bus, 0x14, 0); - dev = pci_locate_device_on_bus( - PCI_ID(PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_SB800_SM), - bus); - - printk(BIOS_DEBUG, "SB800 - %s - %s - End.\n", __FILE__, __func__); - return (dev >> 15) & 0x1f; -} -#endif - - /** * @brief South Bridge CIMx romstage entry, * wrapper of sbPowerOnInit entry point. diff --git a/src/southbridge/amd/cimx/sb800/sb_cimx.h b/src/southbridge/amd/cimx/sb800/sb_cimx.h index 3bec5d8..f07b0e0 100644 --- a/src/southbridge/amd/cimx/sb800/sb_cimx.h +++ b/src/southbridge/amd/cimx/sb800/sb_cimx.h @@ -40,12 +40,4 @@ int acpi_is_wakeup_early(void); */ void sb800_clk_output_48Mhz(void); -#if CONFIG_RAMINIT_SYSINFO -/** - * @brief Get SouthBridge device number, called by finalize_node_setup() - * @param[in] bus target bus number - * @return southbridge device number - */ -u32 get_sbdn(u32 bus); -#endif #endif

1 0

Patch merged into coreboot/master: e8a9134 GOOGLE/SNOW: fix stupid paren error
by gerrit＠coreboot.org April 22, 2013

April 22, 2013

the following patch was just integrated into master: commit e8a91347b182281f18c969229f02e07102e21898 Author: Ronald G. Minnich <rminnich(a)gmail.com> Date: Mon Apr 22 10:46:53 2013 -0700 GOOGLE/SNOW: fix stupid paren error This simple error led to corrupted graphics. How annoying. Change-Id: I2295c0df0f1d16014a603dc5d66bd4d72f3fb7c9 Signed-off-by: Ronald G. Minnich <rminnich(a)gmail.com> Reviewed-on: http://review.coreboot.org/3120 Reviewed-by: David Hendricks <dhendrix(a)chromium.org> Tested-by: build bot (Jenkins) Build-Tested: build bot (Jenkins) at Mon Apr 22 20:29:29 2013, giving +1 Reviewed-By: David Hendricks <dhendrix(a)chromium.org> at Mon Apr 22 20:16:02 2013, giving +2 See http://review.coreboot.org/3120 for details. -gerrit

1 0

New patch to review for coreboot: b5326c9 GOOGLE/SNOW: fix stupid paren error
by Ronald G. Minnich April 22, 2013

April 22, 2013

Ronald G. Minnich (rminnich(a)gmail.com) just uploaded a new patch set to gerrit, which you can find at http://review.coreboot.org/3120 -gerrit commit b5326c99d6f1bbefe8ff499073675a4365dec529 Author: Ronald G. Minnich <rminnich(a)gmail.com> Date: Mon Apr 22 10:46:53 2013 -0700 GOOGLE/SNOW: fix stupid paren error This simple error led to corrupted graphics. How annoying. Change-Id: I2295c0df0f1d16014a603dc5d66bd4d72f3fb7c9 Signed-off-by: Ronald G. Minnich <rminnich(a)gmail.com> --- src/cpu/samsung/exynos5250/cpu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/cpu/samsung/exynos5250/cpu.c b/src/cpu/samsung/exynos5250/cpu.c index 09b670e..e122413 100644 --- a/src/cpu/samsung/exynos5250/cpu.c +++ b/src/cpu/samsung/exynos5250/cpu.c @@ -68,7 +68,7 @@ static void exynos_displayport_init(device_t dev) vi.cmap = (void *)lcdbase; lcdbase += 64*KiB; - mmio_resource(dev, 1, lcdbase/KiB, fb_size + (KiB-1)/KiB); + mmio_resource(dev, 1, lcdbase/KiB, (fb_size + KiB - 1)/KiB); printk(BIOS_DEBUG, "Initializing exynos VGA, base %p\n",(void *)lcdbase); ret = lcd_ctrl_init(&vi, &panel, (void *)lcdbase);

1 0

New patch to review for coreboot: e3999d5 AMD CIMx SB800: Remove declaration of unused `get_sbdn()`
by Paul Menzel April 22, 2013

April 22, 2013

Paul Menzel (paulepanter(a)users.sourceforge.net) just uploaded a new patch set to gerrit, which you can find at http://review.coreboot.org/3119 -gerrit commit e3999d58a62006f4b2b884d68544351b6023c4ec Author: Paul Menzel <paulepanter(a)users.sourceforge.net> Date: Fri Apr 19 10:23:03 2013 +0200 AMD CIMx SB800: Remove declaration of unused `get_sbdn()` No SB800 based board implements the function `get_sbdn()`, which was probably copied from SB700 code. So remove the declaration. Change-Id: I85fe3db0abf2648b6eb7a659f2112d130f856afa Signed-off-by: Paul Menzel <paulepanter(a)users.sourceforge.net> --- src/southbridge/amd/cimx/sb800/sb_cimx.h | 8 -------- 1 file changed, 8 deletions(-) diff --git a/src/southbridge/amd/cimx/sb800/sb_cimx.h b/src/southbridge/amd/cimx/sb800/sb_cimx.h index 3bec5d8..f07b0e0 100644 --- a/src/southbridge/amd/cimx/sb800/sb_cimx.h +++ b/src/southbridge/amd/cimx/sb800/sb_cimx.h @@ -40,12 +40,4 @@ int acpi_is_wakeup_early(void); */ void sb800_clk_output_48Mhz(void); -#if CONFIG_RAMINIT_SYSINFO -/** - * @brief Get SouthBridge device number, called by finalize_node_setup() - * @param[in] bus target bus number - * @return southbridge device number - */ -u32 get_sbdn(u32 bus); -#endif #endif

1 0