January 2013 - U-Boot

[U-Boot] [PATCH V2] drivers/block/systemace - added missing "else" in "ace_writew"
by Alexey Brodkin 09 Jan '13

09 Jan '13

System ACE compact flash controller supports either 8-bit (default) or 16-bit data transfers. And in corresponding driver we need to implement read/write of 16-bit data words properly for both modes of operation. In existing code if width==8 both branches get executed which may cause unexpected behavior of SystemAce controller. Addition of "else" fixes described issue and execution is done as expected for both (8-bit and 16-bit) data bus widths. Signed-off-by: Alexey Brodkin <alexey.brodkin(a)gmail.com> --- drivers/block/systemace.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/block/systemace.c b/drivers/block/systemace.c index 247cf06..27cacdf 100644 --- a/drivers/block/systemace.c +++ b/drivers/block/systemace.c @@ -66,7 +66,8 @@ static void ace_writew(u16 val, unsigned off) writeb(val >> 8, base + off + 1); #endif } - out16(base + off, val); + else + out16(base + off, val); } static u16 ace_readw(unsigned off) -- 1.7.10.4

2 1

[U-Boot] problem displaying bitmap on i.MX6
by Carolyn Smith 09 Jan '13

09 Jan '13

Hello, I have a synchronous LCD panel connected to an LVDS channel on an i.MX6 and am trying to display a bitmap during the startup of U-Boot. The bitmap is 800x480 and is being correctly unpacked into memory as a 565RGB image. The display_info structure is static struct display_info_t const displays[] = { { .pixfmt = IPU_PIX_FMT_RGB666, .enable = enable_rgb, .mode = { .name = "wvga-rgb", .refresh = 60, .xres = 800, .yres = 480, .pixclock = 33614, /* 35088, */ .left_margin = 80, .right_margin = 48, .upper_margin = 6, .lower_margin = 3, .hsync_len = 32, .vsync_len = 7, .sync = FB_SYNC_HOR_HIGH_ACT, .vmode = FB_VMODE_NONINTERLACED } } }; The colors of the bitmap are being displayed properly but it is being stretched horizontally such that (roughly) the left-most 350 columns of the image fill the 800 column wide screen. The pixclock value was changed from 35088 to 33614 and that got us closer to correct operation but we still have the stretching problem. Could the stretching be caused by the pixclock still being wrong? Is there something else I should be looking at that might be causing this? Any suggestions would be welcome. Thank you, Carolyn

1 0

[U-Boot] [PATCH] mx51evk: Let Add DVI output support
by Fabio Estevam 09 Jan '13

09 Jan '13

Add DVI output support and make it the default video output. Currently the CLAA WVGA panel is supported, but this panel has to be purchased separately, so using the DVI output as the default would allow more people to try the splash screen feature on a mx51evk. If someone still wants to use the CLAA WVGA, just set the panel variable as: set panel claa Signed-off-by: Fabio Estevam <fabio.estevam(a)freescale.com> --- board/freescale/mx51evk/mx51evk.c | 2 -- board/freescale/mx51evk/mx51evk_video.c | 39 ++++++++++++++++++++++++++++--- 2 files changed, 36 insertions(+), 5 deletions(-) diff --git a/board/freescale/mx51evk/mx51evk.c b/board/freescale/mx51evk/mx51evk.c index d1ef431..54c16b1 100644 --- a/board/freescale/mx51evk/mx51evk.c +++ b/board/freescale/mx51evk/mx51evk.c @@ -489,8 +489,6 @@ int board_init(void) /* address of boot parameters */ gd->bd->bi_boot_params = PHYS_SDRAM_1 + 0x100; - lcd_enable(); - return 0; } diff --git a/board/freescale/mx51evk/mx51evk_video.c b/board/freescale/mx51evk/mx51evk_video.c index f036cf7..1583131 100644 --- a/board/freescale/mx51evk/mx51evk_video.c +++ b/board/freescale/mx51evk/mx51evk_video.c @@ -48,6 +48,22 @@ static struct fb_videomode const claa_wvga = { .vmode = FB_VMODE_NONINTERLACED }; +static struct fb_videomode const dvi = { + .name = "DVI panel", + .refresh = 60, + .xres = 1024, + .yres = 768, + .pixclock = 15385, + .left_margin = 220, + .right_margin = 40, + .upper_margin = 21, + .lower_margin = 7, + .hsync_len = 60, + .vsync_len = 10, + .sync = 0, + .vmode = FB_VMODE_NONINTERLACED +}; + void setup_iomux_lcd(void) { /* DI2_PIN15 */ @@ -73,9 +89,26 @@ void setup_iomux_lcd(void) gpio_direction_output(MX51EVK_LCD_BACKLIGHT, 1); } -void lcd_enable(void) +int board_video_skip(void) { - int ret = ipuv3_fb_init(&claa_wvga, 1, IPU_PIX_FMT_RGB565); + int ret; + char const *e = getenv("panel"); + + if (e) { + if (strcmp(e, "claa") == 0) { + ret = ipuv3_fb_init(&claa_wvga, 1, IPU_PIX_FMT_RGB565); + if (ret) + printf("claa cannot be configured: %d\n", ret); + return ret; + } + } + + /* + * 'panel' env variable not found or has different value than 'claa' + * Defaulting to dvi output. + */ + ret = ipuv3_fb_init(&dvi, 0, IPU_PIX_FMT_RGB24); if (ret) - printf("LCD cannot be configured: %d\n", ret); + printf("dvi cannot be configured: %d\n", ret); + return ret; } -- 1.7.9.5

1 0

[U-Boot] [PATCH v4 0/6] Improve default environment for easy use
by Otavio Salvador 09 Jan '13

09 Jan '13

This patchset improves the default environment of machines currently supported in OpenEmbedded and Yocto to make it compatible with mainline and Freescale BSP Linux kernel. A new patch has been add to change the format in mx6qsabrelite in environment setting and already merged patches removed from this new pull request. Changes in v4: - Use a warning, instead of error when failing to fetch fdt - Drop exit use - Fix netboot - Use fdt instead of ftd Changes in v3: - Add loadfdt - Use loadfdt - Call exit to abort script Changes in v2: - New patch - reformat environment change to fit in 80 cols - Allow use of dynamic/static ip - Allow force use, or not, of fdt - Change 'auto' to 'try' Otavio Salvador (6): mx6qsabrelite: Use tabs to environment setting mx28evk: We shouldn't hardcode a rootfs filesystem type mx28evk: Add support to dynamically choose between ftd use or not mx53loco: Add support to dynamically choose between ftd use or not mx6qsabrelite: Add support to dynamically choose between ftd use or not mx6qsabre{auto,sd}: Add support to dynamically choose between ftd use or not include/configs/mx28evk.h | 47 ++++++++++++++--- include/configs/mx53loco.h | 40 +++++++++++++- include/configs/mx6qsabre_common.h | 41 +++++++++++++-- include/configs/mx6qsabreauto.h | 1 + include/configs/mx6qsabrelite.h | 101 ++++++++++++++++++++++++------------ include/configs/mx6qsabresd.h | 1 + 6 files changed, 186 insertions(+), 45 deletions(-) -- 1.7.10.4

2 8

Re: [U-Boot] uboot and Ethernet Controller issue
by Wolfgang Denk 09 Jan '13

09 Jan '13

Dear Sandeep C R, In message <CAEX5Uh-0BrxKHjNU-o7Oix9UjQDn1w34o0XNFPhRQibXDuLjSg(a)mail.gmail.com> you wrote: > > The porting was done long back and now linux (2.6.24) also was ported. The > observation is still there with the Linux also. Please let me know if it > has to do with some configuration of 82551. Sorry, I fail to understand why you are now porting Linux v2.6.24, which is also 5 years old... ... > > > I have developed a PowerPC MPC7410 based processor board with Tsi107 as > > the > > > Host bridge and intel 82551 the ethernet controller. We have ported > > > u-boot-1.3.1 into it as its boot loader. > > > > U-Boot v1.3.1? But this is more than 5 years old!! > > > > Why did you chose such a prehistoric version? > > > > > about 90 minutes everytime. Is it something to do with the config of > > > ethernet controller? Please advice. > > > > Please update to current code (like v2013.01-rc2 or so), and try > > again. As your code is not known tous, it is impossible to even guess what might be the cause. Also, we cannot remember all bugs fixed in U-Boot and Linux for the last 5 years - for example, for U-Boot since v1.3.1 we have added 14238 commits from 727 developers; a total of 1619599 lines has been added, 1030461 removed, and 589138 changed. It would really be best to update and use current code instead. The chances for free community help are very small for 5 year old out-of-tree code. Sorry. Best regards, Wolfgang Denk -- DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd(a)denx.de No journaling file system can recover your data if the disk dies. - Steve Rago in <D4Cw1p.L9E(a)plc.com>

1 0

[U-Boot] [PATCH] fs: fat: Fix mkcksum() function parameters
by Marek Vasut 09 Jan '13

09 Jan '13

The mkcksum() function now takes one parameter, the pointer to 11-byte wide character array, which it then operates on. Currently, the function is wrongly passed (dir_entry)->name, which is only 8-byte wide character array. Though by further inspecting the dir_entry structure, it can be noticed that the name[8] entry is immediatelly followed by ext[3] entry. Thus, name[8] and ext[3] in the dir_entry structure actually work as this 11-byte wide array since they're placed right next to each other by current compiler behavior. Depending on this is obviously wrong, thus fix this by correctly passing both (dir_entry)->name and (dir_entry)->ext to the mkcksum() function and adjust the function appropriately. Signed-off-by: Marek Vasut <marex(a)denx.de> Cc: Tom Rini <trini(a)ti.com> --- fs/fat/fat.c | 20 ++++++++++++-------- fs/fat/fat_write.c | 2 +- 2 files changed, 13 insertions(+), 9 deletions(-) diff --git a/fs/fat/fat.c b/fs/fat/fat.c index 41ae15e..8870292 100644 --- a/fs/fat/fat.c +++ b/fs/fat/fat.c @@ -567,15 +567,16 @@ get_vfatname(fsdata *mydata, int curclust, __u8 *cluster, } /* Calculate short name checksum */ -static __u8 mkcksum(const char *str) +static __u8 mkcksum(const char name[8], const char ext[3]) { int i; __u8 ret = 0; - for (i = 0; i < 11; i++) { - ret = (((ret & 1) << 7) | ((ret & 0xfe) >> 1)) + str[i]; - } + for (i = 0; i < sizeof(name); i++) + ret = (((ret & 1) << 7) | ((ret & 0xfe) >> 1)) + name[i]; + for (i = 0; i < sizeof(ext); i++) + ret = (((ret & 1) << 7) | ((ret & 0xfe) >> 1)) + ext[i]; return ret; } @@ -678,7 +679,8 @@ static dir_entry *get_dentfromdir(fsdata *mydata, int startsect, return NULL; } #ifdef CONFIG_SUPPORT_VFAT - if (dols && mkcksum(dentptr->name) == prevcksum) { + __u8 csum = mkcksum(dentptr->name, dentptr->ext); + if (dols && csum == prevcksum) { prevcksum = 0xffff; dentptr++; continue; @@ -946,13 +948,16 @@ do_fat_read_at(const char *filename, unsigned long pos, void *buffer, for (i = 0; i < DIRENTSPERBLOCK; i++) { char s_name[14], l_name[VFAT_MAXLEN_BYTES]; + __u8 csum; l_name[0] = '\0'; if (dentptr->name[0] == DELETED_FLAG) { dentptr++; continue; } - if ((dentptr->attr & ATTR_VOLUME)) { + + csum = mkcksum(dentptr->name, dentptr->ext); + if (dentptr->attr & ATTR_VOLUME) { #ifdef CONFIG_SUPPORT_VFAT if ((dentptr->attr & ATTR_VFAT) == ATTR_VFAT && (dentptr->name[0] & LAST_LONG_ENTRY_MASK)) { @@ -1015,8 +1020,7 @@ do_fat_read_at(const char *filename, unsigned long pos, void *buffer, goto exit; } #ifdef CONFIG_SUPPORT_VFAT - else if (dols == LS_ROOT && - mkcksum(dentptr->name) == prevcksum) { + else if (dols == LS_ROOT && csum == prevcksum) { prevcksum = 0xffff; dentptr++; continue; diff --git a/fs/fat/fat_write.c b/fs/fat/fat_write.c index 5829adf..97a18cb 100644 --- a/fs/fat/fat_write.c +++ b/fs/fat/fat_write.c @@ -335,7 +335,7 @@ fill_dir_slot(fsdata *mydata, dir_entry **dentptr, const char *l_name) /* Get short file name and checksum value */ strncpy(s_name, (*dentptr)->name, 16); - checksum = mkcksum(s_name); + checksum = mkcksum(*dentptr->name, *dentptr->ext); do { memset(slotptr, 0x00, sizeof(dir_slot)); -- 1.7.10.4

4 5

[U-Boot] [PATCH 2/2] calimain: Generate random MAC address for factory tests
by Christian Riesch 09 Jan '13

09 Jan '13

Signed-off-by: Christian Riesch <christian.riesch(a)omicron.at> --- board/omicron/calimain/calimain.c | 31 ++++++++++++++++++++++++++++++- include/configs/calimain.h | 2 ++ 2 files changed, 32 insertions(+), 1 deletion(-) diff --git a/board/omicron/calimain/calimain.c b/board/omicron/calimain/calimain.c index 1060a1f..80e3893 100644 --- a/board/omicron/calimain/calimain.c +++ b/board/omicron/calimain/calimain.c @@ -1,5 +1,5 @@ /* - * Copyright (C) 2011 OMICRON electronics GmbH + * Copyright (C) 2011-2013 OMICRON electronics GmbH * * Based on da850evm.c. Original Copyrights follow: * @@ -136,6 +136,35 @@ int board_init(void) return 0; } +/* seed random number generator with uninitialized SRAM content */ +static void srand_sram(void) +{ + int *p; + int seed = 0; + + for (p = (int *) 0x80000000; p < (int *) 0x8001ffff; p++) + seed ^= *p; + + srand(seed); +} + +int board_late_init(void) +{ + uchar enetaddr[6]; + + if (!eth_getenv_enetaddr("ethaddr", enetaddr)) { + srand_sram(); + eth_random_enetaddr(enetaddr); + if (eth_setenv_enetaddr("ethaddr", enetaddr)) { + printf("Failed to set random ethernet address\n"); + } else { + printf("Setting random ethernet address %pM.\n", + enetaddr); + } + } + return 0; +} + #ifdef CONFIG_DRIVER_TI_EMAC /* * Initializes on-board ethernet controllers. diff --git a/include/configs/calimain.h b/include/configs/calimain.h index 5c2b35d..8cea0d9 100644 --- a/include/configs/calimain.h +++ b/include/configs/calimain.h @@ -30,6 +30,7 @@ #define CONFIG_DRIVER_TI_EMAC #define MACH_TYPE_CALIMAIN 3528 #define CONFIG_MACH_TYPE MACH_TYPE_CALIMAIN +#define CONFIG_BOARD_LATE_INIT /* * SoC Configuration @@ -202,6 +203,7 @@ #define CONFIG_BOOTP_DNS2 #define CONFIG_BOOTP_SEND_HOSTNAME #define CONFIG_NET_RETRY_COUNT 10 +#define CONFIG_RANDOM_MACADDR #endif /* -- 1.7.9.5

2 3

[U-Boot] [PATCH 1/2] net: Remove call of srand from eth_random_enetaddr()
by Christian Riesch 09 Jan '13

09 Jan '13

Currently eth_random_enetaddr() seeds the random number generator with get_timer(0). Some boards might want to use other sources for the seed, therefore move the call of srand() to the board specific code. Signed-off-by: Christian Riesch <christian.riesch(a)omicron.at> Cc: Michael Walle <michael(a)walle.cc> Cc: Joe Hershberger <joe.hershberger(a)gmail.com> --- board/buffalo/lsxl/lsxl.c | 1 + include/net.h | 3 +++ net/eth.c | 2 -- 3 files changed, 4 insertions(+), 2 deletions(-) diff --git a/board/buffalo/lsxl/lsxl.c b/board/buffalo/lsxl/lsxl.c index 57776fb..b7eb0dc 100644 --- a/board/buffalo/lsxl/lsxl.c +++ b/board/buffalo/lsxl/lsxl.c @@ -248,6 +248,7 @@ static void rescue_mode(void) printf("Entering rescue mode..\n"); #ifdef CONFIG_RANDOM_MACADDR if (!eth_getenv_enetaddr("ethaddr", enetaddr)) { + srand(get_timer(0)); eth_random_enetaddr(enetaddr); if (eth_setenv_enetaddr("ethaddr", enetaddr)) { printf("Failed to set ethernet address\n"); diff --git a/include/net.h b/include/net.h index 970d4d1..5fc3693 100644 --- a/include/net.h +++ b/include/net.h @@ -141,6 +141,9 @@ extern int eth_getenv_enetaddr_by_index(const char *base_name, int index, * * In these cases, we generate a random locally administered ethernet address. * + * Remember to seed the random number generator with srand() before calling + * this functon. + * * Args: * enetaddr - returns 6 byte hardware address */ diff --git a/net/eth.c b/net/eth.c index 321d5b1..dc4cc20 100644 --- a/net/eth.c +++ b/net/eth.c @@ -84,8 +84,6 @@ void eth_random_enetaddr(uchar *enetaddr) { uint32_t rval; - srand(get_timer(0)); - rval = rand(); enetaddr[0] = rval & 0xff; enetaddr[1] = (rval >> 8) & 0xff; -- 1.7.9.5

2 3

[U-Boot] Selecting from multiple device trees at runtime
by Curt Brune 09 Jan '13

09 Jan '13

Hello, I am following up on this thread: "Merging device trees at runtime for module-based systems" http://lists.denx.de/pipermail/u-boot/2012-November/139618.html I do not have a modular based system, so I do not need the full flexibility of merging DT fragments at runtime. However, I am very interested in being able to select a full DT from a list of DT's at runtime. I believe this is a degenerate case of the more general N x M modular case. The background: I have a number of different platforms running u-boot. They all use the same Linux kernel and initramfs, but different DT. What I would love is to have a single multi-file uImage I could use on all my platforms. The idea is to introduce a new image type that is a list of device tree blobs. The uImage would contain a list of dtb's and u-boot would select the correct one at runtime. u-boot could iterate through the list inspecting the "model" property of the root node. Thoughts on that approach? Cheers, Curt

4 12

[U-Boot] [PATCH] video:cache:fix: Enable dcache flush at LCD subsystem for Exynos based boards
by Lukasz Majewski 09 Jan '13

09 Jan '13

lcd_set_flush_dcache(1) function is called from exynos_fb.c (exynos_lcd_init()) file to enable lcd_sync() function. Tested-at: TRATS (Exynos 4210) Tested-by: Lukasz Majewski <l.majewski(a)samsung.com> Signed-off-by: Lukasz Majewski <l.majewski(a)samsung.com> Signed-off-by: Kyungmin Park <kyungmin.park(a)samsung.com> Cc: Minkyu Kang <mk7.kang(a)samsung.com> --- drivers/video/exynos_fb.c | 2 ++ 1 files changed, 2 insertions(+), 0 deletions(-) diff --git a/drivers/video/exynos_fb.c b/drivers/video/exynos_fb.c index d9a3f9a..17e1069 100644 --- a/drivers/video/exynos_fb.c +++ b/drivers/video/exynos_fb.c @@ -63,6 +63,8 @@ static void exynos_lcd_init_mem(void *lcdbase, vidinfo_t *vid) static void exynos_lcd_init(vidinfo_t *vid) { exynos_fimd_lcd_init(vid); + /* Enable flushing after LCD writes if requested */ + lcd_set_flush_dcache(1); } static void draw_logo(void) -- 1.7.2.3

2 3