March 2006 - U-Boot

[U-Boot-Users] Valid EMAC address
by Mathieu Deschamps 29 Mar '06

29 Mar '06

Hello , I've just found that an invalid EMAC address -I guess- I've given to test, was the reason for all NET_COMMANDS (ping, tftpboot and the like) to gently fail without any errors whereas NET_DEBUG and ET_DEBUG were definied. Indeed, I've change supposed invalid addr 47:55:47:55:47:55:aa to 12:43:56:78:90:aa and it worked. So here I come and apolozige in advance if somebody already answered theses, I haven't found anything relevant in archives : How to determin if a EMAC is valid/legal ? Why u-boot doesn't implement such a validiy test -at least when debugging is defined- in order to defeat that strange developper reflex, looking for _the_ inexisting bug ? Best Regards, Mathieu Deschamps -- Com2gether Design Center Electronic and Embedded Engineering Services www.com2gether.net

4 5

AW: [U-Boot-Users] Re: combined C/ARM assembly example
by Jonas Mark (ST-FIR/ENG1) * 28 Mar '06

28 Mar '06

Hello Shakthi, > unsigned long *ptr1, *ptr2; > > /* not here */ > > asm volatile ("ldr r0,= 0x30800000"); > asm volatile ("ldr r3,= 0x30800004"); > asm volatile ("ldr r1,= 0xff000000"); > asm volatile ("movs r2, r1, lsl #1"); > asm volatile ("mrs r1, cpsr"); > asm volatile ("str r1, [r0, #0]"); > asm volatile ("str r2, [r3, #0]"); > > /* placed here */ > > ptr1 = (unsigned long *) (0x30800000); > ptr2 = (unsigned long *) (0x30800004); What you are facing is called "register clobbering". See http://gcc.gnu.org/onlinedocs/gcc-2.95.3/gcc_4.html#SEC93 and look for the keyword "clobber". At first the behavior of GCC's inline assembler might look evil but once you understood its power you are going to love it. Best Regards, Mark

1 0

[U-Boot-Users] Re: combined C/ARM assembly example
by Shakthi Kannan 28 Mar '06

28 Mar '06

Hi again, Got this working. Repositioned, ptr1, ptr2 assignment after the assembly usage: #include <common.h> #include <exports.h> int main (int argc, char *argv[]) { unsigned long *ptr1, *ptr2; /* not here */ asm volatile ("ldr r0,= 0x30800000"); asm volatile ("ldr r3,= 0x30800004"); asm volatile ("ldr r1,= 0xff000000"); asm volatile ("movs r2, r1, lsl #1"); asm volatile ("mrs r1, cpsr"); asm volatile ("str r1, [r0, #0]"); asm volatile ("str r2, [r3, #0]"); /* placed here */ ptr1 = (unsigned long *) (0x30800000); ptr2 = (unsigned long *) (0x30800004); printf ("*ptr1 = %8lx\n", *ptr1); printf ("*ptr2 = %8lx\n", *ptr2); return (0); } Regards, SK -- Shakthi Kannan, MS Software Engineer, Hexaware Technologies [E]: shakthimaan at yahoo dot com [M]: (91) 98407-87007 [W]: http://www.shakthimaan.com [L]: Chennai, India __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com

1 0

[U-Boot-Users] using static variables
by David Ho 27 Mar '06

27 Mar '06

Hi, I would appreciate if an u-boot expert can explain the use of static variables in u-boot. My understanding is that when u-boot is still running in flash static variables are not modifiable. Once u-boot moves to RAM, what will happen to static variables? I don't quite understand how the mechanism work before and after relocation. I asked this question because I need a saved copy of a write only latch/register, which I have to initialize early. I have added a field in the global_data struct previously for this. Seeing that introduced changes in the common code I would like to choose a method I can use to localize it in my board specific code. Will I be able to use static variables in board_early_init_r? Are there some other mehods I have not noticed? Regards, David

4 5

[U-Boot-Users] setup cfi flash for linux
by Hiroki Kaminaga 27 Mar '06

27 Mar '06

Hi, I'm implementing snapshot boot for OSK5912 target. It's basically extract system snapshot image, do init/setup at uboot, and jump to kernel resume entry point. Detail here: http://linux.omap.com/pipermail/linux-omap-open-source/2006-March/006599.ht… Currently, I get this message from linux: * * * waiting for erase at 00240000 to complete timed out. status = 0, Xstatus = 80. mtdblock: erase of region [0x0, 0x20000] on "filesystem" failed end_request: I/O error, dev mtdblock3, sector 256 * * * (apart from this, prompt is shown, command gets executed, and ping/telnet from host is working fine.) read/write from/to flash is working in uboot, (md.[bwl], mw.[bwl]), so what is linux complaining? Can someone tell me, is linux doing additional setups for cfi flash? Thanks in advcance. (Hiroki Kaminaga) t --

2 1

[U-Boot-Users] Réf. : Re: [U-Boot-Users] get_ram_size bug and ask for comments
by Matthieu CASTET 27 Mar '06

27 Mar '06

Hi Wolfgang, it seems the code I have was modified. I look in the original cvs version and it seems that the base address isn't restored either. [...] addr = base; save[i] = *addr; <-- save value at base address *addr = 0; <--- modify it. if ((val = *addr) != 0) { <--- error case : restore all values [...] } <-- we forget to restore the value and start to restore value @base+1 with save[i-1]. for (cnt = 1; cnt < maxsize / sizeof (long); cnt <<= 1) { addr = base + cnt; /* pointer arith! */ val = *addr; *addr = save[--i]; if (val != ~cnt) { size = cnt * sizeof (long); /* Restore the original data before leaving the function. */ for (cnt <<= 1; cnt < maxsize / sizeof (long); cnt <<= 1) { addr = base + cnt; *addr = save[--i]; } return (size); } } regards, Matthieu

1 0

[U-Boot-Users] Usbtty with Windows problem
by Vovan 25 Mar '06

25 Mar '06

Hello , I work with OMAP310 based board and I need to get u-boot usbtty console work with Windows. In linux it works fine, but with windows driver (usbser.sys) it hangs when sending from u-boot to host big amount of data, greater than USBTTY_BUFFER_SIZE. When reconnecting again - the rest of data is recieved. Do anybody know where to start ? Usbmon shows that after USBTTY_BUFFER_SIZE data is received to host all the rest usb packets are empty. So console hangs even when typing HELP command in it. I tried to make uboot look like ACM modem, (added acm class desctiptors to USB configuration) but it didn`t help. -- Best regards, Vovan mailto:vovan888@gmail.com

1 0

[U-Boot-Users] Cannot erase flash sector
by Joe Culler 25 Mar '06

25 Mar '06

Hi there, I have a question about erasing flash sectors, please tell me what steps I did wrong, thank you. The following is my boad using u-boot command "flinfo": Bank # 1: SGS Thomson PSD4256G (1 M, uniform sector) Size: 1 MB in 16 Sectors Sector Start Addresses: 20000000 (RO) 20010000 (RO) 20020000 20030000 20040000 20050000 20060000 20070000 20080000 20090000 200A0000 E 200B0000 E 200C0000 E 200D0000 E 200E0000 E 200F0000 (RO) If I want to erase sector from 20020000 to 20090000 for example, I use the command "erase 20020000 2009ffff" to erase flash, it seems to be ok from erase message: Erase Flash from 0x20020000 to 0x2009ffff .............. done Erased 8 sectors But when I use "flinfo" command to see whether those sectors are erased or not, it shows the same above message: Bank # 1: SGS Thomson PSD4256G (1 M, uniform sector) Size: 1 MB in 16 Sectors Sector Start Addresses: 20000000 (RO) 20010000 (RO) 20020000 20030000 20040000 20050000 20060000 20070000 20080000 20090000 200A0000 E 200B0000 E 200C0000 E 200D0000 E 200E0000 E 200F0000 (RO) If I understand correct, sectors from 20020000 to 20090000 should be displayed (E), but those aren't. I don't know what step I did wrong. Regards, Joe.

2 4

[U-Boot-Users] mpc tsec driver and generic PHYs
by David Updegraff 24 Mar '06

24 Mar '06

Hi. With a patch like this, generic 10/100 PHYs seem to behave with TSEC, so you dont have to ennumerate all possible. Scalped from linux phy stuff. -dbu.

1 1

[U-Boot-Users] get_ram_size bug and ask for comments
by Matthieu CASTET 24 Mar '06

24 Mar '06

Hi, the current get_ram_size algorithm doesn't restore the first word of the memory. Something like [1] should be done to prevent that. Also is that possible to get some comment s on the algorithm ? Why save2 is never used to restore values ? Thanks Matthieu [1] --- ../../cvs/uboot/common/memsize.c 2006-03-01 10:37:06.000000000 +0100 +++ memsize.c 2006-03-24 14:41:07.116217000 +0100 @@ -60,6 +60,7 @@ } return (0); } + *addr = save[i]; for (cnt = 1024; cnt < maxsize / sizeof (long); cnt <<= 1) { addr = base + cnt; /* pointer arith! */

2 1