November 2005 - U-Boot

Re: [U-Boot-Users] is there any other toolchains for u-boot of mips arch?
by Grant Likely 23 Nov '05

23 Nov '05

Some quick etiquette points if your going to use the mailing list: 1. Put your replies *below* what your replying to, not above. 2. Always CC the mailing list on replies 3. Don't just describe what you saw; attach logs also. zhuzhenhua wrote: > i do not get clear about it > $ export PATH=/path/to/mips-linux-as/bin:$PATH > > /path/to/mips-linux-as/bin is what path? If mips-linux-as is in the path /opt/mips-linux/bin, then: $ export PATH=/opt/mips-linux/bin:$PATH Basically, your tools should be in your path so you can run the tools from the command line like this: $ mips-linux-as > > do you mean change my toolchain dir name to mips-linux-as? I think your confused. What is the full path of your cross compiler? Here's an example: My cross toolchain is: powerpc-linux It lives in: /opt/powerpc-linux The path to the cross gcc is: /opt/powerpc-linux/bin/powerpc-linux-gcc Therefore I set in my environment: $ PATH=/opt/powerpc-linux/bin:$PATH $ CROSS_COMPILE=powerpc-linux- $ export PATH CROSS_COMPILE

2 1

[U-Boot-Users] if the cache is on,where we need to add the cache flush code?
by zhuzhenhua 23 Nov '05

23 Nov '05

in the cpu/mips/start.S if the cache is on, where need to add the cache flush code? only flush dcache? or both dcacahe and icache? i add cache flush after the copy loop, i wonder whether i need to add cache flush after the GOT table relocate and bss clean as follow is the relocate_code in cpu/mips/start.S, i add cache flush one place,is that right? best regards zhuzhenhua ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ relocate_code: move sp, a0 /* Set new stack pointer */ li t0, CFG_MONITOR_BASE la t3, in_ram lw t2, -12(t3) /* t2 <-- uboot_end_data */ move t1, a2 /* * Fix GOT pointer: * * New GOT-PTR = (old GOT-PTR - CFG_MONITOR_BASE) + Destination Address */ move t6, gp sub gp, CFG_MONITOR_BASE add gp, a2 /* gp now adjusted */ sub t6, gp, t6 /* t6 <-- relocation offset */ /* * t0 = source address * t1 = target address * t2 = source end address */ /* On the purple board we copy the code earlier in a special way * in order to solve flash problems */ #ifndef CONFIG_PURPLE 1: lw t3, 0(t0) sw t3, 0(t1) addu t0, 4 ble t0, t2, 1b addu t1, 4 /* delay slot */ #endif /* If caches were enabled, we would have to flush them here. */ /*~~~~~~~~~~~~~~~~ as follow is my cache flush code */ #ifdef CONFIG_MYBOARD li t1, (16*1024) #cache total size li t2, 16 #cache line size li t3, 0x80000000 addu t4, t1, t3 cacheloop1: cache 0, 0(t3) /* Index_Invalidate_I */ cache 1, 0(t3) /* Index_Writeback_Inv_D */ addu t3, t2 bne t3, t4, cacheloop1 nop #endif /* Jump to where we've relocated ourselves. */ addi t0, a2, in_ram - _start j t0 nop .word uboot_end_data .word uboot_end .word num_got_entries in_ram: /* Now we want to update GOT. */ lw t3, -4(t0) /* t3 <-- num_got_entries */ addi t4, gp, 8 /* Skipping first two entries. */ li t2, 2 1: lw t1, 0(t4) beqz t1, 2f add t1, t6 sw t1, 0(t4) 2: addi t2, 1 blt t2, t3, 1b addi t4, 4 /* delay slot */ /* Clear BSS. */ lw t1, -12(t0) /* t1 <-- uboot_end_data */ lw t2, -8(t0) /* t2 <-- uboot_end */ add t1, t6 /* adjust pointers */ add t2, t6 sub t1, 4 1: addi t1, 4 bltl t1, t2, 1b sw zero, 0(t1) /* delay slot */ #if 0 nop sync nop #endif move a0, a1 la t9, board_init_r j t9 move a1, a2 /* delay slot */ .end relocate_code /* Exception handlers. */ romReserved: b romReserved romExcHandle: b romExcHandle

1 0

[U-Boot-Users] why the = will effect gd in board_init_f but do not effect in board_init_r
by zhuzhenhua 23 Nov '05

23 Nov '05

in my disassemble file , i find the operation of k0 in board_init_f but not find in board_init_r, why? is there any other toolchain to compile u-boot for mips? bfc008e8 <board_init_f>: bfc008e8: 3c1c0001 lui gp,0x1 bfc008ec: 279cdef8 addiu gp,gp,-8456 bfc008f0: 0399e021 addu gp,gp,t9 bfc008f4: 27bdffa8 addiu sp,sp,-88 bfc008f8: afbc0010 sw gp,16(sp) bfc008fc: 8f820124 lw v0,292(gp) bfc00900: afb10044 sw s1,68(sp) bfc00904: afbf0054 sw ra,84(sp) bfc00908: afbc0050 sw gp,80(sp) bfc0090c: afb3004c sw s3,76(sp) bfc00910: afb20048 sw s2,72(sp) bfc00914: afb00040 sw s0,64(sp) bfc00918: 3c114040 lui s1,0x4040 bfc0091c: 02228821 addu s1,s1,v0 bfc00920: 27ba0018 addiu k0,sp,24 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ bfc00bb4 <board_init_r>: bfc00bb4: 3c1c0001 lui gp,0x1 bfc00bb8: 279cdc2c addiu gp,gp,-9172 bfc00bbc: 0399e021 addu gp,gp,t9 bfc00bc0: 27bdffd8 addiu sp,sp,-40 bfc00bc4: afbc0010 sw gp,16(sp) bfc00bc8: afb00018 sw s0,24(sp) bfc00bcc: 00a08021 move s0,a1 bfc00bd0: 00802821 move a1,a0 bfc00bd4: 8f84000c lw a0,12(gp) bfc00bd8: 00000000 nop bfc00bdc: 2484c520 addiu a0,a0,-15072 bfc00be0: afbf0020 sw ra,32(sp) bfc00be4: afbc001c sw gp,28(sp) bfc00be8: 8f9901b0 lw t9,432(gp) bfc00bec: 0320f809 jalr t9 bfc00bf0: 00000000 nop bfc00bf4: 8fbc0010 lw gp,16(sp) bfc00bf8: 8f84000c lw a0,12(gp) bfc00bfc: 00000000 nop bfc00c00: 2484c52c addiu a0,a0,-15060 bfc00c04: 8f990074 lw t9,116(gp) bfc00c08: 0320f809 jalr t9 bfc00c0c: 00000000 nop bfc00c10: 8fbc0010 lw gp,16(sp) bfc00c14: 8f420004 lw v0,4(k0) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ this is gd->flags |= GD_FLG_RELOC; but at this time k0 is still as in board_init_f, bfc00c18: 3c034040 lui v1,0x4040 bfc00c1c: 00701821 addu v1,v1,s0 bfc00c20: 8f84000c lw a0,12(gp)

2 3

Re: [U-Boot-Users] SHA1 Hash calculation
by Tom Saarnio 23 Nov '05

23 Nov '05

>> >> I am wondering whether it would be worth to implement a SHA1 hash >> calculation over given memory address range. In this way one could >> assure for example that the binary image version is correct or that >> binary image is intact. > >Is the existing CRC32 checksum not good enough for your purposes? The CRC32 is not collision-proof in similar fashion that SHA1. In real life, if two binary images give the same SHA1 checksum, you can say that these binaries are identical. With CRC32 this is not so - or at least I would not count on that :-) br, Tom

2 1

[U-Boot-Users] SHA1 Hash calculation
by Tom Saarnio 23 Nov '05

23 Nov '05

Hi! I am wondering whether it would be worth to implement a SHA1 hash calculation over given memory address range. In this way one could assure for example that the binary image version is correct or that binary image is intact. For those who are not familiar waith SHA1 (or other hash algorithms), the SHA1 algorithm computes a 160 bit "fingerprint" of given byte stream. The algorithm is designed so that collisions between any given byte stream is "unlikely" as it is very unlikely that two persons have identical fingerprints or dna. Wikipedia: http://en.wikipedia.org/wiki/SHA-1 Source code: http://www. cr0.net:8040/code/crypto/sha1/ Any comments? br, Tom

2 1

[U-Boot-Users] bootp resetting serverip
by Tehn Yit Chin 23 Nov '05

23 Nov '05

Hi all, Our environment is such that the TFTP server is a different machine to the DHCP server. So when I use the bootp command, serverip is set to the ip address of the DHCP server. This will not as it will attempt to transfer the file from the DHCP server instead of the TFTP server. To overcome this issue, I have modify the BootpCopyNetParams() to check if serverip is empty before copying. The patch is below and attached. --- old_u-boot/net/bootp.c 2005-11-23 15:40:52.000000000 +1100 +++ u-boot/net/bootp.c 2005-11-23 15:26:52.000000000 +1100 @@ -121,8 +121,15 @@ NetCopyIP(&NetOurIP, &bp->bp_yiaddr); NetCopyIP(&tmp_ip, &bp->bp_siaddr); - if (tmp_ip != 0) - NetCopyIP(&NetServerIP, &bp->bp_siaddr); + if (tmp_ip != 0) { + + /* check that serverip is empty before setting the value as + ** the DHCP server could be a different machine. + */ + if (getenv("serverip") == NULL) { + NetCopyIP(&NetServerIP, &bp->bp_siaddr); + } + } memcpy (NetServerEther, ((Ethernet_t *)NetRxPkt)->et_src, 6); if (strlen(bp->bp_file) > 0) copy_filename (BootFile, bp->bp_file, sizeof(BootFile)); So if serverip is not empty, u-boot will assume that you know the ip address of the server, and have set it explicitly. Happy for rocks to be thrown if a different solution exits. cheers, -- Tehn Yit Chin Software Engineer, Grey Innovation Pty. Ltd. --- old_u-boot/net/bootp.c 2005-11-23 15:40:52.000000000 +1100 +++ u-boot/net/bootp.c 2005-11-23 15:26:52.000000000 +1100 @@ -121,8 +121,15 @@ NetCopyIP(&NetOurIP, &bp->bp_yiaddr); NetCopyIP(&tmp_ip, &bp->bp_siaddr); - if (tmp_ip != 0) - NetCopyIP(&NetServerIP, &bp->bp_siaddr); + if (tmp_ip != 0) { + + /* check that serverip is empty before setting the value as + ** the DHCP server could be a different machine. + */ + if (getenv("serverip") == NULL) { + NetCopyIP(&NetServerIP, &bp->bp_siaddr); + } + } memcpy (NetServerEther, ((Ethernet_t *)NetRxPkt)->et_src, 6); if (strlen(bp->bp_file) > 0) copy_filename (BootFile, bp->bp_file, sizeof(BootFile));

2 1

[U-Boot-Users] is there any other toolchains for u-boot of mips arch?
by zhuzhenhua 23 Nov '05

23 Nov '05

to use other toolchain? what need to do? i change the CROSS_COMPILE in Makefile and output the bin directory of the new toolchain and only get the error likes that /bin/sh: line 1: mips-linux-as : command not found regards zhuzhenhua

3 2

[U-Boot-Users] in board_init_r the gd still keep the value as in board_init_f?
by zhuzhenhua 23 Nov '05

23 Nov '05

to everyone in board.c(lib_mips/) i debug in it and in board_init_r, i find that after gd=id ,the gd still keep the value as in board_init_f, and not equal to id, so it now hang at mem_malloc_init(). i see in the gd is defined as k0 register, so i watch the k0 value to see if the gd is correct. does someone meet this situation? regards zhuzhenhua

1 0

[U-Boot-Users] can gd cause hang in board_init_r?
by zhuzhenhua 23 Nov '05

23 Nov '05

in board_init_r(lib_mips/board.c) i add three serial out lines serial_puts("before gd\n"); printf("gd =%x \n", gd); serial_puts("after cmd relocate!\n"); sometimes, it only output first line,sometimes the first two lines, sometimes the three lines are all outputed, are these caused by gd operation? or other reasons? thanks for any hints Regards zhuzhenhua my board_init_r as follows void board_init_r (gd_t *id, ulong dest_addr) { DECLARE_GLOBAL_DATA_PTR; cmd_tbl_t *cmdtp; ulong size; extern void malloc_bin_reloc (void); #ifndef CFG_ENV_IS_NOWHERE extern char * env_name_spec; #endif char *s, *e; bd_t *bd; int i; serial_puts("before gd\n"); gd = id; printf("gd =%x \n", gd); gd->flags |= GD_FLG_RELOC; /* tell others: relocation done */ //debug ("Now running in RAM - U-Boot at: %08lx\n", dest_addr); gd->reloc_off = dest_addr - CFG_MONITOR_BASE; monitor_flash_len = (ulong)&uboot_end_data - dest_addr; /* * We have to relocate the command table manually */ for (cmdtp = &__u_boot_cmd_start; cmdtp != &__u_boot_cmd_end; cmdtp++) { ulong addr; addr = (ulong) (cmdtp->cmd) + gd->reloc_off; #if 0 printf ("Command \"%s\": 0x%08lx => 0x%08lx\n", cmdtp->name, (ulong) (cmdtp->cmd), addr); #endif cmdtp->cmd = (int (*)(struct cmd_tbl_s *, int, int, char *[]))addr; addr = (ulong)(cmdtp->name) + gd->reloc_off; cmdtp->name = (char *)addr; if (cmdtp->usage) { addr = (ulong)(cmdtp->usage) + gd->reloc_off; cmdtp->usage = (char *)addr; } #ifdef CFG_LONGHELP if (cmdtp->help) { addr = (ulong)(cmdtp->help) + gd->reloc_off; cmdtp->help = (char *)addr; } #endif } serial_puts("after cmd relocate!\n"); #if 0 ... #endif while(1); }

2 2

[U-Boot-Users] How to build a usable xipuImage?
by phantom yang 22 Nov '05

22 Nov '05

Hi, I trying to implement "Kernel XIP" I have done the xipuImage patch and set CONFIG_XIP_PHYS_ADDR=0x00100000. And have wrote xipuImage on 0x00100000 (reference to omap5912 memory map). My build xipuImage command: arm-linux-objcopy -O binary -R .note -R .comment -S vmlinux arch/arm/boot/xipImage mkimage -x -A arm -O linux -T kernel -C none -a 0x00100000 -e 0x00100040 -n '2.6.12-omap1-xip' -d arch/arm/boot/xipImage arch/arm/boot/xipuImage Because of the message, "xipImage has invalid buffer for XIP" I marked the header check from mkimage.c and rebuild mkimage: /* for (p=ptr; p < ptr+sizeof(image_header_t); p++) { if ( *p != 0xff ) { fprintf (stderr, "%s: Bad file: \"%s\" has invalid buffer for XIP\n", cmdname, datafile); exit (EXIT_FAILURE); } } */ While booting: ## Booting image at 00100000 ... Image Name: 2.6.12-omap1-xip Image Type: ARM Linux Kernel Image (uncompressed) Data Size: 2321352 Bytes = 2.2 MB Load Address: 00100000 Entry Point: 00100040 Verifying Checksum ... OK XIP Kernel Image ... OK Starting kernel ... [halt!!] -------------------------------------------------- If I don't remove header check. And build xipuImage using: arm-linux-objcopy -O binary -R .note -R .comment -S vmlinux arch/arm/boot/xipImage mkimage -A arm -O linux -T kernel -C none -a 0x00100000 -e 0x00100040 -n '2.6.12-omap1-xip' -d arch/arm/boot/xipImage arch/arm/boot/xipuImage While booting: ## Booting image at 00100000 ... Image Name: 2.6.12-omap1-xip-4 Image Type: ARM Linux Kernel Image (uncompressed) Data Size: 2321352 Bytes = 2.2 MB Load Address: 00100000 Entry Point: 00100040 Verifying Checksum ... OK OK Starting kernel ... [halt!!] Thus, how to build a usable xinuImage? Thanks, phantom yang -- ____________________________________________________ Get your own Hello Kitty email @ www.sanriotown.com

1 0