July 2007 - coreboot - mail.coreboot.org

[LinuxBIOS] status update - K8T890/VT8237
by Rudolf Marek July 13, 2007

July 13, 2007

Hello all, I just got four words for you: It boots now yeah! Yes it boots my standard Linux kernel (2.6.22) SATA/IDE/USB/soundcard/gigabiteth and my Radeon seems to work. My MB is Asus A8V-E SE. Which might be compatible with A8V-E Duluxe... So someone wants to play too? However there are some problems left: 1) dualchannel on K8 with unbuffered dimms Has someone in plan to fix this? 2) something wrong with PCI/RAM resources LinuxBIOS is telling FILO and FILO tells linux this memmap: 00001000-0009ffff : System RAM 000c0000-000effff : System RAM 00100000-1fffffff : System RAM But if I use my K8 resource dumper: MMIO map: #0 0x0000000000 - 0x000000ffff Access: / Dstnode:0 DstLink 0 MMIO map: #1 0x0000000000 - 0x000000ffff Access: / Dstnode:0 DstLink 0 MMIO map: #2 0x0000000000 - 0x000000ffff Access: / Dstnode:0 DstLink 0 MMIO map: #3 0x0000000000 - 0x000000ffff Access: / Dstnode:0 DstLink 0 MMIO map: #4 0x0000000000 - 0x000000ffff Access: / Dstnode:0 DstLink 0 MMIO map: #5 0x00000a0000 - 0x00000cffff Access: R/W Dstnode:0 DstLink 0 MMIO map: #6 0x00fc000000 - 0x00fc2fffff Access: R/W Dstnode:0 DstLink 0 MMIO map: #7 0x00e0000000 - 0x00f7ffffff Access: R/W Dstnode:0 DstLink 0 There is MMIO region up to cffff. I could not long time catch this bug - the system hung disk IO when the disk buffers hit this region. I thought there is something wrong with V-Link or PCI master. But once I spot this I solved it by passing my own memmap to Linux Kernel. Question is: any idea why this happened? I'm using default resource map and reasonably old SVN. This seems to be completely undocumented in K8 code... Perhaps there is something wrong my my memory resource?? 3) ROM strapping of some registers does not work properly I will try to solve it. 4) Fine tuning of some VIA registers. I will try to solve it. 5) ACPI off, GPIO in SIO, Reset button does not work. PCIe regs tune. Thanks, Rudolf

4 10

[LinuxBIOS] Booting Windows - LinuxBIOS
by Augusto Pedroza July 13, 2007

July 13, 2007

For some reason BX_ELTORITO_BOOT was disabled when using ADLO. Does anyone know why? At this point I need qmeu + linuxbios_+ ADLO to be able to install windwos in a qemu image from the installation CD. file: rombios.c Where can I find what are the current cmos settings? qemu + bochsbios is able to boot from the CD because bootseq receives 3 on the following piece of code. When using linuxBIOS, bootseq receives 2. // if BX_ELTORITO_BOOT is not defined, old behavior // check bit 5 in CMOS reg 0x2d. load either 0x00 or 0x80 into DL // in preparation for the intial INT 13h (0=floppy A:, 0x80=C:) // 0: system boot sequence, first drive C: then A: // 1: system boot sequence, first drive A: then C: // else BX_ELTORITO_BOOT is defined // CMOS regs 0x3D and 0x38 contain the boot sequence: // CMOS reg 0x3D & 0x0f : 1st boot device // CMOS reg 0x3D & 0xf0 : 2nd boot device // CMOS reg 0x38 & 0xf0 : 3rd boot device // boot device codes: // 0x00 : not defined // 0x01 : first floppy // 0x02 : first harddrive // 0x03 : first cdrom // else : boot failure #if BX_ELTORITO_BOOT bootseq=inb_cmos(0x3d); bootseq|=((inb_cmos(0x38) & 0xf0) << 4); Looks like I need to change some CMOS settings. Thanks, -- Augusto Pedroza

1 0

[LinuxBIOS] DIMM Page Size
by Joseph Smith July 13, 2007

July 13, 2007

Anyone know what the mathmatical calculation is to determine a Dimms "Page Size"?? I found this but it does not seem to work: Page size = (2^col)*bus_width where, col = number of column address lines (SPD byte 4: Number of Column Addresses) bus_width = number of data(DQ) lines (???SPD byte 13: SDRAM Device Width???) Thanks - Joe

6 13

[LinuxBIOS] PATCH: mkdirp() error on new directory outside given parent
by Peter Stuge July 13, 2007

July 13, 2007

Have a look. I haven't gotten lar to build, but the function works in a separate test program. //Peter

4 9

[LinuxBIOS] r453 - LinuxBIOSv3/util/lar
by svn＠openbios.org July 13, 2007

July 13, 2007

Author: stuge Date: 2007-07-13 12:40:31 +0200 (Fri, 13 Jul 2007) New Revision: 453 Modified: LinuxBIOSv3/util/lar/extract.c LinuxBIOSv3/util/lar/lib.c LinuxBIOSv3/util/lar/lib.h Log: Replaces mkdirp() with mkdirp_below() that aborts directory creation and returns an error if any part of dirpath is located outside the specified parent directory. Use the parent "/" to allow new directories anywhere. Note that dirpath is relative to the working directory, not to parent. Signed-off-by: Peter Stuge <peter(a)stuge.se> Acked-by: Stefan Reinauer <stepan(a)coresystems.de> Modified: LinuxBIOSv3/util/lar/extract.c =================================================================== --- LinuxBIOSv3/util/lar/extract.c 2007-07-12 20:03:02 UTC (rev 452) +++ LinuxBIOSv3/util/lar/extract.c 2007-07-13 10:40:31 UTC (rev 453) @@ -119,7 +119,7 @@ if (pos) { pos[1] = 0; /* printf("Pathname %s\n",pathname); */ - mkdirp(pathname, 0755); + mkdirp_below(".", pathname, 0755); } free(pathname); Modified: LinuxBIOSv3/util/lar/lib.c =================================================================== --- LinuxBIOSv3/util/lar/lib.c 2007-07-12 20:03:02 UTC (rev 452) +++ LinuxBIOSv3/util/lar/lib.c 2007-07-13 10:40:31 UTC (rev 453) @@ -24,6 +24,7 @@ #include <strings.h> #include <unistd.h> #include <dirent.h> +#include <errno.h> #include <sys/stat.h> #include <sys/types.h> @@ -33,42 +34,82 @@ static struct file *files = NULL; -int mkdirp(const char *dirpath, mode_t mode) +/** + * Create a new directory including any missing parent directories. + * + * NOTE: This function does not do complete path resolution as described in + * Linux path_resolution(2) and hence will fail for complex paths: + * + * e.g.: mkdirp_below("subdir", "subdir/../subdir/x", 0777); + * + * This call should create subdir/x, but since subdir/.. is outside subdir, + * the function returns an error. + * + * @param parent Return an error if a new directory would be created outside + * this directory. Pass "/" to allow new directories to be created anywhere. + * @param dirpath The new directory that should be created, the path can be + * either absolute or relative to the current working directory. (It is not + * relative to parent.) + * @param mode Permissions to use for newly created directories. + */ +int mkdirp_below(const char *parent, const char *dirpath, mode_t mode) { - char *pos, *currpath, *path; - char cwd[MAX_PATH]; - int ret = 0; + int ret = -1; + size_t dirsep, parlen, sublen; + char c, *r, *path = NULL, *subdir, rpar[PATH_MAX], rsub[PATH_MAX]; + if (!dirpath) { + fprintf(stderr, "mkdirp_below: No new directory specified\n"); + goto done; + } + path = strdup(dirpath); if (!path) { - fprintf(stderr, "Out of memory.\n"); - exit(1); + perror("Duplicate new directory failed:"); + goto done; } - currpath = path; - - if (!getcwd(cwd, MAX_PATH)) { - free(path); - fprintf(stderr, "Error getting cwd.\n"); - return -1; + if (NULL == realpath(parent, rpar)) { + fprintf(stderr, "realpath(%s) failed: %s\n", parent, + strerror(errno)); + goto done; } + parlen = strlen(rpar); - do { - pos = index(currpath, '/'); - if (pos) - *pos = 0; + for (subdir = path, dirsep = 0; subdir[dirsep]; subdir += dirsep) { + dirsep = strcspn(subdir, "/\\"); + if (!dirsep) { + subdir++; + continue; + } - /* printf("cp=%s\n", currpath); */ - mkdir(currpath, mode); - ret = chdir(currpath); + c = subdir[dirsep]; + subdir[dirsep] = 0; + r = realpath(path, rsub); + sublen = strlen(rsub); + if (NULL == r) { + if(ENOENT != errno) { + fprintf(stderr, "realpath(%s) failed: %s\n", + path, strerror(errno)); + goto done; + } + } else if (sublen < parlen || strncmp(rpar, rsub, parlen)) { + fprintf(stderr, "Abort: %s is outside %s\n", dirpath, + parent); + goto done; + } + if(-1 == mkdir(path, mode) && EEXIST != errno) { + fprintf(stderr, "mkdir(%s): %s\n", path, + strerror(errno)); + goto done; + } + subdir[dirsep] = c; + } + ret = 0; - if (pos) - currpath = pos + 1; - } while (pos && !ret && strlen(currpath)); - - chdir(cwd); - free(path); - +done: + if (path) + free(path); return ret; } Modified: LinuxBIOSv3/util/lar/lib.h =================================================================== --- LinuxBIOSv3/util/lar/lib.h 2007-07-12 20:03:02 UTC (rev 452) +++ LinuxBIOSv3/util/lar/lib.h 2007-07-13 10:40:31 UTC (rev 453) @@ -41,7 +41,7 @@ char *get_bootblock(void); /* prototypes for lib.c functions */ -int mkdirp(const char *dirpath, mode_t mode); +int mkdirp_below(const char *parent, const char *dirpath, mode_t mode); int add_files(const char *name); int add_file_or_directory(const char *name);

1 0

Re: [LinuxBIOS] Does my MoBo supported?
by Rudolf Marek July 13, 2007

July 13, 2007

Hello Peter, Peter Lemenkov wrote: > Hello Rudolf! > > 2007/7/13, Rudolf Marek <r.marek(a)assembler.cz>: >> >> > http://lemenkov.googlepages.com/lspsi_xxx.txt >> >> Hmm for first quick view I cant tell. Perhaps not everything. Do you >> want to >> write support for your chipset yourself? > > Yes. It would be a good experience for me (especially if I succeeded), > but I'm afraid of my lack of knowledge. Well All you need to have some experience with programming and PC architecture. You will need to know how PCI, AGP, APIC, PIC, LPC and SIO works ;) > So far I haven't any works concerned with embedded devices and such > in-deep system hacking and looks like I need some mentoring :) A list is a good menthor, if someone replies ;) > All what I have right now is my experience in C/Perl/Python > programming + already mentioned mainboard + LinuxBIOS svn copy :) Perhaps you will need the datasheets for the chipset. You can get them from VIA under NDA, tell them that you want to this for the LinuxBIOS (GPL) project. Then they will alllow you to publish the source code. It will take some weeks to proceed. Maybe you can just for a start hack something with qemu. Rudolf

1 0

[LinuxBIOS] [PATCH][v3] Move lar into global util/ directory + svn:externals
by Uwe Hermann July 13, 2007

July 13, 2007

OK, not really a patch, but I think we can move lar into the global util/ directory and suck it into v3 via svn:externals. That's a first step to make it independent of v3 (and to allow Jordan to use it in buildrom). I'm waiting for comments or ACKs before doing this, though. Uwe. -- http://www.hermann-uwe.de | http://www.holsham-traders.de http://www.crazy-hacks.org | http://www.unmaintained-free-software.org

4 8

[LinuxBIOS] Does my MoBo supported?
by Peter Lemenkov July 12, 2007

July 12, 2007

Hello All! I've got a small number of motherboards on which I would like to install LinuxBIOS instead of proprietary BLOBs. One of that cars identified as Gigabyte 7VT880-RZ (based on VIA KT880 + VT8237R) Here is output from lspci and dmidecode respectively: http://lemenkov.googlepages.com/lspci.txt http://lemenkov.googlepages.com/dmidecode.txt Can I do something with this M/b? I don't mean bulking it, actually :) -- With best regards!

5 7

[LinuxBIOS] ticket 7 (flashing ICH-x mainboard)
by Jouni Mettälä July 12, 2007

July 12, 2007

Based on intel 82802ab/82802ac datasheet. There is Top Block Lock. TBL# pin disables writing of highest addressable block when it is low. When #TBL pin was desoldered and connected to Vcc (Is it right way to put it high?) Flashing with flashrom was succesful. Verify was ok. It wasn't possible to write that chip with flashrom or uniflash before that. There is also WP# pin to enable or disable writing other blocks. Datasheet is http://download.intel.com/design/chipsets/datashts/29065804.pdf

3 2

Re: [LinuxBIOS] Does my MoBo supported?
by Rudolf Marek July 12, 2007

July 12, 2007

> > http://lemenkov.googlepages.com/lspsi_vvv.txt > > I included dmidecode log also: > > http://lemenkov.googlepages.com/dmidecode.txt I need -xxx and not -vvv ;) R.

1 0