diff options
Diffstat (limited to 'ubi-utils')
| -rw-r--r-- | ubi-utils/scripts/bin2nand2bin_test.sh | 184 | ||||
| -rw-r--r-- | ubi-utils/scripts/inject_biterror.pl | 94 | ||||
| -rw-r--r-- | ubi-utils/src/nand2bin.c | 192 | ||||
| -rw-r--r-- | ubi-utils/src/nandcorr.c | 20 | ||||
| -rw-r--r-- | ubi-utils/src/nandecc.h | 5 |
5 files changed, 425 insertions, 70 deletions
diff --git a/ubi-utils/scripts/bin2nand2bin_test.sh b/ubi-utils/scripts/bin2nand2bin_test.sh new file mode 100644 index 0000000..a17c91b --- /dev/null +++ b/ubi-utils/scripts/bin2nand2bin_test.sh @@ -0,0 +1,184 @@ +#!/bin/sh +# +# Testcase for nand2bin and bin2nand. Generate testdata and inject +# biterrors. Convert data back and compare with original data. +# +# Conversion: +# bin -> bin2nand -> mif -> nand2bin -> img +# + +inject_biterror=./scripts/inject_biterror.pl + +pagesize=2048 +oobsize=64 + +# Create test data +dd if=/dev/urandom of=testblock.bin bs=131072 count=1 + +echo "Test conversion without bitflips ..." + +echo -n "Convert bin to mif ... " +bin2nand --pagesize=${pagesize} -o testblock.mif testblock.bin +if [ $? -ne "0" ]; then + echo "failed!" + exit 1 +else + echo "ok" +fi + +echo -n "Convert mif to bin ... " +nand2bin --pagesize=${pagesize} -o testblock.img testblock.mif +if [ $? -ne "0" ]; then + echo "failed!" + exit 1 +else + echo "ok" +fi + +echo -n "Comparing data ... " +diff testblock.bin testblock.img +if [ $? -ne "0" ]; then + echo "failed!" + exit 1 +else + echo "ok" +fi + +echo "Test conversion with uncorrectable ECC erors ..." +echo -n "Inject biterror at offset $ioffs ... " +${inject_biterror} --offset=0 --bitmask=0x81 \ + --input=testblock.mif \ + --output=testblock_bitflip.mif +if [ $? -ne "0" ]; then + echo "failed!" + exit 1 +else + echo "ok" +fi + +echo "Convert mif to bin ... " +rm testblock.img +nand2bin --correct-ecc --pagesize=${pagesize} -o testblock.img \ + testblock_bitflip.mif +if [ $? -ne "0" ]; then + echo "failed!" + exit 1 +else + echo "ok" +fi + +echo -n "Comparing data, must fail due to uncorrectable ECC ... " +diff testblock.bin testblock.img +if [ $? -ne "0" ]; then + echo "ok" # Must fail! +else + echo "failed!" + exit 1 +fi + +echo "Test bitflips in data ... " +for offs in `seq 0 255` ; do + + cp testblock.mif testblock_bitflip.mif + + for xoffs in 0 256 512 768 ; do + let ioffs=$offs+$xoffs + + cp testblock_bitflip.mif testblock_bitflip_tmp.mif + echo -n "Inject biterror at offset $ioffs ... " + ${inject_biterror} --offset=${ioffs} --bitmask=0x01 \ + --input=testblock_bitflip_tmp.mif \ + --output=testblock_bitflip.mif + if [ $? -ne "0" ]; then + echo "failed!" + exit 1 + else + echo "ok" + fi + done + + echo "Convert mif to bin ... " + rm testblock.img + nand2bin --correct-ecc --pagesize=${pagesize} -o testblock.img \ + testblock_bitflip.mif + if [ $? -ne "0" ]; then + echo "failed!" + exit 1 + else + echo "ok" + fi + + echo -n "Comparing data ... " + diff testblock.bin testblock.img + if [ $? -ne "0" ]; then + hexdump testblock.bin > testblock.bin.txt + hexdump testblock.img > testblock.img.txt + echo "Use tkdiff testblock.bin.txt testblock.img.txt to compare" + echo "failed!" + exit 1 + else + echo "ok" + fi + + # Without correction + echo "Convert mif to bin ... " + rm testblock.img + nand2bin --pagesize=${pagesize} -o testblock.img \ + testblock_bitflip.mif + if [ $? -ne "0" ]; then + echo "failed!" + exit 1 + else + echo "ok" + fi + + echo -n "Comparing data must differ, correction is disabled ... " + diff testblock.bin testblock.img + if [ $? -ne "0" ]; then + echo "ok" # must fail + else + echo "failed!" + exit 1 + fi +done + +echo "Test bitflips in OOB data ... " +for offs in `seq 0 $oobsize` ; do + + let ioffs=$pagesize+$offs + + echo -n "Inject biterror at offset $ioffs ... " + ${inject_biterror} --offset=${ioffs} --bitmask=0x01 \ + --input=testblock.mif \ + --output=testblock_bitflip.mif + if [ $? -ne "0" ]; then + echo "failed!" + exit 1 + else + echo "ok" + fi + + echo "Convert mif to bin ... " + rm testblock.img + nand2bin --correct-ecc --pagesize=${pagesize} -o testblock.img \ + testblock_bitflip.mif + if [ $? -ne "0" ]; then + echo "failed!" + exit 1 + else + echo "ok" + fi + + echo -n "Comparing data ... " + diff testblock.bin testblock.img + if [ $? -ne "0" ]; then + hexdump testblock.bin > testblock.bin.txt + hexdump testblock.img > testblock.img.txt + echo "Use tkdiff testblock.bin.txt testblock.img.txt to compare" + echo "failed!" + exit 1 + else + echo "ok" + fi +done + diff --git a/ubi-utils/scripts/inject_biterror.pl b/ubi-utils/scripts/inject_biterror.pl new file mode 100644 index 0000000..b4a862a --- /dev/null +++ b/ubi-utils/scripts/inject_biterror.pl @@ -0,0 +1,94 @@ +#!/usr/bin/perl -w +# +# 2007 Frank Haverkamp <haver@vnet.ibm.com> +# +# Program for bit-error injection. I am sure that perl experts do it +# in 1 line. Please let me know how it is done right ;-). +# + +use strict; +use warnings; +use Getopt::Long; +use Pod::Usage; + +my $i; +my $help; +my $result; +my $offset = 0; +my $bitmask = 0x01; +my $in = "input.mif"; +my $out = "output.mif"; + +$result = GetOptions ("offset=i" => \$offset, # numeric + "bitmask=o" => \$bitmask, # numeric + "input=s" => \$in, # string + "output=s" => \$out, # string + "help|?" => \$help) or pod2usage(2); + +pod2usage(1) if $help; + +my $buf; + +open(my $in_fh, "<", $in) + or die "Cannot open file $in: $!"; +binmode $in_fh; + +open(my $out_fh, ">", $out) or + die "Cannot open file $out: $!"; +binmode $out_fh; + +$i = 0; +while (sysread($in_fh, $buf, 1)) { + + $buf = pack('C', unpack('C', $buf) ^ $bitmask) if ($i == $offset); + syswrite($out_fh, $buf, 1) or + die "Cannot write to offset $offset: $!"; + $i++; +} + +close $in_fh; +close $out_fh; + +__END__ + +=head1 NAME + +inject_biterrors.pl + +=head1 SYNOPSIS + +inject_biterror.pl [options] + +=head1 OPTIONS + +=over 8 + +=item B<--help> + +Print a brief help message and exits. + +=item B<--offset>=I<offset> + +Byte-offset where bit-error should be injected. + +=item B<--bitmask>=I<bitmask> + +Bit-mask where to inject errors in the byte. + +=item B<--input>=I<input-file> + +Input file. + +=item B<--output>=I<output-file> + +Output file. + +=back + +=head1 DESCRIPTION + +B<inject_biterrors.pl> will read the given input file and inject +biterrors at the I<offset> specified. The location of the biterrors +are defined by the I<bitmask> parameter. + +=cut diff --git a/ubi-utils/src/nand2bin.c b/ubi-utils/src/nand2bin.c index 8db4e77..be62e30 100644 --- a/ubi-utils/src/nand2bin.c +++ b/ubi-utils/src/nand2bin.c @@ -24,6 +24,7 @@ * 1.4 Fixed OOB output file * 1.5 Added verbose output and option to set blocksize. * Added split block mode for more convenient analysis. + * 1.6 Fixed ECC error detection and correction. */ #include <config.h> @@ -43,7 +44,7 @@ #include "config.h" #include "nandecc.h" -#define PROGRAM_VERSION "1.5" +#define PROGRAM_VERSION "1.6" #define MAXPATH 1024 #define MIN(x,y) ((x)<(y)?(x):(y)) @@ -55,6 +56,7 @@ struct args { size_t blocksize; int split_blocks; size_t in_len; /* size of input file */ + int correct_ecc; /* special stuff needed to get additional arguments */ char *arg1; @@ -68,6 +70,7 @@ static struct args myargs = { .blocksize = 128 * 1024, .in_len = 0, .split_blocks = 0, + .correct_ecc = 0, .arg1 = NULL, .options = NULL, }; @@ -81,6 +84,7 @@ static const char *optionsstr = " -p, --pagesize=<pagesize> NAND pagesize\n" " -b, --blocksize=<blocksize> NAND blocksize\n" " -s, --split-blocks generate binaries for each block\n" +" -e, --correct-ecc Correct data according to ECC info\n" " -v, --verbose verbose output\n" " -?, --help Give this help list\n" " --usage Give a short usage message\n"; @@ -92,12 +96,13 @@ static const char *usage = static int verbose = 0; -struct option long_options[] = { +static struct option long_options[] = { { .name = "output", .has_arg = 1, .flag = NULL, .val = 'o' }, { .name = "oob", .has_arg = 1, .flag = NULL, .val = 'O' }, { .name = "pagesize", .has_arg = 1, .flag = NULL, .val = 'p' }, { .name = "blocksize", .has_arg = 1, .flag = NULL, .val = 'b' }, { .name = "split-blocks", .has_arg = 0, .flag = NULL, .val = 's' }, + { .name = "correct-ecc", .has_arg = 0, .flag = NULL, .val = 'e' }, { .name = "verbose", .has_arg = 0, .flag = NULL, .val = 'v' }, { .name = "help", .has_arg = 0, .flag = NULL, .val = '?' }, { .name = "usage", .has_arg = 0, .flag = NULL, .val = 0 }, @@ -109,7 +114,7 @@ struct option long_options[] = { * k, K, kib, KiB for kilobyte * m, M, mib, MiB for megabyte */ -uint32_t str_to_num(char *str) +static uint32_t str_to_num(char *str) { char *s = str; ulong num = strtoul(s, &s, 0); @@ -137,58 +142,61 @@ uint32_t str_to_num(char *str) * @return error * */ -static int -parse_opt(int argc, char **argv, struct args *args) +static int parse_opt(int argc, char **argv, struct args *args) { while (1) { int key; - key = getopt_long(argc, argv, "b:o:O:p:sv?", long_options, NULL); + key = getopt_long(argc, argv, "b:eo:O:p:sv?", long_options, NULL); if (key == -1) break; switch (key) { - case 'p': /* --pagesize<pagesize> */ - args->pagesize = str_to_num(optarg); - break; - - case 'b': /* --blocksize<blocksize> */ - args->blocksize = str_to_num(optarg); - break; - - case 'v': /* --verbose */ - verbose++; - break; - - case 's': /* --split-blocks */ - args->split_blocks = 1; - break; - - case 'o': /* --output=<output.bin> */ - args->output_file = optarg; - break; - - case 'O': /* --oob=<oob.bin> */ - args->oob_file = optarg; - break; - - case '?': /* help */ - printf("Usage: nand2bin [OPTION...] input.mif\n"); - printf("%s", doc); - printf("%s", optionsstr); - printf("\nReport bugs to %s\n", - PACKAGE_BUGREPORT); - exit(0); - break; - - case 'V': - printf("%s\n", PROGRAM_VERSION); - exit(0); - break; - - default: - printf("%s", usage); - exit(-1); + case 'p': /* --pagesize<pagesize> */ + args->pagesize = str_to_num(optarg); + break; + + case 'b': /* --blocksize<blocksize> */ + args->blocksize = str_to_num(optarg); + break; + + case 'v': /* --verbose */ + verbose++; + break; + + case 's': /* --split-blocks */ + args->split_blocks = 1; + break; + + case 'e': /* --correct-ecc */ + args->correct_ecc = 1; + break; + + case 'o': /* --output=<output.bin> */ + args->output_file = optarg; + break; + + case 'O': /* --oob=<oob.bin> */ + args->oob_file = optarg; + break; + + case '?': /* help */ + printf("Usage: nand2bin [OPTION...] input.mif\n"); + printf("%s", doc); + printf("%s", optionsstr); + printf("\nReport bugs to %s\n", + PACKAGE_BUGREPORT); + exit(0); + break; + + case 'V': + printf("%s\n", PROGRAM_VERSION); + exit(0); + break; + + default: + printf("%s", usage); + exit(-1); } } @@ -209,8 +217,7 @@ static int calc_oobsize(size_t pagesize) return 0; } -static inline void -hexdump(FILE *fp, const uint8_t *buf, ssize_t size) +static inline void hexdump(FILE *fp, const uint8_t *buf, ssize_t size) { int k; @@ -221,8 +228,7 @@ hexdump(FILE *fp, const uint8_t *buf, ssize_t size) } } -static int -process_page(uint8_t* buf, uint8_t *oobbuf, size_t pagesize) +static int process_page(uint8_t* buf, uint8_t *oobbuf, size_t pagesize) { int eccpoi, oobsize; size_t i; @@ -262,6 +268,17 @@ static int decompose_image(struct args *args, FILE *in_fp, uint8_t *oob = malloc(oobsize); uint8_t *calc_oob = malloc(oobsize); uint8_t *calc_buf = malloc(args->pagesize); + uint8_t *page_buf; + int pages_per_block = args->blocksize / args->pagesize; + int eccpoi = 0, eccpoi_start; + unsigned int i; + int badpos = bad_marker_offs_in_oob(args->pagesize); + + switch (args->pagesize) { + case 2048: eccpoi_start = 64 / 2; break; + case 512: eccpoi_start = 16 / 2; break; + default: exit(EXIT_FAILURE); + } if (!buf) exit(EXIT_FAILURE); @@ -273,6 +290,7 @@ static int decompose_image(struct args *args, FILE *in_fp, exit(EXIT_FAILURE); while (!feof(in_fp)) { + /* read page by page */ read = fread(buf, 1, args->pagesize, in_fp); if (ferror(in_fp)) { fprintf(stderr, "I/O Error."); @@ -281,13 +299,69 @@ static int decompose_image(struct args *args, FILE *in_fp, if (read != (ssize_t)args->pagesize) break; - rc = fwrite(buf, 1, args->pagesize, bin_fp); - if (ferror(bin_fp)) { + read = fread(oob, 1, oobsize, in_fp); + if (ferror(in_fp)) { fprintf(stderr, "I/O Error."); exit(EXIT_FAILURE); } - read = fread(oob, 1, oobsize, in_fp); - if (ferror(in_fp)) { + + page_buf = buf; /* default is unmodified data */ + + if ((page == 0 || page == 1) && (oob[badpos] != 0xff)) { + if (verbose) + printf("Block %d is bad\n", + page / pages_per_block); + goto write_data; + } + if (args->correct_ecc) + page_buf = calc_buf; + + process_page(buf, calc_oob, args->pagesize); + memcpy(calc_buf, buf, args->pagesize); + + /* + * Our oob format uses only the last 3 bytes out of 4. + * The first byte is 0x00 when the ECC is generated by + * our toolset and 0xff when generated by Linux. This + * is to be fixed when we want nand2bin work for other + * ECC layouts too. + */ + for (i = 0, eccpoi = eccpoi_start; i < args->pagesize; + i += 256, eccpoi += 4) + oob[eccpoi] = calc_oob[eccpoi] = 0xff; + + if (verbose && memcmp(oob, calc_oob, oobsize) != 0) { + printf("\nECC compare mismatch found at block %d page %d!\n", + page / pages_per_block, page % pages_per_block); + + printf("Read out OOB Data:\n"); + hexdump(stdout, oob, oobsize); + + printf("Calculated OOB Data:\n"); + hexdump(stdout, calc_oob, oobsize); + } + + /* Do correction on subpage base */ + for (i = 0, eccpoi = eccpoi_start; i < args->pagesize; + i += 256, eccpoi += 4) { + rc = nand_correct_data(calc_buf + i, &oob[eccpoi + 1], + &calc_oob[eccpoi + 1]); + + if (rc == -1) + fprintf(stdout, "Uncorrectable ECC error at " + "block %d page %d/%d\n", + page / pages_per_block, + page % pages_per_block, i / 256); + else if (rc > 0) + fprintf(stdout, "Correctable ECC error at " + "block %d page %d/%d\n", + page / pages_per_block, + page % pages_per_block, i / 256); + } + + write_data: + rc = fwrite(page_buf, 1, args->pagesize, bin_fp); + if (ferror(bin_fp)) { fprintf(stderr, "I/O Error."); exit(EXIT_FAILURE); } @@ -297,14 +371,6 @@ static int decompose_image(struct args *args, FILE *in_fp, exit(EXIT_FAILURE); } - process_page(buf, calc_oob, args->pagesize); - memcpy(calc_buf, buf, args->pagesize); - - rc = nand_correct_data(calc_buf, oob); - if ((rc != -1) || (memcmp(buf, calc_buf, args->pagesize) != 0)) { - fprintf(stdout, "possible ECC error at page %d\n", - page); - } page++; } free(calc_buf); diff --git a/ubi-utils/src/nandcorr.c b/ubi-utils/src/nandcorr.c index 7f1a547..caa07e2 100644 --- a/ubi-utils/src/nandcorr.c +++ b/ubi-utils/src/nandcorr.c @@ -36,8 +36,15 @@ static int countbits(uint32_t byte) return res; } -int nand_correct_data(uint8_t *dat, const uint8_t *fail_ecc) - +/** + * @dat: data which should be corrected + * @read_ecc: ecc information read from flash + * @calc_ecc: calculated ecc information from the data + * @return: number of corrected bytes + * or -1 when no correction is possible + */ +int nand_correct_data(uint8_t *dat, const uint8_t *read_ecc, + const uint8_t *calc_ecc) { uint8_t s0, s1, s2; @@ -47,9 +54,12 @@ int nand_correct_data(uint8_t *dat, const uint8_t *fail_ecc) * Be careful, the index magic is due to a pointer to a * uint32_t. */ - s0 = fail_ecc[1]; - s1 = fail_ecc[2]; - s2 = fail_ecc[3]; + s0 = calc_ecc[0] ^ read_ecc[0]; + s1 = calc_ecc[1] ^ read_ecc[1]; + s2 = calc_ecc[2] ^ read_ecc[2]; + + if ((s0 | s1 | s2) == 0) + return 0; /* Check for a single bit error */ if( ((s0 ^ (s0 >> 1)) & 0x55) == 0x55 && diff --git a/ubi-utils/src/nandecc.h b/ubi-utils/src/nandecc.h index fb5d529..bcf1982 100644 --- a/ubi-utils/src/nandecc.h +++ b/ubi-utils/src/nandecc.h @@ -22,7 +22,8 @@ #include <stdint.h> -extern int nand_calculate_ecc(const uint8_t *dat, uint8_t *ecc_code); -extern int nand_correct_data(uint8_t *dat, const uint8_t *fail_ecc); +int nand_calculate_ecc(const uint8_t *dat, uint8_t *ecc_code); +int nand_correct_data(uint8_t *dat, const uint8_t *read_ecc, + const uint8_t *calc_ecc); #endif |