#include <linux/crc32.h>
#include <linux/buffer_head.h>
#include <linux/bitops.h>
+#include <linux/debugfs.h>
+#include <linux/module.h>
+#include <linux/fs.h>
#include <asm/byteorder.h>
+#include <cluster/masklog.h>
+
#include "ocfs2.h"
#include "blockcheck.h"
-
/*
* We use the following conventions:
*
* p = # parity bits
* c = # total code bits (d + p)
*/
-static int calc_parity_bits(unsigned int d)
-{
- unsigned int p;
-
- /*
- * Bits required for Single Error Correction is as follows:
- *
- * d + p + 1 <= 2^p
- *
- * We're restricting ourselves to 31 bits of parity, that should be
- * sufficient.
- */
- for (p = 1; p < 32; p++)
- {
- if ((d + p + 1) <= (1 << p))
- return p;
- }
- return 0;
-}
/*
* Calculate the bit offset in the hamming code buffer based on the bit's
* so it's a parity bit. 2 is a power of two (2^1), so it's a parity bit.
* 3 is not a power of two. So bit 1 of the data buffer ends up as bit 3
* in the code buffer.
+ *
+ * The caller can pass in *p if it wants to keep track of the most recent
+ * number of parity bits added. This allows the function to start the
+ * calculation at the last place.
*/
-static unsigned int calc_code_bit(unsigned int i)
+static unsigned int calc_code_bit(unsigned int i, unsigned int *p_cache)
{
- unsigned int b, p;
+ unsigned int b, p = 0;
/*
* Data bits are 0-based, but we're talking code bits, which
*/
b = i + 1;
+ /* Use the cache if it is there */
+ if (p_cache)
+ p = *p_cache;
+ b += p;
+
/*
* For every power of two below our bit number, bump our bit.
*
- * We compare with (b + 1) becuase we have to compare with what b
+ * We compare with (b + 1) because we have to compare with what b
* would be _if_ it were bumped up by the parity bit. Capice?
+ *
+ * p is set above.
*/
- for (p = 0; (1 << p) < (b + 1); p++)
+ for (; (1 << p) < (b + 1); p++)
b++;
+ if (p_cache)
+ *p_cache = p;
+
return b;
}
*/
u32 ocfs2_hamming_encode(u32 parity, void *data, unsigned int d, unsigned int nr)
{
- unsigned int p = calc_parity_bits(nr + d);
- unsigned int i, j, b;
+ unsigned int i, b, p = 0;
- BUG_ON(!p);
+ BUG_ON(!d);
/*
* b is the hamming code bit number. Hamming code specifies a
* i is the offset in this hunk, nr + i is the total bit
* offset.
*/
- b = calc_code_bit(nr + i);
+ b = calc_code_bit(nr + i, &p);
- for (j = 0; j < p; j++)
- {
- /*
- * Data bits in the resultant code are checked by
- * parity bits that are part of the bit number
- * representation. Huh?
- *
- * <wikipedia href="http://en.wikipedia.org/wiki/Hamming_code">
- * In other words, the parity bit at position 2^k
- * checks bits in positions having bit k set in
- * their binary representation. Conversely, for
- * instance, bit 13, i.e. 1101(2), is checked by
- * bits 1000(2) = 8, 0100(2)=4 and 0001(2) = 1.
- * </wikipedia>
- *
- * Note that 'k' is the _code_ bit number. 'b' in
- * our loop.
- */
- if (b & (1 << j))
- parity ^= (1 << j);
- }
+ /*
+ * Data bits in the resultant code are checked by
+ * parity bits that are part of the bit number
+ * representation. Huh?
+ *
+ * <wikipedia href="http://en.wikipedia.org/wiki/Hamming_code">
+ * In other words, the parity bit at position 2^k
+ * checks bits in positions having bit k set in
+ * their binary representation. Conversely, for
+ * instance, bit 13, i.e. 1101(2), is checked by
+ * bits 1000(2) = 8, 0100(2)=4 and 0001(2) = 1.
+ * </wikipedia>
+ *
+ * Note that 'k' is the _code_ bit number. 'b' in
+ * our loop.
+ */
+ parity ^= b;
}
/* While the data buffer was treated as little endian, the
void ocfs2_hamming_fix(void *data, unsigned int d, unsigned int nr,
unsigned int fix)
{
- unsigned int p = calc_parity_bits(nr + d);
unsigned int i, b;
- BUG_ON(!p);
+ BUG_ON(!d);
/*
* If the bit to fix has an hweight of 1, it's a parity bit. One
* nr + d is the bit right past the data hunk we're looking at.
* If fix after that, nothing to do
*/
- if (fix >= calc_code_bit(nr + d))
+ if (fix >= calc_code_bit(nr + d, NULL))
return;
/*
* start b at the offset in the code buffer. See hamming_encode()
* for a more detailed description of 'b'.
*/
- b = calc_code_bit(nr);
+ b = calc_code_bit(nr, NULL);
/* If the fix is before this hunk, nothing to do */
if (fix < b)
return;
ocfs2_hamming_fix(data, blocksize * 8, 0, fix);
}
+
+/*
+ * Debugfs handling.
+ */
+
+#ifdef CONFIG_DEBUG_FS
+
+static int blockcheck_u64_get(void *data, u64 *val)
+{
+ *val = *(u64 *)data;
+ return 0;
+}
+DEFINE_SIMPLE_ATTRIBUTE(blockcheck_fops, blockcheck_u64_get, NULL, "%llu\n");
+
+static struct dentry *blockcheck_debugfs_create(const char *name,
+ struct dentry *parent,
+ u64 *value)
+{
+ return debugfs_create_file(name, S_IFREG | S_IRUSR, parent, value,
+ &blockcheck_fops);
+}
+
+static void ocfs2_blockcheck_debug_remove(struct ocfs2_blockcheck_stats *stats)
+{
+ if (stats) {
+ debugfs_remove(stats->b_debug_check);
+ stats->b_debug_check = NULL;
+ debugfs_remove(stats->b_debug_failure);
+ stats->b_debug_failure = NULL;
+ debugfs_remove(stats->b_debug_recover);
+ stats->b_debug_recover = NULL;
+ debugfs_remove(stats->b_debug_dir);
+ stats->b_debug_dir = NULL;
+ }
+}
+
+static int ocfs2_blockcheck_debug_install(struct ocfs2_blockcheck_stats *stats,
+ struct dentry *parent)
+{
+ int rc = -EINVAL;
+
+ if (!stats)
+ goto out;
+
+ stats->b_debug_dir = debugfs_create_dir("blockcheck", parent);
+ if (!stats->b_debug_dir)
+ goto out;
+
+ stats->b_debug_check =
+ blockcheck_debugfs_create("blocks_checked",
+ stats->b_debug_dir,
+ &stats->b_check_count);
+
+ stats->b_debug_failure =
+ blockcheck_debugfs_create("checksums_failed",
+ stats->b_debug_dir,
+ &stats->b_failure_count);
+
+ stats->b_debug_recover =
+ blockcheck_debugfs_create("ecc_recoveries",
+ stats->b_debug_dir,
+ &stats->b_recover_count);
+ if (stats->b_debug_check && stats->b_debug_failure &&
+ stats->b_debug_recover)
+ rc = 0;
+
+out:
+ if (rc)
+ ocfs2_blockcheck_debug_remove(stats);
+ return rc;
+}
+#else
+static inline int ocfs2_blockcheck_debug_install(struct ocfs2_blockcheck_stats *stats,
+ struct dentry *parent)
+{
+ return 0;
+}
+
+static inline void ocfs2_blockcheck_debug_remove(struct ocfs2_blockcheck_stats *stats)
+{
+}
+#endif /* CONFIG_DEBUG_FS */
+
+/* Always-called wrappers for starting and stopping the debugfs files */
+int ocfs2_blockcheck_stats_debugfs_install(struct ocfs2_blockcheck_stats *stats,
+ struct dentry *parent)
+{
+ return ocfs2_blockcheck_debug_install(stats, parent);
+}
+
+void ocfs2_blockcheck_stats_debugfs_remove(struct ocfs2_blockcheck_stats *stats)
+{
+ ocfs2_blockcheck_debug_remove(stats);
+}
+
+static void ocfs2_blockcheck_inc_check(struct ocfs2_blockcheck_stats *stats)
+{
+ u64 new_count;
+
+ if (!stats)
+ return;
+
+ spin_lock(&stats->b_lock);
+ stats->b_check_count++;
+ new_count = stats->b_check_count;
+ spin_unlock(&stats->b_lock);
+
+ if (!new_count)
+ mlog(ML_NOTICE, "Block check count has wrapped\n");
+}
+
+static void ocfs2_blockcheck_inc_failure(struct ocfs2_blockcheck_stats *stats)
+{
+ u64 new_count;
+
+ if (!stats)
+ return;
+
+ spin_lock(&stats->b_lock);
+ stats->b_failure_count++;
+ new_count = stats->b_failure_count;
+ spin_unlock(&stats->b_lock);
+
+ if (!new_count)
+ mlog(ML_NOTICE, "Checksum failure count has wrapped\n");
+}
+
+static void ocfs2_blockcheck_inc_recover(struct ocfs2_blockcheck_stats *stats)
+{
+ u64 new_count;
+
+ if (!stats)
+ return;
+
+ spin_lock(&stats->b_lock);
+ stats->b_recover_count++;
+ new_count = stats->b_recover_count;
+ spin_unlock(&stats->b_lock);
+
+ if (!new_count)
+ mlog(ML_NOTICE, "ECC recovery count has wrapped\n");
+}
+
+
+
+/*
+ * These are the low-level APIs for using the ocfs2_block_check structure.
+ */
+
/*
* This function generates check information for a block.
* data is the block to be checked. bc is a pointer to the
* Again, the data passed in should be the on-disk endian.
*/
int ocfs2_block_check_validate(void *data, size_t blocksize,
- struct ocfs2_block_check *bc)
+ struct ocfs2_block_check *bc,
+ struct ocfs2_blockcheck_stats *stats)
{
int rc = 0;
struct ocfs2_block_check check;
u32 crc, ecc;
+ ocfs2_blockcheck_inc_check(stats);
+
check.bc_crc32e = le32_to_cpu(bc->bc_crc32e);
check.bc_ecc = le16_to_cpu(bc->bc_ecc);
if (crc == check.bc_crc32e)
goto out;
+ ocfs2_blockcheck_inc_failure(stats);
+ mlog(ML_ERROR,
+ "CRC32 failed: stored: %u, computed %u. Applying ECC.\n",
+ (unsigned int)check.bc_crc32e, (unsigned int)crc);
+
/* Ok, try ECC fixups */
ecc = ocfs2_hamming_encode_block(data, blocksize);
ocfs2_hamming_fix_block(data, blocksize, ecc ^ check.bc_ecc);
/* And check the crc32 again */
crc = crc32_le(~0, data, blocksize);
- if (crc == check.bc_crc32e)
+ if (crc == check.bc_crc32e) {
+ ocfs2_blockcheck_inc_recover(stats);
goto out;
+ }
+
+ mlog(ML_ERROR, "Fixed CRC32 failed: stored: %u, computed %u\n",
+ (unsigned int)check.bc_crc32e, (unsigned int)crc);
rc = -EIO;
* Again, the data passed in should be the on-disk endian.
*/
int ocfs2_block_check_validate_bhs(struct buffer_head **bhs, int nr,
- struct ocfs2_block_check *bc)
+ struct ocfs2_block_check *bc,
+ struct ocfs2_blockcheck_stats *stats)
{
int i, rc = 0;
struct ocfs2_block_check check;
if (!nr)
return 0;
+ ocfs2_blockcheck_inc_check(stats);
+
check.bc_crc32e = le32_to_cpu(bc->bc_crc32e);
check.bc_ecc = le16_to_cpu(bc->bc_ecc);
if (crc == check.bc_crc32e)
goto out;
+ ocfs2_blockcheck_inc_failure(stats);
mlog(ML_ERROR,
"CRC32 failed: stored: %u, computed %u. Applying ECC.\n",
(unsigned int)check.bc_crc32e, (unsigned int)crc);
/* And check the crc32 again */
for (i = 0, crc = ~0; i < nr; i++)
crc = crc32_le(crc, bhs[i]->b_data, bhs[i]->b_size);
- if (crc == check.bc_crc32e)
+ if (crc == check.bc_crc32e) {
+ ocfs2_blockcheck_inc_recover(stats);
goto out;
+ }
mlog(ML_ERROR, "Fixed CRC32 failed: stored: %u, computed %u\n",
(unsigned int)check.bc_crc32e, (unsigned int)crc);
struct ocfs2_block_check *bc)
{
int rc = 0;
+ struct ocfs2_super *osb = OCFS2_SB(sb);
- if (ocfs2_meta_ecc(OCFS2_SB(sb)))
- rc = ocfs2_block_check_validate(data, sb->s_blocksize, bc);
+ if (ocfs2_meta_ecc(osb))
+ rc = ocfs2_block_check_validate(data, sb->s_blocksize, bc,
+ &osb->osb_ecc_stats);
return rc;
}
struct ocfs2_block_check *bc)
{
int rc = 0;
+ struct ocfs2_super *osb = OCFS2_SB(sb);
- if (ocfs2_meta_ecc(OCFS2_SB(sb)))
- rc = ocfs2_block_check_validate_bhs(bhs, nr, bc);
+ if (ocfs2_meta_ecc(osb))
+ rc = ocfs2_block_check_validate_bhs(bhs, nr, bc,
+ &osb->osb_ecc_stats);
return rc;
}
git://ftp.safe.ca / safe / jmp / linux-2.6 / blobdiff