/*===========================================================================*
* lmfs_get_block_ino *
*===========================================================================*/
struct buf *lmfs_get_block_ino(dev_t dev, block_t block, int only_search,
ino_t ino, u64_t ino_off)
{
/* Check to see if the requested block is in the block cache. If so, return
* a pointer to it. If not, evict some other block and fetch it (unless
* 'only_search' is 1). All the blocks in the cache that are not in use
* are linked together in a chain, with 'front' pointing to the least recently
* used block and 'rear' to the most recently used block. If 'only_search' is
* 1, the block being requested will be overwritten in its entirety, so it is
* only necessary to see if it is in the cache; if it is not, any free buffer
* will do. It is not necessary to actually read the block in from disk.
* If 'only_search' is PREFETCH, the block need not be read from the disk,
* and the device is not to be marked on the block, so callers can tell if
* the block returned is valid.
* In addition to the LRU chain, there is also a hash chain to link together
* blocks whose block numbers end with the same bit strings, for fast lookup.
*/
int b;
static struct buf *bp;
u64_t dev_off = (u64_t) block * fs_block_size;
struct buf *prev_ptr;
assert(buf_hash);
assert(buf);
assert(nr_bufs > 0);
ASSERT(fs_block_size > 0);
assert(dev != NO_DEV);
if((ino_off % fs_block_size)) {
printf("cache: unaligned lmfs_get_block_ino ino_off %llu\n",
ino_off);
util_stacktrace();
}
/* Search the hash chain for (dev, block). */
b = BUFHASH(block);
bp = buf_hash[b];
while (bp != NULL) {
if (bp->lmfs_blocknr == block && bp->lmfs_dev == dev) {
if(bp->lmfs_flags & VMMC_EVICTED) {
/* We had it but VM evicted it; invalidate it. */
ASSERT(bp->lmfs_count == 0);
ASSERT(!(bp->lmfs_flags & VMMC_BLOCK_LOCKED));
ASSERT(!(bp->lmfs_flags & VMMC_DIRTY));
bp->lmfs_dev = NO_DEV;
bp->lmfs_bytes = 0;
bp->data = NULL;
break;
}
ASSERT(bp->lmfs_needsetcache == 0);
/* Block needed has been found. */
if (bp->lmfs_count == 0) {
rm_lru(bp);
ASSERT(!(bp->lmfs_flags & VMMC_BLOCK_LOCKED));
bp->lmfs_flags |= VMMC_BLOCK_LOCKED;
}
raisecount(bp);
ASSERT(bp->lmfs_bytes == fs_block_size);
ASSERT(bp->lmfs_dev == dev);
ASSERT(bp->lmfs_dev != NO_DEV);
ASSERT(bp->lmfs_flags & VMMC_BLOCK_LOCKED);
ASSERT(bp->data);
if(ino != VMC_NO_INODE) {
if(bp->lmfs_inode == VMC_NO_INODE
|| bp->lmfs_inode != ino
|| bp->lmfs_inode_offset != ino_off) {
bp->lmfs_inode = ino;
bp->lmfs_inode_offset = ino_off;
bp->lmfs_needsetcache = 1;
}
}
return(bp);
} else {
/* This block is not the one sought. */
bp = bp->lmfs_hash; /* move to next block on hash chain */
}
}
/* Desired block is not on available chain. Find a free block to use. */
if(bp) {
ASSERT(bp->lmfs_flags & VMMC_EVICTED);
} else {
if ((bp = front) == NULL) panic("all buffers in use: %d", nr_bufs);
}
assert(bp);
rm_lru(bp);
/* Remove the block that was just taken from its hash chain. */
b = BUFHASH(bp->lmfs_blocknr);
prev_ptr = buf_hash[b];
if (prev_ptr == bp) {
buf_hash[b] = bp->lmfs_hash;
} else {
/* The block just taken is not on the front of its hash chain. */
while (prev_ptr->lmfs_hash != NULL)
if (prev_ptr->lmfs_hash == bp) {
prev_ptr->lmfs_hash = bp->lmfs_hash; /* found it */
break;
} else {
prev_ptr = prev_ptr->lmfs_hash; /* keep looking */
}
}
freeblock(bp);
bp->lmfs_inode = ino;
bp->lmfs_inode_offset = ino_off;
bp->lmfs_flags = VMMC_BLOCK_LOCKED;
bp->lmfs_needsetcache = 0;
bp->lmfs_dev = dev; /* fill in device number */
bp->lmfs_blocknr = block; /* fill in block number */
ASSERT(bp->lmfs_count == 0);
raisecount(bp);
b = BUFHASH(bp->lmfs_blocknr);
bp->lmfs_hash = buf_hash[b];
buf_hash[b] = bp; /* add to hash list */
assert(dev != NO_DEV);
/* Block is not found in our cache, but we do want it
* if it's in the vm cache.
*/
assert(!bp->data);
assert(!bp->lmfs_bytes);
if(vmcache) {
if((bp->data = vm_map_cacheblock(dev, dev_off, ino, ino_off,
&bp->lmfs_flags, fs_block_size)) != MAP_FAILED) {
bp->lmfs_bytes = fs_block_size;
ASSERT(!bp->lmfs_needsetcache);
return bp;
}
}
bp->data = NULL;
/* Not in the cache; reserve memory for its contents. */
lmfs_alloc_block(bp);
assert(bp->data);
if(only_search == PREFETCH) {
/* PREFETCH: don't do i/o. */
bp->lmfs_dev = NO_DEV;
} else if (only_search == NORMAL) {
read_block(bp);
} else if(only_search == NO_READ) {
/* This block will be overwritten by new contents. */
} else
panic("unexpected only_search value: %d", only_search);
assert(bp->data);
return(bp); /* return the newly acquired block */
}
/*===========================================================================*
* get_block *
*===========================================================================*/
struct buf *get_block(
register dev_t dev, /* on which device is the block? */
register block_t block, /* which block is wanted? */
int only_search /* if NO_READ, don't read, else act normal */
)
{
/* Check to see if the requested block is in the block cache. If so, return
* a pointer to it. If not, evict some other block and fetch it (unless
* 'only_search' is 1). All the blocks in the cache that are not in use
* are linked together in a chain, with 'front' pointing to the least recently
* used block and 'rear' to the most recently used block. If 'only_search' is
* 1, the block being requested will be overwritten in its entirety, so it is
* only necessary to see if it is in the cache; if it is not, any free buffer
* will do. It is not necessary to actually read the block in from disk.
* If 'only_search' is PREFETCH, the block need not be read from the disk,
* and the device is not to be marked on the block, so callers can tell if
* the block returned is valid.
* In addition to the LRU chain, there is also a hash chain to link together
* blocks whose block numbers end with the same bit strings, for fast lookup.
*/
int b;
static struct buf *bp, *prev_ptr;
u64_t yieldid = VM_BLOCKID_NONE, getid = make64(dev, block);
assert(buf_hash);
assert(buf);
assert(nr_bufs > 0);
ASSERT(fs_block_size > 0);
/* Search the hash chain for (dev, block). Do_read() can use
* get_block(NO_DEV ...) to get an unnamed block to fill with zeros when
* someone wants to read from a hole in a file, in which case this search
* is skipped
*/
if (dev != NO_DEV) {
b = BUFHASH(block);
bp = buf_hash[b];
while (bp != NULL) {
if (bp->b_blocknr == block && bp->b_dev == dev) {
/* Block needed has been found. */
if (bp->b_count == 0) rm_lru(bp);
bp->b_count++; /* record that block is in use */
ASSERT(bp->b_bytes == fs_block_size);
ASSERT(bp->b_dev == dev);
ASSERT(bp->b_dev != NO_DEV);
ASSERT(bp->bp);
return(bp);
} else {
/* This block is not the one sought. */
bp = bp->b_hash; /* move to next block on hash chain */
}
}
}
/* Desired block is not on available chain. Take oldest block ('front'). */
if ((bp = front) == NULL) panic("all buffers in use: %d", nr_bufs);
if(bp->b_bytes < fs_block_size) {
ASSERT(!bp->bp);
ASSERT(bp->b_bytes == 0);
if(!(bp->bp = alloc_contig( (size_t) fs_block_size, 0, NULL))) {
printf("MFS: couldn't allocate a new block.\n");
for(bp = front;
bp && bp->b_bytes < fs_block_size; bp = bp->b_next)
;
if(!bp) {
panic("no buffer available");
}
} else {
bp->b_bytes = fs_block_size;
}
}
ASSERT(bp);
ASSERT(bp->bp);
ASSERT(bp->b_bytes == fs_block_size);
ASSERT(bp->b_count == 0);
rm_lru(bp);
/* Remove the block that was just taken from its hash chain. */
b = BUFHASH(bp->b_blocknr);
prev_ptr = buf_hash[b];
if (prev_ptr == bp) {
buf_hash[b] = bp->b_hash;
} else {
/* The block just taken is not on the front of its hash chain. */
while (prev_ptr->b_hash != NULL)
if (prev_ptr->b_hash == bp) {
prev_ptr->b_hash = bp->b_hash; /* found it */
break;
} else {
prev_ptr = prev_ptr->b_hash; /* keep looking */
}
}
/* If the block taken is dirty, make it clean by writing it to the disk.
* Avoid hysteresis by flushing all other dirty blocks for the same device.
*/
if (bp->b_dev != NO_DEV) {
if (ISDIRTY(bp)) flushall(bp->b_dev);
/* Are we throwing out a block that contained something?
* Give it to VM for the second-layer cache.
*/
yieldid = make64(bp->b_dev, bp->b_blocknr);
assert(bp->b_bytes == fs_block_size);
BP_CLEARDEV(bp);
}
/* Fill in block's parameters and add it to the hash chain where it goes. */
if(dev == NO_DEV) BP_CLEARDEV(bp);
else BP_SETDEV(bp, dev);
bp->b_blocknr = block; /* fill in block number */
bp->b_count++; /* record that block is being used */
b = BUFHASH(bp->b_blocknr);
bp->b_hash = buf_hash[b];
buf_hash[b] = bp; /* add to hash list */
if(dev == NO_DEV) {
if(vmcache && cmp64(yieldid, VM_BLOCKID_NONE) != 0) {
vm_yield_block_get_block(yieldid, VM_BLOCKID_NONE,
bp->bp, fs_block_size);
}
return(bp); /* If the caller wanted a NO_DEV block, work is done. */
}
/* Go get the requested block unless searching or prefetching. */
if(only_search == PREFETCH || only_search == NORMAL) {
/* Block is not found in our cache, but we do want it
* if it's in the vm cache.
*/
if(vmcache) {
/* If we can satisfy the PREFETCH or NORMAL request
* from the vm cache, work is done.
*/
if(vm_yield_block_get_block(yieldid, getid,
bp->bp, fs_block_size) == OK) {
return bp;
}
}
}
if(only_search == PREFETCH) {
/* PREFETCH: don't do i/o. */
BP_CLEARDEV(bp);
} else if (only_search == NORMAL) {
read_block(bp);
} else if(only_search == NO_READ) {
/* we want this block, but its contents
* will be overwritten. VM has to forget
* about it.
*/
if(vmcache) {
vm_forgetblock(getid);
}
} else
panic("unexpected only_search value: %d", only_search);
assert(bp->bp);
return(bp); /* return the newly acquired block */
}
/*===========================================================================*
* get_block_ino *
*===========================================================================*/
static int get_block_ino(struct buf **bpp, dev_t dev, block64_t block, int how,
ino_t ino, u64_t ino_off, size_t block_size)
{
/* Check to see if the requested block is in the block cache. The requested
* block is identified by the block number in 'block' on device 'dev', counted
* in the file system block size. The amount of data requested for this block
* is given in 'block_size', which may be less than the file system block size
* iff the requested block is the last (partial) block on a device. Note that
* the given block size does *not* affect the conversion of 'block' to a byte
* offset! Either way, if the block could be obtained, either from the cache
* or by reading from the device, return OK, with a pointer to the buffer
* structure stored in 'bpp'. If not, return a negative error code (and no
* buffer). If necessary, evict some other block and fetch the contents from
* disk (if 'how' is NORMAL). If 'how' is NO_READ, the caller intends to
* overwrite the requested block in its entirety, so it is only necessary to
* see if it is in the cache; if it is not, any free buffer will do. If 'how'
* is PEEK, the function returns the block if it is in the cache or the VM
* cache, and an ENOENT error code otherwise.
* In addition to the LRU chain, there is also a hash chain to link together
* blocks whose block numbers end with the same bit strings, for fast lookup.
*/
int b, r;
static struct buf *bp;
uint64_t dev_off;
struct buf *prev_ptr;
assert(buf_hash);
assert(buf);
assert(nr_bufs > 0);
ASSERT(fs_block_size > 0);
assert(dev != NO_DEV);
assert(block <= UINT64_MAX / fs_block_size);
dev_off = block * fs_block_size;
if((ino_off % fs_block_size)) {
printf("cache: unaligned lmfs_get_block_ino ino_off %llu\n",
ino_off);
util_stacktrace();
}
/* See if the block is in the cache. If so, we can return it right away. */
bp = find_block(dev, block);
if (bp != NULL && !(bp->lmfs_flags & VMMC_EVICTED)) {
ASSERT(bp->lmfs_dev == dev);
ASSERT(bp->lmfs_dev != NO_DEV);
/* The block must have exactly the requested number of bytes. */
if (bp->lmfs_bytes != block_size)
return EIO;
/* Block needed has been found. */
if (bp->lmfs_count == 0) {
rm_lru(bp);
ASSERT(bp->lmfs_needsetcache == 0);
ASSERT(!(bp->lmfs_flags & VMMC_BLOCK_LOCKED));
/* FIXME: race condition against the VMMC_EVICTED check */
bp->lmfs_flags |= VMMC_BLOCK_LOCKED;
}
raisecount(bp);
ASSERT(bp->lmfs_flags & VMMC_BLOCK_LOCKED);
ASSERT(bp->data);
if(ino != VMC_NO_INODE) {
if(bp->lmfs_inode == VMC_NO_INODE
|| bp->lmfs_inode != ino
|| bp->lmfs_inode_offset != ino_off) {
bp->lmfs_inode = ino;
bp->lmfs_inode_offset = ino_off;
bp->lmfs_needsetcache = 1;
}
}
*bpp = bp;
return OK;
}
/* We had the block in the cache but VM evicted it; invalidate it. */
if (bp != NULL) {
assert(bp->lmfs_flags & VMMC_EVICTED);
ASSERT(bp->lmfs_count == 0);
ASSERT(!(bp->lmfs_flags & VMMC_BLOCK_LOCKED));
ASSERT(!(bp->lmfs_flags & VMMC_DIRTY));
bp->lmfs_dev = NO_DEV;
bp->lmfs_bytes = 0;
bp->data = NULL;
}
/* Desired block is not on available chain. Find a free block to use. */
if(bp) {
ASSERT(bp->lmfs_flags & VMMC_EVICTED);
} else {
if ((bp = front) == NULL) panic("all buffers in use: %d", nr_bufs);
}
assert(bp);
rm_lru(bp);
/* Remove the block that was just taken from its hash chain. */
b = BUFHASH(bp->lmfs_blocknr);
prev_ptr = buf_hash[b];
if (prev_ptr == bp) {
buf_hash[b] = bp->lmfs_hash;
} else {
/* The block just taken is not on the front of its hash chain. */
while (prev_ptr->lmfs_hash != NULL)
if (prev_ptr->lmfs_hash == bp) {
prev_ptr->lmfs_hash = bp->lmfs_hash; /* found it */
break;
} else {
prev_ptr = prev_ptr->lmfs_hash; /* keep looking */
}
}
freeblock(bp);
bp->lmfs_inode = ino;
bp->lmfs_inode_offset = ino_off;
bp->lmfs_flags = VMMC_BLOCK_LOCKED;
bp->lmfs_needsetcache = 0;
bp->lmfs_dev = dev; /* fill in device number */
bp->lmfs_blocknr = block; /* fill in block number */
ASSERT(bp->lmfs_count == 0);
raisecount(bp);
b = BUFHASH(bp->lmfs_blocknr);
bp->lmfs_hash = buf_hash[b];
buf_hash[b] = bp; /* add to hash list */
assert(dev != NO_DEV);
/* The block is not found in our cache, but we do want it if it's in the VM
* cache. The exception is NO_READ, purely for context switching performance
* reasons. NO_READ is used for 1) newly allocated blocks, 2) blocks being
* prefetched, and 3) blocks about to be fully overwritten. In the first two
* cases, VM will not have the block in its cache anyway, and for the third
* we save on one VM call only if the block is in the VM cache.
*/
assert(!bp->data);
assert(!bp->lmfs_bytes);
if (how != NO_READ && vmcache) {
if((bp->data = vm_map_cacheblock(dev, dev_off, ino, ino_off,
&bp->lmfs_flags, roundup(block_size, PAGE_SIZE))) != MAP_FAILED) {
bp->lmfs_bytes = block_size;
ASSERT(!bp->lmfs_needsetcache);
*bpp = bp;
return OK;
}
}
bp->data = NULL;
/* The block is not in the cache, and VM does not know about it. If we were
* requested to search for the block only, we can now return failure to the
* caller. Return the block to the pool without allocating data pages, since
* these would be freed upon recycling the block anyway.
*/
if (how == PEEK) {
bp->lmfs_dev = NO_DEV;
put_block(bp, ONE_SHOT);
return ENOENT;
}
/* Not in the cache; reserve memory for its contents. */
lmfs_alloc_block(bp, block_size);
assert(bp->data);
if (how == NORMAL) {
/* Try to read the block. Return an error code on failure. */
if ((r = read_block(bp, block_size)) != OK) {
put_block(bp, 0);
return r;
}
} else if(how == NO_READ) {
/* This block will be overwritten by new contents. */
} else
panic("unexpected 'how' value: %d", how);
assert(bp->data);
*bpp = bp; /* return the newly acquired block */
return OK;
}