diff options
Diffstat (limited to '')
| -rw-r--r-- | fs/bcachefs/compress.c | 732 |
1 files changed, 732 insertions, 0 deletions
diff --git a/fs/bcachefs/compress.c b/fs/bcachefs/compress.c new file mode 100644 index 0000000000..51af8ea230 --- /dev/null +++ b/fs/bcachefs/compress.c @@ -0,0 +1,732 @@ +// SPDX-License-Identifier: GPL-2.0 +#include "bcachefs.h" +#include "checksum.h" +#include "compress.h" +#include "extents.h" +#include "super-io.h" + +#include <linux/lz4.h> +#include <linux/zlib.h> +#include <linux/zstd.h> + +/* Bounce buffer: */ +struct bbuf { + void *b; + enum { + BB_NONE, + BB_VMAP, + BB_KMALLOC, + BB_MEMPOOL, + } type; + int rw; +}; + +static struct bbuf __bounce_alloc(struct bch_fs *c, unsigned size, int rw) +{ + void *b; + + BUG_ON(size > c->opts.encoded_extent_max); + + b = kmalloc(size, GFP_NOFS|__GFP_NOWARN); + if (b) + return (struct bbuf) { .b = b, .type = BB_KMALLOC, .rw = rw }; + + b = mempool_alloc(&c->compression_bounce[rw], GFP_NOFS); + if (b) + return (struct bbuf) { .b = b, .type = BB_MEMPOOL, .rw = rw }; + + BUG(); +} + +static bool bio_phys_contig(struct bio *bio, struct bvec_iter start) +{ + struct bio_vec bv; + struct bvec_iter iter; + void *expected_start = NULL; + + __bio_for_each_bvec(bv, bio, iter, start) { + if (expected_start && + expected_start != page_address(bv.bv_page) + bv.bv_offset) + return false; + + expected_start = page_address(bv.bv_page) + + bv.bv_offset + bv.bv_len; + } + + return true; +} + +static struct bbuf __bio_map_or_bounce(struct bch_fs *c, struct bio *bio, + struct bvec_iter start, int rw) +{ + struct bbuf ret; + struct bio_vec bv; + struct bvec_iter iter; + unsigned nr_pages = 0; + struct page *stack_pages[16]; + struct page **pages = NULL; + void *data; + + BUG_ON(start.bi_size > c->opts.encoded_extent_max); + + if (!PageHighMem(bio_iter_page(bio, start)) && + bio_phys_contig(bio, start)) + return (struct bbuf) { + .b = page_address(bio_iter_page(bio, start)) + + bio_iter_offset(bio, start), + .type = BB_NONE, .rw = rw + }; + + /* check if we can map the pages contiguously: */ + __bio_for_each_segment(bv, bio, iter, start) { + if (iter.bi_size != start.bi_size && + bv.bv_offset) + goto bounce; + + if (bv.bv_len < iter.bi_size && + bv.bv_offset + bv.bv_len < PAGE_SIZE) + goto bounce; + + nr_pages++; + } + + BUG_ON(DIV_ROUND_UP(start.bi_size, PAGE_SIZE) > nr_pages); + + pages = nr_pages > ARRAY_SIZE(stack_pages) + ? kmalloc_array(nr_pages, sizeof(struct page *), GFP_NOFS) + : stack_pages; + if (!pages) + goto bounce; + + nr_pages = 0; + __bio_for_each_segment(bv, bio, iter, start) + pages[nr_pages++] = bv.bv_page; + + data = vmap(pages, nr_pages, VM_MAP, PAGE_KERNEL); + if (pages != stack_pages) + kfree(pages); + + if (data) + return (struct bbuf) { + .b = data + bio_iter_offset(bio, start), + .type = BB_VMAP, .rw = rw + }; +bounce: + ret = __bounce_alloc(c, start.bi_size, rw); + + if (rw == READ) + memcpy_from_bio(ret.b, bio, start); + + return ret; +} + +static struct bbuf bio_map_or_bounce(struct bch_fs *c, struct bio *bio, int rw) +{ + return __bio_map_or_bounce(c, bio, bio->bi_iter, rw); +} + +static void bio_unmap_or_unbounce(struct bch_fs *c, struct bbuf buf) +{ + switch (buf.type) { + case BB_NONE: + break; + case BB_VMAP: + vunmap((void *) ((unsigned long) buf.b & PAGE_MASK)); + break; + case BB_KMALLOC: + kfree(buf.b); + break; + case BB_MEMPOOL: + mempool_free(buf.b, &c->compression_bounce[buf.rw]); + break; + } +} + +static inline void zlib_set_workspace(z_stream *strm, void *workspace) +{ +#ifdef __KERNEL__ + strm->workspace = workspace; +#endif +} + +static int __bio_uncompress(struct bch_fs *c, struct bio *src, + void *dst_data, struct bch_extent_crc_unpacked crc) +{ + struct bbuf src_data = { NULL }; + size_t src_len = src->bi_iter.bi_size; + size_t dst_len = crc.uncompressed_size << 9; + void *workspace; + int ret; + + src_data = bio_map_or_bounce(c, src, READ); + + switch (crc.compression_type) { + case BCH_COMPRESSION_TYPE_lz4_old: + case BCH_COMPRESSION_TYPE_lz4: + ret = LZ4_decompress_safe_partial(src_data.b, dst_data, + src_len, dst_len, dst_len); + if (ret != dst_len) + goto err; + break; + case BCH_COMPRESSION_TYPE_gzip: { + z_stream strm = { + .next_in = src_data.b, + .avail_in = src_len, + .next_out = dst_data, + .avail_out = dst_len, + }; + + workspace = mempool_alloc(&c->decompress_workspace, GFP_NOFS); + + zlib_set_workspace(&strm, workspace); + zlib_inflateInit2(&strm, -MAX_WBITS); + ret = zlib_inflate(&strm, Z_FINISH); + + mempool_free(workspace, &c->decompress_workspace); + + if (ret != Z_STREAM_END) + goto err; + break; + } + case BCH_COMPRESSION_TYPE_zstd: { + ZSTD_DCtx *ctx; + size_t real_src_len = le32_to_cpup(src_data.b); + + if (real_src_len > src_len - 4) + goto err; + + workspace = mempool_alloc(&c->decompress_workspace, GFP_NOFS); + ctx = zstd_init_dctx(workspace, zstd_dctx_workspace_bound()); + + ret = zstd_decompress_dctx(ctx, + dst_data, dst_len, + src_data.b + 4, real_src_len); + + mempool_free(workspace, &c->decompress_workspace); + + if (ret != dst_len) + goto err; + break; + } + default: + BUG(); + } + ret = 0; +out: + bio_unmap_or_unbounce(c, src_data); + return ret; +err: + ret = -EIO; + goto out; +} + +int bch2_bio_uncompress_inplace(struct bch_fs *c, struct bio *bio, + struct bch_extent_crc_unpacked *crc) +{ + struct bbuf data = { NULL }; + size_t dst_len = crc->uncompressed_size << 9; + + /* bio must own its pages: */ + BUG_ON(!bio->bi_vcnt); + BUG_ON(DIV_ROUND_UP(crc->live_size, PAGE_SECTORS) > bio->bi_max_vecs); + + if (crc->uncompressed_size << 9 > c->opts.encoded_extent_max || + crc->compressed_size << 9 > c->opts.encoded_extent_max) { + bch_err(c, "error rewriting existing data: extent too big"); + return -EIO; + } + + data = __bounce_alloc(c, dst_len, WRITE); + + if (__bio_uncompress(c, bio, data.b, *crc)) { + if (!c->opts.no_data_io) + bch_err(c, "error rewriting existing data: decompression error"); + bio_unmap_or_unbounce(c, data); + return -EIO; + } + + /* + * XXX: don't have a good way to assert that the bio was allocated with + * enough space, we depend on bch2_move_extent doing the right thing + */ + bio->bi_iter.bi_size = crc->live_size << 9; + + memcpy_to_bio(bio, bio->bi_iter, data.b + (crc->offset << 9)); + + crc->csum_type = 0; + crc->compression_type = 0; + crc->compressed_size = crc->live_size; + crc->uncompressed_size = crc->live_size; + crc->offset = 0; + crc->csum = (struct bch_csum) { 0, 0 }; + + bio_unmap_or_unbounce(c, data); + return 0; +} + +int bch2_bio_uncompress(struct bch_fs *c, struct bio *src, + struct bio *dst, struct bvec_iter dst_iter, + struct bch_extent_crc_unpacked crc) +{ + struct bbuf dst_data = { NULL }; + size_t dst_len = crc.uncompressed_size << 9; + int ret; + + if (crc.uncompressed_size << 9 > c->opts.encoded_extent_max || + crc.compressed_size << 9 > c->opts.encoded_extent_max) + return -EIO; + + dst_data = dst_len == dst_iter.bi_size + ? __bio_map_or_bounce(c, dst, dst_iter, WRITE) + : __bounce_alloc(c, dst_len, WRITE); + + ret = __bio_uncompress(c, src, dst_data.b, crc); + if (ret) + goto err; + + if (dst_data.type != BB_NONE && + dst_data.type != BB_VMAP) + memcpy_to_bio(dst, dst_iter, dst_data.b + (crc.offset << 9)); +err: + bio_unmap_or_unbounce(c, dst_data); + return ret; +} + +static int attempt_compress(struct bch_fs *c, + void *workspace, + void *dst, size_t dst_len, + void *src, size_t src_len, + struct bch_compression_opt compression) +{ + enum bch_compression_type compression_type = + __bch2_compression_opt_to_type[compression.type]; + + switch (compression_type) { + case BCH_COMPRESSION_TYPE_lz4: + if (compression.level < LZ4HC_MIN_CLEVEL) { + int len = src_len; + int ret = LZ4_compress_destSize( + src, dst, + &len, dst_len, + workspace); + if (len < src_len) + return -len; + + return ret; + } else { + int ret = LZ4_compress_HC( + src, dst, + src_len, dst_len, + compression.level, + workspace); + + return ret ?: -1; + } + case BCH_COMPRESSION_TYPE_gzip: { + z_stream strm = { + .next_in = src, + .avail_in = src_len, + .next_out = dst, + .avail_out = dst_len, + }; + + zlib_set_workspace(&strm, workspace); + zlib_deflateInit2(&strm, + compression.level + ? clamp_t(unsigned, compression.level, + Z_BEST_SPEED, Z_BEST_COMPRESSION) + : Z_DEFAULT_COMPRESSION, + Z_DEFLATED, -MAX_WBITS, DEF_MEM_LEVEL, + Z_DEFAULT_STRATEGY); + + if (zlib_deflate(&strm, Z_FINISH) != Z_STREAM_END) + return 0; + + if (zlib_deflateEnd(&strm) != Z_OK) + return 0; + + return strm.total_out; + } + case BCH_COMPRESSION_TYPE_zstd: { + /* + * rescale: + * zstd max compression level is 22, our max level is 15 + */ + unsigned level = min((compression.level * 3) / 2, zstd_max_clevel()); + ZSTD_parameters params = zstd_get_params(level, c->opts.encoded_extent_max); + ZSTD_CCtx *ctx = zstd_init_cctx(workspace, c->zstd_workspace_size); + + /* + * ZSTD requires that when we decompress we pass in the exact + * compressed size - rounding it up to the nearest sector + * doesn't work, so we use the first 4 bytes of the buffer for + * that. + * + * Additionally, the ZSTD code seems to have a bug where it will + * write just past the end of the buffer - so subtract a fudge + * factor (7 bytes) from the dst buffer size to account for + * that. + */ + size_t len = zstd_compress_cctx(ctx, + dst + 4, dst_len - 4 - 7, + src, src_len, + ¶ms); + if (zstd_is_error(len)) + return 0; + + *((__le32 *) dst) = cpu_to_le32(len); + return len + 4; + } + default: + BUG(); + } +} + +static unsigned __bio_compress(struct bch_fs *c, + struct bio *dst, size_t *dst_len, + struct bio *src, size_t *src_len, + struct bch_compression_opt compression) +{ + struct bbuf src_data = { NULL }, dst_data = { NULL }; + void *workspace; + enum bch_compression_type compression_type = + __bch2_compression_opt_to_type[compression.type]; + unsigned pad; + int ret = 0; + + BUG_ON(compression_type >= BCH_COMPRESSION_TYPE_NR); + BUG_ON(!mempool_initialized(&c->compress_workspace[compression_type])); + + /* If it's only one block, don't bother trying to compress: */ + if (src->bi_iter.bi_size <= c->opts.block_size) + return BCH_COMPRESSION_TYPE_incompressible; + + dst_data = bio_map_or_bounce(c, dst, WRITE); + src_data = bio_map_or_bounce(c, src, READ); + + workspace = mempool_alloc(&c->compress_workspace[compression_type], GFP_NOFS); + + *src_len = src->bi_iter.bi_size; + *dst_len = dst->bi_iter.bi_size; + + /* + * XXX: this algorithm sucks when the compression code doesn't tell us + * how much would fit, like LZ4 does: + */ + while (1) { + if (*src_len <= block_bytes(c)) { + ret = -1; + break; + } + + ret = attempt_compress(c, workspace, + dst_data.b, *dst_len, + src_data.b, *src_len, + compression); + if (ret > 0) { + *dst_len = ret; + ret = 0; + break; + } + + /* Didn't fit: should we retry with a smaller amount? */ + if (*src_len <= *dst_len) { + ret = -1; + break; + } + + /* + * If ret is negative, it's a hint as to how much data would fit + */ + BUG_ON(-ret >= *src_len); + + if (ret < 0) + *src_len = -ret; + else + *src_len -= (*src_len - *dst_len) / 2; + *src_len = round_down(*src_len, block_bytes(c)); + } + + mempool_free(workspace, &c->compress_workspace[compression_type]); + + if (ret) + goto err; + + /* Didn't get smaller: */ + if (round_up(*dst_len, block_bytes(c)) >= *src_len) + goto err; + + pad = round_up(*dst_len, block_bytes(c)) - *dst_len; + + memset(dst_data.b + *dst_len, 0, pad); + *dst_len += pad; + + if (dst_data.type != BB_NONE && + dst_data.type != BB_VMAP) + memcpy_to_bio(dst, dst->bi_iter, dst_data.b); + + BUG_ON(!*dst_len || *dst_len > dst->bi_iter.bi_size); + BUG_ON(!*src_len || *src_len > src->bi_iter.bi_size); + BUG_ON(*dst_len & (block_bytes(c) - 1)); + BUG_ON(*src_len & (block_bytes(c) - 1)); + ret = compression_type; +out: + bio_unmap_or_unbounce(c, src_data); + bio_unmap_or_unbounce(c, dst_data); + return ret; +err: + ret = BCH_COMPRESSION_TYPE_incompressible; + goto out; +} + +unsigned bch2_bio_compress(struct bch_fs *c, + struct bio *dst, size_t *dst_len, + struct bio *src, size_t *src_len, + unsigned compression_opt) +{ + unsigned orig_dst = dst->bi_iter.bi_size; + unsigned orig_src = src->bi_iter.bi_size; + unsigned compression_type; + + /* Don't consume more than BCH_ENCODED_EXTENT_MAX from @src: */ + src->bi_iter.bi_size = min_t(unsigned, src->bi_iter.bi_size, + c->opts.encoded_extent_max); + /* Don't generate a bigger output than input: */ + dst->bi_iter.bi_size = min(dst->bi_iter.bi_size, src->bi_iter.bi_size); + + compression_type = + __bio_compress(c, dst, dst_len, src, src_len, + bch2_compression_decode(compression_opt)); + + dst->bi_iter.bi_size = orig_dst; + src->bi_iter.bi_size = orig_src; + return compression_type; +} + +static int __bch2_fs_compress_init(struct bch_fs *, u64); + +#define BCH_FEATURE_none 0 + +static const unsigned bch2_compression_opt_to_feature[] = { +#define x(t, n) [BCH_COMPRESSION_OPT_##t] = BCH_FEATURE_##t, + BCH_COMPRESSION_OPTS() +#undef x +}; + +#undef BCH_FEATURE_none + +static int __bch2_check_set_has_compressed_data(struct bch_fs *c, u64 f) +{ + int ret = 0; + + if ((c->sb.features & f) == f) + return 0; + + mutex_lock(&c->sb_lock); + + if ((c->sb.features & f) == f) { + mutex_unlock(&c->sb_lock); + return 0; + } + + ret = __bch2_fs_compress_init(c, c->sb.features|f); + if (ret) { + mutex_unlock(&c->sb_lock); + return ret; + } + + c->disk_sb.sb->features[0] |= cpu_to_le64(f); + bch2_write_super(c); + mutex_unlock(&c->sb_lock); + + return 0; +} + +int bch2_check_set_has_compressed_data(struct bch_fs *c, + unsigned compression_opt) +{ + unsigned compression_type = bch2_compression_decode(compression_opt).type; + + BUG_ON(compression_type >= ARRAY_SIZE(bch2_compression_opt_to_feature)); + + return compression_type + ? __bch2_check_set_has_compressed_data(c, + 1ULL << bch2_compression_opt_to_feature[compression_type]) + : 0; +} + +void bch2_fs_compress_exit(struct bch_fs *c) +{ + unsigned i; + + mempool_exit(&c->decompress_workspace); + for (i = 0; i < ARRAY_SIZE(c->compress_workspace); i++) + mempool_exit(&c->compress_workspace[i]); + mempool_exit(&c->compression_bounce[WRITE]); + mempool_exit(&c->compression_bounce[READ]); +} + +static int __bch2_fs_compress_init(struct bch_fs *c, u64 features) +{ + size_t decompress_workspace_size = 0; + ZSTD_parameters params = zstd_get_params(zstd_max_clevel(), + c->opts.encoded_extent_max); + + /* + * ZSTD is lying: if we allocate the size of the workspace it says it + * requires, it returns memory allocation errors + */ + c->zstd_workspace_size = zstd_cctx_workspace_bound(¶ms.cParams); + + struct { + unsigned feature; + enum bch_compression_type type; + size_t compress_workspace; + size_t decompress_workspace; + } compression_types[] = { + { BCH_FEATURE_lz4, BCH_COMPRESSION_TYPE_lz4, + max_t(size_t, LZ4_MEM_COMPRESS, LZ4HC_MEM_COMPRESS), + 0 }, + { BCH_FEATURE_gzip, BCH_COMPRESSION_TYPE_gzip, + zlib_deflate_workspacesize(MAX_WBITS, DEF_MEM_LEVEL), + zlib_inflate_workspacesize(), }, + { BCH_FEATURE_zstd, BCH_COMPRESSION_TYPE_zstd, + c->zstd_workspace_size, + zstd_dctx_workspace_bound() }, + }, *i; + bool have_compressed = false; + + for (i = compression_types; + i < compression_types + ARRAY_SIZE(compression_types); + i++) + have_compressed |= (features & (1 << i->feature)) != 0; + + if (!have_compressed) + return 0; + + if (!mempool_initialized(&c->compression_bounce[READ]) && + mempool_init_kvpmalloc_pool(&c->compression_bounce[READ], + 1, c->opts.encoded_extent_max)) + return -BCH_ERR_ENOMEM_compression_bounce_read_init; + + if (!mempool_initialized(&c->compression_bounce[WRITE]) && + mempool_init_kvpmalloc_pool(&c->compression_bounce[WRITE], + 1, c->opts.encoded_extent_max)) + return -BCH_ERR_ENOMEM_compression_bounce_write_init; + + for (i = compression_types; + i < compression_types + ARRAY_SIZE(compression_types); + i++) { + decompress_workspace_size = + max(decompress_workspace_size, i->decompress_workspace); + + if (!(features & (1 << i->feature))) + continue; + + if (mempool_initialized(&c->compress_workspace[i->type])) + continue; + + if (mempool_init_kvpmalloc_pool( + &c->compress_workspace[i->type], + 1, i->compress_workspace)) + return -BCH_ERR_ENOMEM_compression_workspace_init; + } + + if (!mempool_initialized(&c->decompress_workspace) && + mempool_init_kvpmalloc_pool(&c->decompress_workspace, + 1, decompress_workspace_size)) + return -BCH_ERR_ENOMEM_decompression_workspace_init; + + return 0; +} + +static u64 compression_opt_to_feature(unsigned v) +{ + unsigned type = bch2_compression_decode(v).type; + + return BIT_ULL(bch2_compression_opt_to_feature[type]); +} + +int bch2_fs_compress_init(struct bch_fs *c) +{ + u64 f = c->sb.features; + + f |= compression_opt_to_feature(c->opts.compression); + f |= compression_opt_to_feature(c->opts.background_compression); + + return __bch2_fs_compress_init(c, f); +} + +int bch2_opt_compression_parse(struct bch_fs *c, const char *_val, u64 *res, + struct printbuf *err) +{ + char *val = kstrdup(_val, GFP_KERNEL); + char *p = val, *type_str, *level_str; + struct bch_compression_opt opt = { 0 }; + int ret; + + if (!val) + return -ENOMEM; + + type_str = strsep(&p, ":"); + level_str = p; + + ret = match_string(bch2_compression_opts, -1, type_str); + if (ret < 0 && err) + prt_str(err, "invalid compression type"); + if (ret < 0) + goto err; + + opt.type = ret; + + if (level_str) { + unsigned level; + + ret = kstrtouint(level_str, 10, &level); + if (!ret && !opt.type && level) + ret = -EINVAL; + if (!ret && level > 15) + ret = -EINVAL; + if (ret < 0 && err) + prt_str(err, "invalid compression level"); + if (ret < 0) + goto err; + + opt.level = level; + } + + *res = bch2_compression_encode(opt); +err: + kfree(val); + return ret; +} + +void bch2_compression_opt_to_text(struct printbuf *out, u64 v) +{ + struct bch_compression_opt opt = bch2_compression_decode(v); + + if (opt.type < BCH_COMPRESSION_OPT_NR) + prt_str(out, bch2_compression_opts[opt.type]); + else + prt_printf(out, "(unknown compression opt %u)", opt.type); + if (opt.level) + prt_printf(out, ":%u", opt.level); +} + +void bch2_opt_compression_to_text(struct printbuf *out, + struct bch_fs *c, + struct bch_sb *sb, + u64 v) +{ + return bch2_compression_opt_to_text(out, v); +} + +int bch2_opt_compression_validate(u64 v, struct printbuf *err) +{ + if (!bch2_compression_opt_valid(v)) { + prt_printf(err, "invalid compression opt %llu", v); + return -BCH_ERR_invalid_sb_opt_compression; + } + + return 0; +} |