1 // SPDX-License-Identifier: GPL-2.0-only
11  * MS-SMB2 "3.1.4.4 Compressing the Message"
12  * MS-SMB2 "3.1.5.3 Decompressing the Chained Message"
13  * MS-XCA - for details of the supported algorithms
39  * Parsing the sample goes from "low-hanging fruits" (fastest algorithms, likely compressible)
48  * has_low_entropy() - Compute Shannon entropy of the sampled data.
53  *	   compress the data) is below the minimum threshold.
56  * There _is_ an entropy level here that's > 65 (minimum threshold) that would indicate a
65 	const size_t threshold = 65, max_entropy = 8 * ilog2(16);  in has_low_entropy()  local
74 		sum += p * (len - p2);  in has_low_entropy()
79 	return ((sum * 100 / max_entropy) <= threshold);  in has_low_entropy()
86  * calc_byte_distribution() - Compute byte distribution on the sampled data.
91  * BYTE_DIST_BAD:	A "hard no" for compression -- a computed uniform distribution of
93  * BYTE_DIST_GOOD:	High probability (normal (Gaussian) distribution) of the data being
95  * BYTE_DIST_MAYBE:	When computed byte distribution resulted in "low > n < high"
100 	const size_t low = 64, high = 200, threshold = slen * 90 / 100;  in calc_byte_distribution()  local
107 	if (sum > threshold)  in calc_byte_distribution()
110 	for (; i < high && bkt[i].count > 0; i++) {  in calc_byte_distribution()
112 		if (sum > threshold)  in calc_byte_distribution()
119 	if (i >= high)  in calc_byte_distribution()
132 			/* Too many non-ASCII (0-63) bytes. */  in is_mostly_ascii()
151 	if (a->count > b->count)  in cmp_bkt()
152 		return -1;  in cmp_bkt()
166 	loff_t start = iter->xarray_start + iter->iov_offset;  in collect_sample()
172 	last = (start + max - 1) / PAGE_SIZE;  in collect_sample()
174 		nr = xa_extract(iter->xarray, (void **)folios, index, last, ARRAY_SIZE(folios),  in collect_sample()
177 			return -EIO;  in collect_sample()
182 			foff = start - folio_pos(folio);  in collect_sample()
188 				len = min_t(size_t, max, PAGE_SIZE - off);  in collect_sample()
197 				if (len2 < SZ_2K || s >= max - SZ_2K)  in collect_sample()
200 				max -= len;  in collect_sample()
215  * is_compressible() - Determines if a chunk of data is compressible.
232 	/* Preventive double check -- already checked in should_compress(). */  in is_compressible()
237 	if (len - read_size > max)  in is_compressible()
295 	const struct smb2_hdr *shdr = rq->rq_iov->iov_base;  in should_compress()
297 	if (unlikely(!tcon || !tcon->ses || !tcon->ses->server))  in should_compress()
300 	if (!tcon->ses->server->compression.enabled)  in should_compress()
303 	if (!(tcon->share_flags & SMB2_SHAREFLAG_COMPRESS_DATA))  in should_compress()
306 	if (shdr->Command == SMB2_WRITE) {  in should_compress()
307 		const struct smb2_write_req *wreq = rq->rq_iov->iov_base;  in should_compress()
309 		if (le32_to_cpu(wreq->Length) < SMB_COMPRESS_MIN_LEN)  in should_compress()
312 		return is_compressible(&rq->rq_iter);  in should_compress()
315 	return (shdr->Command == SMB2_READ);  in should_compress()
325 	if (!server || !rq || !rq->rq_iov || !rq->rq_iov->iov_base)  in smb_compress()
326 		return -EINVAL;  in smb_compress()
328 	if (rq->rq_iov->iov_len != sizeof(struct smb2_write_req))  in smb_compress()
329 		return -EINVAL;  in smb_compress()
331 	slen = iov_iter_count(&rq->rq_iter);  in smb_compress()
334 		ret = -ENOMEM;  in smb_compress()
339 	iter = rq->rq_iter;  in smb_compress()
342 		ret = -EIO;  in smb_compress()
353 		ret = -ENOMEM;  in smb_compress()
367 		hdr.Offset = cpu_to_le32(rq->rq_iov[0].iov_len);  in smb_compress()
371 		iov[1] = rq->rq_iov[0];  in smb_compress()
378 	} else if (ret == -EMSGSIZE || dlen >= slen) {  in smb_compress()