Merge tag 'v6.16-p1' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6

Pull crypto updates from Herbert Xu:
 "API:
   - Fix memcpy_sglist to handle partially overlapping SG lists
   - Use memcpy_sglist to replace null skcipher
   - Rename CRYPTO_TESTS to CRYPTO_BENCHMARK
   - Flip CRYPTO_MANAGER_DISABLE_TEST into CRYPTO_SELFTESTS
   - Hide CRYPTO_MANAGER
   - Add delayed freeing of driver crypto_alg structures

  Compression:
   - Allocate large buffers on first use instead of initialisation in scomp
   - Drop destination linearisation buffer in scomp
   - Move scomp stream allocation into acomp
   - Add acomp scatter-gather walker
   - Remove request chaining
   - Add optional async request allocation

  Hashing:
   - Remove request chaining
   - Add optional async request allocation
   - Move partial block handling into API
   - Add ahash support to hmac
   - Fix shash documentation to disallow usage in hard IRQs

  Algorithms:
   - Remove unnecessary SIMD fallback code on x86 and arm/arm64
   - Drop avx10_256 xts(aes)/ctr(aes) on x86
   - Improve avx-512 optimisations for xts(aes)
   - Move chacha arch implementations into lib/crypto
   - Move poly1305 into lib/crypto and drop unused Crypto API algorithm
   - Disable powerpc/poly1305 as it has no SIMD fallback
   - Move sha256 arch implementations into lib/crypto
   - Convert deflate to acomp
   - Set block size correctly in cbcmac

  Drivers:
   - Do not use sg_dma_len before mapping in sun8i-ss
   - Fix warm-reboot failure by making shutdown do more work in qat
   - Add locking in zynqmp-sha
   - Remove cavium/zip
   - Add support for PCI device 0x17D8 to ccp
   - Add qat_6xxx support in qat
   - Add support for RK3576 in rockchip-rng
   - Add support for i.MX8QM in caam

  Others:
   - Fix irq_fpu_usable/kernel_fpu_begin inconsistency during CPU bring-up
   - Add new SEV/SNP platform shutdown API in ccp"

* tag 'v6.16-p1' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6: (382 commits)
  x86/fpu: Fix irq_fpu_usable() to return false during CPU onlining
  crypto: qat - add missing header inclusion
  crypto: api - Redo lookup on EEXIST
  Revert "crypto: testmgr - Add hash export format testing"
  crypto: marvell/cesa - Do not chain submitted requests
  crypto: powerpc/poly1305 - add depends on BROKEN for now
  Revert "crypto: powerpc/poly1305 - Add SIMD fallback"
  crypto: ccp - Add missing tee info reg for teev2
  crypto: ccp - Add missing bootloader info reg for pspv5
  crypto: sun8i-ce - move fallback ahash_request to the end of the struct
  crypto: octeontx2 - Use dynamic allocated memory region for lmtst
  crypto: octeontx2 - Initialize cptlfs device info once
  crypto: xts - Only add ecb if it is not already there
  crypto: lrw - Only add ecb if it is not already there
  crypto: testmgr - Add hash export format testing
  crypto: testmgr - Use ahash for generic tfm
  crypto: hmac - Add ahash support
  crypto: testmgr - Ignore EEXIST on shash allocation
  crypto: algapi - Add driver template support to crypto_inst_setname
  crypto: shash - Set reqsize in shash_alg
  ...

2 files changed, 123 insertions, 142 deletions

diff --git a/fs/bcachefs/checksum.c b/fs/bcachefs/checksum.c
index ff5ab8ada777..d3e2e4f776c6 100644
--- a/fs/bcachefs/checksum.c
+++ b/fs/bcachefs/checksum.c
@@ -91,7 +91,7 @@ static void bch2_checksum_update(struct bch2_checksum_state *state, const void *
 	}
 }
 
-static void bch2_chacha20_init(u32 state[CHACHA_STATE_WORDS],
+static void bch2_chacha20_init(struct chacha_state *state,
 			       const struct bch_key *key, struct nonce nonce)
 {
 	u32 key_words[CHACHA_KEY_SIZE / sizeof(u32)];
@@ -109,11 +109,11 @@ static void bch2_chacha20_init(u32 state[CHACHA_STATE_WORDS],
 void bch2_chacha20(const struct bch_key *key, struct nonce nonce,
 		   void *data, size_t len)
 {
-	u32 state[CHACHA_STATE_WORDS];
+	struct chacha_state state;
 
-	bch2_chacha20_init(state, key, nonce);
-	chacha20_crypt(state, data, data, len);
-	memzero_explicit(state, sizeof(state));
+	bch2_chacha20_init(&state, key, nonce);
+	chacha20_crypt(&state, data, data, len);
+	chacha_zeroize_state(&state);
 }
 
 static void bch2_poly1305_init(struct poly1305_desc_ctx *desc,
@@ -257,14 +257,14 @@ int __bch2_encrypt_bio(struct bch_fs *c, unsigned type,
 {
 	struct bio_vec bv;
 	struct bvec_iter iter;
-	u32 chacha_state[CHACHA_STATE_WORDS];
+	struct chacha_state chacha_state;
 	int ret = 0;
 
 	if (bch2_fs_inconsistent_on(!c->chacha20_key_set,
 				    c, "attempting to encrypt without encryption key"))
 		return -BCH_ERR_no_encryption_key;
 
-	bch2_chacha20_init(chacha_state, &c->chacha20_key, nonce);
+	bch2_chacha20_init(&chacha_state, &c->chacha20_key, nonce);
 
 	bio_for_each_segment(bv, bio, iter) {
 		void *p;
@@ -280,10 +280,10 @@ int __bch2_encrypt_bio(struct bch_fs *c, unsigned type,
 		}
 
 		p = bvec_kmap_local(&bv);
-		chacha20_crypt(chacha_state, p, p, bv.bv_len);
+		chacha20_crypt(&chacha_state, p, p, bv.bv_len);
 		kunmap_local(p);
 	}
-	memzero_explicit(chacha_state, sizeof(chacha_state));
+	chacha_zeroize_state(&chacha_state);
 	return ret;
 }
 
diff --git a/fs/ubifs/compress.c b/fs/ubifs/compress.c
index ea6f06adcd43..059a02691edd 100644
--- a/fs/ubifs/compress.c
+++ b/fs/ubifs/compress.c
@@ -19,6 +19,11 @@
 #include <linux/highmem.h>
 #include "ubifs.h"
 
+union ubifs_in_ptr {
+	const void *buf;
+	struct folio *folio;
+};
+
 /* Fake description object for the "none" compressor */
 static struct ubifs_compressor none_compr = {
 	.compr_type = UBIFS_COMPR_NONE,
@@ -68,28 +73,61 @@ static struct ubifs_compressor zstd_compr = {
 /* All UBIFS compressors */
 struct ubifs_compressor *ubifs_compressors[UBIFS_COMPR_TYPES_CNT];
 
-static int ubifs_compress_req(const struct ubifs_info *c,
-			      struct acomp_req *req,
-			      void *out_buf, int *out_len,
-			      const char *compr_name)
+static void ubifs_compress_common(int *compr_type, union ubifs_in_ptr in_ptr,
+				  size_t in_offset, int in_len, bool in_folio,
+				  void *out_buf, int *out_len)
 {
-	struct crypto_wait wait;
-	int in_len = req->slen;
+	struct ubifs_compressor *compr = ubifs_compressors[*compr_type];
 	int dlen = *out_len;
 	int err;
 
+	if (*compr_type == UBIFS_COMPR_NONE)
+		goto no_compr;
+
+	/* If the input data is small, do not even try to compress it */
+	if (in_len < UBIFS_MIN_COMPR_LEN)
+		goto no_compr;
+
 	dlen = min(dlen, in_len - UBIFS_MIN_COMPRESS_DIFF);
 
-	crypto_init_wait(&wait);
-	acomp_request_set_callback(req, CRYPTO_TFM_REQ_MAY_BACKLOG,
-				   crypto_req_done, &wait);
-	acomp_request_set_dst_dma(req, out_buf, dlen);
-	err = crypto_acomp_compress(req);
-	err = crypto_wait_req(err, &wait);
-	*out_len = req->dlen;
-	acomp_request_free(req);
+	do {
+		ACOMP_REQUEST_ON_STACK(req, compr->cc);
+		DECLARE_CRYPTO_WAIT(wait);
+
+		acomp_request_set_callback(req, 0, NULL, NULL);
+		if (in_folio)
+			acomp_request_set_src_folio(req, in_ptr.folio,
+						    in_offset, in_len);
+		else
+			acomp_request_set_src_dma(req, in_ptr.buf, in_len);
+		acomp_request_set_dst_dma(req, out_buf, dlen);
+		err = crypto_acomp_compress(req);
+		dlen = req->dlen;
+		if (err != -EAGAIN)
+			break;
+
+		req = ACOMP_REQUEST_CLONE(req, GFP_NOFS | __GFP_NOWARN);
+		acomp_request_set_callback(req, CRYPTO_TFM_REQ_MAY_BACKLOG,
+					   crypto_req_done, &wait);
+		err = crypto_acomp_compress(req);
+		err = crypto_wait_req(err, &wait);
+		dlen = req->dlen;
+		acomp_request_free(req);
+	} while (0);
+
+	*out_len = dlen;
+	if (err)
+		goto no_compr;
 
-	return err;
+	return;
+
+no_compr:
+	if (in_folio)
+		memcpy_from_folio(out_buf, in_ptr.folio, in_offset, in_len);
+	else
+		memcpy(out_buf, in_ptr.buf, in_len);
+	*out_len = in_len;
+	*compr_type = UBIFS_COMPR_NONE;
 }
 
 /**
@@ -114,32 +152,10 @@ static int ubifs_compress_req(const struct ubifs_info *c,
 void ubifs_compress(const struct ubifs_info *c, const void *in_buf,
 		    int in_len, void *out_buf, int *out_len, int *compr_type)
 {
-	int err;
-	struct ubifs_compressor *compr = ubifs_compressors[*compr_type];
-
-	if (*compr_type == UBIFS_COMPR_NONE)
-		goto no_compr;
+	union ubifs_in_ptr in_ptr = { .buf = in_buf };
 
-	/* If the input data is small, do not even try to compress it */
-	if (in_len < UBIFS_MIN_COMPR_LEN)
-		goto no_compr;
-
-	{
-		ACOMP_REQUEST_ALLOC(req, compr->cc, GFP_NOFS | __GFP_NOWARN);
-
-		acomp_request_set_src_dma(req, in_buf, in_len);
-		err = ubifs_compress_req(c, req, out_buf, out_len, compr->name);
-	}
-
-	if (err)
-		goto no_compr;
-
-	return;
-
-no_compr:
-	memcpy(out_buf, in_buf, in_len);
-	*out_len = in_len;
-	*compr_type = UBIFS_COMPR_NONE;
+	ubifs_compress_common(compr_type, in_ptr, 0, in_len, false,
+			      out_buf, out_len);
 }
 
 /**
@@ -166,55 +182,71 @@ void ubifs_compress_folio(const struct ubifs_info *c, struct folio *in_folio,
 			  size_t in_offset, int in_len, void *out_buf,
 			  int *out_len, int *compr_type)
 {
-	int err;
-	struct ubifs_compressor *compr = ubifs_compressors[*compr_type];
+	union ubifs_in_ptr in_ptr = { .folio = in_folio };
 
-	if (*compr_type == UBIFS_COMPR_NONE)
-		goto no_compr;
-
-	/* If the input data is small, do not even try to compress it */
-	if (in_len < UBIFS_MIN_COMPR_LEN)
-		goto no_compr;
+	ubifs_compress_common(compr_type, in_ptr, in_offset, in_len, true,
+			      out_buf, out_len);
+}
 
-	{
-		ACOMP_REQUEST_ALLOC(req, compr->cc, GFP_NOFS | __GFP_NOWARN);
+static int ubifs_decompress_common(const struct ubifs_info *c,
+				   const void *in_buf, int in_len,
+				   void *out_ptr, size_t out_offset,
+				   int *out_len, bool out_folio,
+				   int compr_type)
+{
+	struct ubifs_compressor *compr;
+	int dlen = *out_len;
+	int err;
 
-		acomp_request_set_src_folio(req, in_folio, in_offset, in_len);
-		err = ubifs_compress_req(c, req, out_buf, out_len, compr->name);
+	if (unlikely(compr_type < 0 || compr_type >= UBIFS_COMPR_TYPES_CNT)) {
+		ubifs_err(c, "invalid compression type %d", compr_type);
+		return -EINVAL;
 	}
 
-	if (err)
-		goto no_compr;
-
-	return;
+	compr = ubifs_compressors[compr_type];
 
-no_compr:
-	memcpy_from_folio(out_buf, in_folio, in_offset, in_len);
-	*out_len = in_len;
-	*compr_type = UBIFS_COMPR_NONE;
-}
+	if (unlikely(!compr->capi_name)) {
+		ubifs_err(c, "%s compression is not compiled in", compr->name);
+		return -EINVAL;
+	}
 
-static int ubifs_decompress_req(const struct ubifs_info *c,
-				struct acomp_req *req,
-				const void *in_buf, int in_len, int *out_len,
-				const char *compr_name)
-{
-	struct crypto_wait wait;
-	int err;
+	if (compr_type == UBIFS_COMPR_NONE) {
+		if (out_folio)
+			memcpy_to_folio(out_ptr, out_offset, in_buf, in_len);
+		else
+			memcpy(out_ptr, in_buf, in_len);
+		*out_len = in_len;
+		return 0;
+	}
 
-	crypto_init_wait(&wait);
-	acomp_request_set_callback(req, CRYPTO_TFM_REQ_MAY_BACKLOG,
-				   crypto_req_done, &wait);
-	acomp_request_set_src_dma(req, in_buf, in_len);
-	err = crypto_acomp_decompress(req);
-	err = crypto_wait_req(err, &wait);
-	*out_len = req->dlen;
+	do {
+		ACOMP_REQUEST_ON_STACK(req, compr->cc);
+		DECLARE_CRYPTO_WAIT(wait);
 
+		acomp_request_set_callback(req, CRYPTO_TFM_REQ_MAY_BACKLOG,
+					   crypto_req_done, &wait);
+		acomp_request_set_src_dma(req, in_buf, in_len);
+		if (out_folio)
+			acomp_request_set_dst_folio(req, out_ptr, out_offset,
+						    dlen);
+		else
+			acomp_request_set_dst_dma(req, out_ptr, dlen);
+		err = crypto_acomp_decompress(req);
+		dlen = req->dlen;
+		if (err != -EAGAIN)
+			break;
+
+		req = ACOMP_REQUEST_CLONE(req, GFP_NOFS | __GFP_NOWARN);
+		err = crypto_acomp_decompress(req);
+		err = crypto_wait_req(err, &wait);
+		dlen = req->dlen;
+		acomp_request_free(req);
+	} while (0);
+
+	*out_len = dlen;
 	if (err)
 		ubifs_err(c, "cannot decompress %d bytes, compressor %s, error %d",
-			  in_len, compr_name, err);
-
-	acomp_request_free(req);
+			  in_len, compr->name, err);
 
 	return err;
 }
@@ -235,33 +267,8 @@ static int ubifs_decompress_req(const struct ubifs_info *c,
 int ubifs_decompress(const struct ubifs_info *c, const void *in_buf,
 		     int in_len, void *out_buf, int *out_len, int compr_type)
 {
-	struct ubifs_compressor *compr;
-
-	if (unlikely(compr_type < 0 || compr_type >= UBIFS_COMPR_TYPES_CNT)) {
-		ubifs_err(c, "invalid compression type %d", compr_type);
-		return -EINVAL;
-	}
-
-	compr = ubifs_compressors[compr_type];
-
-	if (unlikely(!compr->capi_name)) {
-		ubifs_err(c, "%s compression is not compiled in", compr->name);
-		return -EINVAL;
-	}
-
-	if (compr_type == UBIFS_COMPR_NONE) {
-		memcpy(out_buf, in_buf, in_len);
-		*out_len = in_len;
-		return 0;
-	}
-
-	{
-		ACOMP_REQUEST_ALLOC(req, compr->cc, GFP_NOFS | __GFP_NOWARN);
-
-		acomp_request_set_dst_dma(req, out_buf, *out_len);
-		return ubifs_decompress_req(c, req, in_buf, in_len, out_len,
-					    compr->name);
-	}
+	return ubifs_decompress_common(c, in_buf, in_len, out_buf, 0, out_len,
+				       false, compr_type);
 }
 
 /**
@@ -283,34 +290,8 @@ int ubifs_decompress_folio(const struct ubifs_info *c, const void *in_buf,
 			   int in_len, struct folio *out_folio,
 			   size_t out_offset, int *out_len, int compr_type)
 {
-	struct ubifs_compressor *compr;
-
-	if (unlikely(compr_type < 0 || compr_type >= UBIFS_COMPR_TYPES_CNT)) {
-		ubifs_err(c, "invalid compression type %d", compr_type);
-		return -EINVAL;
-	}
-
-	compr = ubifs_compressors[compr_type];
-
-	if (unlikely(!compr->capi_name)) {
-		ubifs_err(c, "%s compression is not compiled in", compr->name);
-		return -EINVAL;
-	}
-
-	if (compr_type == UBIFS_COMPR_NONE) {
-		memcpy_to_folio(out_folio, out_offset, in_buf, in_len);
-		*out_len = in_len;
-		return 0;
-	}
-
-	{
-		ACOMP_REQUEST_ALLOC(req, compr->cc, GFP_NOFS | __GFP_NOWARN);
-
-		acomp_request_set_dst_folio(req, out_folio, out_offset,
-					    *out_len);
-		return ubifs_decompress_req(c, req, in_buf, in_len, out_len,
-					    compr->name);
-	}
+	return ubifs_decompress_common(c, in_buf, in_len, out_folio,
+				       out_offset, out_len, true, compr_type);
 }
 
 /**