From: Guenter Roeck <linux@roeck-us.net>
To: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: linux-crypto@vger.kernel.org, herbert@gondor.apana.org.au,
ebiggers@kernel.org
Subject: Re: [PATCH 15/17] crypto: arm/aes-ce - implement ciphertext stealing for CBC
Date: Mon, 30 Sep 2019 09:32:41 -0700 [thread overview]
Message-ID: <20190930163241.GA14355@roeck-us.net> (raw)
In-Reply-To: <20190821143253.30209-16-ard.biesheuvel@linaro.org>
On Wed, Aug 21, 2019 at 05:32:51PM +0300, Ard Biesheuvel wrote:
> Instead of relying on the CTS template to wrap the accelerated CBC
> skcipher, implement the ciphertext stealing part directly.
>
> Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
For arm:allmodconfig built with gcc 9.2.0, this patch results in
arch/arm/crypto/aes-ce-core.S: Assembler messages:
arch/arm/crypto/aes-ce-core.S:299: Error: selected processor does not support `movw ip,:lower16:.Lcts_permute_table' in ARM mode
arch/arm/crypto/aes-ce-core.S:300: Error: selected processor does not support `movt ip,:upper16:.Lcts_permute_table' in ARM mode
arch/arm/crypto/aes-ce-core.S:337: Error: selected processor does not support `movw ip,:lower16:.Lcts_permute_table' in ARM mode
arch/arm/crypto/aes-ce-core.S:338: Error: selected processor does not support `movt ip,:upper16:.Lcts_permute_table' in ARM mode
arch/arm/crypto/aes-ce-core.S:552: Error: selected processor does not support `movw ip,:lower16:.Lcts_permute_table' in ARM mode
arch/arm/crypto/aes-ce-core.S:553: Error: selected processor does not support `movt ip,:upper16:.Lcts_permute_table' in ARM mode
arch/arm/crypto/aes-ce-core.S:638: Error: selected processor does not support `movw ip,:lower16:.Lcts_permute_table' in ARM mode
arch/arm/crypto/aes-ce-core.S:639: Error: selected processor does not support `movt ip,:upper16:.Lcts_permute_table' in ARM mode
Any idea how to avoid that ?
Guenter
> ---
> arch/arm/crypto/aes-ce-core.S | 85 +++++++++
> arch/arm/crypto/aes-ce-glue.c | 188 ++++++++++++++++++--
> 2 files changed, 256 insertions(+), 17 deletions(-)
>
> diff --git a/arch/arm/crypto/aes-ce-core.S b/arch/arm/crypto/aes-ce-core.S
> index 763e51604ab6..b978cdf133af 100644
> --- a/arch/arm/crypto/aes-ce-core.S
> +++ b/arch/arm/crypto/aes-ce-core.S
> @@ -284,6 +284,91 @@ ENTRY(ce_aes_cbc_decrypt)
> pop {r4-r6, pc}
> ENDPROC(ce_aes_cbc_decrypt)
>
> +
> + /*
> + * ce_aes_cbc_cts_encrypt(u8 out[], u8 const in[], u32 const rk[],
> + * int rounds, int bytes, u8 const iv[])
> + * ce_aes_cbc_cts_decrypt(u8 out[], u8 const in[], u32 const rk[],
> + * int rounds, int bytes, u8 const iv[])
> + */
> +
> +ENTRY(ce_aes_cbc_cts_encrypt)
> + push {r4-r6, lr}
> + ldrd r4, r5, [sp, #16]
> +
> + movw ip, :lower16:.Lcts_permute_table
> + movt ip, :upper16:.Lcts_permute_table
> + sub r4, r4, #16
> + add lr, ip, #32
> + add ip, ip, r4
> + sub lr, lr, r4
> + vld1.8 {q5}, [ip]
> + vld1.8 {q6}, [lr]
> +
> + add ip, r1, r4
> + vld1.8 {q0}, [r1] @ overlapping loads
> + vld1.8 {q3}, [ip]
> +
> + vld1.8 {q1}, [r5] @ get iv
> + prepare_key r2, r3
> +
> + veor q0, q0, q1 @ xor with iv
> + bl aes_encrypt
> +
> + vtbl.8 d4, {d0-d1}, d10
> + vtbl.8 d5, {d0-d1}, d11
> + vtbl.8 d2, {d6-d7}, d12
> + vtbl.8 d3, {d6-d7}, d13
> +
> + veor q0, q0, q1
> + bl aes_encrypt
> +
> + add r4, r0, r4
> + vst1.8 {q2}, [r4] @ overlapping stores
> + vst1.8 {q0}, [r0]
> +
> + pop {r4-r6, pc}
> +ENDPROC(ce_aes_cbc_cts_encrypt)
> +
> +ENTRY(ce_aes_cbc_cts_decrypt)
> + push {r4-r6, lr}
> + ldrd r4, r5, [sp, #16]
> +
> + movw ip, :lower16:.Lcts_permute_table
> + movt ip, :upper16:.Lcts_permute_table
> + sub r4, r4, #16
> + add lr, ip, #32
> + add ip, ip, r4
> + sub lr, lr, r4
> + vld1.8 {q5}, [ip]
> + vld1.8 {q6}, [lr]
> +
> + add ip, r1, r4
> + vld1.8 {q0}, [r1] @ overlapping loads
> + vld1.8 {q1}, [ip]
> +
> + vld1.8 {q3}, [r5] @ get iv
> + prepare_key r2, r3
> +
> + bl aes_decrypt
> +
> + vtbl.8 d4, {d0-d1}, d10
> + vtbl.8 d5, {d0-d1}, d11
> + vtbx.8 d0, {d2-d3}, d12
> + vtbx.8 d1, {d2-d3}, d13
> +
> + veor q1, q1, q2
> + bl aes_decrypt
> + veor q0, q0, q3 @ xor with iv
> +
> + add r4, r0, r4
> + vst1.8 {q1}, [r4] @ overlapping stores
> + vst1.8 {q0}, [r0]
> +
> + pop {r4-r6, pc}
> +ENDPROC(ce_aes_cbc_cts_decrypt)
> +
> +
> /*
> * aes_ctr_encrypt(u8 out[], u8 const in[], u32 const rk[], int rounds,
> * int blocks, u8 ctr[])
> diff --git a/arch/arm/crypto/aes-ce-glue.c b/arch/arm/crypto/aes-ce-glue.c
> index c215792a2494..cdb1a07e7ad0 100644
> --- a/arch/arm/crypto/aes-ce-glue.c
> +++ b/arch/arm/crypto/aes-ce-glue.c
> @@ -35,6 +35,10 @@ asmlinkage void ce_aes_cbc_encrypt(u8 out[], u8 const in[], u32 const rk[],
> int rounds, int blocks, u8 iv[]);
> asmlinkage void ce_aes_cbc_decrypt(u8 out[], u8 const in[], u32 const rk[],
> int rounds, int blocks, u8 iv[]);
> +asmlinkage void ce_aes_cbc_cts_encrypt(u8 out[], u8 const in[], u32 const rk[],
> + int rounds, int bytes, u8 const iv[]);
> +asmlinkage void ce_aes_cbc_cts_decrypt(u8 out[], u8 const in[], u32 const rk[],
> + int rounds, int bytes, u8 const iv[]);
>
> asmlinkage void ce_aes_ctr_encrypt(u8 out[], u8 const in[], u32 const rk[],
> int rounds, int blocks, u8 ctr[]);
> @@ -210,48 +214,182 @@ static int ecb_decrypt(struct skcipher_request *req)
> return err;
> }
>
> -static int cbc_encrypt(struct skcipher_request *req)
> +static int cbc_encrypt_walk(struct skcipher_request *req,
> + struct skcipher_walk *walk)
> {
> struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req);
> struct crypto_aes_ctx *ctx = crypto_skcipher_ctx(tfm);
> - struct skcipher_walk walk;
> unsigned int blocks;
> - int err;
> + int err = 0;
>
> - err = skcipher_walk_virt(&walk, req, false);
> -
> - while ((blocks = (walk.nbytes / AES_BLOCK_SIZE))) {
> + while ((blocks = (walk->nbytes / AES_BLOCK_SIZE))) {
> kernel_neon_begin();
> - ce_aes_cbc_encrypt(walk.dst.virt.addr, walk.src.virt.addr,
> + ce_aes_cbc_encrypt(walk->dst.virt.addr, walk->src.virt.addr,
> ctx->key_enc, num_rounds(ctx), blocks,
> - walk.iv);
> + walk->iv);
> kernel_neon_end();
> - err = skcipher_walk_done(&walk, walk.nbytes % AES_BLOCK_SIZE);
> + err = skcipher_walk_done(walk, walk->nbytes % AES_BLOCK_SIZE);
> }
> return err;
> }
>
> -static int cbc_decrypt(struct skcipher_request *req)
> +static int cbc_encrypt(struct skcipher_request *req)
> {
> - struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req);
> - struct crypto_aes_ctx *ctx = crypto_skcipher_ctx(tfm);
> struct skcipher_walk walk;
> - unsigned int blocks;
> int err;
>
> err = skcipher_walk_virt(&walk, req, false);
> + if (err)
> + return err;
> + return cbc_encrypt_walk(req, &walk);
> +}
>
> - while ((blocks = (walk.nbytes / AES_BLOCK_SIZE))) {
> +static int cbc_decrypt_walk(struct skcipher_request *req,
> + struct skcipher_walk *walk)
> +{
> + struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req);
> + struct crypto_aes_ctx *ctx = crypto_skcipher_ctx(tfm);
> + unsigned int blocks;
> + int err = 0;
> +
> + while ((blocks = (walk->nbytes / AES_BLOCK_SIZE))) {
> kernel_neon_begin();
> - ce_aes_cbc_decrypt(walk.dst.virt.addr, walk.src.virt.addr,
> + ce_aes_cbc_decrypt(walk->dst.virt.addr, walk->src.virt.addr,
> ctx->key_dec, num_rounds(ctx), blocks,
> - walk.iv);
> + walk->iv);
> kernel_neon_end();
> - err = skcipher_walk_done(&walk, walk.nbytes % AES_BLOCK_SIZE);
> + err = skcipher_walk_done(walk, walk->nbytes % AES_BLOCK_SIZE);
> }
> return err;
> }
>
> +static int cbc_decrypt(struct skcipher_request *req)
> +{
> + struct skcipher_walk walk;
> + int err;
> +
> + err = skcipher_walk_virt(&walk, req, false);
> + if (err)
> + return err;
> + return cbc_decrypt_walk(req, &walk);
> +}
> +
> +static int cts_cbc_encrypt(struct skcipher_request *req)
> +{
> + struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req);
> + struct crypto_aes_ctx *ctx = crypto_skcipher_ctx(tfm);
> + int cbc_blocks = DIV_ROUND_UP(req->cryptlen, AES_BLOCK_SIZE) - 2;
> + struct scatterlist *src = req->src, *dst = req->dst;
> + struct scatterlist sg_src[2], sg_dst[2];
> + struct skcipher_request subreq;
> + struct skcipher_walk walk;
> + int err;
> +
> + skcipher_request_set_tfm(&subreq, tfm);
> + skcipher_request_set_callback(&subreq, skcipher_request_flags(req),
> + NULL, NULL);
> +
> + if (req->cryptlen <= AES_BLOCK_SIZE) {
> + if (req->cryptlen < AES_BLOCK_SIZE)
> + return -EINVAL;
> + cbc_blocks = 1;
> + }
> +
> + if (cbc_blocks > 0) {
> + skcipher_request_set_crypt(&subreq, req->src, req->dst,
> + cbc_blocks * AES_BLOCK_SIZE,
> + req->iv);
> +
> + err = skcipher_walk_virt(&walk, &subreq, false) ?:
> + cbc_encrypt_walk(&subreq, &walk);
> + if (err)
> + return err;
> +
> + if (req->cryptlen == AES_BLOCK_SIZE)
> + return 0;
> +
> + dst = src = scatterwalk_ffwd(sg_src, req->src, subreq.cryptlen);
> + if (req->dst != req->src)
> + dst = scatterwalk_ffwd(sg_dst, req->dst,
> + subreq.cryptlen);
> + }
> +
> + /* handle ciphertext stealing */
> + skcipher_request_set_crypt(&subreq, src, dst,
> + req->cryptlen - cbc_blocks * AES_BLOCK_SIZE,
> + req->iv);
> +
> + err = skcipher_walk_virt(&walk, &subreq, false);
> + if (err)
> + return err;
> +
> + kernel_neon_begin();
> + ce_aes_cbc_cts_encrypt(walk.dst.virt.addr, walk.src.virt.addr,
> + ctx->key_enc, num_rounds(ctx), walk.nbytes,
> + walk.iv);
> + kernel_neon_end();
> +
> + return skcipher_walk_done(&walk, 0);
> +}
> +
> +static int cts_cbc_decrypt(struct skcipher_request *req)
> +{
> + struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req);
> + struct crypto_aes_ctx *ctx = crypto_skcipher_ctx(tfm);
> + int cbc_blocks = DIV_ROUND_UP(req->cryptlen, AES_BLOCK_SIZE) - 2;
> + struct scatterlist *src = req->src, *dst = req->dst;
> + struct scatterlist sg_src[2], sg_dst[2];
> + struct skcipher_request subreq;
> + struct skcipher_walk walk;
> + int err;
> +
> + skcipher_request_set_tfm(&subreq, tfm);
> + skcipher_request_set_callback(&subreq, skcipher_request_flags(req),
> + NULL, NULL);
> +
> + if (req->cryptlen <= AES_BLOCK_SIZE) {
> + if (req->cryptlen < AES_BLOCK_SIZE)
> + return -EINVAL;
> + cbc_blocks = 1;
> + }
> +
> + if (cbc_blocks > 0) {
> + skcipher_request_set_crypt(&subreq, req->src, req->dst,
> + cbc_blocks * AES_BLOCK_SIZE,
> + req->iv);
> +
> + err = skcipher_walk_virt(&walk, &subreq, false) ?:
> + cbc_decrypt_walk(&subreq, &walk);
> + if (err)
> + return err;
> +
> + if (req->cryptlen == AES_BLOCK_SIZE)
> + return 0;
> +
> + dst = src = scatterwalk_ffwd(sg_src, req->src, subreq.cryptlen);
> + if (req->dst != req->src)
> + dst = scatterwalk_ffwd(sg_dst, req->dst,
> + subreq.cryptlen);
> + }
> +
> + /* handle ciphertext stealing */
> + skcipher_request_set_crypt(&subreq, src, dst,
> + req->cryptlen - cbc_blocks * AES_BLOCK_SIZE,
> + req->iv);
> +
> + err = skcipher_walk_virt(&walk, &subreq, false);
> + if (err)
> + return err;
> +
> + kernel_neon_begin();
> + ce_aes_cbc_cts_decrypt(walk.dst.virt.addr, walk.src.virt.addr,
> + ctx->key_dec, num_rounds(ctx), walk.nbytes,
> + walk.iv);
> + kernel_neon_end();
> +
> + return skcipher_walk_done(&walk, 0);
> +}
> +
> static int ctr_encrypt(struct skcipher_request *req)
> {
> struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req);
> @@ -486,6 +624,22 @@ static struct skcipher_alg aes_algs[] = { {
> .setkey = ce_aes_setkey,
> .encrypt = cbc_encrypt,
> .decrypt = cbc_decrypt,
> +}, {
> + .base.cra_name = "__cts(cbc(aes))",
> + .base.cra_driver_name = "__cts-cbc-aes-ce",
> + .base.cra_priority = 300,
> + .base.cra_flags = CRYPTO_ALG_INTERNAL,
> + .base.cra_blocksize = AES_BLOCK_SIZE,
> + .base.cra_ctxsize = sizeof(struct crypto_aes_ctx),
> + .base.cra_module = THIS_MODULE,
> +
> + .min_keysize = AES_MIN_KEY_SIZE,
> + .max_keysize = AES_MAX_KEY_SIZE,
> + .ivsize = AES_BLOCK_SIZE,
> + .walksize = 2 * AES_BLOCK_SIZE,
> + .setkey = ce_aes_setkey,
> + .encrypt = cts_cbc_encrypt,
> + .decrypt = cts_cbc_decrypt,
> }, {
> .base.cra_name = "__ctr(aes)",
> .base.cra_driver_name = "__ctr-aes-ce",
next prev parent reply other threads:[~2019-09-30 16:32 UTC|newest]
Thread overview: 37+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-08-21 14:32 [PATCH 00/17] crypto: arm/aes - XTS ciphertext stealing and other updates Ard Biesheuvel
2019-08-21 14:32 ` [PATCH 01/17] crypto: arm/aes - fix round key prototypes Ard Biesheuvel
2019-08-21 14:32 ` [PATCH 02/17] crypto: arm/aes-ce - yield the SIMD unit between scatterwalk steps Ard Biesheuvel
2019-08-21 14:32 ` [PATCH 03/17] crypto: arm/aes-ce - switch to 4x interleave Ard Biesheuvel
2019-08-21 14:32 ` [PATCH 04/17] crypto: arm/aes-ce - replace tweak mask literal with composition Ard Biesheuvel
2019-08-21 14:32 ` [PATCH 05/17] crypto: arm/aes-neonbs " Ard Biesheuvel
2019-08-21 14:32 ` [PATCH 06/17] crypto: arm64/aes-neonbs " Ard Biesheuvel
2019-08-21 14:32 ` [PATCH 07/17] crypto: arm64/aes-neon - limit exposed routines if faster driver is enabled Ard Biesheuvel
2019-08-21 14:32 ` [PATCH 08/17] crypto: skcipher - add the ability to abort a skcipher walk Ard Biesheuvel
2019-08-30 8:03 ` Herbert Xu
2019-08-31 18:01 ` Ard Biesheuvel
2019-09-03 6:54 ` crypto: skcipher - Unmap pages after an external error Herbert Xu
2019-09-03 7:05 ` crypto: ablkcipher " Herbert Xu
2019-09-03 7:09 ` crypto: blkcipher " Herbert Xu
2019-09-03 13:50 ` crypto: skcipher " Eric Biggers
2019-09-03 22:36 ` Herbert Xu
2019-09-05 5:22 ` Eric Biggers
2019-09-05 5:40 ` Herbert Xu
2019-09-06 1:57 ` Eric Biggers
2019-09-06 2:15 ` Herbert Xu
2019-09-06 3:13 ` [v2 PATCH] " Herbert Xu
2019-09-07 0:52 ` Ard Biesheuvel
2019-09-07 1:19 ` Herbert Xu
2019-09-07 1:32 ` Ard Biesheuvel
2019-09-07 1:56 ` Herbert Xu
2019-09-07 2:14 ` Ard Biesheuvel
2019-08-21 14:32 ` [PATCH 09/17] crypto: arm64/aes-cts-cbc-ce - performance tweak Ard Biesheuvel
2019-08-21 14:32 ` [PATCH 10/17] crypto: arm64/aes-cts-cbc - move request context data to the stack Ard Biesheuvel
2019-08-21 14:32 ` [PATCH 11/17] crypto: arm64/aes - implement support for XTS ciphertext stealing Ard Biesheuvel
2019-08-21 14:32 ` [PATCH 12/17] crypto: arm64/aes-neonbs - implement ciphertext stealing for XTS Ard Biesheuvel
2019-08-21 14:32 ` [PATCH 13/17] crypto: arm/aes-ce " Ard Biesheuvel
2019-08-21 14:32 ` [PATCH 14/17] crypto: arm/aes-neonbs " Ard Biesheuvel
2019-08-21 14:32 ` [PATCH 15/17] crypto: arm/aes-ce - implement ciphertext stealing for CBC Ard Biesheuvel
2019-09-30 16:32 ` Guenter Roeck [this message]
2019-09-30 18:19 ` Ard Biesheuvel
2019-08-21 14:32 ` [PATCH 16/17] crypto: testmgr - add test vectors for XTS ciphertext stealing Ard Biesheuvel
2019-08-21 14:32 ` [PATCH 17/17] crypto: testmgr - Add additional AES-XTS vectors for covering CTS Ard Biesheuvel
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20190930163241.GA14355@roeck-us.net \
--to=linux@roeck-us.net \
--cc=ard.biesheuvel@linaro.org \
--cc=ebiggers@kernel.org \
--cc=herbert@gondor.apana.org.au \
--cc=linux-crypto@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).