diff options
Diffstat (limited to 'drivers/crypto')
31 files changed, 1935 insertions, 353 deletions
diff --git a/drivers/crypto/Kconfig b/drivers/crypto/Kconfig index 7d74d092aa8..308c7fb92a6 100644 --- a/drivers/crypto/Kconfig +++ b/drivers/crypto/Kconfig @@ -298,21 +298,15 @@ config CRYPTO_DEV_TEGRA_AES will be called tegra-aes. config CRYPTO_DEV_NX - tristate "Support for Power7+ in-Nest cryptographic acceleration" + bool "Support for IBM Power7+ in-Nest cryptographic acceleration" depends on PPC64 && IBMVIO - select CRYPTO_AES - select CRYPTO_CBC - select CRYPTO_ECB - select CRYPTO_CCM - select CRYPTO_GCM - select CRYPTO_AUTHENC - select CRYPTO_XCBC - select CRYPTO_SHA256 - select CRYPTO_SHA512 + default n help - Support for Power7+ in-Nest cryptographic acceleration. This - module supports acceleration for AES and SHA2 algorithms. If you - choose 'M' here, this module will be called nx_crypto. + Support for Power7+ in-Nest cryptographic acceleration. + +if CRYPTO_DEV_NX + source "drivers/crypto/nx/Kconfig" +endif config CRYPTO_DEV_UX500 tristate "Driver for ST-Ericsson UX500 crypto hardware acceleration" @@ -340,7 +334,7 @@ config CRYPTO_DEV_ATMEL_AES select CRYPTO_AES select CRYPTO_ALGAPI select CRYPTO_BLKCIPHER - select CONFIG_AT_HDMAC + select AT_HDMAC help Some Atmel processors have AES hw accelerator. Select this if you want to use the Atmel module for diff --git a/drivers/crypto/amcc/crypto4xx_core.c b/drivers/crypto/amcc/crypto4xx_core.c index 802e85102c3..f88e3d8f6b6 100644 --- a/drivers/crypto/amcc/crypto4xx_core.c +++ b/drivers/crypto/amcc/crypto4xx_core.c @@ -1226,6 +1226,7 @@ static int __init crypto4xx_probe(struct platform_device *ofdev) core_dev->dev->ce_base = of_iomap(ofdev->dev.of_node, 0); if (!core_dev->dev->ce_base) { dev_err(dev, "failed to of_iomap\n"); + rc = -ENOMEM; goto err_iomap; } diff --git a/drivers/crypto/atmel-aes.c b/drivers/crypto/atmel-aes.c index 6bb20fffbf4..8061336e07e 100644 --- a/drivers/crypto/atmel-aes.c +++ b/drivers/crypto/atmel-aes.c @@ -24,15 +24,10 @@ #include <linux/platform_device.h> #include <linux/device.h> -#include <linux/module.h> #include <linux/init.h> #include <linux/errno.h> #include <linux/interrupt.h> -#include <linux/kernel.h> -#include <linux/clk.h> #include <linux/irq.h> -#include <linux/io.h> -#include <linux/platform_device.h> #include <linux/scatterlist.h> #include <linux/dma-mapping.h> #include <linux/delay.h> @@ -1017,7 +1012,6 @@ static int atmel_aes_register_algs(struct atmel_aes_dev *dd) int err, i, j; for (i = 0; i < ARRAY_SIZE(aes_algs); i++) { - INIT_LIST_HEAD(&aes_algs[i].cra_list); err = crypto_register_alg(&aes_algs[i]); if (err) goto err_aes_algs; @@ -1026,7 +1020,6 @@ static int atmel_aes_register_algs(struct atmel_aes_dev *dd) atmel_aes_hw_version_init(dd); if (dd->hw_version >= 0x130) { - INIT_LIST_HEAD(&aes_cfb64_alg[0].cra_list); err = crypto_register_alg(&aes_cfb64_alg[0]); if (err) goto err_aes_cfb64_alg; diff --git a/drivers/crypto/atmel-sha.c b/drivers/crypto/atmel-sha.c index f938b9d79b6..bcdf55fdc62 100644 --- a/drivers/crypto/atmel-sha.c +++ b/drivers/crypto/atmel-sha.c @@ -24,15 +24,10 @@ #include <linux/platform_device.h> #include <linux/device.h> -#include <linux/module.h> #include <linux/init.h> #include <linux/errno.h> #include <linux/interrupt.h> -#include <linux/kernel.h> -#include <linux/clk.h> #include <linux/irq.h> -#include <linux/io.h> -#include <linux/platform_device.h> #include <linux/scatterlist.h> #include <linux/dma-mapping.h> #include <linux/delay.h> diff --git a/drivers/crypto/atmel-tdes.c b/drivers/crypto/atmel-tdes.c index eb2b61e57e2..7495f98c722 100644 --- a/drivers/crypto/atmel-tdes.c +++ b/drivers/crypto/atmel-tdes.c @@ -24,15 +24,10 @@ #include <linux/platform_device.h> #include <linux/device.h> -#include <linux/module.h> #include <linux/init.h> #include <linux/errno.h> #include <linux/interrupt.h> -#include <linux/kernel.h> -#include <linux/clk.h> #include <linux/irq.h> -#include <linux/io.h> -#include <linux/platform_device.h> #include <linux/scatterlist.h> #include <linux/dma-mapping.h> #include <linux/delay.h> @@ -1044,7 +1039,6 @@ static int atmel_tdes_register_algs(struct atmel_tdes_dev *dd) int err, i, j; for (i = 0; i < ARRAY_SIZE(tdes_algs); i++) { - INIT_LIST_HEAD(&tdes_algs[i].cra_list); err = crypto_register_alg(&tdes_algs[i]); if (err) goto err_tdes_algs; diff --git a/drivers/crypto/caam/caamalg.c b/drivers/crypto/caam/caamalg.c index 0c1ea8492ef..b2a0a0726a5 100644 --- a/drivers/crypto/caam/caamalg.c +++ b/drivers/crypto/caam/caamalg.c @@ -205,7 +205,7 @@ static void init_sh_desc_key_aead(u32 *desc, struct caam_ctx *ctx, { u32 *key_jump_cmd; - init_sh_desc(desc, HDR_SHARE_WAIT); + init_sh_desc(desc, HDR_SHARE_SERIAL); /* Skip if already shared */ key_jump_cmd = append_jump(desc, JUMP_JSL | JUMP_TEST_ALL | @@ -224,7 +224,7 @@ static int aead_set_sh_desc(struct crypto_aead *aead) struct aead_tfm *tfm = &aead->base.crt_aead; struct caam_ctx *ctx = crypto_aead_ctx(aead); struct device *jrdev = ctx->jrdev; - bool keys_fit_inline = 0; + bool keys_fit_inline = false; u32 *key_jump_cmd, *jump_cmd; u32 geniv, moveiv; u32 *desc; @@ -239,7 +239,7 @@ static int aead_set_sh_desc(struct crypto_aead *aead) if (DESC_AEAD_ENC_LEN + DESC_JOB_IO_LEN + ctx->split_key_pad_len + ctx->enckeylen <= CAAM_DESC_BYTES_MAX) - keys_fit_inline = 1; + keys_fit_inline = true; /* aead_encrypt shared descriptor */ desc = ctx->sh_desc_enc; @@ -297,12 +297,12 @@ static int aead_set_sh_desc(struct crypto_aead *aead) if (DESC_AEAD_DEC_LEN + DESC_JOB_IO_LEN + ctx->split_key_pad_len + ctx->enckeylen <= CAAM_DESC_BYTES_MAX) - keys_fit_inline = 1; + keys_fit_inline = true; desc = ctx->sh_desc_dec; /* aead_decrypt shared descriptor */ - init_sh_desc(desc, HDR_SHARE_WAIT); + init_sh_desc(desc, HDR_SHARE_SERIAL); /* Skip if already shared */ key_jump_cmd = append_jump(desc, JUMP_JSL | JUMP_TEST_ALL | @@ -365,7 +365,7 @@ static int aead_set_sh_desc(struct crypto_aead *aead) if (DESC_AEAD_GIVENC_LEN + DESC_JOB_IO_LEN + ctx->split_key_pad_len + ctx->enckeylen <= CAAM_DESC_BYTES_MAX) - keys_fit_inline = 1; + keys_fit_inline = true; /* aead_givencrypt shared descriptor */ desc = ctx->sh_desc_givenc; @@ -564,7 +564,7 @@ static int ablkcipher_setkey(struct crypto_ablkcipher *ablkcipher, /* ablkcipher_encrypt shared descriptor */ desc = ctx->sh_desc_enc; - init_sh_desc(desc, HDR_SHARE_WAIT); + init_sh_desc(desc, HDR_SHARE_SERIAL); /* Skip if already shared */ key_jump_cmd = append_jump(desc, JUMP_JSL | JUMP_TEST_ALL | JUMP_COND_SHRD); @@ -605,7 +605,7 @@ static int ablkcipher_setkey(struct crypto_ablkcipher *ablkcipher, /* ablkcipher_decrypt shared descriptor */ desc = ctx->sh_desc_dec; - init_sh_desc(desc, HDR_SHARE_WAIT); + init_sh_desc(desc, HDR_SHARE_SERIAL); /* Skip if already shared */ key_jump_cmd = append_jump(desc, JUMP_JSL | JUMP_TEST_ALL | JUMP_COND_SHRD); @@ -1354,10 +1354,10 @@ static struct aead_edesc *aead_giv_edesc_alloc(struct aead_givcrypt_request contig &= ~GIV_SRC_CONTIG; if (dst_nents || iv_dma + ivsize != sg_dma_address(req->dst)) contig &= ~GIV_DST_CONTIG; - if (unlikely(req->src != req->dst)) { - dst_nents = dst_nents ? : 1; - sec4_sg_len += 1; - } + if (unlikely(req->src != req->dst)) { + dst_nents = dst_nents ? : 1; + sec4_sg_len += 1; + } if (!(contig & GIV_SRC_CONTIG)) { assoc_nents = assoc_nents ? : 1; src_nents = src_nents ? : 1; @@ -1650,7 +1650,11 @@ struct caam_alg_template { }; static struct caam_alg_template driver_algs[] = { - /* single-pass ipsec_esp descriptor */ + /* + * single-pass ipsec_esp descriptor + * authencesn(*,*) is also registered, although not present + * explicitly here. + */ { .name = "authenc(hmac(md5),cbc(aes))", .driver_name = "authenc-hmac-md5-cbc-aes-caam", @@ -2213,7 +2217,9 @@ static int __init caam_algapi_init(void) for (i = 0; i < ARRAY_SIZE(driver_algs); i++) { /* TODO: check if h/w supports alg */ struct caam_crypto_alg *t_alg; + bool done = false; +authencesn: t_alg = caam_alg_alloc(ctrldev, &driver_algs[i]); if (IS_ERR(t_alg)) { err = PTR_ERR(t_alg); @@ -2227,8 +2233,25 @@ static int __init caam_algapi_init(void) dev_warn(ctrldev, "%s alg registration failed\n", t_alg->crypto_alg.cra_driver_name); kfree(t_alg); - } else + } else { list_add_tail(&t_alg->entry, &priv->alg_list); + if (driver_algs[i].type == CRYPTO_ALG_TYPE_AEAD && + !memcmp(driver_algs[i].name, "authenc", 7) && + !done) { + char *name; + + name = driver_algs[i].name; + memmove(name + 10, name + 7, strlen(name) - 7); + memcpy(name + 7, "esn", 3); + + name = driver_algs[i].driver_name; + memmove(name + 10, name + 7, strlen(name) - 7); + memcpy(name + 7, "esn", 3); + + done = true; + goto authencesn; + } + } } if (!list_empty(&priv->alg_list)) dev_info(ctrldev, "%s algorithms registered in /proc/crypto\n", diff --git a/drivers/crypto/caam/caamhash.c b/drivers/crypto/caam/caamhash.c index 895aaf2bca9..32aba7a6150 100644 --- a/drivers/crypto/caam/caamhash.c +++ b/drivers/crypto/caam/caamhash.c @@ -225,7 +225,7 @@ static inline void init_sh_desc_key_ahash(u32 *desc, struct caam_hash_ctx *ctx) { u32 *key_jump_cmd; - init_sh_desc(desc, HDR_SHARE_WAIT); + init_sh_desc(desc, HDR_SHARE_SERIAL); if (ctx->split_key_len) { /* Skip if already shared */ @@ -311,7 +311,7 @@ static int ahash_set_sh_desc(struct crypto_ahash *ahash) /* ahash_update shared descriptor */ desc = ctx->sh_desc_update; - init_sh_desc(desc, HDR_SHARE_WAIT); + init_sh_desc(desc, HDR_SHARE_SERIAL); /* Import context from software */ append_cmd(desc, CMD_SEQ_LOAD | LDST_SRCDST_BYTE_CONTEXT | @@ -430,6 +430,10 @@ static u32 hash_digest_key(struct caam_hash_ctx *ctx, const u8 *key_in, int ret = 0; desc = kmalloc(CAAM_CMD_SZ * 6 + CAAM_PTR_SZ * 2, GFP_KERNEL | GFP_DMA); + if (!desc) { + dev_err(jrdev, "unable to allocate key input memory\n"); + return -ENOMEM; + } init_job_desc(desc, 0); @@ -1736,8 +1740,11 @@ static void __exit caam_algapi_hash_exit(void) struct caam_hash_alg *t_alg, *n; dev_node = of_find_compatible_node(NULL, NULL, "fsl,sec-v4.0"); - if (!dev_node) - return; + if (!dev_node) { + dev_node = of_find_compatible_node(NULL, NULL, "fsl,sec4.0"); + if (!dev_node) + return; + } pdev = of_find_device_by_node(dev_node); if (!pdev) @@ -1812,8 +1819,11 @@ static int __init caam_algapi_hash_init(void) int i = 0, err = 0; dev_node = of_find_compatible_node(NULL, NULL, "fsl,sec-v4.0"); - if (!dev_node) - return -ENODEV; + if (!dev_node) { + dev_node = of_find_compatible_node(NULL, NULL, "fsl,sec4.0"); + if (!dev_node) + return -ENODEV; + } pdev = of_find_device_by_node(dev_node); if (!pdev) diff --git a/drivers/crypto/caam/caamrng.c b/drivers/crypto/caam/caamrng.c index e2bfe161dec..d1939a9539c 100644 --- a/drivers/crypto/caam/caamrng.c +++ b/drivers/crypto/caam/caamrng.c @@ -193,7 +193,7 @@ static inline void rng_create_sh_desc(struct caam_rng_ctx *ctx) struct device *jrdev = ctx->jrdev; u32 *desc = ctx->sh_desc; - init_sh_desc(desc, HDR_SHARE_WAIT); + init_sh_desc(desc, HDR_SHARE_SERIAL); /* Propagate errors from shared to job descriptor */ append_cmd(desc, SET_OK_NO_PROP_ERRORS | CMD_LOAD); @@ -284,8 +284,11 @@ static int __init caam_rng_init(void) struct caam_drv_private *priv; dev_node = of_find_compatible_node(NULL, NULL, "fsl,sec-v4.0"); - if (!dev_node) - return -ENODEV; + if (!dev_node) { + dev_node = of_find_compatible_node(NULL, NULL, "fsl,sec4.0"); + if (!dev_node) + return -ENODEV; + } pdev = of_find_device_by_node(dev_node); if (!pdev) diff --git a/drivers/crypto/caam/compat.h b/drivers/crypto/caam/compat.h index 762aeff626a..cf15e781380 100644 --- a/drivers/crypto/caam/compat.h +++ b/drivers/crypto/caam/compat.h @@ -23,6 +23,7 @@ #include <linux/types.h> #include <linux/debugfs.h> #include <linux/circ_buf.h> +#include <linux/string.h> #include <net/xfrm.h> #include <crypto/algapi.h> diff --git a/drivers/crypto/caam/ctrl.c b/drivers/crypto/caam/ctrl.c index 414ba20c05a..bf20dd89170 100644 --- a/drivers/crypto/caam/ctrl.c +++ b/drivers/crypto/caam/ctrl.c @@ -129,7 +129,7 @@ static int instantiate_rng(struct device *jrdev) /* * By default, the TRNG runs for 200 clocks per sample; - * 800 clocks per sample generates better entropy. + * 1600 clocks per sample generates better entropy. */ static void kick_trng(struct platform_device *pdev) { @@ -144,9 +144,9 @@ static void kick_trng(struct platform_device *pdev) /* put RNG4 into program mode */ setbits32(&r4tst->rtmctl, RTMCTL_PRGM); - /* 800 clocks per sample */ + /* 1600 clocks per sample */ val = rd_reg32(&r4tst->rtsdctl); - val = (val & ~RTSDCTL_ENT_DLY_MASK) | (800 << RTSDCTL_ENT_DLY_SHIFT); + val = (val & ~RTSDCTL_ENT_DLY_MASK) | (1600 << RTSDCTL_ENT_DLY_SHIFT); wr_reg32(&r4tst->rtsdctl, val); /* min. freq. count */ wr_reg32(&r4tst->rtfrqmin, 400); diff --git a/drivers/crypto/caam/error.c b/drivers/crypto/caam/error.c index 9955ed9643e..30b8f74833d 100644 --- a/drivers/crypto/caam/error.c +++ b/drivers/crypto/caam/error.c @@ -77,10 +77,8 @@ static void report_ccb_status(u32 status, char *outstr) "Not instantiated", "Test instantiate", "Prediction resistance", - "", "Prediction resistance and test request", "Uninstantiate", - "", "Secure key generation", }; u8 cha_id = (status & JRSTA_CCBERR_CHAID_MASK) >> diff --git a/drivers/crypto/caam/jr.c b/drivers/crypto/caam/jr.c index 53c8c51d588..93d14070141 100644 --- a/drivers/crypto/caam/jr.c +++ b/drivers/crypto/caam/jr.c @@ -63,7 +63,7 @@ static void caam_jr_dequeue(unsigned long devarg) head = ACCESS_ONCE(jrp->head); - spin_lock_bh(&jrp->outlock); + spin_lock(&jrp->outlock); sw_idx = tail = jrp->tail; hw_idx = jrp->out_ring_read_index; @@ -115,7 +115,7 @@ static void caam_jr_dequeue(unsigned long devarg) jrp->tail = tail; } - spin_unlock_bh(&jrp->outlock); + spin_unlock(&jrp->outlock); /* Finally, execute user's callback */ usercall(dev, userdesc, userstatus, userarg); @@ -236,14 +236,14 @@ int caam_jr_enqueue(struct device *dev, u32 *desc, return -EIO; } - spin_lock(&jrp->inplock); + spin_lock_bh(&jrp->inplock); head = jrp->head; tail = ACCESS_ONCE(jrp->tail); if (!rd_reg32(&jrp->rregs->inpring_avail) || CIRC_SPACE(head, tail, JOBR_DEPTH) <= 0) { - spin_unlock(&jrp->inplock); + spin_unlock_bh(&jrp->inplock); dma_unmap_single(dev, desc_dma, desc_size, DMA_TO_DEVICE); return -EBUSY; } @@ -265,7 +265,7 @@ int caam_jr_enqueue(struct device *dev, u32 *desc, wr_reg32(&jrp->rregs->inpring_jobadd, 1); - spin_unlock(&jrp->inplock); + spin_unlock_bh(&jrp->inplock); return 0; } diff --git a/drivers/crypto/caam/key_gen.c b/drivers/crypto/caam/key_gen.c index 002888185f1..f6dba10246c 100644 --- a/drivers/crypto/caam/key_gen.c +++ b/drivers/crypto/caam/key_gen.c @@ -54,6 +54,10 @@ u32 gen_split_key(struct device *jrdev, u8 *key_out, int split_key_len, int ret = 0; desc = kmalloc(CAAM_CMD_SZ * 6 + CAAM_PTR_SZ * 2, GFP_KERNEL | GFP_DMA); + if (!desc) { + dev_err(jrdev, "unable to allocate key input memory\n"); + return -ENOMEM; + } init_job_desc(desc, 0); @@ -120,3 +124,4 @@ u32 gen_split_key(struct device *jrdev, u8 *key_out, int split_key_len, return ret; } +EXPORT_SYMBOL(gen_split_key); diff --git a/drivers/crypto/geode-aes.c b/drivers/crypto/geode-aes.c index f3e36c86b6c..51f196d77f2 100644 --- a/drivers/crypto/geode-aes.c +++ b/drivers/crypto/geode-aes.c @@ -289,7 +289,6 @@ static struct crypto_alg geode_alg = { .cra_blocksize = AES_MIN_BLOCK_SIZE, .cra_ctxsize = sizeof(struct geode_aes_op), .cra_module = THIS_MODULE, - .cra_list = LIST_HEAD_INIT(geode_alg.cra_list), .cra_u = { .cipher = { .cia_min_keysize = AES_MIN_KEY_SIZE, @@ -402,7 +401,6 @@ static struct crypto_alg geode_cbc_alg = { .cra_alignmask = 15, .cra_type = &crypto_blkcipher_type, .cra_module = THIS_MODULE, - .cra_list = LIST_HEAD_INIT(geode_cbc_alg.cra_list), .cra_u = { .blkcipher = { .min_keysize = AES_MIN_KEY_SIZE, @@ -489,7 +487,6 @@ static struct crypto_alg geode_ecb_alg = { .cra_alignmask = 15, .cra_type = &crypto_blkcipher_type, .cra_module = THIS_MODULE, - .cra_list = LIST_HEAD_INIT(geode_ecb_alg.cra_list), .cra_u = { .blkcipher = { .min_keysize = AES_MIN_KEY_SIZE, @@ -588,21 +585,8 @@ static struct pci_driver geode_aes_driver = { .remove = __devexit_p(geode_aes_remove) }; -static int __init -geode_aes_init(void) -{ - return pci_register_driver(&geode_aes_driver); -} - -static void __exit -geode_aes_exit(void) -{ - pci_unregister_driver(&geode_aes_driver); -} +module_pci_driver(geode_aes_driver); MODULE_AUTHOR("Advanced Micro Devices, Inc."); MODULE_DESCRIPTION("Geode LX Hardware AES driver"); MODULE_LICENSE("GPL"); - -module_init(geode_aes_init); -module_exit(geode_aes_exit); diff --git a/drivers/crypto/hifn_795x.c b/drivers/crypto/hifn_795x.c index c9c4befb5a8..fda32968a66 100644 --- a/drivers/crypto/hifn_795x.c +++ b/drivers/crypto/hifn_795x.c @@ -821,8 +821,8 @@ static int hifn_register_rng(struct hifn_device *dev) /* * We must wait at least 256 Pk_clk cycles between two reads of the rng. */ - dev->rng_wait_time = DIV_ROUND_UP(NSEC_PER_SEC, dev->pk_clk_freq) * - 256; + dev->rng_wait_time = DIV_ROUND_UP_ULL(NSEC_PER_SEC, + dev->pk_clk_freq) * 256; dev->rng.name = dev->name; dev->rng.data_present = hifn_rng_data_present, @@ -2611,14 +2611,17 @@ static int __devinit hifn_probe(struct pci_dev *pdev, const struct pci_device_id size = pci_resource_len(pdev, i); dev->bar[i] = ioremap_nocache(addr, size); - if (!dev->bar[i]) + if (!dev->bar[i]) { + err = -ENOMEM; goto err_out_unmap_bars; + } } dev->desc_virt = pci_alloc_consistent(pdev, sizeof(struct hifn_dma), &dev->desc_dma); if (!dev->desc_virt) { dprintk("Failed to allocate descriptor rings.\n"); + err = -ENOMEM; goto err_out_unmap_bars; } memset(dev->desc_virt, 0, sizeof(struct hifn_dma)); diff --git a/drivers/crypto/n2_core.c b/drivers/crypto/n2_core.c index a8bd0310f8f..aab257403b4 100644 --- a/drivers/crypto/n2_core.c +++ b/drivers/crypto/n2_core.c @@ -42,7 +42,7 @@ MODULE_DESCRIPTION("Niagara2 Crypto driver"); MODULE_LICENSE("GPL"); MODULE_VERSION(DRV_MODULE_VERSION); -#define N2_CRA_PRIORITY 300 +#define N2_CRA_PRIORITY 200 static DEFINE_MUTEX(spu_lock); diff --git a/drivers/crypto/nx/Kconfig b/drivers/crypto/nx/Kconfig new file mode 100644 index 00000000000..f82616621ae --- /dev/null +++ b/drivers/crypto/nx/Kconfig @@ -0,0 +1,26 @@ +config CRYPTO_DEV_NX_ENCRYPT + tristate "Encryption acceleration support" + depends on PPC64 && IBMVIO + default y + select CRYPTO_AES + select CRYPTO_CBC + select CRYPTO_ECB + select CRYPTO_CCM + select CRYPTO_GCM + select CRYPTO_AUTHENC + select CRYPTO_XCBC + select CRYPTO_SHA256 + select CRYPTO_SHA512 + help + Support for Power7+ in-Nest encryption acceleration. This + module supports acceleration for AES and SHA2 algorithms. If you + choose 'M' here, this module will be called nx_crypto. + +config CRYPTO_DEV_NX_COMPRESS + tristate "Compression acceleration support" + depends on PPC64 && IBMVIO + default y + help + Support for Power7+ in-Nest compression acceleration. This + module supports acceleration for AES and SHA2 algorithms. If you + choose 'M' here, this module will be called nx_compress. diff --git a/drivers/crypto/nx/Makefile b/drivers/crypto/nx/Makefile index 411ce59c80d..bb770ea45ce 100644 --- a/drivers/crypto/nx/Makefile +++ b/drivers/crypto/nx/Makefile @@ -1,4 +1,4 @@ -obj-$(CONFIG_CRYPTO_DEV_NX) += nx-crypto.o +obj-$(CONFIG_CRYPTO_DEV_NX_ENCRYPT) += nx-crypto.o nx-crypto-objs := nx.o \ nx_debugfs.o \ nx-aes-cbc.o \ @@ -9,3 +9,6 @@ nx-crypto-objs := nx.o \ nx-aes-xcbc.o \ nx-sha256.o \ nx-sha512.o + +obj-$(CONFIG_CRYPTO_DEV_NX_COMPRESS) += nx-compress.o +nx-compress-objs := nx-842.o diff --git a/drivers/crypto/nx/nx-842.c b/drivers/crypto/nx/nx-842.c new file mode 100644 index 00000000000..0ce62573867 --- /dev/null +++ b/drivers/crypto/nx/nx-842.c @@ -0,0 +1,1617 @@ +/* + * Driver for IBM Power 842 compression accelerator + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. + * + * Copyright (C) IBM Corporation, 2012 + * + * Authors: Robert Jennings <rcj@linux.vnet.ibm.com> + * Seth Jennings <sjenning@linux.vnet.ibm.com> + */ + +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/nx842.h> +#include <linux/of.h> +#include <linux/slab.h> + +#include <asm/page.h> +#include <asm/pSeries_reconfig.h> +#include <asm/vio.h> + +#include "nx_csbcpb.h" /* struct nx_csbcpb */ + +#define MODULE_NAME "nx-compress" +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Robert Jennings <rcj@linux.vnet.ibm.com>"); +MODULE_DESCRIPTION("842 H/W Compression driver for IBM Power processors"); + +#define SHIFT_4K 12 +#define SHIFT_64K 16 +#define SIZE_4K (1UL << SHIFT_4K) +#define SIZE_64K (1UL << SHIFT_64K) + +/* IO buffer must be 128 byte aligned */ +#define IO_BUFFER_ALIGN 128 + +struct nx842_header { + int blocks_nr; /* number of compressed blocks */ + int offset; /* offset of the first block (from beginning of header) */ + int sizes[0]; /* size of compressed blocks */ +}; + +static inline int nx842_header_size(const struct nx842_header *hdr) +{ + return sizeof(struct nx842_header) + + hdr->blocks_nr * sizeof(hdr->sizes[0]); +} + +/* Macros for fields within nx_csbcpb */ +/* Check the valid bit within the csbcpb valid field */ +#define NX842_CSBCBP_VALID_CHK(x) (x & BIT_MASK(7)) + +/* CE macros operate on the completion_extension field bits in the csbcpb. + * CE0 0=full completion, 1=partial completion + * CE1 0=CE0 indicates completion, 1=termination (output may be modified) + * CE2 0=processed_bytes is source bytes, 1=processed_bytes is target bytes */ +#define NX842_CSBCPB_CE0(x) (x & BIT_MASK(7)) +#define NX842_CSBCPB_CE1(x) (x & BIT_MASK(6)) +#define NX842_CSBCPB_CE2(x) (x & BIT_MASK(5)) + +/* The NX unit accepts data only on 4K page boundaries */ +#define NX842_HW_PAGE_SHIFT SHIFT_4K +#define NX842_HW_PAGE_SIZE (ASM_CONST(1) << NX842_HW_PAGE_SHIFT) +#define NX842_HW_PAGE_MASK (~(NX842_HW_PAGE_SIZE-1)) + +enum nx842_status { + UNAVAILABLE, + AVAILABLE +}; + +struct ibm_nx842_counters { + atomic64_t comp_complete; + atomic64_t comp_failed; + atomic64_t decomp_complete; + atomic64_t decomp_failed; + atomic64_t swdecomp; + atomic64_t comp_times[32]; + atomic64_t decomp_times[32]; +}; + +static struct nx842_devdata { + struct vio_dev *vdev; + struct device *dev; + struct ibm_nx842_counters *counters; + unsigned int max_sg_len; + unsigned int max_sync_size; + unsigned int max_sync_sg; + enum nx842_status status; +} __rcu *devdata; +static DEFINE_SPINLOCK(devdata_mutex); + +#define NX842_COUNTER_INC(_x) \ +static inline void nx842_inc_##_x( \ + const struct nx842_devdata *dev) { \ + if (dev) \ + atomic64_inc(&dev->counters->_x); \ +} +NX842_COUNTER_INC(comp_complete); +NX842_COUNTER_INC(comp_failed); +NX842_COUNTER_INC(decomp_complete); +NX842_COUNTER_INC(decomp_failed); +NX842_COUNTER_INC(swdecomp); + +#define NX842_HIST_SLOTS 16 + +static void ibm_nx842_incr_hist(atomic64_t *times, unsigned int time) +{ + int bucket = fls(time); + + if (bucket) + bucket = min((NX842_HIST_SLOTS - 1), bucket - 1); + + atomic64_inc(×[bucket]); +} + +/* NX unit operation flags */ +#define NX842_OP_COMPRESS 0x0 +#define NX842_OP_CRC 0x1 +#define NX842_OP_DECOMPRESS 0x2 +#define NX842_OP_COMPRESS_CRC (NX842_OP_COMPRESS | NX842_OP_CRC) +#define NX842_OP_DECOMPRESS_CRC (NX842_OP_DECOMPRESS | NX842_OP_CRC) +#define NX842_OP_ASYNC (1<<23) +#define NX842_OP_NOTIFY (1<<22) +#define NX842_OP_NOTIFY_INT(x) ((x & 0xff)<<8) + +static unsigned long nx842_get_desired_dma(struct vio_dev *viodev) +{ + /* No use of DMA mappings within the driver. */ + return 0; +} + +struct nx842_slentry { + unsigned long ptr; /* Real address (use __pa()) */ + unsigned long len; +}; + +/* pHyp scatterlist entry */ +struct nx842_scatterlist { + int entry_nr; /* number of slentries */ + struct nx842_slentry *entries; /* ptr to array of slentries */ +}; + +/* Does not include sizeof(entry_nr) in the size */ +static inline unsigned long nx842_get_scatterlist_size( + struct nx842_scatterlist *sl) +{ + return sl->entry_nr * sizeof(struct nx842_slentry); +} + +static int nx842_build_scatterlist(unsigned long buf, int len, + struct nx842_scatterlist *sl) +{ + unsigned long nextpage; + struct nx842_slentry *entry; + + sl->entry_nr = 0; + + entry = sl->entries; + while (len) { + entry->ptr = __pa(buf); + nextpage = ALIGN(buf + 1, NX842_HW_PAGE_SIZE); + if (nextpage < buf + len) { + /* we aren't at the end yet */ + if (IS_ALIGNED(buf, NX842_HW_PAGE_SIZE)) + /* we are in the middle (or beginning) */ + entry->len = NX842_HW_PAGE_SIZE; + else + /* we are at the beginning */ + entry->len = nextpage - buf; + } else { + /* at the end */ + entry->len = len; + } + + len -= entry->len; + buf += entry->len; + sl->entry_nr++; + entry++; + } + + return 0; +} + +/* + * Working memory for software decompression + */ +struct sw842_fifo { + union { + char f8[256][8]; + char f4[512][4]; + }; + char f2[256][2]; + unsigned char f84_full; + unsigned char f2_full; + unsigned char f8_count; + unsigned char f2_count; + unsigned int f4_count; +}; + +/* + * Working memory for crypto API + */ +struct nx842_workmem { + char bounce[PAGE_SIZE]; /* bounce buffer for decompression input */ + union { + /* hardware working memory */ + struct { + /* scatterlist */ + char slin[SIZE_4K]; + char slout[SIZE_4K]; + /* coprocessor status/parameter block */ + struct nx_csbcpb csbcpb; + }; + /* software working memory */ + struct sw842_fifo swfifo; /* software decompression fifo */ + }; +}; + +int nx842_get_workmem_size(void) +{ + return sizeof(struct nx842_workmem) + NX842_HW_PAGE_SIZE; +} +EXPORT_SYMBOL_GPL(nx842_get_workmem_size); + +int nx842_get_workmem_size_aligned(void) +{ + return sizeof(struct nx842_workmem); +} +EXPORT_SYMBOL_GPL(nx842_get_workmem_size_aligned); + +static int nx842_validate_result(struct device *dev, + struct cop_status_block *csb) +{ + /* The csb must be valid after returning from vio_h_cop_sync */ + if (!NX842_CSBCBP_VALID_CHK(csb->valid)) { + dev_err(dev, "%s: cspcbp not valid upon completion.\n", + __func__); + dev_dbg(dev, "valid:0x%02x cs:0x%02x cc:0x%02x ce:0x%02x\n", + csb->valid, + csb->crb_seq_number, + csb->completion_code, + csb->completion_extension); + dev_dbg(dev, "processed_bytes:%d address:0x%016lx\n", + csb->processed_byte_count, + (unsigned long)csb->address); + return -EIO; + } + + /* Check return values from the hardware in the CSB */ + switch (csb->completion_code) { + case 0: /* Completed without error */ + break; + case 64: /* Target bytes > Source bytes during compression */ + case 13: /* Output buffer too small */ + dev_dbg(dev, "%s: Compression output larger than input\n", + __func__); + return -ENOSPC; + case 66: /* Input data contains an illegal template field */ + case 67: /* Template indicates data past the end of the input stream */ + dev_dbg(dev, "%s: Bad data for decompression (code:%d)\n", + __func__, csb->completion_code); + return -EINVAL; + default: + dev_dbg(dev, "%s: Unspecified error (code:%d)\n", + __func__, csb->completion_code); + return -EIO; + } + + /* Hardware sanity check */ + if (!NX842_CSBCPB_CE2(csb->completion_extension)) { + dev_err(dev, "%s: No error returned by hardware, but " + "data returned is unusable, contact support.\n" + "(Additional info: csbcbp->processed bytes " + "does not specify processed bytes for the " + "target buffer.)\n", __func__); + return -EIO; + } + + return 0; +} + +/** + * nx842_compress - Compress data using the 842 algorithm + * + * Compression provide by the NX842 coprocessor on IBM Power systems. + * The input buffer is compressed and the result is stored in the + * provided output buffer. + * + * Upon return from this function @outlen contains the length of the + * compressed data. If there is an error then @outlen will be 0 and an + * error will be specified by the return code from this function. + * + * @in: Pointer to input buffer, must be page aligned + * @inlen: Length of input buffer, must be PAGE_SIZE + * @out: Pointer to output buffer + * @outlen: Length of output buffer + * @wrkmem: ptr to buffer for working memory, size determined by + * nx842_get_workmem_size() + * + * Returns: + * 0 Success, output of length @outlen stored in the buffer at @out + * -ENOMEM Unable to allocate internal buffers + * -ENOSPC Output buffer is to small + * -EMSGSIZE XXX Difficult to describe this limitation + * -EIO Internal error + * -ENODEV Hardware unavailable + */ +int nx842_compress(const unsigned char *in, unsigned int inlen, + unsigned char *out, unsigned int *outlen, void *wmem) +{ + struct nx842_header *hdr; + struct nx842_devdata *local_devdata; + struct device *dev = NULL; + struct nx842_workmem *workmem; + struct nx842_scatterlist slin, slout; + struct nx_csbcpb *csbcpb; + int ret = 0, max_sync_size, i, bytesleft, size, hdrsize; + unsigned long inbuf, outbuf, padding; + struct vio_pfo_op op = { + .done = NULL, + .handle = 0, + .timeout = 0, + }; + unsigned long start_time = get_tb(); + + /* + * Make sure input buffer is 64k page aligned. This is assumed since + * this driver is designed for page compression only (for now). This + * is very nice since we can now use direct DDE(s) for the input and + * the alignment is guaranteed. + */ + inbuf = (unsigned long)in; + if (!IS_ALIGNED(inbuf, PAGE_SIZE) || inlen != PAGE_SIZE) + return -EINVAL; + + rcu_read_lock(); + local_devdata = rcu_dereference(devdata); + if (!local_devdata || !local_devdata->dev) { + rcu_read_unlock(); + return -ENODEV; + } + max_sync_size = local_devdata->max_sync_size; + dev = local_devdata->dev; + + /* Create the header */ + hdr = (struct nx842_header *)out; + hdr->blocks_nr = PAGE_SIZE / max_sync_size; + hdrsize = nx842_header_size(hdr); + outbuf = (unsigned long)out + hdrsize; + bytesleft = *outlen - hdrsize; + + /* Init scatterlist */ + workmem = (struct nx842_workmem *)ALIGN((unsigned long)wmem, + NX842_HW_PAGE_SIZE); + slin.entries = (struct nx842_slentry *)workmem->slin; + slout.entries = (struct nx842_slentry *)workmem->slout; + + /* Init operation */ + op.flags = NX842_OP_COMPRESS; + csbcpb = &workmem->csbcpb; + memset(csbcpb, 0, sizeof(*csbcpb)); + op.csbcpb = __pa(csbcpb); + op.out = __pa(slout.entries); + + for (i = 0; i < hdr->blocks_nr; i++) { + /* + * Aligning the output blocks to 128 bytes does waste space, + * but it prevents the need for bounce buffers and memory + * copies. It also simplifies the code a lot. In the worst + * case (64k page, 4k max_sync_size), you lose up to + * (128*16)/64k = ~3% the compression factor. For 64k + * max_sync_size, the loss would be at most 128/64k = ~0.2%. + */ + padding = ALIGN(outbuf, IO_BUFFER_ALIGN) - outbuf; + outbuf += padding; + bytesleft -= padding; + if (i == 0) + /* save offset into first block in header */ + hdr->offset = padding + hdrsize; + + if (bytesleft <= 0) { + ret = -ENOSPC; + goto unlock; + } + + /* + * NOTE: If the default max_sync_size is changed from 4k + * to 64k, remove the "likely" case below, since a + * scatterlist will always be needed. + */ + if (likely(max_sync_size == NX842_HW_PAGE_SIZE)) { + /* Create direct DDE */ + op.in = __pa(inbuf); + op.inlen = max_sync_size; + + } else { + /* Create indirect DDE (scatterlist) */ + nx842_build_scatterlist(inbuf, max_sync_size, &slin); + op.in = __pa(slin.entries); + op.inlen = -nx842_get_scatterlist_size(&slin); + } + + /* + * If max_sync_size != NX842_HW_PAGE_SIZE, an indirect + * DDE is required for the outbuf. + * If max_sync_size == NX842_HW_PAGE_SIZE, outbuf must + * also be page aligned (1 in 128/4k=32 chance) in order + * to use a direct DDE. + * This is unlikely, just use an indirect DDE always. + */ + nx842_build_scatterlist(outbuf, + min(bytesleft, max_sync_size), &slout); + /* op.out set before loop */ + op.outlen = -nx842_get_scatterlist_size(&slout); + + /* Send request to pHyp */ + ret = vio_h_cop_sync(local_devdata->vdev, &op); + + /* Check for pHyp error */ + if (ret) { + dev_dbg(dev, "%s: vio_h_cop_sync error (ret=%d, hret=%ld)\n", + __func__, ret, op.hcall_err); + ret = -EIO; + goto unlock; + } + + /* Check for hardware error */ + ret = nx842_validate_result(dev, &csbcpb->csb); + if (ret && ret != -ENOSPC) + goto unlock; + + /* Handle incompressible data */ + if (unlikely(ret == -ENOSPC)) { + if (bytesleft < max_sync_size) { + /* + * Not enough space left in the output buffer + * to store uncompressed block + */ + goto unlock; + } else { + /* Store incompressible block */ + memcpy((void *)outbuf, (void *)inbuf, + max_sync_size); + hdr->sizes[i] = -max_sync_size; + outbuf += max_sync_size; + bytesleft -= max_sync_size; + /* Reset ret, incompressible data handled */ + ret = 0; + } + } else { + /* Normal case, compression was successful */ + size = csbcpb->csb.processed_byte_count; + dev_dbg(dev, "%s: processed_bytes=%d\n", + __func__, size); + hdr->sizes[i] = size; + outbuf += size; + bytesleft -= size; + } + + inbuf += max_sync_size; + } + + *outlen = (unsigned int)(outbuf - (unsigned long)out); + +unlock: + if (ret) + nx842_inc_comp_failed(local_devdata); + else { + nx842_inc_comp_complete(local_devdata); + ibm_nx842_incr_hist(local_devdata->counters->comp_times, + (get_tb() - start_time) / tb_ticks_per_usec); + } + rcu_read_unlock(); + return ret; +} +EXPORT_SYMBOL_GPL(nx842_compress); + +static int sw842_decompress(const unsigned char *, int, unsigned char *, int *, + const void *); + +/** + * nx842_decompress - Decompress data using the 842 algorithm + * + * Decompression provide by the NX842 coprocessor on IBM Power systems. + * The input buffer is decompressed and the result is stored in the + * provided output buffer. The size allocated to the output buffer is + * provided by the caller of this function in @outlen. Upon return from + * this function @outlen contains the length of the decompressed data. + * If there is an error then @outlen will be 0 and an error will be + * specified by the return code from this function. + * + * @in: Pointer to input buffer, will use bounce buffer if not 128 byte + * aligned + * @inlen: Length of input buffer + * @out: Pointer to output buffer, must be page aligned + * @outlen: Length of output buffer, must be PAGE_SIZE + * @wrkmem: ptr to buffer for working memory, size determined by + * nx842_get_workmem_size() + * + * Returns: + * 0 Success, output of length @outlen stored in the buffer at @out + * -ENODEV Hardware decompression device is unavailable + * -ENOMEM Unable to allocate internal buffers + * -ENOSPC Output buffer is to small + * -EINVAL Bad input data encountered when attempting decompress + * -EIO Internal error + */ +int nx842_decompress(const unsigned char *in, unsigned int inlen, + unsigned char *out, unsigned int *outlen, void *wmem) +{ + struct nx842_header *hdr; + struct nx842_devdata *local_devdata; + struct device *dev = NULL; + struct nx842_workmem *workmem; + struct nx842_scatterlist slin, slout; + struct nx_csbcpb *csbcpb; + int ret = 0, i, size, max_sync_size; + unsigned long inbuf, outbuf; + struct vio_pfo_op op = { + .done = NULL, + .handle = 0, + .timeout = 0, + }; + unsigned long start_time = get_tb(); + + /* Ensure page alignment and size */ + outbuf = (unsigned long)out; + if (!IS_ALIGNED(outbuf, PAGE_SIZE) || *outlen != PAGE_SIZE) + return -EINVAL; + + rcu_read_lock(); + local_devdata = rcu_dereference(devdata); + if (local_devdata) + dev = local_devdata->dev; + + /* Get header */ + hdr = (struct nx842_header *)in; + + workmem = (struct nx842_workmem *)ALIGN((unsigned long)wmem, + NX842_HW_PAGE_SIZE); + + inbuf = (unsigned long)in + hdr->offset; + if (likely(!IS_ALIGNED(inbuf, IO_BUFFER_ALIGN))) { + /* Copy block(s) into bounce buffer for alignment */ + memcpy(workmem->bounce, in + hdr->offset, inlen - hdr->offset); + inbuf = (unsigned long)workmem->bounce; + } + + /* Init scatterlist */ + slin.entries = (struct nx842_slentry *)workmem->slin; + slout.entries = (struct nx842_slentry *)workmem->slout; + + /* Init operation */ + op.flags = NX842_OP_DECOMPRESS; + csbcpb = &workmem->csbcpb; + memset(csbcpb, 0, sizeof(*csbcpb)); + op.csbcpb = __pa(csbcpb); + + /* + * max_sync_size may have changed since compression, + * so we can't read it from the device info. We need + * to derive it from hdr->blocks_nr. + */ + max_sync_size = PAGE_SIZE / hdr->blocks_nr; + + for (i = 0; i < hdr->blocks_nr; i++) { + /* Skip padding */ + inbuf = ALIGN(inbuf, IO_BUFFER_ALIGN); + + if (hdr->sizes[i] < 0) { + /* Negative sizes indicate uncompressed data blocks */ + size = abs(hdr->sizes[i]); + memcpy((void *)outbuf, (void *)inbuf, size); + outbuf += size; + inbuf += size; + continue; + } + + if (!dev) + goto sw; + + /* + * The better the compression, the more likely the "likely" + * case becomes. + */ + if (likely((inbuf & NX842_HW_PAGE_MASK) == + ((inbuf + hdr->sizes[i] - 1) & NX842_HW_PAGE_MASK))) { + /* Create direct DDE */ + op.in = __pa(inbuf); + op.inlen = hdr->sizes[i]; + } else { + /* Create indirect DDE (scatterlist) */ + nx842_build_scatterlist(inbuf, hdr->sizes[i] , &slin); + op.in = __pa(slin.entries); + op.inlen = -nx842_get_scatterlist_size(&slin); + } + + /* + * NOTE: If the default max_sync_size is changed from 4k + * to 64k, remove the "likely" case below, since a + * scatterlist will always be needed. + */ + if (likely(max_sync_size == NX842_HW_PAGE_SIZE)) { + /* Create direct DDE */ + op.out = __pa(outbuf); + op.outlen = max_sync_size; + } else { + /* Create indirect DDE (scatterlist) */ + nx842_build_scatterlist(outbuf, max_sync_size, &slout); + op.out = __pa(slout.entries); + op.outlen = -nx842_get_scatterlist_size(&slout); + } + + /* Send request to pHyp */ + ret = vio_h_cop_sync(local_devdata->vdev, &op); + + /* Check for pHyp error */ + if (ret) { + dev_dbg(dev, "%s: vio_h_cop_sync error (ret=%d, hret=%ld)\n", + __func__, ret, op.hcall_err); + dev = NULL; + goto sw; + } + + /* Check for hardware error */ + ret = nx842_validate_result(dev, &csbcpb->csb); + if (ret) { + dev = NULL; + goto sw; + } + + /* HW decompression success */ + inbuf += hdr->sizes[i]; + outbuf += csbcpb->csb.processed_byte_count; + continue; + +sw: + /* software decompression */ + size = max_sync_size; + ret = sw842_decompress( + (unsigned char *)inbuf, hdr->sizes[i], + (unsigned char *)outbuf, &size, wmem); + if (ret) + pr_debug("%s: sw842_decompress failed with %d\n", + __func__, ret); + + if (ret) { + if (ret != -ENOSPC && ret != -EINVAL && + ret != -EMSGSIZE) + ret = -EIO; + goto unlock; + } + + /* SW decompression success */ + inbuf += hdr->sizes[i]; + outbuf += size; + } + + *outlen = (unsigned int)(outbuf - (unsigned long)out); + +unlock: + if (ret) + /* decompress fail */ + nx842_inc_decomp_failed(local_devdata); + else { + if (!dev) + /* software decompress */ + nx842_inc_swdecomp(local_devdata); + nx842_inc_decomp_complete(local_devdata); + ibm_nx842_incr_hist(local_devdata->counters->decomp_times, + (get_tb() - start_time) / tb_ticks_per_usec); + } + + rcu_read_unlock(); + return ret; +} +EXPORT_SYMBOL_GPL(nx842_decompress); + +/** + * nx842_OF_set_defaults -- Set default (disabled) values for devdata + * + * @devdata - struct nx842_devdata to update + * + * Returns: + * 0 on success + * -ENOENT if @devdata ptr is NULL + */ +static int nx842_OF_set_defaults(struct nx842_devdata *devdata) +{ + if (devdata) { + devdata->max_sync_size = 0; + devdata->max_sync_sg = 0; + devdata->max_sg_len = 0; + devdata->status = UNAVAILABLE; + return 0; + } else + return -ENOENT; +} + +/** + * nx842_OF_upd_status -- Update the device info from OF status prop + * + * The status property indicates if the accelerator is enabled. If the + * device is in the OF tree it indicates that the hardware is present. + * The status field indicates if the device is enabled when the status + * is 'okay'. Otherwise the device driver will be disabled. + * + * @devdata - struct nx842_devdata to update + * @prop - struct property point containing the maxsyncop for the update + * + * Returns: + * 0 - Device is available + * -EINVAL - Device is not available + */ +static int nx842_OF_upd_status(struct nx842_devdata *devdata, + struct property *prop) { + int ret = 0; + const char *status = (const char *)prop->value; + + if (!strncmp(status, "okay", (size_t)prop->length)) { + devdata->status = AVAILABLE; + } else { + dev_info(devdata->dev, "%s: status '%s' is not 'okay'\n", + __func__, status); + devdata->status = UNAVAILABLE; + } + + return ret; +} + +/** + * nx842_OF_upd_maxsglen -- Update the device info from OF maxsglen prop + * + * Definition of the 'ibm,max-sg-len' OF property: + * This field indicates the maximum byte length of a scatter list + * for the platform facility. It is a single cell encoded as with encode-int. + * + * Example: + * # od -x ibm,max-sg-len + * 0000000 0000 0ff0 + * + * In this example, the maximum byte length of a scatter list is + * 0x0ff0 (4,080). + * + * @devdata - struct nx842_devdata to update + * @prop - struct property point containing the maxsyncop for the update + * + * Returns: + * 0 on success + * -EINVAL on failure + */ +static int nx842_OF_upd_maxsglen(struct nx842_devdata *devdata, + struct property *prop) { + int ret = 0; + const int *maxsglen = prop->value; + + if (prop->length != sizeof(*maxsglen)) { + dev_err(devdata->dev, "%s: unexpected format for ibm,max-sg-len property\n", __func__); + dev_dbg(devdata->dev, "%s: ibm,max-sg-len is %d bytes long, expected %lu bytes\n", __func__, + prop->length, sizeof(*maxsglen)); + ret = -EINVAL; + } else { + devdata->max_sg_len = (unsigned int)min(*maxsglen, + (int)NX842_HW_PAGE_SIZE); + } + + return ret; +} + +/** + * nx842_OF_upd_maxsyncop -- Update the device info from OF maxsyncop prop + * + * Definition of the 'ibm,max-sync-cop' OF property: + * Two series of cells. The first series of cells represents the maximums + * that can be synchronously compressed. The second series of cells + * represents the maximums that can be synchronously decompressed. + * 1. The first cell in each series contains the count of the number of + * data length, scatter list elements pairs that follow – each being + * of the form + * a. One cell data byte length + * b. One cell total number of scatter list elements + * + * Example: + * # od -x ibm,max-sync-cop + * 0000000 0000 0001 0000 1000 0000 01fe 0000 0001 + * 0000020 0000 1000 0000 01fe + * + * In this example, compression supports 0x1000 (4,096) data byte length + * and 0x1fe (510) total scatter list elements. Decompression supports + * 0x1000 (4,096) data byte length and 0x1f3 (510) total scatter list + * elements. + * + * @devdata - struct nx842_devdata to update + * @prop - struct property point containing the maxsyncop for the update + * + * Returns: + * 0 on success + * -EINVAL on failure + */ +static int nx842_OF_upd_maxsyncop(struct nx842_devdata *devdata, + struct property *prop) { + int ret = 0; + const struct maxsynccop_t { + int comp_elements; + int comp_data_limit; + int comp_sg_limit; + int decomp_elements; + int decomp_data_limit; + int decomp_sg_limit; + } *maxsynccop; + + if (prop->length != sizeof(*maxsynccop)) { + dev_err(devdata->dev, "%s: unexpected format for ibm,max-sync-cop property\n", __func__); + dev_dbg(devdata->dev, "%s: ibm,max-sync-cop is %d bytes long, expected %lu bytes\n", __func__, prop->length, + sizeof(*maxsynccop)); + ret = -EINVAL; + goto out; + } + + maxsynccop = (const struct maxsynccop_t *)prop->value; + + /* Use one limit rather than separate limits for compression and + * decompression. Set a maximum for this so as not to exceed the + * size that the header can support and round the value down to + * the hardware page size (4K) */ + devdata->max_sync_size = + (unsigned int)min(maxsynccop->comp_data_limit, + maxsynccop->decomp_data_limit); + + devdata->max_sync_size = min_t(unsigned int, devdata->max_sync_size, + SIZE_64K); + + if (devdata->max_sync_size < SIZE_4K) { + dev_err(devdata->dev, "%s: hardware max data size (%u) is " + "less than the driver minimum, unable to use " + "the hardware device\n", + __func__, devdata->max_sync_size); + ret = -EINVAL; + goto out; + } + + devdata->max_sync_sg = (unsigned int)min(maxsynccop->comp_sg_limit, + maxsynccop->decomp_sg_limit); + if (devdata->max_sync_sg < 1) { + dev_err(devdata->dev, "%s: hardware max sg size (%u) is " + "less than the driver minimum, unable to use " + "the hardware device\n", + __func__, devdata->max_sync_sg); + ret = -EINVAL; + goto out; + } + +out: + return ret; +} + +/** + * + * nx842_OF_upd -- Handle OF properties updates for the device. + * + * Set all properties from the OF tree. Optionally, a new property + * can be provided by the @new_prop pointer to overwrite an existing value. + * The device will remain disabled until all values are valid, this function + * will return an error for updates unless all values are valid. + * + * @new_prop: If not NULL, this property is being updated. If NULL, update + * all properties from the current values in the OF tree. + * + * Returns: + * 0 - Success + * -ENOMEM - Could not allocate memory for new devdata structure + * -EINVAL - property value not found, new_prop is not a recognized + * property for the device or property value is not valid. + * -ENODEV - Device is not available + */ +static int nx842_OF_upd(struct property *new_prop) +{ + struct nx842_devdata *old_devdata = NULL; + struct nx842_devdata *new_devdata = NULL; + struct device_node *of_node = NULL; + struct property *status = NULL; + struct property *maxsglen = NULL; + struct property *maxsyncop = NULL; + int ret = 0; + unsigned long flags; + + spin_lock_irqsave(&devdata_mutex, flags); + old_devdata = rcu_dereference_check(devdata, + lockdep_is_held(&devdata_mutex)); + if (old_devdata) + of_node = old_devdata->dev->of_node; + + if (!old_devdata || !of_node) { + pr_err("%s: device is not available\n", __func__); + spin_unlock_irqrestore(&devdata_mutex, flags); + return -ENODEV; + } + + new_devdata = kzalloc(sizeof(*new_devdata), GFP_NOFS); + if (!new_devdata) { + dev_err(old_devdata->dev, "%s: Could not allocate memory for device data\n", __func__); + ret = -ENOMEM; + goto error_out; + } + + memcpy(new_devdata, old_devdata, sizeof(*old_devdata)); + new_devdata->counters = old_devdata->counters; + + /* Set ptrs for existing properties */ + status = of_find_property(of_node, "status", NULL); + maxsglen = of_find_property(of_node, "ibm,max-sg-len", NULL); + maxsyncop = of_find_property(of_node, "ibm,max-sync-cop", NULL); + if (!status || !maxsglen || !maxsyncop) { + dev_err(old_devdata->dev, "%s: Could not locate device properties\n", __func__); + ret = -EINVAL; + goto error_out; + } + + /* Set ptr to new property if provided */ + if (new_prop) { + /* Single property */ + if (!strncmp(new_prop->name, "status", new_prop->length)) { + status = new_prop; + + } else if (!strncmp(new_prop->name, "ibm,max-sg-len", + new_prop->length)) { + maxsglen = new_prop; + + } else if (!strncmp(new_prop->name, "ibm,max-sync-cop", + new_prop->length)) { + maxsyncop = new_prop; + + } else { + /* + * Skip the update, the property being updated + * has no impact. + */ + goto out; + } + } + + /* Perform property updates */ + ret = nx842_OF_upd_status(new_devdata, status); + if (ret) + goto error_out; + + ret = nx842_OF_upd_maxsglen(new_devdata, maxsglen); + if (ret) + goto error_out; + + ret = nx842_OF_upd_maxsyncop(new_devdata, maxsyncop); + if (ret) + goto error_out; + +out: + dev_info(old_devdata->dev, "%s: max_sync_size new:%u old:%u\n", + __func__, new_devdata->max_sync_size, + old_devdata->max_sync_size); + dev_info(old_devdata->dev, "%s: max_sync_sg new:%u old:%u\n", + __func__, new_devdata->max_sync_sg, + old_devdata->max_sync_sg); + dev_info(old_devdata->dev, "%s: max_sg_len new:%u old:%u\n", + __func__, new_devdata->max_sg_len, + old_devdata->max_sg_len); + + rcu_assign_pointer(devdata, new_devdata); + spin_unlock_irqrestore(&devdata_mutex, flags); + synchronize_rcu(); + dev_set_drvdata(new_devdata->dev, new_devdata); + kfree(old_devdata); + return 0; + +error_out: + if (new_devdata) { + dev_info(old_devdata->dev, "%s: device disabled\n", __func__); + nx842_OF_set_defaults(new_devdata); + rcu_assign_pointer(devdata, new_devdata); + spin_unlock_irqrestore(&devdata_mutex, flags); + synchronize_rcu(); + dev_set_drvdata(new_devdata->dev, new_devdata); + kfree(old_devdata); + } else { + dev_err(old_devdata->dev, "%s: could not update driver from hardware\n", __func__); + spin_unlock_irqrestore(&devdata_mutex, flags); + } + + if (!ret) + ret = -EINVAL; + return ret; +} + +/** + * nx842_OF_notifier - Process updates to OF properties for the device + * + * @np: notifier block + * @action: notifier action + * @update: struct pSeries_reconfig_prop_update pointer if action is + * PSERIES_UPDATE_PROPERTY + * + * Returns: + * NOTIFY_OK on success + * NOTIFY_BAD encoded with error number on failure, use + * notifier_to_errno() to decode this value + */ +static int nx842_OF_notifier(struct notifier_block *np, + unsigned long action, + void *update) +{ + struct pSeries_reconfig_prop_update *upd; + struct nx842_devdata *local_devdata; + struct device_node *node = NULL; + + upd = (struct pSeries_reconfig_prop_update *)update; + + rcu_read_lock(); + local_devdata = rcu_dereference(devdata); + if (local_devdata) + node = local_devdata->dev->of_node; + + if (local_devdata && + action == PSERIES_UPDATE_PROPERTY && + !strcmp(upd->node->name, node->name)) { + rcu_read_unlock(); + nx842_OF_upd(upd->property); + } else + rcu_read_unlock(); + + return NOTIFY_OK; +} + +static struct notifier_block nx842_of_nb = { + .notifier_call = nx842_OF_notifier, +}; + +#define nx842_counter_read(_name) \ +static ssize_t nx842_##_name##_show(struct device *dev, \ + struct device_attribute *attr, \ + char *buf) { \ + struct nx842_devdata *local_devdata; \ + int p = 0; \ + rcu_read_lock(); \ + local_devdata = rcu_dereference(devdata); \ + if (local_devdata) \ + p = snprintf(buf, PAGE_SIZE, "%ld\n", \ + atomic64_read(&local_devdata->counters->_name)); \ + rcu_read_unlock(); \ + return p; \ +} + +#define NX842DEV_COUNTER_ATTR_RO(_name) \ + nx842_counter_read(_name); \ + static struct device_attribute dev_attr_##_name = __ATTR(_name, \ + 0444, \ + nx842_##_name##_show,\ + NULL); + +NX842DEV_COUNTER_ATTR_RO(comp_complete); +NX842DEV_COUNTER_ATTR_RO(comp_failed); +NX842DEV_COUNTER_ATTR_RO(decomp_complete); +NX842DEV_COUNTER_ATTR_RO(decomp_failed); +NX842DEV_COUNTER_ATTR_RO(swdecomp); + +static ssize_t nx842_timehist_show(struct device *, + struct device_attribute *, char *); + +static struct device_attribute dev_attr_comp_times = __ATTR(comp_times, 0444, + nx842_timehist_show, NULL); +static struct device_attribute dev_attr_decomp_times = __ATTR(decomp_times, + 0444, nx842_timehist_show, NULL); + +static ssize_t nx842_timehist_show(struct device *dev, + struct device_attribute *attr, char *buf) { + char *p = buf; + struct nx842_devdata *local_devdata; + atomic64_t *times; + int bytes_remain = PAGE_SIZE; + int bytes; + int i; + + rcu_read_lock(); + local_devdata = rcu_dereference(devdata); + if (!local_devdata) { + rcu_read_unlock(); + return 0; + } + + if (attr == &dev_attr_comp_times) + times = local_devdata->counters->comp_times; + else if (attr == &dev_attr_decomp_times) + times = local_devdata->counters->decomp_times; + else { + rcu_read_unlock(); + return 0; + } + + for (i = 0; i < (NX842_HIST_SLOTS - 2); i++) { + bytes = snprintf(p, bytes_remain, "%u-%uus:\t%ld\n", + i ? (2<<(i-1)) : 0, (2<<i)-1, + atomic64_read(×[i])); + bytes_remain -= bytes; + p += bytes; + } + /* The last bucket holds everything over + * 2<<(NX842_HIST_SLOTS - 2) us */ + bytes = snprintf(p, bytes_remain, "%uus - :\t%ld\n", + 2<<(NX842_HIST_SLOTS - 2), + atomic64_read(×[(NX842_HIST_SLOTS - 1)])); + p += bytes; + + rcu_read_unlock(); + return p - buf; +} + +static struct attribute *nx842_sysfs_entries[] = { + &dev_attr_comp_complete.attr, + &dev_attr_comp_failed.attr, + &dev_attr_decomp_complete.attr, + &dev_attr_decomp_failed.attr, + &dev_attr_swdecomp.attr, + &dev_attr_comp_times.attr, + &dev_attr_decomp_times.attr, + NULL, +}; + +static struct attribute_group nx842_attribute_group = { + .name = NULL, /* put in device directory */ + .attrs = nx842_sysfs_entries, +}; + +static int __init nx842_probe(struct vio_dev *viodev, + const struct vio_device_id *id) +{ + struct nx842_devdata *old_devdata, *new_devdata = NULL; + unsigned long flags; + int ret = 0; + + spin_lock_irqsave(&devdata_mutex, flags); + old_devdata = rcu_dereference_check(devdata, + lockdep_is_held(&devdata_mutex)); + + if (old_devdata && old_devdata->vdev != NULL) { + dev_err(&viodev->dev, "%s: Attempt to register more than one instance of the hardware\n", __func__); + ret = -1; + goto error_unlock; + } + + dev_set_drvdata(&viodev->dev, NULL); + + new_devdata = kzalloc(sizeof(*new_devdata), GFP_NOFS); + if (!new_devdata) { + dev_err(&viodev->dev, "%s: Could not allocate memory for device data\n", __func__); + ret = -ENOMEM; + goto error_unlock; + } + + new_devdata->counters = kzalloc(sizeof(*new_devdata->counters), + GFP_NOFS); + if (!new_devdata->counters) { + dev_err(&viodev->dev, "%s: Could not allocate memory for performance counters\n", __func__); + ret = -ENOMEM; + goto error_unlock; + } + + new_devdata->vdev = viodev; + new_devdata->dev = &viodev->dev; + nx842_OF_set_defaults(new_devdata); + + rcu_assign_pointer(devdata, new_devdata); + spin_unlock_irqrestore(&devdata_mutex, flags); + synchronize_rcu(); + kfree(old_devdata); + + pSeries_reconfig_notifier_register(&nx842_of_nb); + + ret = nx842_OF_upd(NULL); + if (ret && ret != -ENODEV) { + dev_err(&viodev->dev, "could not parse device tree. %d\n", ret); + ret = -1; + goto error; + } + + rcu_read_lock(); + if (dev_set_drvdata(&viodev->dev, rcu_dereference(devdata))) { + rcu_read_unlock(); + dev_err(&viodev->dev, "failed to set driver data for device\n"); + ret = -1; + goto error; + } + rcu_read_unlock(); + + if (sysfs_create_group(&viodev->dev.kobj, &nx842_attribute_group)) { + dev_err(&viodev->dev, "could not create sysfs device attributes\n"); + ret = -1; + goto error; + } + + return 0; + +error_unlock: + spin_unlock_irqrestore(&devdata_mutex, flags); + if (new_devdata) + kfree(new_devdata->counters); + kfree(new_devdata); +error: + return ret; +} + +static int __exit nx842_remove(struct vio_dev *viodev) +{ + struct nx842_devdata *old_devdata; + unsigned long flags; + + pr_info("Removing IBM Power 842 compression device\n"); + sysfs_remove_group(&viodev->dev.kobj, &nx842_attribute_group); + + spin_lock_irqsave(&devdata_mutex, flags); + old_devdata = rcu_dereference_check(devdata, + lockdep_is_held(&devdata_mutex)); + pSeries_reconfig_notifier_unregister(&nx842_of_nb); + rcu_assign_pointer(devdata, NULL); + spin_unlock_irqrestore(&devdata_mutex, flags); + synchronize_rcu(); + dev_set_drvdata(&viodev->dev, NULL); + if (old_devdata) + kfree(old_devdata->counters); + kfree(old_devdata); + return 0; +} + +static struct vio_device_id nx842_driver_ids[] = { + {"ibm,compression-v1", "ibm,compression"}, + {"", ""}, +}; + +static struct vio_driver nx842_driver = { + .name = MODULE_NAME, + .probe = nx842_probe, + .remove = nx842_remove, + .get_desired_dma = nx842_get_desired_dma, + .id_table = nx842_driver_ids, +}; + +static int __init nx842_init(void) +{ + struct nx842_devdata *new_devdata; + pr_info("Registering IBM Power 842 compression driver\n"); + + RCU_INIT_POINTER(devdata, NULL); + new_devdata = kzalloc(sizeof(*new_devdata), GFP_KERNEL); + if (!new_devdata) { + pr_err("Could not allocate memory for device data\n"); + return -ENOMEM; + } + new_devdata->status = UNAVAILABLE; + RCU_INIT_POINTER(devdata, new_devdata); + + return vio_register_driver(&nx842_driver); +} + +module_init(nx842_init); + +static void __exit nx842_exit(void) +{ + struct nx842_devdata *old_devdata; + unsigned long flags; + + pr_info("Exiting IBM Power 842 compression driver\n"); + spin_lock_irqsave(&devdata_mutex, flags); + old_devdata = rcu_dereference_check(devdata, + lockdep_is_held(&devdata_mutex)); + rcu_assign_pointer(devdata, NULL); + spin_unlock_irqrestore(&devdata_mutex, flags); + synchronize_rcu(); + if (old_devdata) + dev_set_drvdata(old_devdata->dev, NULL); + kfree(old_devdata); + vio_unregister_driver(&nx842_driver); +} + +module_exit(nx842_exit); + +/********************************* + * 842 software decompressor +*********************************/ +typedef int (*sw842_template_op)(const char **, int *, unsigned char **, + struct sw842_fifo *); + +static int sw842_data8(const char **, int *, unsigned char **, + struct sw842_fifo *); +static int sw842_data4(const char **, int *, unsigned char **, + struct sw842_fifo *); +static int sw842_data2(const char **, int *, unsigned char **, + struct sw842_fifo *); +static int sw842_ptr8(const char **, int *, unsigned char **, + struct sw842_fifo *); +static int sw842_ptr4(const char **, int *, unsigned char **, + struct sw842_fifo *); +static int sw842_ptr2(const char **, int *, unsigned char **, + struct sw842_fifo *); + +/* special templates */ +#define SW842_TMPL_REPEAT 0x1B +#define SW842_TMPL_ZEROS 0x1C +#define SW842_TMPL_EOF 0x1E + +static sw842_template_op sw842_tmpl_ops[26][4] = { + { sw842_data8, NULL}, /* 0 (00000) */ + { sw842_data4, sw842_data2, sw842_ptr2, NULL}, + { sw842_data4, sw842_ptr2, sw842_data2, NULL}, + { sw842_data4, sw842_ptr2, sw842_ptr2, NULL}, + { sw842_data4, sw842_ptr4, NULL}, + { sw842_data2, sw842_ptr2, sw842_data4, NULL}, + { sw842_data2, sw842_ptr2, sw842_data2, sw842_ptr2}, + { sw842_data2, sw842_ptr2, sw842_ptr2, sw842_data2}, + { sw842_data2, sw842_ptr2, sw842_ptr2, sw842_ptr2,}, + { sw842_data2, sw842_ptr2, sw842_ptr4, NULL}, + { sw842_ptr2, sw842_data2, sw842_data4, NULL}, /* 10 (01010) */ + { sw842_ptr2, sw842_data4, sw842_ptr2, NULL}, + { sw842_ptr2, sw842_data2, sw842_ptr2, sw842_data2}, + { sw842_ptr2, sw842_data2, sw842_ptr2, sw842_ptr2}, + { sw842_ptr2, sw842_data2, sw842_ptr4, NULL}, + { sw842_ptr2, sw842_ptr2, sw842_data4, NULL}, + { sw842_ptr2, sw842_ptr2, sw842_data2, sw842_ptr2}, + { sw842_ptr2, sw842_ptr2, sw842_ptr2, sw842_data2}, + { sw842_ptr2, sw842_ptr2, sw842_ptr2, sw842_ptr2}, + { sw842_ptr2, sw842_ptr2, sw842_ptr4, NULL}, + { sw842_ptr4, sw842_data4, NULL}, /* 20 (10100) */ + { sw842_ptr4, sw842_data2, sw842_ptr2, NULL}, + { sw842_ptr4, sw842_ptr2, sw842_data2, NULL}, + { sw842_ptr4, sw842_ptr2, sw842_ptr2, NULL}, + { sw842_ptr4, sw842_ptr4, NULL}, + { sw842_ptr8, NULL} +}; + +/* Software decompress helpers */ + +static uint8_t sw842_get_byte(const char *buf, int bit) +{ + uint8_t tmpl; + uint16_t tmp; + tmp = htons(*(uint16_t *)(buf)); + tmp = (uint16_t)(tmp << bit); + tmp = ntohs(tmp); + memcpy(&tmpl, &tmp, 1); + return tmpl; +} + +static uint8_t sw842_get_template(const char **buf, int *bit) +{ + uint8_t byte; + byte = sw842_get_byte(*buf, *bit); + byte = byte >> 3; + byte &= 0x1F; + *buf += (*bit + 5) / 8; + *bit = (*bit + 5) % 8; + return byte; +} + +/* repeat_count happens to be 5-bit too (like the template) */ +static uint8_t sw842_get_repeat_count(const char **buf, int *bit) +{ + uint8_t byte; + byte = sw842_get_byte(*buf, *bit); + byte = byte >> 2; + byte &= 0x3F; + *buf += (*bit + 6) / 8; + *bit = (*bit + 6) % 8; + return byte; +} + +static uint8_t sw842_get_ptr2(const char **buf, int *bit) +{ + uint8_t ptr; + ptr = sw842_get_byte(*buf, *bit); + (*buf)++; + return ptr; +} + +static uint16_t sw842_get_ptr4(const char **buf, int *bit, + struct sw842_fifo *fifo) +{ + uint16_t ptr; + ptr = htons(*(uint16_t *)(*buf)); + ptr = (uint16_t)(ptr << *bit); + ptr = ptr >> 7; + ptr &= 0x01FF; + *buf += (*bit + 9) / 8; + *bit = (*bit + 9) % 8; + return ptr; +} + +static uint8_t sw842_get_ptr8(const char **buf, int *bit, + struct sw842_fifo *fifo) +{ + return sw842_get_ptr2(buf, bit); +} + +/* Software decompress template ops */ + +static int sw842_data8(const char **inbuf, int *inbit, + unsigned char **outbuf, struct sw842_fifo *fifo) +{ + int ret; + + ret = sw842_data4(inbuf, inbit, outbuf, fifo); + if (ret) + return ret; + ret = sw842_data4(inbuf, inbit, outbuf, fifo); + return ret; +} + +static int sw842_data4(const char **inbuf, int *inbit, + unsigned char **outbuf, struct sw842_fifo *fifo) +{ + int ret; + + ret = sw842_data2(inbuf, inbit, outbuf, fifo); + if (ret) + return ret; + ret = sw842_data2(inbuf, inbit, outbuf, fifo); + return ret; +} + +static int sw842_data2(const char **inbuf, int *inbit, + unsigned char **outbuf, struct sw842_fifo *fifo) +{ + **outbuf = sw842_get_byte(*inbuf, *inbit); + (*inbuf)++; + (*outbuf)++; + **outbuf = sw842_get_byte(*inbuf, *inbit); + (*inbuf)++; + (*outbuf)++; + return 0; +} + +static int sw842_ptr8(const char **inbuf, int *inbit, + unsigned char **outbuf, struct sw842_fifo *fifo) +{ + uint8_t ptr; + ptr = sw842_get_ptr8(inbuf, inbit, fifo); + if (!fifo->f84_full && (ptr >= fifo->f8_count)) + return 1; + memcpy(*outbuf, fifo->f8[ptr], 8); + *outbuf += 8; + return 0; +} + +static int sw842_ptr4(const char **inbuf, int *inbit, + unsigned char **outbuf, struct sw842_fifo *fifo) +{ + uint16_t ptr; + ptr = sw842_get_ptr4(inbuf, inbit, fifo); + if (!fifo->f84_full && (ptr >= fifo->f4_count)) + return 1; + memcpy(*outbuf, fifo->f4[ptr], 4); + *outbuf += 4; + return 0; +} + +static int sw842_ptr2(const char **inbuf, int *inbit, + unsigned char **outbuf, struct sw842_fifo *fifo) +{ + uint8_t ptr; + ptr = sw842_get_ptr2(inbuf, inbit); + if (!fifo->f2_full && (ptr >= fifo->f2_count)) + return 1; + memcpy(*outbuf, fifo->f2[ptr], 2); + *outbuf += 2; + return 0; +} + +static void sw842_copy_to_fifo(const char *buf, struct sw842_fifo *fifo) +{ + unsigned char initial_f2count = fifo->f2_count; + + memcpy(fifo->f8[fifo->f8_count], buf, 8); + fifo->f4_count += 2; + fifo->f8_count += 1; + + if (!fifo->f84_full && fifo->f4_count >= 512) { + fifo->f84_full = 1; + fifo->f4_count /= 512; + } + + memcpy(fifo->f2[fifo->f2_count++], buf, 2); + memcpy(fifo->f2[fifo->f2_count++], buf + 2, 2); + memcpy(fifo->f2[fifo->f2_count++], buf + 4, 2); + memcpy(fifo->f2[fifo->f2_count++], buf + 6, 2); + if (fifo->f2_count < initial_f2count) + fifo->f2_full = 1; +} + +static int sw842_decompress(const unsigned char *src, int srclen, + unsigned char *dst, int *destlen, + const void *wrkmem) +{ + uint8_t tmpl; + const char *inbuf; + int inbit = 0; + unsigned char *outbuf, *outbuf_end, *origbuf, *prevbuf; + const char *inbuf_end; + sw842_template_op op; + int opindex; + int i, repeat_count; + struct sw842_fifo *fifo; + int ret = 0; + + fifo = &((struct nx842_workmem *)(wrkmem))->swfifo; + memset(fifo, 0, sizeof(*fifo)); + + origbuf = NULL; + inbuf = src; + inbuf_end = src + srclen; + outbuf = dst; + outbuf_end = dst + *destlen; + + while ((tmpl = sw842_get_template(&inbuf, &inbit)) != SW842_TMPL_EOF) { + if (inbuf >= inbuf_end) { + ret = -EINVAL; + goto out; + } + + opindex = 0; + prevbuf = origbuf; + origbuf = outbuf; + switch (tmpl) { + case SW842_TMPL_REPEAT: + if (prevbuf == NULL) { + ret = -EINVAL; + goto out; + } + + repeat_count = sw842_get_repeat_count(&inbuf, + &inbit) + 1; + + /* Did the repeat count advance past the end of input */ + if (inbuf > inbuf_end) { + ret = -EINVAL; + goto out; + } + + for (i = 0; i < repeat_count; i++) { + /* Would this overflow the output buffer */ + if ((outbuf + 8) > outbuf_end) { + ret = -ENOSPC; + goto out; + } + + memcpy(outbuf, prevbuf, 8); + sw842_copy_to_fifo(outbuf, fifo); + outbuf += 8; + } + break; + + case SW842_TMPL_ZEROS: + /* Would this overflow the output buffer */ + if ((outbuf + 8) > outbuf_end) { + ret = -ENOSPC; + goto out; + } + + memset(outbuf, 0, 8); + sw842_copy_to_fifo(outbuf, fifo); + outbuf += 8; + break; + + default: + if (tmpl > 25) { + ret = -EINVAL; + goto out; + } + + /* Does this go past the end of the input buffer */ + if ((inbuf + 2) > inbuf_end) { + ret = -EINVAL; + goto out; + } + + /* Would this overflow the output buffer */ + if ((outbuf + 8) > outbuf_end) { + ret = -ENOSPC; + goto out; + } + + while (opindex < 4 && + (op = sw842_tmpl_ops[tmpl][opindex++]) + != NULL) { + ret = (*op)(&inbuf, &inbit, &outbuf, fifo); + if (ret) { + ret = -EINVAL; + goto out; + } + sw842_copy_to_fifo(origbuf, fifo); + } + } + } + +out: + if (!ret) + *destlen = (unsigned int)(outbuf - dst); + else + *destlen = 0; + + return ret; +} diff --git a/drivers/crypto/nx/nx-aes-cbc.c b/drivers/crypto/nx/nx-aes-cbc.c index 69ed796ee32..a76d4c4f29f 100644 --- a/drivers/crypto/nx/nx-aes-cbc.c +++ b/drivers/crypto/nx/nx-aes-cbc.c @@ -127,7 +127,6 @@ struct crypto_alg nx_cbc_aes_alg = { .cra_ctxsize = sizeof(struct nx_crypto_ctx), .cra_type = &crypto_blkcipher_type, .cra_module = THIS_MODULE, - .cra_list = LIST_HEAD_INIT(nx_cbc_aes_alg.cra_list), .cra_init = nx_crypto_ctx_aes_cbc_init, .cra_exit = nx_crypto_ctx_exit, .cra_blkcipher = { diff --git a/drivers/crypto/nx/nx-aes-ccm.c b/drivers/crypto/nx/nx-aes-ccm.c index 7aeac678b9c..ef5eae6d140 100644 --- a/drivers/crypto/nx/nx-aes-ccm.c +++ b/drivers/crypto/nx/nx-aes-ccm.c @@ -430,7 +430,6 @@ struct crypto_alg nx_ccm_aes_alg = { .cra_ctxsize = sizeof(struct nx_crypto_ctx), .cra_type = &crypto_aead_type, .cra_module = THIS_MODULE, - .cra_list = LIST_HEAD_INIT(nx_ccm_aes_alg.cra_list), .cra_init = nx_crypto_ctx_aes_ccm_init, .cra_exit = nx_crypto_ctx_exit, .cra_aead = { @@ -453,7 +452,6 @@ struct crypto_alg nx_ccm4309_aes_alg = { .cra_ctxsize = sizeof(struct nx_crypto_ctx), .cra_type = &crypto_nivaead_type, .cra_module = THIS_MODULE, - .cra_list = LIST_HEAD_INIT(nx_ccm4309_aes_alg.cra_list), .cra_init = nx_crypto_ctx_aes_ccm_init, .cra_exit = nx_crypto_ctx_exit, .cra_aead = { diff --git a/drivers/crypto/nx/nx-aes-ctr.c b/drivers/crypto/nx/nx-aes-ctr.c index 52d4eb05e8f..b6286f14680 100644 --- a/drivers/crypto/nx/nx-aes-ctr.c +++ b/drivers/crypto/nx/nx-aes-ctr.c @@ -141,7 +141,6 @@ struct crypto_alg nx_ctr_aes_alg = { .cra_ctxsize = sizeof(struct nx_crypto_ctx), .cra_type = &crypto_blkcipher_type, .cra_module = THIS_MODULE, - .cra_list = LIST_HEAD_INIT(nx_ctr_aes_alg.cra_list), .cra_init = nx_crypto_ctx_aes_ctr_init, .cra_exit = nx_crypto_ctx_exit, .cra_blkcipher = { @@ -163,7 +162,6 @@ struct crypto_alg nx_ctr3686_aes_alg = { .cra_ctxsize = sizeof(struct nx_crypto_ctx), .cra_type = &crypto_blkcipher_type, .cra_module = THIS_MODULE, - .cra_list = LIST_HEAD_INIT(nx_ctr3686_aes_alg.cra_list), .cra_init = nx_crypto_ctx_aes_ctr_init, .cra_exit = nx_crypto_ctx_exit, .cra_blkcipher = { diff --git a/drivers/crypto/nx/nx-aes-ecb.c b/drivers/crypto/nx/nx-aes-ecb.c index 7b77bc2d1df..ba5f1611336 100644 --- a/drivers/crypto/nx/nx-aes-ecb.c +++ b/drivers/crypto/nx/nx-aes-ecb.c @@ -126,7 +126,6 @@ struct crypto_alg nx_ecb_aes_alg = { .cra_ctxsize = sizeof(struct nx_crypto_ctx), .cra_type = &crypto_blkcipher_type, .cra_module = THIS_MODULE, - .cra_list = LIST_HEAD_INIT(nx_ecb_aes_alg.cra_list), .cra_init = nx_crypto_ctx_aes_ecb_init, .cra_exit = nx_crypto_ctx_exit, .cra_blkcipher = { diff --git a/drivers/crypto/nx/nx-aes-gcm.c b/drivers/crypto/nx/nx-aes-gcm.c index 9ab1c7341da..c8109edc5cf 100644 --- a/drivers/crypto/nx/nx-aes-gcm.c +++ b/drivers/crypto/nx/nx-aes-gcm.c @@ -316,7 +316,6 @@ struct crypto_alg nx_gcm_aes_alg = { .cra_ctxsize = sizeof(struct nx_crypto_ctx), .cra_type = &crypto_aead_type, .cra_module = THIS_MODULE, - .cra_list = LIST_HEAD_INIT(nx_gcm_aes_alg.cra_list), .cra_init = nx_crypto_ctx_aes_gcm_init, .cra_exit = nx_crypto_ctx_exit, .cra_aead = { @@ -338,7 +337,6 @@ struct crypto_alg nx_gcm4106_aes_alg = { .cra_ctxsize = sizeof(struct nx_crypto_ctx), .cra_type = &crypto_nivaead_type, .cra_module = THIS_MODULE, - .cra_list = LIST_HEAD_INIT(nx_gcm4106_aes_alg.cra_list), .cra_init = nx_crypto_ctx_aes_gcm_init, .cra_exit = nx_crypto_ctx_exit, .cra_aead = { diff --git a/drivers/crypto/omap-aes.c b/drivers/crypto/omap-aes.c index 63e57b57a12..093a8af59cb 100644 --- a/drivers/crypto/omap-aes.c +++ b/drivers/crypto/omap-aes.c @@ -876,7 +876,6 @@ static int omap_aes_probe(struct platform_device *pdev) for (i = 0; i < ARRAY_SIZE(algs); i++) { pr_debug("i: %d\n", i); - INIT_LIST_HEAD(&algs[i].cra_list); err = crypto_register_alg(&algs[i]); if (err) goto err_algs; diff --git a/drivers/crypto/padlock-aes.c b/drivers/crypto/padlock-aes.c index 37b2e9406af..633ba945e15 100644 --- a/drivers/crypto/padlock-aes.c +++ b/drivers/crypto/padlock-aes.c @@ -328,7 +328,6 @@ static struct crypto_alg aes_alg = { .cra_ctxsize = sizeof(struct aes_ctx), .cra_alignmask = PADLOCK_ALIGNMENT - 1, .cra_module = THIS_MODULE, - .cra_list = LIST_HEAD_INIT(aes_alg.cra_list), .cra_u = { .cipher = { .cia_min_keysize = AES_MIN_KEY_SIZE, @@ -408,7 +407,6 @@ static struct crypto_alg ecb_aes_alg = { .cra_alignmask = PADLOCK_ALIGNMENT - 1, .cra_type = &crypto_blkcipher_type, .cra_module = THIS_MODULE, - .cra_list = LIST_HEAD_INIT(ecb_aes_alg.cra_list), .cra_u = { .blkcipher = { .min_keysize = AES_MIN_KEY_SIZE, @@ -491,7 +489,6 @@ static struct crypto_alg cbc_aes_alg = { .cra_alignmask = PADLOCK_ALIGNMENT - 1, .cra_type = &crypto_blkcipher_type, .cra_module = THIS_MODULE, - .cra_list = LIST_HEAD_INIT(cbc_aes_alg.cra_list), .cra_u = { .blkcipher = { .min_keysize = AES_MIN_KEY_SIZE, diff --git a/drivers/crypto/s5p-sss.c b/drivers/crypto/s5p-sss.c index bc986f80608..a22714412cd 100644 --- a/drivers/crypto/s5p-sss.c +++ b/drivers/crypto/s5p-sss.c @@ -626,7 +626,6 @@ static int s5p_aes_probe(struct platform_device *pdev) crypto_init_queue(&pdata->queue, CRYPTO_QUEUE_LEN); for (i = 0; i < ARRAY_SIZE(algs); i++) { - INIT_LIST_HEAD(&algs[i].cra_list); err = crypto_register_alg(&algs[i]); if (err) goto err_algs; diff --git a/drivers/crypto/talitos.c b/drivers/crypto/talitos.c index efff788d2f1..da1112765a4 100644 --- a/drivers/crypto/talitos.c +++ b/drivers/crypto/talitos.c @@ -38,6 +38,7 @@ #include <linux/spinlock.h> #include <linux/rtnetlink.h> #include <linux/slab.h> +#include <linux/string.h> #include <crypto/algapi.h> #include <crypto/aes.h> @@ -714,8 +715,13 @@ badkey: /* * talitos_edesc - s/w-extended descriptor + * @assoc_nents: number of segments in associated data scatterlist * @src_nents: number of segments in input scatterlist * @dst_nents: number of segments in output scatterlist + * @assoc_chained: whether assoc is chained or not + * @src_chained: whether src is chained or not + * @dst_chained: whether dst is chained or not + * @iv_dma: dma address of iv for checking continuity and link table * @dma_len: length of dma mapped link_tbl space * @dma_link_tbl: bus physical address of link_tbl * @desc: h/w descriptor @@ -726,10 +732,13 @@ badkey: * of link_tbl data */ struct talitos_edesc { + int assoc_nents; int src_nents; int dst_nents; - int src_is_chained; - int dst_is_chained; + bool assoc_chained; + bool src_chained; + bool dst_chained; + dma_addr_t iv_dma; int dma_len; dma_addr_t dma_link_tbl; struct talitos_desc desc; @@ -738,7 +747,7 @@ struct talitos_edesc { static int talitos_map_sg(struct device *dev, struct scatterlist *sg, unsigned int nents, enum dma_data_direction dir, - int chained) + bool chained) { if (unlikely(chained)) while (sg) { @@ -768,13 +777,13 @@ static void talitos_sg_unmap(struct device *dev, unsigned int dst_nents = edesc->dst_nents ? : 1; if (src != dst) { - if (edesc->src_is_chained) + if (edesc->src_chained) talitos_unmap_sg_chain(dev, src, DMA_TO_DEVICE); else dma_unmap_sg(dev, src, src_nents, DMA_TO_DEVICE); if (dst) { - if (edesc->dst_is_chained) + if (edesc->dst_chained) talitos_unmap_sg_chain(dev, dst, DMA_FROM_DEVICE); else @@ -782,7 +791,7 @@ static void talitos_sg_unmap(struct device *dev, DMA_FROM_DEVICE); } } else - if (edesc->src_is_chained) + if (edesc->src_chained) talitos_unmap_sg_chain(dev, src, DMA_BIDIRECTIONAL); else dma_unmap_sg(dev, src, src_nents, DMA_BIDIRECTIONAL); @@ -797,7 +806,13 @@ static void ipsec_esp_unmap(struct device *dev, unmap_single_talitos_ptr(dev, &edesc->desc.ptr[2], DMA_TO_DEVICE); unmap_single_talitos_ptr(dev, &edesc->desc.ptr[0], DMA_TO_DEVICE); - dma_unmap_sg(dev, areq->assoc, 1, DMA_TO_DEVICE); + if (edesc->assoc_chained) + talitos_unmap_sg_chain(dev, areq->assoc, DMA_TO_DEVICE); + else + /* assoc_nents counts also for IV in non-contiguous cases */ + dma_unmap_sg(dev, areq->assoc, + edesc->assoc_nents ? edesc->assoc_nents - 1 : 1, + DMA_TO_DEVICE); talitos_sg_unmap(dev, edesc, areq->src, areq->dst); @@ -825,9 +840,10 @@ static void ipsec_esp_encrypt_done(struct device *dev, ipsec_esp_unmap(dev, edesc, areq); /* copy the generated ICV to dst */ - if (edesc->dma_len) { + if (edesc->dst_nents) { icvdata = &edesc->link_tbl[edesc->src_nents + - edesc->dst_nents + 2]; + edesc->dst_nents + 2 + + edesc->assoc_nents]; sg = sg_last(areq->dst, edesc->dst_nents); memcpy((char *)sg_virt(sg) + sg->length - ctx->authsize, icvdata, ctx->authsize); @@ -857,7 +873,8 @@ static void ipsec_esp_decrypt_swauth_done(struct device *dev, /* auth check */ if (edesc->dma_len) icvdata = &edesc->link_tbl[edesc->src_nents + - edesc->dst_nents + 2]; + edesc->dst_nents + 2 + + edesc->assoc_nents]; else icvdata = &edesc->link_tbl[0]; @@ -932,10 +949,9 @@ static int sg_to_link_tbl(struct scatterlist *sg, int sg_count, * fill in and submit ipsec_esp descriptor */ static int ipsec_esp(struct talitos_edesc *edesc, struct aead_request *areq, - u8 *giv, u64 seq, - void (*callback) (struct device *dev, - struct talitos_desc *desc, - void *context, int error)) + u64 seq, void (*callback) (struct device *dev, + struct talitos_desc *desc, + void *context, int error)) { struct crypto_aead *aead = crypto_aead_reqtfm(areq); struct talitos_ctx *ctx = crypto_aead_ctx(aead); @@ -950,12 +966,42 @@ static int ipsec_esp(struct talitos_edesc *edesc, struct aead_request *areq, /* hmac key */ map_single_talitos_ptr(dev, &desc->ptr[0], ctx->authkeylen, &ctx->key, 0, DMA_TO_DEVICE); + /* hmac data */ - map_single_talitos_ptr(dev, &desc->ptr[1], areq->assoclen + ivsize, - sg_virt(areq->assoc), 0, DMA_TO_DEVICE); + desc->ptr[1].len = cpu_to_be16(areq->assoclen + ivsize); + if (edesc->assoc_nents) { + int tbl_off = edesc->src_nents + edesc->dst_nents + 2; + struct talitos_ptr *tbl_ptr = &edesc->link_tbl[tbl_off]; + + to_talitos_ptr(&desc->ptr[1], edesc->dma_link_tbl + tbl_off * + sizeof(struct talitos_ptr)); + desc->ptr[1].j_extent = DESC_PTR_LNKTBL_JUMP; + + /* assoc_nents - 1 entries for assoc, 1 for IV */ + sg_count = sg_to_link_tbl(areq->assoc, edesc->assoc_nents - 1, + areq->assoclen, tbl_ptr); + + /* add IV to link table */ + tbl_ptr += sg_count - 1; + tbl_ptr->j_extent = 0; + tbl_ptr++; + to_talitos_ptr(tbl_ptr, edesc->iv_dma); + tbl_ptr->len = cpu_to_be16(ivsize); + tbl_ptr->j_extent = DESC_PTR_LNKTBL_RETURN; + + dma_sync_single_for_device(dev, edesc->dma_link_tbl, + edesc->dma_len, DMA_BIDIRECTIONAL); + } else { + to_talitos_ptr(&desc->ptr[1], sg_dma_address(areq->assoc)); + desc->ptr[1].j_extent = 0; + } + /* cipher iv */ - map_single_talitos_ptr(dev, &desc->ptr[2], ivsize, giv ?: areq->iv, 0, - DMA_TO_DEVICE); + to_talitos_ptr(&desc->ptr[2], edesc->iv_dma); + desc->ptr[2].len = cpu_to_be16(ivsize); + desc->ptr[2].j_extent = 0; + /* Sync needed for the aead_givencrypt case */ + dma_sync_single_for_device(dev, edesc->iv_dma, ivsize, DMA_TO_DEVICE); /* cipher key */ map_single_talitos_ptr(dev, &desc->ptr[3], ctx->enckeylen, @@ -974,7 +1020,7 @@ static int ipsec_esp(struct talitos_edesc *edesc, struct aead_request *areq, sg_count = talitos_map_sg(dev, areq->src, edesc->src_nents ? : 1, (areq->src == areq->dst) ? DMA_BIDIRECTIONAL : DMA_TO_DEVICE, - edesc->src_is_chained); + edesc->src_chained); if (sg_count == 1) { to_talitos_ptr(&desc->ptr[4], sg_dma_address(areq->src)); @@ -1006,32 +1052,30 @@ static int ipsec_esp(struct talitos_edesc *edesc, struct aead_request *areq, if (areq->src != areq->dst) sg_count = talitos_map_sg(dev, areq->dst, edesc->dst_nents ? : 1, - DMA_FROM_DEVICE, - edesc->dst_is_chained); + DMA_FROM_DEVICE, edesc->dst_chained); if (sg_count == 1) { to_talitos_ptr(&desc->ptr[5], sg_dma_address(areq->dst)); } else { - struct talitos_ptr *link_tbl_ptr = - &edesc->link_tbl[edesc->src_nents + 1]; + int tbl_off = edesc->src_nents + 1; + struct talitos_ptr *tbl_ptr = &edesc->link_tbl[tbl_off]; to_talitos_ptr(&desc->ptr[5], edesc->dma_link_tbl + - (edesc->src_nents + 1) * - sizeof(struct talitos_ptr)); + tbl_off * sizeof(struct talitos_ptr)); sg_count = sg_to_link_tbl(areq->dst, sg_count, cryptlen, - link_tbl_ptr); + tbl_ptr); /* Add an entry to the link table for ICV data */ - link_tbl_ptr += sg_count - 1; - link_tbl_ptr->j_extent = 0; - sg_count++; - link_tbl_ptr++; - link_tbl_ptr->j_extent = DESC_PTR_LNKTBL_RETURN; - link_tbl_ptr->len = cpu_to_be16(authsize); + tbl_ptr += sg_count - 1; + tbl_ptr->j_extent = 0; + tbl_ptr++; + tbl_ptr->j_extent = DESC_PTR_LNKTBL_RETURN; + tbl_ptr->len = cpu_to_be16(authsize); /* icv data follows link tables */ - to_talitos_ptr(link_tbl_ptr, edesc->dma_link_tbl + - (edesc->src_nents + edesc->dst_nents + 2) * + to_talitos_ptr(tbl_ptr, edesc->dma_link_tbl + + (tbl_off + edesc->dst_nents + 1 + + edesc->assoc_nents) * sizeof(struct talitos_ptr)); desc->ptr[5].j_extent |= DESC_PTR_LNKTBL_JUMP; dma_sync_single_for_device(ctx->dev, edesc->dma_link_tbl, @@ -1053,17 +1097,17 @@ static int ipsec_esp(struct talitos_edesc *edesc, struct aead_request *areq, /* * derive number of elements in scatterlist */ -static int sg_count(struct scatterlist *sg_list, int nbytes, int *chained) +static int sg_count(struct scatterlist *sg_list, int nbytes, bool *chained) { struct scatterlist *sg = sg_list; int sg_nents = 0; - *chained = 0; + *chained = false; while (nbytes > 0) { sg_nents++; nbytes -= sg->length; if (!sg_is_last(sg) && (sg + 1)->length == 0) - *chained = 1; + *chained = true; sg = scatterwalk_sg_next(sg); } @@ -1132,17 +1176,21 @@ static size_t sg_copy_end_to_buffer(struct scatterlist *sgl, unsigned int nents, * allocate and map the extended descriptor */ static struct talitos_edesc *talitos_edesc_alloc(struct device *dev, + struct scatterlist *assoc, struct scatterlist *src, struct scatterlist *dst, - int hash_result, + u8 *iv, + unsigned int assoclen, unsigned int cryptlen, unsigned int authsize, + unsigned int ivsize, int icv_stashing, u32 cryptoflags) { struct talitos_edesc *edesc; - int src_nents, dst_nents, alloc_len, dma_len; - int src_chained, dst_chained = 0; + int assoc_nents = 0, src_nents, dst_nents, alloc_len, dma_len; + bool assoc_chained = false, src_chained = false, dst_chained = false; + dma_addr_t iv_dma = 0; gfp_t flags = cryptoflags & CRYPTO_TFM_REQ_MAY_SLEEP ? GFP_KERNEL : GFP_ATOMIC; @@ -1151,10 +1199,29 @@ static struct talitos_edesc *talitos_edesc_alloc(struct device *dev, return ERR_PTR(-EINVAL); } + if (iv) + iv_dma = dma_map_single(dev, iv, ivsize, DMA_TO_DEVICE); + + if (assoc) { + /* + * Currently it is assumed that iv is provided whenever assoc + * is. + */ + BUG_ON(!iv); + + assoc_nents = sg_count(assoc, assoclen, &assoc_chained); + talitos_map_sg(dev, assoc, assoc_nents, DMA_TO_DEVICE, + assoc_chained); + assoc_nents = (assoc_nents == 1) ? 0 : assoc_nents; + + if (assoc_nents || sg_dma_address(assoc) + assoclen != iv_dma) + assoc_nents = assoc_nents ? assoc_nents + 1 : 2; + } + src_nents = sg_count(src, cryptlen + authsize, &src_chained); src_nents = (src_nents == 1) ? 0 : src_nents; - if (hash_result) { + if (!dst) { dst_nents = 0; } else { if (dst == src) { @@ -1172,9 +1239,9 @@ static struct talitos_edesc *talitos_edesc_alloc(struct device *dev, * and the ICV data itself */ alloc_len = sizeof(struct talitos_edesc); - if (src_nents || dst_nents) { - dma_len = (src_nents + dst_nents + 2) * - sizeof(struct talitos_ptr) + authsize; + if (assoc_nents || src_nents || dst_nents) { + dma_len = (src_nents + dst_nents + 2 + assoc_nents) * + sizeof(struct talitos_ptr) + authsize; alloc_len += dma_len; } else { dma_len = 0; @@ -1183,14 +1250,20 @@ static struct talitos_edesc *talitos_edesc_alloc(struct device *dev, edesc = kmalloc(alloc_len, GFP_DMA | flags); if (!edesc) { + talitos_unmap_sg_chain(dev, assoc, DMA_TO_DEVICE); + if (iv_dma) + dma_unmap_single(dev, iv_dma, ivsize, DMA_TO_DEVICE); dev_err(dev, "could not allocate edescriptor\n"); return ERR_PTR(-ENOMEM); } + edesc->assoc_nents = assoc_nents; edesc->src_nents = src_nents; edesc->dst_nents = dst_nents; - edesc->src_is_chained = src_chained; - edesc->dst_is_chained = dst_chained; + edesc->assoc_chained = assoc_chained; + edesc->src_chained = src_chained; + edesc->dst_chained = dst_chained; + edesc->iv_dma = iv_dma; edesc->dma_len = dma_len; if (dma_len) edesc->dma_link_tbl = dma_map_single(dev, &edesc->link_tbl[0], @@ -1200,14 +1273,16 @@ static struct talitos_edesc *talitos_edesc_alloc(struct device *dev, return edesc; } -static struct talitos_edesc *aead_edesc_alloc(struct aead_request *areq, +static struct talitos_edesc *aead_edesc_alloc(struct aead_request *areq, u8 *iv, int icv_stashing) { struct crypto_aead *authenc = crypto_aead_reqtfm(areq); struct talitos_ctx *ctx = crypto_aead_ctx(authenc); + unsigned int ivsize = crypto_aead_ivsize(authenc); - return talitos_edesc_alloc(ctx->dev, areq->src, areq->dst, 0, - areq->cryptlen, ctx->authsize, icv_stashing, + return talitos_edesc_alloc(ctx->dev, areq->assoc, areq->src, areq->dst, + iv, areq->assoclen, areq->cryptlen, + ctx->authsize, ivsize, icv_stashing, areq->base.flags); } @@ -1218,14 +1293,14 @@ static int aead_encrypt(struct aead_request *req) struct talitos_edesc *edesc; /* allocate extended descriptor */ - edesc = aead_edesc_alloc(req, 0); + edesc = aead_edesc_alloc(req, req->iv, 0); if (IS_ERR(edesc)) return PTR_ERR(edesc); /* set encrypt */ edesc->desc.hdr = ctx->desc_hdr_template | DESC_HDR_MODE0_ENCRYPT; - return ipsec_esp(edesc, req, NULL, 0, ipsec_esp_encrypt_done); + return ipsec_esp(edesc, req, 0, ipsec_esp_encrypt_done); } static int aead_decrypt(struct aead_request *req) @@ -1241,7 +1316,7 @@ static int aead_decrypt(struct aead_request *req) req->cryptlen -= authsize; /* allocate extended descriptor */ - edesc = aead_edesc_alloc(req, 1); + edesc = aead_edesc_alloc(req, req->iv, 1); if (IS_ERR(edesc)) return PTR_ERR(edesc); @@ -1257,9 +1332,7 @@ static int aead_decrypt(struct aead_request *req) /* reset integrity check result bits */ edesc->desc.hdr_lo = 0; - return ipsec_esp(edesc, req, NULL, 0, - ipsec_esp_decrypt_hwauth_done); - + return ipsec_esp(edesc, req, 0, ipsec_esp_decrypt_hwauth_done); } /* Have to check the ICV with software */ @@ -1268,7 +1341,8 @@ static int aead_decrypt(struct aead_request *req) /* stash incoming ICV for later cmp with ICV generated by the h/w */ if (edesc->dma_len) icvdata = &edesc->link_tbl[edesc->src_nents + - edesc->dst_nents + 2]; + edesc->dst_nents + 2 + + edesc->assoc_nents]; else icvdata = &edesc->link_tbl[0]; @@ -1277,7 +1351,7 @@ static int aead_decrypt(struct aead_request *req) memcpy(icvdata, (char *)sg_virt(sg) + sg->length - ctx->authsize, ctx->authsize); - return ipsec_esp(edesc, req, NULL, 0, ipsec_esp_decrypt_swauth_done); + return ipsec_esp(edesc, req, 0, ipsec_esp_decrypt_swauth_done); } static int aead_givencrypt(struct aead_givcrypt_request *req) @@ -1288,7 +1362,7 @@ static int aead_givencrypt(struct aead_givcrypt_request *req) struct talitos_edesc *edesc; /* allocate extended descriptor */ - edesc = aead_edesc_alloc(areq, 0); + edesc = aead_edesc_alloc(areq, req->giv, 0); if (IS_ERR(edesc)) return PTR_ERR(edesc); @@ -1299,8 +1373,7 @@ static int aead_givencrypt(struct aead_givcrypt_request *req) /* avoid consecutive packets going out with same IV */ *(__be64 *)req->giv ^= cpu_to_be64(req->seq); - return ipsec_esp(edesc, areq, req->giv, req->seq, - ipsec_esp_encrypt_done); + return ipsec_esp(edesc, areq, req->seq, ipsec_esp_encrypt_done); } static int ablkcipher_setkey(struct crypto_ablkcipher *cipher, @@ -1356,7 +1429,7 @@ static int common_nonsnoop(struct talitos_edesc *edesc, struct device *dev = ctx->dev; struct talitos_desc *desc = &edesc->desc; unsigned int cryptlen = areq->nbytes; - unsigned int ivsize; + unsigned int ivsize = crypto_ablkcipher_ivsize(cipher); int sg_count, ret; /* first DWORD empty */ @@ -1365,9 +1438,9 @@ static int common_nonsnoop(struct talitos_edesc *edesc, desc->ptr[0].j_extent = 0; /* cipher iv */ - ivsize = crypto_ablkcipher_ivsize(cipher); - map_single_talitos_ptr(dev, &desc->ptr[1], ivsize, areq->info, 0, - DMA_TO_DEVICE); + to_talitos_ptr(&desc->ptr[1], edesc->iv_dma); + desc->ptr[1].len = cpu_to_be16(ivsize); + desc->ptr[1].j_extent = 0; /* cipher key */ map_single_talitos_ptr(dev, &desc->ptr[2], ctx->keylen, @@ -1382,7 +1455,7 @@ static int common_nonsnoop(struct talitos_edesc *edesc, sg_count = talitos_map_sg(dev, areq->src, edesc->src_nents ? : 1, (areq->src == areq->dst) ? DMA_BIDIRECTIONAL : DMA_TO_DEVICE, - edesc->src_is_chained); + edesc->src_chained); if (sg_count == 1) { to_talitos_ptr(&desc->ptr[3], sg_dma_address(areq->src)); @@ -1409,8 +1482,7 @@ static int common_nonsnoop(struct talitos_edesc *edesc, if (areq->src != areq->dst) sg_count = talitos_map_sg(dev, areq->dst, edesc->dst_nents ? : 1, - DMA_FROM_DEVICE, - edesc->dst_is_chained); + DMA_FROM_DEVICE, edesc->dst_chained); if (sg_count == 1) { to_talitos_ptr(&desc->ptr[4], sg_dma_address(areq->dst)); @@ -1450,9 +1522,11 @@ static struct talitos_edesc *ablkcipher_edesc_alloc(struct ablkcipher_request * { struct crypto_ablkcipher *cipher = crypto_ablkcipher_reqtfm(areq); struct talitos_ctx *ctx = crypto_ablkcipher_ctx(cipher); + unsigned int ivsize = crypto_ablkcipher_ivsize(cipher); - return talitos_edesc_alloc(ctx->dev, areq->src, areq->dst, 0, - areq->nbytes, 0, 0, areq->base.flags); + return talitos_edesc_alloc(ctx->dev, NULL, areq->src, areq->dst, + areq->info, 0, areq->nbytes, 0, ivsize, 0, + areq->base.flags); } static int ablkcipher_encrypt(struct ablkcipher_request *areq) @@ -1578,8 +1652,7 @@ static int common_nonsnoop_hash(struct talitos_edesc *edesc, sg_count = talitos_map_sg(dev, req_ctx->psrc, edesc->src_nents ? : 1, - DMA_TO_DEVICE, - edesc->src_is_chained); + DMA_TO_DEVICE, edesc->src_chained); if (sg_count == 1) { to_talitos_ptr(&desc->ptr[3], sg_dma_address(req_ctx->psrc)); @@ -1631,8 +1704,8 @@ static struct talitos_edesc *ahash_edesc_alloc(struct ahash_request *areq, struct talitos_ctx *ctx = crypto_ahash_ctx(tfm); struct talitos_ahash_req_ctx *req_ctx = ahash_request_ctx(areq); - return talitos_edesc_alloc(ctx->dev, req_ctx->psrc, NULL, 1, - nbytes, 0, 0, areq->base.flags); + return talitos_edesc_alloc(ctx->dev, NULL, req_ctx->psrc, NULL, NULL, 0, + nbytes, 0, 0, 0, areq->base.flags); } static int ahash_init(struct ahash_request *areq) @@ -1690,7 +1763,7 @@ static int ahash_process_req(struct ahash_request *areq, unsigned int nbytes) unsigned int nbytes_to_hash; unsigned int to_hash_later; unsigned int nsg; - int chained; + bool chained; if (!req_ctx->last && (nbytes + req_ctx->nbuf <= blocksize)) { /* Buffer up to one whole block */ @@ -1902,21 +1975,18 @@ struct talitos_alg_template { }; static struct talitos_alg_template driver_algs[] = { - /* AEAD algorithms. These use a single-pass ipsec_esp descriptor */ + /* + * AEAD algorithms. These use a single-pass ipsec_esp descriptor. + * authencesn(*,*) is also registered, although not present + * explicitly here. + */ { .type = CRYPTO_ALG_TYPE_AEAD, .alg.crypto = { .cra_name = "authenc(hmac(sha1),cbc(aes))", .cra_driver_name = "authenc-hmac-sha1-cbc-aes-talitos", .cra_blocksize = AES_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AEAD | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_aead_type, .cra_aead = { - .setkey = aead_setkey, - .setauthsize = aead_setauthsize, - .encrypt = aead_encrypt, - .decrypt = aead_decrypt, - .givencrypt = aead_givencrypt, - .geniv = "<built-in>", .ivsize = AES_BLOCK_SIZE, .maxauthsize = SHA1_DIGEST_SIZE, } @@ -1935,14 +2005,7 @@ static struct talitos_alg_template driver_algs[] = { .cra_driver_name = "authenc-hmac-sha1-cbc-3des-talitos", .cra_blocksize = DES3_EDE_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AEAD | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_aead_type, .cra_aead = { - .setkey = aead_setkey, - .setauthsize = aead_setauthsize, - .encrypt = aead_encrypt, - .decrypt = aead_decrypt, - .givencrypt = aead_givencrypt, - .geniv = "<built-in>", .ivsize = DES3_EDE_BLOCK_SIZE, .maxauthsize = SHA1_DIGEST_SIZE, } @@ -1962,14 +2025,7 @@ static struct talitos_alg_template driver_algs[] = { .cra_driver_name = "authenc-hmac-sha224-cbc-aes-talitos", .cra_blocksize = AES_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AEAD | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_aead_type, .cra_aead = { - .setkey = aead_setkey, - .setauthsize = aead_setauthsize, - .encrypt = aead_encrypt, - .decrypt = aead_decrypt, - .givencrypt = aead_givencrypt, - .geniv = "<built-in>", .ivsize = AES_BLOCK_SIZE, .maxauthsize = SHA224_DIGEST_SIZE, } @@ -1988,14 +2044,7 @@ static struct talitos_alg_template driver_algs[] = { .cra_driver_name = "authenc-hmac-sha224-cbc-3des-talitos", .cra_blocksize = DES3_EDE_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AEAD | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_aead_type, .cra_aead = { - .setkey = aead_setkey, - .setauthsize = aead_setauthsize, - .encrypt = aead_encrypt, - .decrypt = aead_decrypt, - .givencrypt = aead_givencrypt, - .geniv = "<built-in>", .ivsize = DES3_EDE_BLOCK_SIZE, .maxauthsize = SHA224_DIGEST_SIZE, } @@ -2015,14 +2064,7 @@ static struct talitos_alg_template driver_algs[] = { .cra_driver_name = "authenc-hmac-sha256-cbc-aes-talitos", .cra_blocksize = AES_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AEAD | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_aead_type, .cra_aead = { - .setkey = aead_setkey, - .setauthsize = aead_setauthsize, - .encrypt = aead_encrypt, - .decrypt = aead_decrypt, - .givencrypt = aead_givencrypt, - .geniv = "<built-in>", .ivsize = AES_BLOCK_SIZE, .maxauthsize = SHA256_DIGEST_SIZE, } @@ -2041,14 +2083,7 @@ static struct talitos_alg_template driver_algs[] = { .cra_driver_name = "authenc-hmac-sha256-cbc-3des-talitos", .cra_blocksize = DES3_EDE_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AEAD | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_aead_type, .cra_aead = { - .setkey = aead_setkey, - .setauthsize = aead_setauthsize, - .encrypt = aead_encrypt, - .decrypt = aead_decrypt, - .givencrypt = aead_givencrypt, - .geniv = "<built-in>", .ivsize = DES3_EDE_BLOCK_SIZE, .maxauthsize = SHA256_DIGEST_SIZE, } @@ -2068,14 +2103,7 @@ static struct talitos_alg_template driver_algs[] = { .cra_driver_name = "authenc-hmac-sha384-cbc-aes-talitos", .cra_blocksize = AES_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AEAD | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_aead_type, .cra_aead = { - .setkey = aead_setkey, - .setauthsize = aead_setauthsize, - .encrypt = aead_encrypt, - .decrypt = aead_decrypt, - .givencrypt = aead_givencrypt, - .geniv = "<built-in>", .ivsize = AES_BLOCK_SIZE, .maxauthsize = SHA384_DIGEST_SIZE, } @@ -2094,14 +2122,7 @@ static struct talitos_alg_template driver_algs[] = { .cra_driver_name = "authenc-hmac-sha384-cbc-3des-talitos", .cra_blocksize = DES3_EDE_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AEAD | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_aead_type, .cra_aead = { - .setkey = aead_setkey, - .setauthsize = aead_setauthsize, - .encrypt = aead_encrypt, - .decrypt = aead_decrypt, - .givencrypt = aead_givencrypt, - .geniv = "<built-in>", .ivsize = DES3_EDE_BLOCK_SIZE, .maxauthsize = SHA384_DIGEST_SIZE, } @@ -2121,14 +2142,7 @@ static struct talitos_alg_template driver_algs[] = { .cra_driver_name = "authenc-hmac-sha512-cbc-aes-talitos", .cra_blocksize = AES_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AEAD | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_aead_type, .cra_aead = { - .setkey = aead_setkey, - .setauthsize = aead_setauthsize, - .encrypt = aead_encrypt, - .decrypt = aead_decrypt, - .givencrypt = aead_givencrypt, - .geniv = "<built-in>", .ivsize = AES_BLOCK_SIZE, .maxauthsize = SHA512_DIGEST_SIZE, } @@ -2147,14 +2161,7 @@ static struct talitos_alg_template driver_algs[] = { .cra_driver_name = "authenc-hmac-sha512-cbc-3des-talitos", .cra_blocksize = DES3_EDE_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AEAD | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_aead_type, .cra_aead = { - .setkey = aead_setkey, - .setauthsize = aead_setauthsize, - .encrypt = aead_encrypt, - .decrypt = aead_decrypt, - .givencrypt = aead_givencrypt, - .geniv = "<built-in>", .ivsize = DES3_EDE_BLOCK_SIZE, .maxauthsize = SHA512_DIGEST_SIZE, } @@ -2174,14 +2181,7 @@ static struct talitos_alg_template driver_algs[] = { .cra_driver_name = "authenc-hmac-md5-cbc-aes-talitos", .cra_blocksize = AES_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AEAD | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_aead_type, .cra_aead = { - .setkey = aead_setkey, - .setauthsize = aead_setauthsize, - .encrypt = aead_encrypt, - .decrypt = aead_decrypt, - .givencrypt = aead_givencrypt, - .geniv = "<built-in>", .ivsize = AES_BLOCK_SIZE, .maxauthsize = MD5_DIGEST_SIZE, } @@ -2200,14 +2200,7 @@ static struct talitos_alg_template driver_algs[] = { .cra_driver_name = "authenc-hmac-md5-cbc-3des-talitos", .cra_blocksize = DES3_EDE_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AEAD | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_aead_type, .cra_aead = { - .setkey = aead_setkey, - .setauthsize = aead_setauthsize, - .encrypt = aead_encrypt, - .decrypt = aead_decrypt, - .givencrypt = aead_givencrypt, - .geniv = "<built-in>", .ivsize = DES3_EDE_BLOCK_SIZE, .maxauthsize = MD5_DIGEST_SIZE, } @@ -2229,12 +2222,7 @@ static struct talitos_alg_template driver_algs[] = { .cra_blocksize = AES_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_ablkcipher_type, .cra_ablkcipher = { - .setkey = ablkcipher_setkey, - .encrypt = ablkcipher_encrypt, - .decrypt = ablkcipher_decrypt, - .geniv = "eseqiv", .min_keysize = AES_MIN_KEY_SIZE, .max_keysize = AES_MAX_KEY_SIZE, .ivsize = AES_BLOCK_SIZE, @@ -2251,12 +2239,7 @@ static struct talitos_alg_template driver_algs[] = { .cra_blocksize = DES3_EDE_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_ablkcipher_type, .cra_ablkcipher = { - .setkey = ablkcipher_setkey, - .encrypt = ablkcipher_encrypt, - .decrypt = ablkcipher_decrypt, - .geniv = "eseqiv", .min_keysize = DES3_EDE_KEY_SIZE, .max_keysize = DES3_EDE_KEY_SIZE, .ivsize = DES3_EDE_BLOCK_SIZE, @@ -2270,11 +2253,6 @@ static struct talitos_alg_template driver_algs[] = { /* AHASH algorithms. */ { .type = CRYPTO_ALG_TYPE_AHASH, .alg.hash = { - .init = ahash_init, - .update = ahash_update, - .final = ahash_final, - .finup = ahash_finup, - .digest = ahash_digest, .halg.digestsize = MD5_DIGEST_SIZE, .halg.base = { .cra_name = "md5", @@ -2282,7 +2260,6 @@ static struct talitos_alg_template driver_algs[] = { .cra_blocksize = MD5_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AHASH | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_ahash_type } }, .desc_hdr_template = DESC_HDR_TYPE_COMMON_NONSNOOP_NO_AFEU | @@ -2291,11 +2268,6 @@ static struct talitos_alg_template driver_algs[] = { }, { .type = CRYPTO_ALG_TYPE_AHASH, .alg.hash = { - .init = ahash_init, - .update = ahash_update, - .final = ahash_final, - .finup = ahash_finup, - .digest = ahash_digest, .halg.digestsize = SHA1_DIGEST_SIZE, .halg.base = { .cra_name = "sha1", @@ -2303,7 +2275,6 @@ static struct talitos_alg_template driver_algs[] = { .cra_blocksize = SHA1_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AHASH | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_ahash_type } }, .desc_hdr_template = DESC_HDR_TYPE_COMMON_NONSNOOP_NO_AFEU | @@ -2312,11 +2283,6 @@ static struct talitos_alg_template driver_algs[] = { }, { .type = CRYPTO_ALG_TYPE_AHASH, .alg.hash = { - .init = ahash_init, - .update = ahash_update, - .final = ahash_final, - .finup = ahash_finup, - .digest = ahash_digest, .halg.digestsize = SHA224_DIGEST_SIZE, .halg.base = { .cra_name = "sha224", @@ -2324,7 +2290,6 @@ static struct talitos_alg_template driver_algs[] = { .cra_blocksize = SHA224_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AHASH | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_ahash_type } }, .desc_hdr_template = DESC_HDR_TYPE_COMMON_NONSNOOP_NO_AFEU | @@ -2333,11 +2298,6 @@ static struct talitos_alg_template driver_algs[] = { }, { .type = CRYPTO_ALG_TYPE_AHASH, .alg.hash = { - .init = ahash_init, - .update = ahash_update, - .final = ahash_final, - .finup = ahash_finup, - .digest = ahash_digest, .halg.digestsize = SHA256_DIGEST_SIZE, .halg.base = { .cra_name = "sha256", @@ -2345,7 +2305,6 @@ static struct talitos_alg_template driver_algs[] = { .cra_blocksize = SHA256_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AHASH | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_ahash_type } }, .desc_hdr_template = DESC_HDR_TYPE_COMMON_NONSNOOP_NO_AFEU | @@ -2354,11 +2313,6 @@ static struct talitos_alg_template driver_algs[] = { }, { .type = CRYPTO_ALG_TYPE_AHASH, .alg.hash = { - .init = ahash_init, - .update = ahash_update, - .final = ahash_final, - .finup = ahash_finup, - .digest = ahash_digest, .halg.digestsize = SHA384_DIGEST_SIZE, .halg.base = { .cra_name = "sha384", @@ -2366,7 +2320,6 @@ static struct talitos_alg_template driver_algs[] = { .cra_blocksize = SHA384_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AHASH | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_ahash_type } }, .desc_hdr_template = DESC_HDR_TYPE_COMMON_NONSNOOP_NO_AFEU | @@ -2375,11 +2328,6 @@ static struct talitos_alg_template driver_algs[] = { }, { .type = CRYPTO_ALG_TYPE_AHASH, .alg.hash = { - .init = ahash_init, - .update = ahash_update, - .final = ahash_final, - .finup = ahash_finup, - .digest = ahash_digest, .halg.digestsize = SHA512_DIGEST_SIZE, .halg.base = { .cra_name = "sha512", @@ -2387,7 +2335,6 @@ static struct talitos_alg_template driver_algs[] = { .cra_blocksize = SHA512_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AHASH | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_ahash_type } }, .desc_hdr_template = DESC_HDR_TYPE_COMMON_NONSNOOP_NO_AFEU | @@ -2396,12 +2343,6 @@ static struct talitos_alg_template driver_algs[] = { }, { .type = CRYPTO_ALG_TYPE_AHASH, .alg.hash = { - .init = ahash_init, - .update = ahash_update, - .final = ahash_final, - .finup = ahash_finup, - .digest = ahash_digest, - .setkey = ahash_setkey, .halg.digestsize = MD5_DIGEST_SIZE, .halg.base = { .cra_name = "hmac(md5)", @@ -2409,7 +2350,6 @@ static struct talitos_alg_template driver_algs[] = { .cra_blocksize = MD5_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AHASH | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_ahash_type } }, .desc_hdr_template = DESC_HDR_TYPE_COMMON_NONSNOOP_NO_AFEU | @@ -2418,12 +2358,6 @@ static struct talitos_alg_template driver_algs[] = { }, { .type = CRYPTO_ALG_TYPE_AHASH, .alg.hash = { - .init = ahash_init, - .update = ahash_update, - .final = ahash_final, - .finup = ahash_finup, - .digest = ahash_digest, - .setkey = ahash_setkey, .halg.digestsize = SHA1_DIGEST_SIZE, .halg.base = { .cra_name = "hmac(sha1)", @@ -2431,7 +2365,6 @@ static struct talitos_alg_template driver_algs[] = { .cra_blocksize = SHA1_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AHASH | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_ahash_type } }, .desc_hdr_template = DESC_HDR_TYPE_COMMON_NONSNOOP_NO_AFEU | @@ -2440,12 +2373,6 @@ static struct talitos_alg_template driver_algs[] = { }, { .type = CRYPTO_ALG_TYPE_AHASH, .alg.hash = { - .init = ahash_init, - .update = ahash_update, - .final = ahash_final, - .finup = ahash_finup, - .digest = ahash_digest, - .setkey = ahash_setkey, .halg.digestsize = SHA224_DIGEST_SIZE, .halg.base = { .cra_name = "hmac(sha224)", @@ -2453,7 +2380,6 @@ static struct talitos_alg_template driver_algs[] = { .cra_blocksize = SHA224_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AHASH | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_ahash_type } }, .desc_hdr_template = DESC_HDR_TYPE_COMMON_NONSNOOP_NO_AFEU | @@ -2462,12 +2388,6 @@ static struct talitos_alg_template driver_algs[] = { }, { .type = CRYPTO_ALG_TYPE_AHASH, .alg.hash = { - .init = ahash_init, - .update = ahash_update, - .final = ahash_final, - .finup = ahash_finup, - .digest = ahash_digest, - .setkey = ahash_setkey, .halg.digestsize = SHA256_DIGEST_SIZE, .halg.base = { .cra_name = "hmac(sha256)", @@ -2475,7 +2395,6 @@ static struct talitos_alg_template driver_algs[] = { .cra_blocksize = SHA256_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AHASH | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_ahash_type } }, .desc_hdr_template = DESC_HDR_TYPE_COMMON_NONSNOOP_NO_AFEU | @@ -2484,12 +2403,6 @@ static struct talitos_alg_template driver_algs[] = { }, { .type = CRYPTO_ALG_TYPE_AHASH, .alg.hash = { - .init = ahash_init, - .update = ahash_update, - .final = ahash_final, - .finup = ahash_finup, - .digest = ahash_digest, - .setkey = ahash_setkey, .halg.digestsize = SHA384_DIGEST_SIZE, .halg.base = { .cra_name = "hmac(sha384)", @@ -2497,7 +2410,6 @@ static struct talitos_alg_template driver_algs[] = { .cra_blocksize = SHA384_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AHASH | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_ahash_type } }, .desc_hdr_template = DESC_HDR_TYPE_COMMON_NONSNOOP_NO_AFEU | @@ -2506,12 +2418,6 @@ static struct talitos_alg_template driver_algs[] = { }, { .type = CRYPTO_ALG_TYPE_AHASH, .alg.hash = { - .init = ahash_init, - .update = ahash_update, - .final = ahash_final, - .finup = ahash_finup, - .digest = ahash_digest, - .setkey = ahash_setkey, .halg.digestsize = SHA512_DIGEST_SIZE, .halg.base = { .cra_name = "hmac(sha512)", @@ -2519,7 +2425,6 @@ static struct talitos_alg_template driver_algs[] = { .cra_blocksize = SHA512_BLOCK_SIZE, .cra_flags = CRYPTO_ALG_TYPE_AHASH | CRYPTO_ALG_ASYNC, - .cra_type = &crypto_ahash_type } }, .desc_hdr_template = DESC_HDR_TYPE_COMMON_NONSNOOP_NO_AFEU | @@ -2677,14 +2582,34 @@ static struct talitos_crypto_alg *talitos_alg_alloc(struct device *dev, case CRYPTO_ALG_TYPE_ABLKCIPHER: alg = &t_alg->algt.alg.crypto; alg->cra_init = talitos_cra_init; + alg->cra_type = &crypto_ablkcipher_type; + alg->cra_ablkcipher.setkey = ablkcipher_setkey; + alg->cra_ablkcipher.encrypt = ablkcipher_encrypt; + alg->cra_ablkcipher.decrypt = ablkcipher_decrypt; + alg->cra_ablkcipher.geniv = "eseqiv"; break; case CRYPTO_ALG_TYPE_AEAD: alg = &t_alg->algt.alg.crypto; alg->cra_init = talitos_cra_init_aead; + alg->cra_type = &crypto_aead_type; + alg->cra_aead.setkey = aead_setkey; + alg->cra_aead.setauthsize = aead_setauthsize; + alg->cra_aead.encrypt = aead_encrypt; + alg->cra_aead.decrypt = aead_decrypt; + alg->cra_aead.givencrypt = aead_givencrypt; + alg->cra_aead.geniv = "<built-in>"; break; case CRYPTO_ALG_TYPE_AHASH: alg = &t_alg->algt.alg.hash.halg.base; alg->cra_init = talitos_cra_init_ahash; + alg->cra_type = &crypto_ahash_type; + t_alg->algt.alg.hash.init = ahash_init; + t_alg->algt.alg.hash.update = ahash_update; + t_alg->algt.alg.hash.final = ahash_final; + t_alg->algt.alg.hash.finup = ahash_finup; + t_alg->algt.alg.hash.digest = ahash_digest; + t_alg->algt.alg.hash.setkey = ahash_setkey; + if (!(priv->features & TALITOS_FTR_HMAC_OK) && !strncmp(alg->cra_name, "hmac", 4)) { kfree(t_alg); @@ -2896,7 +2821,9 @@ static int talitos_probe(struct platform_device *ofdev) if (hw_supports(dev, driver_algs[i].desc_hdr_template)) { struct talitos_crypto_alg *t_alg; char *name = NULL; + bool authenc = false; +authencesn: t_alg = talitos_alg_alloc(dev, &driver_algs[i]); if (IS_ERR(t_alg)) { err = PTR_ERR(t_alg); @@ -2911,6 +2838,8 @@ static int talitos_probe(struct platform_device *ofdev) err = crypto_register_alg( &t_alg->algt.alg.crypto); name = t_alg->algt.alg.crypto.cra_driver_name; + authenc = authenc ? !authenc : + !(bool)memcmp(name, "authenc", 7); break; case CRYPTO_ALG_TYPE_AHASH: err = crypto_register_ahash( @@ -2923,8 +2852,25 @@ static int talitos_probe(struct platform_device *ofdev) dev_err(dev, "%s alg registration failed\n", name); kfree(t_alg); - } else + } else { list_add_tail(&t_alg->entry, &priv->alg_list); + if (authenc) { + struct crypto_alg *alg = + &driver_algs[i].alg.crypto; + + name = alg->cra_name; + memmove(name + 10, name + 7, + strlen(name) - 7); + memcpy(name + 7, "esn", 3); + + name = alg->cra_driver_name; + memmove(name + 10, name + 7, + strlen(name) - 7); + memcpy(name + 7, "esn", 3); + + goto authencesn; + } + } } } if (!list_empty(&priv->alg_list)) diff --git a/drivers/crypto/tegra-aes.c b/drivers/crypto/tegra-aes.c index ac236f6724f..37185e6630c 100644 --- a/drivers/crypto/tegra-aes.c +++ b/drivers/crypto/tegra-aes.c @@ -969,6 +969,7 @@ static int tegra_aes_probe(struct platform_device *pdev) aes_wq = alloc_workqueue("tegra_aes_wq", WQ_HIGHPRI | WQ_UNBOUND, 1); if (!aes_wq) { dev_err(dev, "alloc_workqueue failed\n"); + err = -ENOMEM; goto out; } @@ -1004,8 +1005,6 @@ static int tegra_aes_probe(struct platform_device *pdev) aes_dev = dd; for (i = 0; i < ARRAY_SIZE(algs); i++) { - INIT_LIST_HEAD(&algs[i].cra_list); - algs[i].cra_priority = 300; algs[i].cra_ctxsize = sizeof(struct tegra_aes_ctx); algs[i].cra_module = THIS_MODULE; diff --git a/drivers/crypto/ux500/cryp/cryp_core.c b/drivers/crypto/ux500/cryp/cryp_core.c index 1c307e1b840..bc615cc5626 100644 --- a/drivers/crypto/ux500/cryp/cryp_core.c +++ b/drivers/crypto/ux500/cryp/cryp_core.c @@ -32,7 +32,7 @@ #include <plat/ste_dma40.h> -#include <mach/crypto-ux500.h> +#include <linux/platform_data/crypto-ux500.h> #include <mach/hardware.h> #include "cryp_p.h" @@ -1486,6 +1486,7 @@ static int ux500_cryp_probe(struct platform_device *pdev) if (!res_irq) { dev_err(dev, "[%s]: IORESOURCE_IRQ unavailable", __func__); + ret = -ENODEV; goto out_power; } diff --git a/drivers/crypto/ux500/hash/hash_core.c b/drivers/crypto/ux500/hash/hash_core.c index 08d5032cb56..632c3339895 100644 --- a/drivers/crypto/ux500/hash/hash_core.c +++ b/drivers/crypto/ux500/hash/hash_core.c @@ -31,7 +31,7 @@ #include <crypto/scatterwalk.h> #include <crypto/algapi.h> -#include <mach/crypto-ux500.h> +#include <linux/platform_data/crypto-ux500.h> #include <mach/hardware.h> #include "hash_alg.h" @@ -1991,7 +1991,6 @@ static int __init ux500_hash_mod_init(void) static void __exit ux500_hash_mod_fini(void) { platform_driver_unregister(&hash_driver); - return; } module_init(ux500_hash_mod_init); |