Re: [PATCHv3 06/10] crypto: add rocksoft 64b crc framework

[Eric Biggers <ebiggers@kernel.org>]

On Tue, Feb 22, 2022 at 08:31:40AM -0800, Keith Busch wrote:
> +config CRYPTO_CRC64_ROCKSOFT
> +	tristate "Rocksoft Model CRC64 algorithm"
> +	depends on CRC64
> +	select CRYPTO_HASH
> +	help
> +	  Rocksoft Model CRC64 computation is being cast as a crypto
> +	  transform. This allows for faster crc64 transforms to be used
> +	  if they are available.

The first sentence of this help text doesn't make sense.

> diff --git a/crypto/crc64_rocksoft_generic.c b/crypto/crc64_rocksoft_generic.c
> new file mode 100644
> index 000000000000..55bad1939614
> --- /dev/null
> +++ b/crypto/crc64_rocksoft_generic.c
> @@ -0,0 +1,104 @@
> +// SPDX-License-Identifier: GPL-2.0-only
> +/*
> + * Cryptographic API.

The "Cryptographic API" line doesn't provide any helpful information.

> +static int chksum_final(struct shash_desc *desc, u8 *out)
> +{
> +	struct chksum_desc_ctx *ctx = shash_desc_ctx(desc);
> +
> +	*(u64 *)out = ctx->crc;
> +	return 0;
> +}
> +
> +static int __chksum_finup(u64 crc, const u8 *data, unsigned int len, u8 *out)
> +{
> +	*(u64 *)out = crc64_rocksoft_generic(crc, data, len);
> +	return 0;
> +}

These 64-bit writes violate alignment rules and will give the wrong result on
big endian CPUs.  They need to use put_unaligned_le64().

> +static int __init crc64_rocksoft_x86_mod_init(void)
> +{
> +	return crypto_register_shash(&alg);
> +}
> +
> +static void __exit crc64_rocksoft_x86_mod_fini(void)
> +{
> +	crypto_unregister_shash(&alg);
> +}

This has nothing to do with x86.

> +config CRC64_ROCKSOFT
> +	tristate "CRC calculation for the Rocksoft^TM model CRC64"

I'm sure what the rules for trademarks are, but kernel source code usually
doesn't have the trademark symbol/abbreviation scattered everywhere.

> +	select CRYPTO
> +	select CRYPTO_CRC64_ROCKSOFT
> +	help
> +	  This option is only needed if a module that's not in the
> +	  kernel tree needs to calculate CRC checks for use with the
> +	  rocksoft model parameters.

Out-of-tree modules can't be the reason to have a kconfig option.  What is the
real reason?

> +u64 crc64_rocksoft(const unsigned char *buffer, size_t len)
> +{
> +	return crc64_rocksoft_update(~0ULL, buffer, len);
> +}
> +EXPORT_SYMBOL(crc64_rocksoft);

Isn't this missing the bitwise inversion at the end?

> +MODULE_AUTHOR("Keith Busch <kbusch@kernel.org>");
> +MODULE_DESCRIPTION("Rocksoft model CRC64 calculation (library API)");
> +MODULE_LICENSE("GPL");
> +MODULE_SOFTDEP("pre: crc64");

Shouldn't the MODULE_SOFTDEP be on crc64-rocksoft?

- Eric