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

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

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

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

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

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

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

Add explicit array bounds to the function prototypes for the parameters
that didn't already get handled by the conversion to use chacha_state:

- chacha_block_*():
  Change 'u8 *out' or 'u8 *stream' to u8 out[CHACHA_BLOCK_SIZE].

- hchacha_block_*():
  Change 'u32 *out' or 'u32 *stream' to u32 out[HCHACHA_OUT_WORDS].

- chacha_init():
  Change 'const u32 *key' to 'const u32 key[CHACHA_KEY_WORDS]'.
  Change 'const u8 *iv' to 'const u8 iv[CHACHA_IV_SIZE]'.

No functional changes.  This just makes it clear when fixed-size arrays
are expected.

Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

The ChaCha state matrix is 16 32-bit words.  Currently it is represented
in the code as a raw u32 array, or even just a pointer to u32.  This
weak typing is error-prone.  Instead, introduce struct chacha_state:

    struct chacha_state {
            u32 x[16];
    };

Convert all ChaCha and HChaCha functions to use struct chacha_state.
No functional changes.

Signed-off-by: Eric Biggers <ebiggers@google.com>
Acked-by: Kent Overstreet <kent.overstreet@linux.dev>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

Make the architecture-optimized CRC code do its CPU feature checks in
subsys_initcalls instead of arch_initcalls.  This makes it consistent
with arch/*/lib/crypto/ and ensures that it runs after initcalls that
possibly could be a prerequisite for kernel-mode FPU, such as x86's
xfd_update_static_branch() and loongarch's init_euen_mask().

Note: as far as I can tell, x86's xfd_update_static_branch() isn't
*actually* needed for kernel-mode FPU.  loongarch's init_euen_mask() is
needed to enable save/restore of the vector registers, but loongarch
doesn't yet have any CRC or crypto code that uses vector registers
anyway.  Regardless, let's be consistent with arch/*/lib/crypto/ and
robust against any potential future dependency on an arch_initcall.

Link: https://lore.kernel.org/r/20250510035959.87995-1-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>

Add CRYPTO_ARCH_HAVE_LIB_SHA256_SIMD and a SIMD block function
so that the caller can decide whether to use SIMD.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

Export the block functions as GPL only, there is no reason
to let arbitrary modules use these internal functions.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

This reverts commit c4741b23059794bd99beef0f700103b0d983b3fd.

Crypto API self-tests no longer run at registration time and now
occur either at late_initcall or upon the first use.

Therefore the premise of the above commit no longer exists.  Revert
it and subsequent additions of subsys_initcall and arch_initcall.

Note that lib/crypto calls will stay at subsys_initcall (or rather
downgraded from arch_initcall) because they may need to occur
before Crypto API registration.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

Instead of providing crypto_shash algorithms for the arch-optimized
SHA-256 code, instead implement the SHA-256 library.  This is much
simpler, it makes the SHA-256 library functions be arch-optimized, and
it fixes the longstanding issue where the arch-optimized SHA-256 was
disabled by default.  SHA-256 still remains available through
crypto_shash, but individual architectures no longer need to handle it.

To merge the scalar, NEON, and CE code all into one module cleanly, add
!CPU_V7M as a direct dependency of the CE code.  Previously, !CPU_V7M
was only a direct dependency of the scalar and NEON code.  The result is
still the same because CPU_V7M implies !KERNEL_MODE_NEON, so !CPU_V7M
was already an indirect dependency of the CE code.

To match sha256_blocks_arch(), change the type of the nblocks parameter
of the assembly functions from int to size_t.  The assembly functions
actually already treated it as size_t.

While renaming the assembly files, also fix the naming quirk where
"sha2" meant sha256.  (SHA-512 is also part of SHA-2.)

Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

Now that every architecture provides a block function, use that
to implement the lib/poly1305 and remove the old per-arch code.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

Add block-only interface.

Also remove the unnecessary SIMD fallback path.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>