x86/crc-t10dif: implement crc_t10dif using new template

Instantiate crc-pclmul-template.S for crc_t10dif and delete the original
PCLMULQDQ optimized implementation.  This has the following advantages:

- Less CRC-variant-specific code.
- VPCLMULQDQ support, greatly improving performance on sufficiently long
  messages on newer CPUs.
- A faster reduction from 128 bits to the final CRC.
- Support for i386.

Benchmark results on AMD Ryzen 9 9950X (Zen 5) using crc_kunit:

        Length     Before        After
	------     ------        -----
	     1     440 MB/s      386 MB/s
	    16    1865 MB/s     2008 MB/s
	    64    4343 MB/s     6917 MB/s
	   127    5440 MB/s     8909 MB/s
	   128    5533 MB/s    12150 MB/s
	   200    5908 MB/s    14423 MB/s
	   256   15870 MB/s    21288 MB/s
	   511   14219 MB/s    25840 MB/s
	   512   18361 MB/s    37797 MB/s
	  1024   19941 MB/s    61374 MB/s
	  3173   20461 MB/s    74909 MB/s
	  4096   21310 MB/s    78919 MB/s
	 16384   21663 MB/s    85012 MB/s

Acked-by: Ard Biesheuvel <ardb@kernel.org>
Acked-by: Keith Busch <kbusch@kernel.org>
Reviewed-by: "Martin K. Petersen" <martin.petersen@oracle.com>
Link: https://lore.kernel.org/r/20250210174540.161705-6-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>

Author: ebiggers

arch/x86/Kconfig

@@ -77,7 +77,7 @@ config X86
 	select ARCH_HAS_CPU_FINALIZE_INIT
 	select ARCH_HAS_CPU_PASID		if IOMMU_SVA
 	select ARCH_HAS_CRC32
-	select ARCH_HAS_CRC_T10DIF		if X86_64
+	select ARCH_HAS_CRC_T10DIF
 	select ARCH_HAS_CURRENT_STACK_POINTER
 	select ARCH_HAS_DEBUG_VIRTUAL
 	select ARCH_HAS_DEBUG_VM_PGTABLE	if !X86_PAE

arch/x86/lib/Makefile

@@ -43,7 +43,7 @@ crc32-x86-y := crc32-glue.o crc32-pclmul.o
 crc32-x86-$(CONFIG_64BIT) += crc32c-3way.o
 
 obj-$(CONFIG_CRC_T10DIF_ARCH) += crc-t10dif-x86.o
-crc-t10dif-x86-y := crc-t10dif-glue.o crct10dif-pcl-asm_64.o
+crc-t10dif-x86-y := crc-t10dif-glue.o crc16-msb-pclmul.o
 
 obj-y += msr.o msr-reg.o msr-reg-export.o hweight.o
 obj-y += iomem.o

arch/x86/lib/crc-pclmul-consts.h

@@ -2,11 +2,57 @@
 /*
  * CRC constants generated by:
  *
- *	./scripts/gen-crc-consts.py x86_pclmul crc32_lsb_0xedb88320
+ *	./scripts/gen-crc-consts.py x86_pclmul crc16_msb_0x8bb7,crc32_lsb_0xedb88320
  *
  * Do not edit manually.
  */
 
+/*
+ * CRC folding constants generated for most-significant-bit-first CRC-16 using
+ * G(x) = x^16 + x^15 + x^11 + x^9 + x^8 + x^7 + x^5 + x^4 + x^2 + x^1 + x^0
+ */
+static const struct {
+	u8 bswap_mask[16];
+	u64 fold_across_2048_bits_consts[2];
+	u64 fold_across_1024_bits_consts[2];
+	u64 fold_across_512_bits_consts[2];
+	u64 fold_across_256_bits_consts[2];
+	u64 fold_across_128_bits_consts[2];
+	u8 shuf_table[48];
+	u64 barrett_reduction_consts[2];
+} crc16_msb_0x8bb7_consts ____cacheline_aligned __maybe_unused = {
+	.bswap_mask = {15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0},
+	.fold_across_2048_bits_consts = {
+		0xdccf000000000000,	/* LO64_TERMS: (x^2000 mod G) * x^48 */
+		0x4b0b000000000000,	/* HI64_TERMS: (x^2064 mod G) * x^48 */
+	},
+	.fold_across_1024_bits_consts = {
+		0x9d9d000000000000,	/* LO64_TERMS: (x^976 mod G) * x^48 */
+		0x7cf5000000000000,	/* HI64_TERMS: (x^1040 mod G) * x^48 */
+	},
+	.fold_across_512_bits_consts = {
+		0x044c000000000000,	/* LO64_TERMS: (x^464 mod G) * x^48 */
+		0xe658000000000000,	/* HI64_TERMS: (x^528 mod G) * x^48 */
+	},
+	.fold_across_256_bits_consts = {
+		0x6ee3000000000000,	/* LO64_TERMS: (x^208 mod G) * x^48 */
+		0xe7b5000000000000,	/* HI64_TERMS: (x^272 mod G) * x^48 */
+	},
+	.fold_across_128_bits_consts = {
+		0x2d56000000000000,	/* LO64_TERMS: (x^80 mod G) * x^48 */
+		0x06df000000000000,	/* HI64_TERMS: (x^144 mod G) * x^48 */
+	},
+	.shuf_table = {
+		-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
+		 0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15,
+		-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
+	},
+	.barrett_reduction_consts = {
+		0x8bb7000000000000,	/* LO64_TERMS: (G - x^16) * x^48 */
+		0xf65a57f81d33a48a,	/* HI64_TERMS: (floor(x^79 / G) * x) - x^64 */
+	},
+};
+
 /*
  * CRC folding constants generated for least-significant-bit-first CRC-32 using
  * G(x) = x^32 + x^26 + x^23 + x^22 + x^16 + x^12 + x^11 + x^10 + x^8 + x^7 +

arch/x86/lib/crc-t10dif-glue.c

@@ -1,37 +1,32 @@
 // SPDX-License-Identifier: GPL-2.0-or-later
 /*
- * CRC-T10DIF using PCLMULQDQ instructions
+ * CRC-T10DIF using [V]PCLMULQDQ instructions
  *
  * Copyright 2024 Google LLC
  */
 
-#include <asm/cpufeatures.h>
-#include <asm/simd.h>
-#include <crypto/internal/simd.h>
 #include <linux/crc-t10dif.h>
 #include <linux/module.h>
+#include "crc-pclmul-template.h"
 
 static DEFINE_STATIC_KEY_FALSE(have_pclmulqdq);
 
-asmlinkage u16 crc_t10dif_pcl(u16 init_crc, const u8 *buf, size_t len);
+DECLARE_CRC_PCLMUL_FUNCS(crc16_msb, u16);
 
 u16 crc_t10dif_arch(u16 crc, const u8 *p, size_t len)
 {
-	if (len >= 16 &&
-	    static_key_enabled(&have_pclmulqdq) && crypto_simd_usable()) {
-		kernel_fpu_begin();
-		crc = crc_t10dif_pcl(crc, p, len);
-		kernel_fpu_end();
-		return crc;
-	}
+	CRC_PCLMUL(crc, p, len, crc16_msb, crc16_msb_0x8bb7_consts,
+		   have_pclmulqdq);
 	return crc_t10dif_generic(crc, p, len);
 }
 EXPORT_SYMBOL(crc_t10dif_arch);
 
 static int __init crc_t10dif_x86_init(void)
 {
-	if (boot_cpu_has(X86_FEATURE_PCLMULQDQ))
+	if (boot_cpu_has(X86_FEATURE_PCLMULQDQ)) {
 		static_branch_enable(&have_pclmulqdq);
+		INIT_CRC_PCLMUL(crc16_msb);
+	}
 	return 0;
 }
 arch_initcall(crc_t10dif_x86_init);
@@ -41,5 +36,5 @@ static void __exit crc_t10dif_x86_exit(void)
 }
 module_exit(crc_t10dif_x86_exit);
 
-MODULE_DESCRIPTION("CRC-T10DIF using PCLMULQDQ instructions");
+MODULE_DESCRIPTION("CRC-T10DIF using [V]PCLMULQDQ instructions");
 MODULE_LICENSE("GPL");

arch/x86/lib/crc16-msb-pclmul.S

@@ -0,0 +1,6 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+// Copyright 2025 Google LLC
+
+#include "crc-pclmul-template.S"
+
+DEFINE_CRC_PCLMUL_FUNCS(crc16_msb, /* bits= */ 16, /* lsb= */ 0)