btrfs-progs: crypto: add PCL based implementation for crc32c

Copy faster implementation of crc32c from linux kernel as of 6.5-rc7
(x86_64, arch/x86/crypto/crc32c-pcl-intel-asm_64.S). This needs
assembler build support, so detect target architecture so
cross-compilation still works.

Add a special CPU flag so the old and new implementations can be
benchmarked and verified separately.

Sample benchmark:

CPU flags: 0x1ff
CPU features: SSE2 SSSE3 SSE41 SSE42 SHA AVX AVX2 CRC32C_PCL
Block size:     4096
Iterations:     1000000
Implementation: builtin
Units:          CPU cycles

      NULL-NOP: cycles:     77177218, cycles/i       77
   NULL-MEMCPY: cycles:    226313072, cycles/i      226,    62133.395 MiB/s
    CRC32C-ref: cycles:  24418596066, cycles/i    24418,      575.859 MiB/s
     CRC32C-NI: cycles:   1188335920, cycles/i     1188,    11833.073 MiB/s
    CRC32C-PCL: cycles:    463193456, cycles/i      463,    30358.037 MiB/s
        XXHASH: cycles:    851606646, cycles/i      851,    16511.916 MiB/s
    SHA256-ref: cycles:  74476234956, cycles/i    74476,      188.808 MiB/s
     SHA256-NI: cycles:  34198637428, cycles/i    34198,      411.177 MiB/s
    BLAKE2-ref: cycles:  14761411664, cycles/i    14761,      952.597 MiB/s
   BLAKE2-SSE2: cycles:  18101896796, cycles/i    18101,      776.807 MiB/s
  BLAKE2-SSE41: cycles:  12599091062, cycles/i    12599,     1116.087 MiB/s
   BLAKE2-AVX2: cycles:   9668247506, cycles/i     9668,     1454.418 MiB/s

The new implementation is about 2.5x faster.

Note: there new version does not work on musl because of linkage
problems (relocations in .rodata), so it's still using the old
implementation.

Signed-off-by: David Sterba <[email protected]>

Author: kdave

Makefile

@@ -88,6 +88,8 @@ DISABLE_WARNING_FLAGS := $(call cc-disable-warning, format-truncation) \
 ENABLE_WARNING_FLAGS := $(call cc-option, -Wimplicit-fallthrough) \
 			$(call cc-option, -Wmissing-prototypes)
 
+ASFLAGS =
+
 # Common build flags
 CFLAGS = $(SUBST_CFLAGS) \
 	 -std=gnu11 \
@@ -383,6 +385,14 @@ CRYPTO_OBJECTS = crypto/sha224-256.o crypto/blake2b-ref.o crypto/blake2b-sse2.o
 CRYPTO_CFLAGS = -DCRYPTOPROVIDER_BUILTIN=1
 endif
 
+ifeq ($(TARGET_CPU),x86_64)
+# FIXME: linkage is broken on musl for some reason
+ifeq ($(HAVE_GLIBC),1)
+CRYPTO_OBJECTS += crypto/crc32c-pcl-intel-asm_64.o
+ASFLAGS += -fPIC
+endif
+endif
+
 CHECKER_FLAGS += $(btrfs_convert_cflags)
 
 # collect values of the variables above
@@ -450,6 +460,13 @@ endif
 		-MT $($(dir $@).deps/$(notdir $@):.o.d=.static.o) \
 		-MT $(dir $@).deps/$(notdir $@) $(CFLAGS) $<
 
+.S.o:
+	@echo "    [AS]     $@"
+	$(Q)$(CC) $(CFLAGS) $(ASFLAGS) -c $< -o $@
+
+%.static.o: %.S
+	@echo "    [AS]     $@"
+	$(Q)$(CC) $(CFLAGS) $(ASFLAGS) -c $< -o $@
 #
 # Pick from per-file variables, btrfs_*_cflags
 #

Makefile.inc.in

@@ -28,6 +28,8 @@ HAVE_CFLAG_msse2 = @HAVE_CFLAG_msse2@
 HAVE_CFLAG_msse41 = @HAVE_CFLAG_msse41@
 HAVE_CFLAG_mavx2 = @HAVE_CFLAG_mavx2@
 HAVE_CFLAG_msha = @HAVE_CFLAG_msha@
+TARGET_CPU = @target_cpu@
+HAVE_GLIBC = @HAVE_GLIBC@
 
 SUBST_CFLAGS = @CFLAGS@
 SUBST_LDFLAGS = @LDFLAGS@

common/cpu-utils.c

@@ -58,6 +58,7 @@ void cpu_print_flags(void) {
 	FLAG(SHA);
 	FLAG(AVX);
 	FLAG(AVX2);
+	FLAG(CRC32C_PCL);
 	putchar(10);
 }
 #undef FLAG
@@ -88,6 +89,7 @@ void cpu_detect_flags(void)
 	if (b & (1UL << 29))
 		__cpu_flags |= CPU_FLAG_SHA;
 
+	__cpu_flags |= CPU_FLAG_CRC32C_PCL;
 	__cpu_flags_orig = __cpu_flags;
 }
 

common/cpu-utils.h

@@ -37,6 +37,9 @@ enum cpu_feature {
 	ENUM_CPU_BIT(CPU_FLAG_SHA),
 	ENUM_CPU_BIT(CPU_FLAG_AVX),
 	ENUM_CPU_BIT(CPU_FLAG_AVX2),
+
+	/* Special features */
+	ENUM_CPU_BIT(CPU_FLAG_CRC32C_PCL),
 };
 
 #undef ENUM_CPU_BIT

configure.ac

@@ -40,6 +40,7 @@ AC_PREFIX_DEFAULT([/usr/local])
 
 AC_PROG_CC
 AC_CANONICAL_HOST
+AC_CANONICAL_TARGET
 AC_C_CONST
 AC_C_VOLATILE
 AC_C_BIGENDIAN
@@ -79,6 +80,9 @@ AC_CHECK_FUNCS([reallocarray])
 
 AC_CHECK_FUNCS([clock_gettime])
 
+AX_CHECK_DEFINE([features.h],[__GLIBC__],[HAVE_GLIBC=1],[HAVE_GLIBC=0])
+AC_SUBST([HAVE_GLIBC])
+
 AX_GCC_BUILTIN([__builtin_add_overflow])
 AX_GCC_BUILTIN([__builtin_sub_overflow])
 AX_GCC_BUILTIN([__builtin_mul_overflow])