lib/crc: Prepare for arch-optimized code in subdirs of lib/crc/

Rework how lib/crc/ supports arch-optimized code.  First, instead of the
arch-optimized CRC code being in arch/$(SRCARCH)/lib/, it will now be in
lib/crc/$(SRCARCH)/.  Second, the API functions (e.g. crc32c()),
arch-optimized functions (e.g. crc32c_arch()), and generic functions
(e.g. crc32c_base()) will now be part of a single module for each CRC
type, allowing better inlining and dead code elimination.  The second
change is made possible by the first.

As an example, consider CONFIG_CRC32=m on x86.  We'll now have just
crc32.ko instead of both crc32-x86.ko and crc32.ko.  The two modules
were already coupled together and always both got loaded together via
direct symbol dependency, so the separation provided no benefit.

Note: later I'd like to apply the same design to lib/crypto/ too, where
often the API functions are out-of-line so this will work even better.
In those cases, for each algorithm we currently have 3 modules all
coupled together, e.g. libsha256.ko, libsha256-generic.ko, and
sha256-x86.ko.  We should have just one, inline things properly, and
rely on the compiler's dead code elimination to decide the inclusion of
the generic code instead of manually setting it via kconfig.

Having arch-specific code outside arch/ was somewhat controversial when
Zinc proposed it back in 2018.  But I don't think the concerns are
warranted.  It's better from a technical perspective, as it enables the
improvements mentioned above.  This model is already successfully used
in other places in the kernel such as lib/raid6/.  The community of each
architecture still remains free to work on the code, even if it's not in
arch/.  At the time there was also a desire to put the library code in
the same files as the old-school crypto API, but that was a mistake; now
that the library is separate, that's no longer a constraint either.

Reviewed-by: "Martin K. Petersen" <martin.petersen@oracle.com>
Acked-by: Ingo Molnar <mingo@kernel.org>
Acked-by: "Jason A. Donenfeld" <Jason@zx2c4.com>
Link: https://lore.kernel.org/r/20250607200454.73587-3-ebiggers@kernel.org
Link: https://lore.kernel.org/r/20250612054514.142728-1-ebiggers@kernel.org
Link: https://lore.kernel.org/r/20250621012221.4351-1-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@kernel.org>

10 files changed, 159 insertions, 89 deletions

diff --git a/Documentation/core-api/kernel-api.rst b/Documentation/core-api/kernel-api.rst
index c4642d9f13a9..9c8370891a39 100644
--- a/Documentation/core-api/kernel-api.rst
+++ b/Documentation/core-api/kernel-api.rst
@@ -148,7 +148,7 @@ CRC Functions
 .. kernel-doc:: lib/crc/crc16.c
    :export:
 
-.. kernel-doc:: lib/crc/crc32.c
+.. kernel-doc:: lib/crc/crc32-main.c
 
 .. kernel-doc:: lib/crc/crc-ccitt.c
    :export:
diff --git a/MAINTAINERS b/MAINTAINERS
index b0ec17dfbe03..d75676e17ac7 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -6359,7 +6359,6 @@ L:	linux-crypto@vger.kernel.org
 S:	Maintained
 T:	git https://git.kernel.org/pub/scm/linux/kernel/git/ebiggers/linux.git crc-next
 F:	Documentation/staging/crc*
-F:	arch/*/lib/crc*
 F:	include/linux/crc*
 F:	lib/crc/
 F:	scripts/gen-crc-consts.py
diff --git a/include/linux/crc-t10dif.h b/include/linux/crc-t10dif.h
index a559fdff3f7e..ecc8bc2dd7f4 100644
--- a/include/linux/crc-t10dif.h
+++ b/include/linux/crc-t10dif.h
@@ -4,15 +4,7 @@
 
 #include <linux/types.h>
 
-u16 crc_t10dif_arch(u16 crc, const u8 *p, size_t len);
-u16 crc_t10dif_generic(u16 crc, const u8 *p, size_t len);
-
-static inline u16 crc_t10dif_update(u16 crc, const u8 *p, size_t len)
-{
-	if (IS_ENABLED(CONFIG_CRC_T10DIF_ARCH))
-		return crc_t10dif_arch(crc, p, len);
-	return crc_t10dif_generic(crc, p, len);
-}
+u16 crc_t10dif_update(u16 crc, const u8 *p, size_t len);
 
 static inline u16 crc_t10dif(const u8 *p, size_t len)
 {
diff --git a/include/linux/crc32.h b/include/linux/crc32.h
index 36bbc0405aa0..22dbe7144eb4 100644
--- a/include/linux/crc32.h
+++ b/include/linux/crc32.h
@@ -5,33 +5,9 @@
 #include <linux/types.h>
 #include <linux/bitrev.h>
 
-u32 crc32_le_arch(u32 crc, const u8 *p, size_t len);
-u32 crc32_le_base(u32 crc, const u8 *p, size_t len);
-u32 crc32_be_arch(u32 crc, const u8 *p, size_t len);
-u32 crc32_be_base(u32 crc, const u8 *p, size_t len);
-u32 crc32c_arch(u32 crc, const u8 *p, size_t len);
-u32 crc32c_base(u32 crc, const u8 *p, size_t len);
-
-static inline u32 crc32_le(u32 crc, const void *p, size_t len)
-{
-	if (IS_ENABLED(CONFIG_CRC32_ARCH))
-		return crc32_le_arch(crc, p, len);
-	return crc32_le_base(crc, p, len);
-}
-
-static inline u32 crc32_be(u32 crc, const void *p, size_t len)
-{
-	if (IS_ENABLED(CONFIG_CRC32_ARCH))
-		return crc32_be_arch(crc, p, len);
-	return crc32_be_base(crc, p, len);
-}
-
-static inline u32 crc32c(u32 crc, const void *p, size_t len)
-{
-	if (IS_ENABLED(CONFIG_CRC32_ARCH))
-		return crc32c_arch(crc, p, len);
-	return crc32c_base(crc, p, len);
-}
+u32 crc32_le(u32 crc, const void *p, size_t len);
+u32 crc32_be(u32 crc, const void *p, size_t len);
+u32 crc32c(u32 crc, const void *p, size_t len);
 
 /*
  * crc32_optimizations() returns flags that indicate which CRC32 library
diff --git a/include/linux/crc64.h b/include/linux/crc64.h
index b6aa290a7931..fc0c06ab1993 100644
--- a/include/linux/crc64.h
+++ b/include/linux/crc64.h
@@ -4,11 +4,6 @@
 
 #include <linux/types.h>
 
-u64 crc64_be_arch(u64 crc, const u8 *p, size_t len);
-u64 crc64_be_generic(u64 crc, const u8 *p, size_t len);
-u64 crc64_nvme_arch(u64 crc, const u8 *p, size_t len);
-u64 crc64_nvme_generic(u64 crc, const u8 *p, size_t len);
-
 /**
  * crc64_be - Calculate bitwise big-endian ECMA-182 CRC64
  * @crc: seed value for computation. 0 or (u64)~0 for a new CRC calculation,
@@ -16,12 +11,7 @@ u64 crc64_nvme_generic(u64 crc, const u8 *p, size_t len);
  * @p: pointer to buffer over which CRC64 is run
  * @len: length of buffer @p
  */
-static inline u64 crc64_be(u64 crc, const void *p, size_t len)
-{
-	if (IS_ENABLED(CONFIG_CRC64_ARCH))
-		return crc64_be_arch(crc, p, len);
-	return crc64_be_generic(crc, p, len);
-}
+u64 crc64_be(u64 crc, const void *p, size_t len);
 
 /**
  * crc64_nvme - Calculate CRC64-NVME
@@ -33,11 +23,6 @@ static inline u64 crc64_be(u64 crc, const void *p, size_t len)
  * This computes the CRC64 defined in the NVME NVM Command Set Specification,
  * *including the bitwise inversion at the beginning and end*.
  */
-static inline u64 crc64_nvme(u64 crc, const void *p, size_t len)
-{
-	if (IS_ENABLED(CONFIG_CRC64_ARCH))
-		return ~crc64_nvme_arch(~crc, p, len);
-	return ~crc64_nvme_generic(~crc, p, len);
-}
+u64 crc64_nvme(u64 crc, const void *p, size_t len);
 
 #endif /* _LINUX_CRC64_H */
diff --git a/lib/crc/Kconfig b/lib/crc/Kconfig
index e0e7168b74c7..04c7aeb0fc8e 100644
--- a/lib/crc/Kconfig
+++ b/lib/crc/Kconfig
@@ -48,8 +48,8 @@ config ARCH_HAS_CRC_T10DIF
 	bool
 
 config CRC_T10DIF_ARCH
-	tristate
-	default CRC_T10DIF if ARCH_HAS_CRC_T10DIF && CRC_OPTIMIZATIONS
+	bool
+	depends on CRC_T10DIF && CRC_OPTIMIZATIONS
 
 config CRC32
 	tristate
@@ -62,8 +62,8 @@ config ARCH_HAS_CRC32
 	bool
 
 config CRC32_ARCH
-	tristate
-	default CRC32 if ARCH_HAS_CRC32 && CRC_OPTIMIZATIONS
+	bool
+	depends on CRC32 && CRC_OPTIMIZATIONS
 
 config CRC64
 	tristate
@@ -75,11 +75,12 @@ config ARCH_HAS_CRC64
 	bool
 
 config CRC64_ARCH
-	tristate
-	default CRC64 if ARCH_HAS_CRC64 && CRC_OPTIMIZATIONS
+	bool
+	depends on CRC64 && CRC_OPTIMIZATIONS
 
 config CRC_OPTIMIZATIONS
 	bool "Enable optimized CRC implementations" if EXPERT
+	depends on !UML
 	default y
 	help
 	  Disabling this option reduces code size slightly by disabling the
diff --git a/lib/crc/Makefile b/lib/crc/Makefile
index ff4c30dda452..926edc3b035f 100644
--- a/lib/crc/Makefile
+++ b/lib/crc/Makefile
@@ -8,16 +8,32 @@ obj-$(CONFIG_CRC8) += crc8.o
 obj-$(CONFIG_CRC16) += crc16.o
 obj-$(CONFIG_CRC_CCITT) += crc-ccitt.o
 obj-$(CONFIG_CRC_ITU_T) += crc-itu-t.o
+
 obj-$(CONFIG_CRC_T10DIF) += crc-t10dif.o
+crc-t10dif-y := crc-t10dif-main.o
+ifeq ($(CONFIG_CRC_T10DIF_ARCH),y)
+CFLAGS_crc-t10dif-main.o += -I$(src)/$(SRCARCH)
+endif
+
 obj-$(CONFIG_CRC32) += crc32.o
+crc32-y := crc32-main.o
+ifeq ($(CONFIG_CRC32_ARCH),y)
+CFLAGS_crc32-main.o += -I$(src)/$(SRCARCH)
+endif
+
 obj-$(CONFIG_CRC64) += crc64.o
+crc64-y := crc64-main.o
+ifeq ($(CONFIG_CRC64_ARCH),y)
+CFLAGS_crc64-main.o += -I$(src)/$(SRCARCH)
+endif
+
 obj-y += tests/
 
 hostprogs := gen_crc32table gen_crc64table
 clean-files := crc32table.h crc64table.h
 
-$(obj)/crc32.o: $(obj)/crc32table.h
-$(obj)/crc64.o: $(obj)/crc64table.h
+$(obj)/crc32-main.o: $(obj)/crc32table.h
+$(obj)/crc64-main.o: $(obj)/crc64table.h
 
 quiet_cmd_crc32 = GEN     $@
       cmd_crc32 = $< > $@
diff --git a/lib/crc/crc-t10dif.c b/lib/crc/crc-t10dif-main.c
index 311c2ab829f1..bc91f3888184 100644
--- a/lib/crc/crc-t10dif.c
+++ b/lib/crc/crc-t10dif-main.c
@@ -50,16 +50,39 @@ static const u16 t10_dif_crc_table[256] = {
 	0xF0D8, 0x7B6F, 0x6C01, 0xE7B6, 0x42DD, 0xC96A, 0xDE04, 0x55B3
 };
 
-u16 crc_t10dif_generic(u16 crc, const u8 *p, size_t len)
+static inline u16 __maybe_unused
+crc_t10dif_generic(u16 crc, const u8 *p, size_t len)
 {
-	size_t i;
+	while (len--)
+		crc = (crc << 8) ^ t10_dif_crc_table[(crc >> 8) ^ *p++];
+	return crc;
+}
 
-	for (i = 0; i < len; i++)
-		crc = (crc << 8) ^ t10_dif_crc_table[(crc >> 8) ^ p[i]];
+#ifdef CONFIG_CRC_T10DIF_ARCH
+#include "crc-t10dif.h" /* $(SRCARCH)/crc-t10dif.h */
+#else
+#define crc_t10dif_arch crc_t10dif_generic
+#endif
 
-	return crc;
+u16 crc_t10dif_update(u16 crc, const u8 *p, size_t len)
+{
+	return crc_t10dif_arch(crc, p, len);
+}
+EXPORT_SYMBOL(crc_t10dif_update);
+
+#ifdef crc_t10dif_mod_init_arch
+static int __init crc_t10dif_mod_init(void)
+{
+	crc_t10dif_mod_init_arch();
+	return 0;
+}
+subsys_initcall(crc_t10dif_mod_init);
+
+static void __exit crc_t10dif_mod_exit(void)
+{
 }
-EXPORT_SYMBOL(crc_t10dif_generic);
+module_exit(crc_t10dif_mod_exit);
+#endif
 
-MODULE_DESCRIPTION("T10 DIF CRC calculation");
+MODULE_DESCRIPTION("CRC-T10DIF library functions");
 MODULE_LICENSE("GPL");
diff --git a/lib/crc/crc32.c b/lib/crc/crc32-main.c
index 6811b37df2aa..7843512ffef4 100644
--- a/lib/crc/crc32.c
+++ b/lib/crc/crc32-main.c
@@ -30,30 +30,75 @@
 
 #include "crc32table.h"
 
-MODULE_AUTHOR("Matt Domsch <Matt_Domsch@dell.com>");
-MODULE_DESCRIPTION("Various CRC32 calculations");
-MODULE_LICENSE("GPL");
-
-u32 crc32_le_base(u32 crc, const u8 *p, size_t len)
+static inline u32 __maybe_unused
+crc32_le_base(u32 crc, const u8 *p, size_t len)
 {
 	while (len--)
 		crc = (crc >> 8) ^ crc32table_le[(crc & 255) ^ *p++];
 	return crc;
 }
-EXPORT_SYMBOL(crc32_le_base);
 
-u32 crc32c_base(u32 crc, const u8 *p, size_t len)
+static inline u32 __maybe_unused
+crc32_be_base(u32 crc, const u8 *p, size_t len)
 {
 	while (len--)
-		crc = (crc >> 8) ^ crc32ctable_le[(crc & 255) ^ *p++];
+		crc = (crc << 8) ^ crc32table_be[(crc >> 24) ^ *p++];
 	return crc;
 }
-EXPORT_SYMBOL(crc32c_base);
 
-u32 crc32_be_base(u32 crc, const u8 *p, size_t len)
+static inline u32 __maybe_unused
+crc32c_base(u32 crc, const u8 *p, size_t len)
 {
 	while (len--)
-		crc = (crc << 8) ^ crc32table_be[(crc >> 24) ^ *p++];
+		crc = (crc >> 8) ^ crc32ctable_le[(crc & 255) ^ *p++];
 	return crc;
 }
-EXPORT_SYMBOL(crc32_be_base);
+
+#ifdef CONFIG_CRC32_ARCH
+#include "crc32.h" /* $(SRCARCH)/crc32.h */
+
+u32 crc32_optimizations(void)
+{
+	return crc32_optimizations_arch();
+}
+EXPORT_SYMBOL(crc32_optimizations);
+#else
+#define crc32_le_arch crc32_le_base
+#define crc32_be_arch crc32_be_base
+#define crc32c_arch crc32c_base
+#endif
+
+u32 crc32_le(u32 crc, const void *p, size_t len)
+{
+	return crc32_le_arch(crc, p, len);
+}
+EXPORT_SYMBOL(crc32_le);
+
+u32 crc32_be(u32 crc, const void *p, size_t len)
+{
+	return crc32_be_arch(crc, p, len);
+}
+EXPORT_SYMBOL(crc32_be);
+
+u32 crc32c(u32 crc, const void *p, size_t len)
+{
+	return crc32c_arch(crc, p, len);
+}
+EXPORT_SYMBOL(crc32c);
+
+#ifdef crc32_mod_init_arch
+static int __init crc32_mod_init(void)
+{
+	crc32_mod_init_arch();
+	return 0;
+}
+subsys_initcall(crc32_mod_init);
+
+static void __exit crc32_mod_exit(void)
+{
+}
+module_exit(crc32_mod_exit);
+#endif
+
+MODULE_DESCRIPTION("CRC32 library functions");
+MODULE_LICENSE("GPL");
diff --git a/lib/crc/crc64.c b/lib/crc/crc64-main.c
index 5b1b17057f0a..038afc7f4d44 100644
--- a/lib/crc/crc64.c
+++ b/lib/crc/crc64-main.c
@@ -38,21 +38,54 @@
 #include <linux/crc64.h>
 #include "crc64table.h"
 
-MODULE_DESCRIPTION("CRC64 calculations");
-MODULE_LICENSE("GPL v2");
-
-u64 crc64_be_generic(u64 crc, const u8 *p, size_t len)
+static inline u64 __maybe_unused
+crc64_be_generic(u64 crc, const u8 *p, size_t len)
 {
 	while (len--)
 		crc = (crc << 8) ^ crc64table[(crc >> 56) ^ *p++];
 	return crc;
 }
-EXPORT_SYMBOL_GPL(crc64_be_generic);
 
-u64 crc64_nvme_generic(u64 crc, const u8 *p, size_t len)
+static inline u64 __maybe_unused
+crc64_nvme_generic(u64 crc, const u8 *p, size_t len)
 {
 	while (len--)
 		crc = (crc >> 8) ^ crc64nvmetable[(crc & 0xff) ^ *p++];
 	return crc;
 }
-EXPORT_SYMBOL_GPL(crc64_nvme_generic);
+
+#ifdef CONFIG_CRC64_ARCH
+#include "crc64.h" /* $(SRCARCH)/crc64.h */
+#else
+#define crc64_be_arch crc64_be_generic
+#define crc64_nvme_arch crc64_nvme_generic
+#endif
+
+u64 crc64_be(u64 crc, const void *p, size_t len)
+{
+	return crc64_be_arch(crc, p, len);
+}
+EXPORT_SYMBOL_GPL(crc64_be);
+
+u64 crc64_nvme(u64 crc, const void *p, size_t len)
+{
+	return ~crc64_nvme_arch(~crc, p, len);
+}
+EXPORT_SYMBOL_GPL(crc64_nvme);
+
+#ifdef crc64_mod_init_arch
+static int __init crc64_mod_init(void)
+{
+	crc64_mod_init_arch();
+	return 0;
+}
+subsys_initcall(crc64_mod_init);
+
+static void __exit crc64_mod_exit(void)
+{
+}
+module_exit(crc64_mod_exit);
+#endif
+
+MODULE_DESCRIPTION("CRC64 library functions");
+MODULE_LICENSE("GPL");