lib: rework bitreverse
Subsystems (7)
cover lib: rework bitreverse
This series is a resend for Jinjie Ruan's "arch/riscv: Add bitrev.h file to support rev8 and brev8" [1], my follow-up "lib: compile generic bitrev based on GENERIC_BITREVERSE" [2], and the fix for a build error reported by Nathan Chancellor [3].
No changes, except for combining pieces together and rebasing on top of the tree.
[1] https://lore.kernel.org/all/[email protected]/
[2] https://lore.kernel.org/all/[email protected]/
[3] https://lore.kernel.org/all/20260429202922.GA3575295@ax162/
Build-tested against x86 tinyconfig and defconfig, having disabled and enabiled CRC32 and BITREVERSE, correspondingly.
Jinjie Ruan (3):
lib/bitrev: Introduce GENERIC_BITREVERSE and cleanup Kconfig
bitops: Define generic __bitrev8/16/32 for reuse arch/riscv: Add bitrev.h file to support rev8 and brev8
Yury Norov (3):
lib: include crc32.h conditionally on CONFIG_CRC32
lib: compile generic bitrev.c conditionally on GENERIC_BITREVERSE
MAINTAINERS: BITOPS: include bitrev.[ch]
MAINTAINERS | 2 ++ arch/riscv/Kconfig | 2 ++ arch/riscv/include/asm/bitrev.h | 51 +++++++++++++++++++++++++++ include/asm-generic/bitops/__bitrev.h | 25 +++++++++++++ include/linux/bitrev.h | 20 +++-------- include/linux/etherdevice.h | 4 +++ lib/Kconfig | 18 ++++++++++ lib/Makefile | 2 +- lib/bitrev.c | 3 -- 9 files changed, 107 insertions(+), 20 deletions(-) create mode 100644 arch/riscv/include/asm/bitrev.h create mode 100644 include/asm-generic/bitops/__bitrev.h
Signature · 1 line
2.51.0
PATCH 1/6 lib: include crc32.h conditionally on CONFIG_CRC32
Currently, bitreverse API is either declared based on CONFIG_HAVE_ARCH_BITREVERSE, wired to arch implementation, or if the arch has no bitreverse, based on generic implementation.
So, regardless of CONFIG_BITREVERSE=n, the corresponding API is always declared. If that happens, the functions become declared but not implemented, which is an error.
The following patches of the series make it possible to have bitreverse API undeclared if CONFIG_BITREVERSE=n, thus spotting the problem when building the tinyconfig:
$ make -skj"$(nproc)" ARCH=s390 CROSS_COMPILE=s390-linux- mrproper tinyconfig fs/select.o
In file included from include/linux/crc32.h:6,
from include/linux/etherdevice.h:23,
from include/linux/if_vlan.h:11,
from include/linux/filter.h:21,
from include/net/xdp.h:10,
from include/net/busy_poll.h:19,
from fs/select.c:33:
include/linux/etherdevice.h: In function 'eth_hw_addr_crc':
include/linux/bitrev.h:16:20: error: implicit declaration of function 'generic___bitrev32' [-Wimplicit-function-declaration]
16 | #define __bitrev32 generic___bitrev32
| ^~~~~~~~~~~~~~~~~~
include/linux/bitrev.h:67:9: note: in expansion of macro '__bitrev32'
67 | __bitrev32(__x); \
| ^~~~~~~~~~
include/linux/crc32.h:107:36: note: in expansion of macro 'bitrev32'
107 | #define ether_crc(length, data) bitrev32(crc32_le(~0, data, length))
| ^~~~~~~~
include/linux/etherdevice.h:292:16: note: in expansion of macro 'ether_crc'
292 | return ether_crc(ETH_ALEN, ha->addr);
| ^~~~~~~~~
make[5]: *** [scripts/Makefile.build:289: fs/select.o] Error 1
...The current unconditionally enabled codebase doesn't use CRC32, neither bitrev functionality, and if generic___bitrev32 prototype is provided, the compilation and linkage phases are passed OK.
The only header requiring the crc32 and bitreverse prototypes is include/linux/etherdevice.h. Thus, protect inclusion of corresponding headers in the etherdevice with CONFIG_CRC32, together with the only function depending on it.
--- include/linux/etherdevice.h | 4 ++++ 1 file changed, 4 insertions(+)
include/linux/etherdevice.h +4 -0
@@ -20,7 +20,9 @@#include <linux/if_ether.h> #include <linux/netdevice.h> #include <linux/random.h> +#ifdef CONFIG_CRC32 #include <linux/crc32.h> +#endif #include <linux/unaligned.h> #include <asm/bitsperlong.h>
@@ -281,6 +283,7 @@ static inline void eth_hw_addr_random(struct net_device *dev) dev->addr_assign_type =NET_ADDR_RANDOM ; } +#ifdef CONFIG_CRC32 /** * eth_hw_addr_crc - Calculate CRC from netdev_hw_addr * @ha: pointer to hardware address
@@ -291,6 +294,7 @@ static inline u32 eth_hw_addr_crc(struct netdev_hw_addr *ha) {return ether_crc (ETH_ALEN , ha->addr); } +#endif /** * ether_addr_copy - Copy an Ethernet address
Signature · 1 line
2.51.0
PATCH 2/6 lib/bitrev: Introduce GENERIC_BITREVERSE and cleanup Kconfig
In reply to Yury NorovFrom: Jinjie Ruan <[email protected]>
Currently, the bit reversal lookup table is controlled by !HAVE_ARCH_BITREVERSE. This makes it difficult for architectures to provide a hardware-accelerated implementation while still falling back to the generic table for specific configurations.
Introduce CONFIG_GENERIC_BITREVERSE to explicitly manage the generic lookup table implementation. By using 'def_bool !HAVE_ARCH_BITREVERSE' with a dependency on 'BITREVERSE', we ensure that:
1. The table is only compiled when needed.
2. The .config is not polluted with useless options when BITREVERSE is disabled.
3. Avoids bloating the .data section for architectures that have full hardware bit-reverse support and don't need the table.
Update lib/bitrev.c to use CONFIG_GENERIC_BITREVERSE instead of checking the absence of HAVE_ARCH_BITREVERSE. This provides a cleaner interface for architectures like RISC-V that may want to selectively use the generic implementation as a fallback.
--- lib/Kconfig | 18 ++++++++++++++++++ lib/bitrev.c | 4 ++-- 2 files changed, 20 insertions(+), 2 deletions(-)
lib/Kconfig +18 -0
@@ -62,6 +62,24 @@ config HAVE_ARCH_BITREVERSE This option enables the use of hardware bit-reversal instructions on architectures which support such operations. +config GENERIC_BITREVERSE + def_bool !HAVE_ARCH_BITREVERSE + depends on BITREVERSE + help + This option provides the standard software-based bit reversal + implementation using a lookup table. + + Architecture-specific implementations (HAVE_ARCH_BITREVERSE) + and this generic version are not necessarily mutually exclusive + at the configuration level, but selecting this ensures that + the generic `bitrev8/16/32` functions are available when the + CPU does not provide native instructions (like RISC-V's ZBKB + extension). + + If you are an architecture maintainer and your CPU has native + bit-reversal instructions, you should select HAVE_ARCH_BITREVERSE + to skip this table-based implementation. + config ARCH_HAS_STRNCPY_FROM_USER bool
lib/bitrev.c +2 -2
@@ -1,5 +1,5 @@// SPDX-License-Identifier: GPL-2.0-only -#ifndef CONFIG_HAVE_ARCH_BITREVERSE +#ifdef CONFIG_GENERIC_BITREVERSE #include <linux/types.h> #include <linux/module.h> #include <linux/bitrev.h>
@@ -44,4 +44,4 @@ const u8 byte_rev_table[256] = { };EXPORT_SYMBOL_GPL (byte_rev_table); -#endif /* CONFIG_HAVE_ARCH_BITREVERSE */ +#endif /* CONFIG_GENERIC_BITREVERSE */
Signature · 1 line
2.51.0
PATCH 3/6 bitops: Define generic __bitrev8/16/32 for reuse
In reply to Yury NorovFrom: Jinjie Ruan <[email protected]>
Define generic __bitrev8/16/32 using the implementation in <linux/bitrev.h>, so they can be reused in <asm/bitrev.h>, such as RISCV.
--- include/asm-generic/bitops/__bitrev.h | 25 +++++++++++++++++++++++++ include/linux/bitrev.h | 20 ++++---------------- 2 files changed, 29 insertions(+), 16 deletions(-) create mode 100644 include/asm-generic/bitops/__bitrev.h
include/asm-generic/bitops/__bitrev.h new file +25 -0
@@ -0,0 +1,25 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +#ifndef _ASM_GENERIC_BITOPS___BITREV_H_ +#define _ASM_GENERIC_BITOPS___BITREV_H_ + +#ifdef CONFIG_GENERIC_BITREVERSE +#include <asm/types.h> + +extern u8 const byte_rev_table[256 ]; +static __always_inline __attribute_const__ u8 generic___bitrev8(u8 byte) +{ +return byte_rev_table[byte]; +} + +static __always_inline __attribute_const__ u16 generic___bitrev16(u16 x) +{ +return (generic___bitrev8 (x &0xff ) <<8 ) |generic___bitrev8 (x >>8 ); +} + +static __always_inline __attribute_const__ u32 generic___bitrev32(u32 x) +{ +return (generic___bitrev16 (x &0xffff ) <<16 ) |generic___bitrev16 (x >>16 ); +} +#endif /* CONFIG_GENERIC_BITREVERSE */ + +#endif /* _ASM_GENERIC_BITOPS___BITREV_H_ */
include/linux/bitrev.h +4 -16
@@ -12,22 +12,10 @@#define __bitrev8 __arch_bitrev8#else -extern u8 const byte_rev_table[256 ]; -static inline u8 __bitrev8 (u8 byte) -{ -return byte_rev_table[byte]; -} - -static inline u16 __bitrev16 (u16 x) -{ -return (__bitrev8 (x &0xff ) <<8 ) |__bitrev8 (x >>8 ); -} - -static inline u32 __bitrev32 (u32 x) -{ -return (__bitrev16 (x &0xffff ) <<16 ) |__bitrev16 (x >>16 ); -} - +#include <asm-generic/bitops/__bitrev.h> +#define __bitrev32 generic___bitrev32 +#define __bitrev16 generic___bitrev16 +#define __bitrev8 generic___bitrev8#endif /* CONFIG_HAVE_ARCH_BITREVERSE */ #define __bitrev8x4 (x) (__bitrev32(swab32(x)))
Signature · 1 line
2.51.0
PATCH 4/6 arch/riscv: Add bitrev.h file to support rev8 and brev8
In reply to Yury NorovFrom: Jinjie Ruan <[email protected]>
The RISC-V Bit-manipulation Extension for Cryptography (Zbkb) provides the 'brev8' instruction, which reverses the bits within each byte. Combined with the 'rev8' instruction (from Zbb or Zbkb), which reverses the byte order of a register, we can efficiently implement 16-bit, 32-bit, and (on RV64) 64-bit bit reversal.
This is significantly faster than the default software table-lookup implementation in lib/bitrev.c, as it replaces memory accesses and multiple arithmetic operations with just two or three hardware instructions.
Select HAVE_ARCH_BITREVERSE as well as GENERIC_BITREVERSE, and provide <asm/bitrev.h> to utilize these instructions when the Zbkb extension is available at runtime via the alternatives mechanism.
--- arch/riscv/Kconfig | 2 ++ arch/riscv/include/asm/bitrev.h | 51 +++++++++++++++++++++++++++++++++ 2 files changed, 53 insertions(+) create mode 100644 arch/riscv/include/asm/bitrev.h
arch/riscv/Kconfig +2 -0
@@ -104,6 +104,7 @@ config RISCV select FUNCTION_ALIGNMENT_8B if DYNAMIC_FTRACE_WITH_CALL_OPS select GENERIC_ARCH_TOPOLOGY select GENERIC_ATOMIC64 if !64BIT + select GENERIC_BITREVERSE select GENERIC_CLOCKEVENTS_BROADCAST if SMP select GENERIC_CPU_DEVICES select GENERIC_CPU_VULNERABILITIES
@@ -128,6 +129,7 @@ config RISCV select HAS_IOPORT if MMU select HAVE_ALIGNED_STRUCT_PAGE select HAVE_ARCH_AUDITSYSCALL + select HAVE_ARCH_BITREVERSE if RISCV_ISA_ZBKB select HAVE_ARCH_HUGE_VMALLOC if HAVE_ARCH_HUGE_VMAP select HAVE_ARCH_HUGE_VMAP if MMU && 64BIT select HAVE_ARCH_JUMP_LABEL
arch/riscv/include/asm/bitrev.h new file +51 -0
51 lines not on this page · raw message
Signature · 1 line
2.51.0
PATCH 5/6 lib: compile generic bitrev.c conditionally on GENERIC_BITREVERSE
In reply to Yury NorovThe file is compiled based on CONFIG_BITREVERSE=y, but everything inside is protected with CONFIG_GENERIC_BITREVERSE.
Make it simpler by switching the Makefile to compile lib/bitrev.c based on the proper config.
--- lib/Makefile | 2 +- lib/bitrev.c | 3 --- 2 files changed, 1 insertion(+), 4 deletions(-)
lib/Makefile +1 -1
@@ -145,7 +145,7 @@ obj-$(CONFIG_DEBUG_PREEMPT) += smp_processor_id.o obj-$(CONFIG_LIST_HARDENED) += list_debug.o obj-$(CONFIG_DEBUG_OBJECTS) += debugobjects.o -obj-$(CONFIG_BITREVERSE) += bitrev.o +obj-$(CONFIG_GENERIC_BITREVERSE) += bitrev.o obj-$(CONFIG_LINEAR_RANGES) += linear_ranges.o obj-$(CONFIG_PACKING) += packing.o obj-$(CONFIG_PACKING_KUNIT_TEST) += packing_test.o
lib/bitrev.c +0 -3
@@ -1,5 +1,4 @@// SPDX-License-Identifier: GPL-2.0-only -#ifdef CONFIG_GENERIC_BITREVERSE #include <linux/types.h> #include <linux/module.h> #include <linux/bitrev.h>
@@ -43,5 +42,3 @@ const u8 byte_rev_table[256] = {0x1f ,0x9f ,0x5f ,0xdf ,0x3f ,0xbf ,0x7f ,0xff , };EXPORT_SYMBOL_GPL (byte_rev_table); - -#endif /* CONFIG_GENERIC_BITREVERSE */
Signature · 1 line
2.51.0
PATCH 6/6 MAINTAINERS: BITOPS: include bitrev.[ch]
In reply to Yury NorovArch bitrev API is covered in MAINTAINERS under the BITOPS entry, while generic bitrev is unmaintained. Move it under BITOPS too.
--- MAINTAINERS | 2 ++ 1 file changed, 2 insertions(+)
MAINTAINERS +2 -0
@@ -4549,7 +4549,9 @@ F: arch/*/lib/bitops.c F: include/asm-generic/bitops F: include/asm-generic/bitops.h F: include/linux/bitops.h +F: include/linux/bitrev.h F: include/linux/count_zeros.h +F: lib/bitrev.c F: lib/hweight.c F: lib/test_bitops.c F: lib/tests/bitops_kunit.c
Signature · 1 line
2.51.0
[email protected]
AI reviewRe: [PATCH 1/6] lib: include crc32.h conditionally on CONFIG_CRC32
In reply to Yury NorovThank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [Low] Using `#ifdef CONFIG_CRC32` breaks the build for modular drivers that rely on `eth_hw_addr_crc()`. --
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] Using #ifdef CONFIG_CRC32 breaks the build for modular drivers that rely on eth_hw_addr_crc(). --
commit 734658b27c8ae06b58089e30d3e39cd61c7d09d6
Author: Yury Norov <[email protected]>
lib: include crc32.h conditionally on CONFIG_CRC32
This commit restricts the inclusion of crc32.h and the definition of eth_hw_addr_crc() in etherdevice.h to only when CONFIG_CRC32 is enabled. This fixes implicit declaration errors when building tinyconfig where CONFIG_BITREVERSE=n.
include/linux/etherdevice.h line 25
Quoted 11 lines
include/linux/etherdevice.h +2
-0@@ -20,7 +20,9 @@#include <linux/if_ether.h> #include <linux/netdevice.h> #include <linux/random.h> +#ifdef CONFIG_CRC32 #include <linux/crc32.h> +#endif
Will this break the build for modular drivers?
CONFIG_CRC32 is a tristate option. When it is configured as a module (=m), the build system defines CONFIG_CRC32_MODULE but not CONFIG_CRC32. Thus, the preprocessor check evaluates to false.
Should this use #if IS_ENABLED(CONFIG_CRC32) instead to support both built-in and modular configurations?
[ ... ]
include/linux/etherdevice.h static inline u32 eth_hw_addr_crc(struct netdev_hw_addr *ha) · line 297
Quoted 13 lines
include/linux/etherdevice.h +2
-0@@ -281,6 +283,7 @@ static inline void eth_hw_addr_random(struct net_device *dev) dev->addr_assign_type =NET_ADDR_RANDOM ; } +#ifdef CONFIG_CRC32 /** * eth_hw_addr_crc - Calculate CRC from netdev_hw_addr * @ha: pointer to hardware address @@ -291,6 +294,7 @@ static inline u32 eth_hw_addr_crc(struct netdev_hw_addr *ha) {return ether_crc (ETH_ALEN , ha->addr); } +#endif
If CONFIG_CRC32 is built as a module, will modular network drivers that call eth_hw_addr_crc() fail to compile due to an implicit declaration error since this definition is hidden?
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
[email protected]
AI reviewRe: [PATCH 2/6] lib/bitrev: Introduce GENERIC_BITREVERSE and cleanup Kconfig
In reply to Yury NorovThank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider: - [High] Build and linking failure for modules using bitreverse when CONFIG_BITREVERSE is compiled as a…
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] Build and linking failure for modules using bitreverse when CONFIG_BITREVERSE is compiled as a module ('m').
- [Medium] Kconfig unmet direct dependencies warning when GENERIC_BITREVERSE is selected by architectures. --
commit 43d61911d4e52b31128989028f6541618dd3291b
Author: Jinjie Ruan <[email protected]>
lib/bitrev: Introduce GENERIC_BITREVERSE and cleanup Kconfig
This commit introduces the GENERIC_BITREVERSE config option to explicitly manage the generic bit reversal lookup table, replacing the previous check for !HAVE_ARCH_BITREVERSE. It updates lib/bitrev.c to use this new symbol and makes it dependent on the BITREVERSE config.
lib/Kconfig config HAVE_ARCH_BITREVERSE · line 67
Quoted 11 lines
lib/Kconfig +3
-0@@ -62,6 +62,24 @@ config HAVE_ARCH_BITREVERSE This option enables the use of hardware bit-reversal instructions on architectures which support such operations. +config GENERIC_BITREVERSE + def_bool !HAVE_ARCH_BITREVERSE + depends on BITREVERSE
Because BITREVERSE is a tristate symbol and GENERIC_BITREVERSE is defined as a bool, what happens when BITREVERSE is configured as a module (=m)?
Kconfig evaluates a bool depending on a module as 'n'. This means when CONFIG_BITREVERSE=m, CONFIG_GENERIC_BITREVERSE evaluates to 'n' and remains undefined.
Will this result in missing declarations and linker failures for any modules relying on the bitreverse library?
Additionally, does this strict dependency cause Kconfig unmet direct dependency warnings if an architecture subsequently uses select GENERIC_BITREVERSE but the kernel configuration has BITREVERSE disabled (such as in allnoconfig)?
Quoted 2 lines
lib/Kconfig +2
-0+ help + This option provides the standard software-based bit reversal
[ ... ]
lib/bitrev.c line 2
Quoted 8 lines
lib/bitrev.c +1
-1@@ -1,5 +1,5 @@// SPDX-License-Identifier: GPL-2.0-only -#ifndef CONFIG_HAVE_ARCH_BITREVERSE +#ifdef CONFIG_GENERIC_BITREVERSE
If CONFIG_GENERIC_BITREVERSE evaluates to 'n' when CONFIG_BITREVERSE=m, this file will be compiled empty. Does this prevent byte_rev_table from being exported for modules that need it, causing a modpost linking failure?
Quoted 2 lines
lib/bitrev.c +0
-0#include <linux/types.h> #include <linux/module.h>
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=2
[email protected]
AI reviewRe: [PATCH 3/6] bitops: Define generic __bitrev8/16/32 for reuse
In reply to Yury NorovThank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [Low] The #ifdef CONFIG_GENERIC_BITREVERSE guard in <asm-generic/bitops/__bitrev.h> introduces build…
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] The #ifdef CONFIG_GENERIC_BITREVERSE guard in <asm-generic/bitops/__bitrev.h> introduces build regressions (implicit function declarations) for modules and uninstantiated inline functions. --
commit d29d1db3f8ae8fed09b1d7cc283d610df9bcb8b6
Author: Jinjie Ruan <[email protected]>
bitops: Define generic __bitrev8/16/32 for reuse
This commit extracts the generic bit reversal functions into a new <asm-generic/bitops/__bitrev.h> header, allowing architectures to reuse them.
include/asm-generic/bitops/__bitrev.h line 5
Quoted 11 lines
include/asm-generic/bitops/__bitrev.h new file +5
-0@@ -0,0 +1,25 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +#ifndef _ASM_GENERIC_BITOPS___BITREV_H_ +#define _ASM_GENERIC_BITOPS___BITREV_H_ + +#ifdef CONFIG_GENERIC_BITREVERSE
Will this cause build regressions for modules when CONFIG_BITREVERSE=m?
If CONFIG_GENERIC_BITREVERSE is a bool depending on CONFIG_BITREVERSE, when CONFIG_BITREVERSE=m, the bool evaluates to n. This causes CONFIG_GENERIC_BITREVERSE to be n, meaning the generic___bitrev* declarations are removed. Any module calling bitrev8() would then fail to compile due to an implicit function declaration.
This also means that compiling bitrev.o based on CONFIG_GENERIC_BITREVERSE prevents it from being built as a module.
Quoted 4 lines
include/asm-generic/bitops/__bitrev.h new file +4
-0+#include <asm/types.h> + +extern u8 const byte_rev_table[256 ]; +static __always_inline __attribute_const__ u8 generic___bitrev8(u8 byte)
[ ... ]
Quoted 18 lines
include/linux/bitrev.h +0
-10@@ -12,22 +12,10 @@#define __bitrev8 __arch_bitrev8#else -extern u8 const byte_rev_table[256 ]; -static inline u8 __bitrev8 (u8 byte) -{ -return byte_rev_table[byte]; -} - -static inline u16 __bitrev16 (u16 x) -{ -return (__bitrev8 (x &0xff ) <<8 ) |__bitrev8 (x >>8 ); -}
include/linux/bitrev.h line 18
Quoted 10 lines
include/linux/bitrev.h +4
-6- -static inline u32 __bitrev32 (u32 x) -{ -return (__bitrev16 (x &0xffff ) <<16 ) |__bitrev16 (x >>16 ); -} - +#include <asm-generic/bitops/__bitrev.h> +#define __bitrev32 generic___bitrev32 +#define __bitrev16 generic___bitrev16 +#define __bitrev8 generic___bitrev8
Does this introduce header fragility if CONFIG_BITREVERSE=n?
If CONFIG_BITREVERSE=n, the guard in <asm-generic/bitops/__bitrev.h> is skipped, but __bitrev32 is still unconditionally mapped to generic___bitrev32 here.
Any C file including <linux/bitrev.h> that contains an uninstantiated static inline function using bitrev32() will fail to compile with -Werror=implicit-function-declaration. Previously, unconditional declarations in <linux/bitrev.h> safely deferred such errors to link-time.
Quoted 1 line
include/linux/bitrev.h +0
-0#endif /* CONFIG_HAVE_ARCH_BITREVERSE */
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=3
[email protected]
AI reviewRe: [PATCH 4/6] arch/riscv: Add bitrev.h file to support rev8 and brev8
In reply to Yury NorovThank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [Low] Selecting GENERIC_BITREVERSE without satisfying its dependency causes a Kconfig unmet dependency…
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] Selecting GENERIC_BITREVERSE without satisfying its dependency causes a Kconfig unmet dependency warning. --
commit b26322371528c694cf7ce4bf22b509fc4ef43133
Author: Jinjie Ruan <[email protected]>
arch/riscv: Add bitrev.h file to support rev8 and brev8
This commit adds <asm/bitrev.h> to efficiently implement bit reversal using the 'rev8' and 'brev8' instructions from the RISC-V Zbkb extension. It updates Kconfig to select GENERIC_BITREVERSE and HAVE_ARCH_BITREVERSE to utilize these hardware instructions.
arch/riscv/Kconfig config RISCV · line 107
Quoted 9 lines
arch/riscv/Kconfig +1
-0@@ -104,6 +104,7 @@ config RISCV select FUNCTION_ALIGNMENT_8B if DYNAMIC_FTRACE_WITH_CALL_OPS select GENERIC_ARCH_TOPOLOGY select GENERIC_ATOMIC64 if !64BIT + select GENERIC_BITREVERSE
Will this select introduce a Kconfig unmet dependency regression?
In lib/Kconfig, GENERIC_BITREVERSE explicitly depends on BITREVERSE.
Selecting a symbol directly without also satisfying its dependencies forces the target symbol on, which causes Kconfig warnings at build time when BITREVERSE is disabled (for example, during allnoconfig builds).
Could this select also include BITREVERSE to satisfy the dependency chain?
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=4
[email protected]
AI reviewRe: [PATCH 5/6] lib: compile generic bitrev.c conditionally on GENERIC_BITREVERSE
In reply to Yury NorovThank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [Medium] The commit breaks loadable module support for `lib/bitrev.c` by forcibly compiling it statically…
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] The commit breaks loadable module support for lib/bitrev.c by forcibly compiling it statically into the core kernel (vmlinux) even when configured as a module (CONFIG_BITREVERSE=m). --
commit 7ea6ceee0bfd2cae761187221063672a5947f4c4
Author: Yury Norov <[email protected]>
lib: compile generic bitrev.c conditionally on GENERIC_BITREVERSE
This commit simplifies the build logic for the generic bit-reversal library. It changes the Makefile to compile lib/bitrev.c based on CONFIG_GENERIC_BITREVERSE instead of CONFIG_BITREVERSE, and removes the corresponding preprocessor conditionals from the source file.
lib/Makefile obj-$(CONFIG_DEBUG_PREEMPT) += smp_processor_id.o · line 148
Quoted 10 lines
lib/Makefile +1
-1@@ -145,7 +145,7 @@ obj-$(CONFIG_DEBUG_PREEMPT) += smp_processor_id.o obj-$(CONFIG_LIST_HARDENED) += list_debug.o obj-$(CONFIG_DEBUG_OBJECTS) += debugobjects.o -obj-$(CONFIG_BITREVERSE) += bitrev.o +obj-$(CONFIG_GENERIC_BITREVERSE) += bitrev.o
Will this break loadable module support for lib/bitrev.c?
CONFIG_BITREVERSE is a tristate, allowing the library to be built as a module. However, CONFIG_GENERIC_BITREVERSE is a boolean. When CONFIG_BITREVERSE is set to modular, CONFIG_GENERIC_BITREVERSE evaluates to y.
By switching the Makefile rule to use CONFIG_GENERIC_BITREVERSE, bitrev.o will be built into the core kernel unconditionally when the generic implementation is used, even if a modular build was requested.
Could this regression lead to unintended kernel bloat by ignoring the configuration intent?
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=5
Quoted 8 lines · Yury Norov
This series is a resend for Jinjie Ruan's "
arch/riscv: Addbitrev.hfile to support rev8 and brev8" [1], my follow-up "lib: compile generic bitrev based onGENERIC_BITREVERSE" [2], and the fix for a build error reported by Nathan Chancellor [3].No changes, except for combining pieces together and rebasing on top of the tree.
Quoted 6 lines
[1] https://lore.kernel.org/all/[email protected]/
[2] https://lore.kernel.org/all/[email protected]/
[3] https://lore.kernel.org/all/20260429202922.GA3575295@ax162/Build-tested against x86 tinyconfig and defconfig, having disabled and enabiled CRC32 and BITREVERSE, correspondingly.
I've got a feedback from sashiko bot.
1. We need to do #if IS_DEFINED(CONFIG_CRC32) instead of #ifdef, because it may be a module.
2. Selecting GENERIC_BITREVERSE without BITREVERSE causes the unmet direct dependency warning:
WARNING: unmet direct dependencies detected for GENERIC_BITREVERSE
Depends on [n]: BITREVERSE [=n]
Selected by [m]:
- MYCONFIG [=m]
I'll send a v2 to fix it
Thanks,
Yury
Arnd Bergmann
maintainerRe: [PATCH 1/6] lib: include crc32.h conditionally on CONFIG_CRC32
In reply to Yury NorovQuoted 7 lines · Yury Norov
Currently, bitreverse API is either declared based on
CONFIG_HAVE_ARCH_BITREVERSE, wired to arch implementation, or if the arch has no bitreverse, based on generic implementation.So, regardless of
CONFIG_BITREVERSE=n, the corresponding API is always declared. If that happens, the functions become declared but not implemented, which is an error.
I'm not following that description. Why is it an error to declare a funtion that is not implemented? Isn't that how optional interfaces tend to work in general?
Quoted 4 lines
The only header requiring the crc32 and bitreverse prototypes is
include/linux/etherdevice.h. Thus, protect inclusion of corresponding headers in the etherdevice withCONFIG_CRC32, together with the only function depending on it.
...
include/linux/etherdevice.h line 27
Quoted 8 lines
#include <
linux/if_ether.h>
#include <linux/netdevice.h>
#include <linux/random.h>include/linux/etherdevice.h +2
-0+#ifdef CONFIG_CRC32 #include <linux/crc32.h> +#endif #include <linux/unaligned.h> #include <asm/bitsperlong.h>
Don't add #ifdef blocks around headers. If the header cannot be included without side-effects, change the linux/crc32.h file instead of its users.
It looks like the problem is the check for CONFIG_GENERIC_BITREVERSE in include/asm-generic/bitops/__bitrev.h, which ends up hinding the generic___bitrev32() helper without need.
Simply removing the #ifdef there should avoid the build failure.
include/linux/etherdevice.h static inline u32 eth_hw_addr_crc(struct netdev_hw_addr *ha) · line 297
Quoted 9 lines
include/linux/etherdevice.h +2
-0+#ifdef CONFIG_CRC32 /** * eth_hw_addr_crc - Calculate CRC from netdev_hw_addr * @ha: pointer to hardware address @@ -291,6 +294,7 @@ static inline u32 eth_hw_addr_crc(struct netdev_hw_addr *ha) {return ether_crc (ETH_ALEN , ha->addr); } +#endif
I see there are only user users of this function, neither of them are performance critical. So the other options would be to either open-code this function in the two callers and remove it entirely, or move it into net/ethernet/eth.c.
Arnd
Re: [PATCH 1/6] lib: include crc32.h conditionally on CONFIG_CRC32
Quoted 37 lines · On Mon, 04 May 2026 10:03:10 +0200 "Arnd Bergmann" <[email protected]> wrote:
Quoted 7 lines · Yury Norov
Currently, bitreverse API is either declared based on
CONFIG_HAVE_ARCH_BITREVERSE, wired to arch implementation, or if the arch has no bitreverse, based on generic implementation.So, regardless of
CONFIG_BITREVERSE=n, the corresponding API is always declared. If that happens, the functions become declared but not implemented, which is an error.I'm not following that description. Why is it an error to declare a funtion that is not implemented? Isn't that how optional interfaces tend to work in general?
Quoted 4 lines
The only header requiring the crc32 and bitreverse prototypes is
include/linux/etherdevice.h. Thus, protect inclusion of corresponding headers in the etherdevice withCONFIG_CRC32, together with the only function depending on it....
include/linux/etherdevice.h line 27
Quoted 8 lines
#include <
linux/if_ether.h>
#include <linux/netdevice.h>
#include <linux/random.h>include/linux/etherdevice.h +2
-0+#ifdef CONFIG_CRC32 #include <linux/crc32.h> +#endif #include <linux/unaligned.h> #include <asm/bitsperlong.h> Don't add #ifdef blocks around headers. If the header cannot be included without side-effects, change the
linux/crc32.hfile instead of its users.It looks like the problem is the check for
CONFIG_GENERIC_BITREVERSEininclude/asm-generic/bitops/__bitrev.h, which ends up hinding thegeneric___bitrev32()helper without need.Simply removing the #ifdef there should avoid the build failure.
Quoted 14 lines
include/linux/etherdevice.h static inline u32 eth_hw_addr_crc(struct netdev_hw_addr *ha) · line 297
Quoted 9 lines
include/linux/etherdevice.h +2
-0+#ifdef CONFIG_CRC32 /** * eth_hw_addr_crc - Calculate CRC from netdev_hw_addr * @ha: pointer to hardware address @@ -291,6 +294,7 @@ static inline u32 eth_hw_addr_crc(struct netdev_hw_addr *ha) {return ether_crc (ETH_ALEN , ha->addr); } +#endif I see there are only user users of this function, neither of them are performance critical. So the other options would be to either open-code this function in the two callers and remove it entirely, or move it into
net/ethernet/eth.c.
Or change to a #define so that only the users need to have the required headers included.
But open-coding in the callers saves anyone trying to read the code having to look at another file to see what is going on.
-- David
Quoted 3 lines
Arnd
Re: [PATCH 1/6] lib: include crc32.h conditionally on CONFIG_CRC32
In reply to Arnd BergmannQuoted 5 lines · Arnd Bergmann
Quoted 4 lines · Yury Norov
Currently, bitreverse API is either declared based on
CONFIG_HAVE_ARCH_BITREVERSE, wired to arch implementation, or if the arch has no bitreverse, based on generic implementation.
Quoted 7 lines
Quoted 3 lines
So, regardless of
CONFIG_BITREVERSE=n, the corresponding API is always declared. If that happens, the functions become declared but not implemented, which is an error.I'm not following that description. Why is it an error to declare a funtion that is not implemented? Isn't that how optional interfaces tend to work in general?
Never heard about such a thing like "optional interface". And git grep tends to second that...
Quoted 5 lines
Quoted 4 lines
The only header requiring the crc32 and bitreverse prototypes is
include/linux/etherdevice.h. Thus, protect inclusion of corresponding headers in the etherdevice withCONFIG_CRC32, together with the only function depending on it....
Quoted 12 lines
include/linux/etherdevice.h line 27
Quoted 8 lines
#include <
linux/if_ether.h>
#include <linux/netdevice.h>
#include <linux/random.h>include/linux/etherdevice.h +2
-0+#ifdef CONFIG_CRC32 #include <linux/crc32.h> +#endif #include <linux/unaligned.h> #include <asm/bitsperlong.h> Don't add #ifdef blocks around headers. If the header cannot be included without side-effects, change the
linux/crc32.hfile instead of its users.
linux/acpi.h does that like many othes. What exactly is wrong with protecting headers inclusion?
Quoted 5 lines
It looks like the problem is the check for
CONFIG_GENERIC_BITREVERSEininclude/asm-generic/bitops/__bitrev.h, which ends up hinding thegeneric___bitrev32()helper without need.Simply removing the #ifdef there should avoid the build failure.
OK, it seems like this is what I don't understand.
We've got an optional feature, like CRC32, which is enabled by CONFIG_CRC32. The most conservative way is to declare everything CRC32-related in the corresponding header, and then protect the header with IS_ENABLED(CONFIG_CRC32).
I understand that from practical perspective, we can declare some simple macros, like header size, unprotected. But what we've got now is a sort of mess: all CRC32-related functions are declared unprotected, and generic headers are good to use them. Compiler is happy while those functions are actually unused. Next, CRC32 depends on BITREVERSE, which is again unprotected, and it may optionally have an arch implementation.
So if arch bitrev() is implemented, you can use part of bitreverse and crc32 APIs despite that they are explicitly disabled - just because they are implemented as macros in unprotected headers. And you cannot use some others - because they are implemented differently, as a real functions.
And this is not covered by any written rule - just the implementation details. And to make it worse, this all is available to drivers, which may simply fail after the next kernel update.
Can you please elaborate on how is that supposed to work. In my naive world, if I disable some feature, I'm pretty sure that my kernel shouldn't build if I try to use it. Can you point to any related kernel docs?
Thanks,
Yury
Quoted 16 lines
include/linux/etherdevice.h static inline u32 eth_hw_addr_crc(struct netdev_hw_addr *ha) · line 297
Quoted 9 lines
include/linux/etherdevice.h +2
-0+#ifdef CONFIG_CRC32 /** * eth_hw_addr_crc - Calculate CRC from netdev_hw_addr * @ha: pointer to hardware address @@ -291,6 +294,7 @@ static inline u32 eth_hw_addr_crc(struct netdev_hw_addr *ha) {return ether_crc (ETH_ALEN , ha->addr); } +#endif I see there are only user users of this function, neither of them are performance critical. So the other options would be to either open-code this function in the two callers and remove it entirely, or move it into
net/ethernet/eth.c.Arnd
Arnd Bergmann
maintainerRe: [PATCH 1/6] lib: include crc32.h conditionally on CONFIG_CRC32
Quoted 8 lines · Yury Norov
Quoted 4 lines · Arnd Bergmann
On Thu, Apr 30, 2026, at 23:13, Yury Norov wrote:
I'm not following that description. Why is it an error to declare a funtion that is not implemented? Isn't that how optional interfaces tend to work in general?Never heard about such a thing like "optional interface". And git grep tends to second that...
I meant any library interface that can be turned on or off
Quoted 18 lines
Quoted 5 lines
Quoted 4 lines
The only header requiring the crc32 and bitreverse prototypes is
include/linux/etherdevice.h. Thus, protect inclusion of corresponding headers in the etherdevice withCONFIG_CRC32, together with the only function depending on it....
Quoted 12 lines
include/linux/etherdevice.h line 27
Quoted 8 lines
#include <
linux/if_ether.h>
#include <linux/netdevice.h>
#include <linux/random.h>include/linux/etherdevice.h +2
-0+#ifdef CONFIG_CRC32 #include <linux/crc32.h> +#endif #include <linux/unaligned.h> #include <asm/bitsperlong.h> Don't add #ifdef blocks around headers. If the header cannot be included without side-effects, change the
linux/crc32.hfile instead of its users.
Quoted 2 lines
linux/acpi.hdoes that like many othes. What exactly is wrong with protecting headers inclusion?
There is no "protecting" here, you just add complexity to the build when headers are sometimes included indirectly and but other times are not, depending on kernel configuration.
It's unlikely to cause problems for the crc32.h header, but the acpi example definitely risks running into circular inclusions when you end up with some other header that depending on configuration ends up including linux/acpi.h while also bring included indirectly from that one.
Quoted 20 lines
Quoted 5 lines
It looks like the problem is the check for
CONFIG_GENERIC_BITREVERSEininclude/asm-generic/bitops/__bitrev.h, which ends up hinding thegeneric___bitrev32()helper without need.Simply removing the #ifdef there should avoid the build failure.
OK, it seems like this is what I don't understand.
We've got an optional feature, like CRC32, which is enabled by
CONFIG_CRC32. The most conservative way is to declare everything CRC32-related in the corresponding header, and then protect the header withIS_ENABLED(CONFIG_CRC32).I understand that from practical perspective, we can declare some simple macros, like header size, unprotected. But what we've got now is a sort of mess: all CRC32-related functions are declared unprotected, and generic headers are good to use them. Compiler is happy while those functions are actually unused. Next, CRC32 depends on BITREVERSE, which is again unprotected, and it may optionally have an arch implementation.
Quoted 4 lines
So if arch
bitrev()is implemented, you can use part of bitreverse and crc32 APIs despite that they are explicitly disabled - just because they are implemented as macros in unprotected headers. And you cannot use some others - because they are implemented differently, as a real functions.
I think you trying to solve a non-problem here. It is extremely rare for any kernel to be built without crc32 or bitrev. Even in randconfig builds, both are usually selected by some driver. The behavior we've always had here is that in the rare one randconfig in 10000 that hits the one driver missing 'select BITREVERSE', we get a link failure with an obvious fix. If you add the extra #ifdef, it gets a little more likely to actually run into the bug, and it shows up a little earlier as a compile failure instead of a link failure, but otherwise, nothing changes.
Arnd
Re: [PATCH 1/6] lib: include crc32.h conditionally on CONFIG_CRC32
Quoted 11 lines · Arnd Bergmann
Quoted 8 lines · Yury Norov
Quoted 4 lines · Arnd Bergmann
On Thu, Apr 30, 2026, at 23:13, Yury Norov wrote:
I'm not following that description. Why is it an error to declare a funtion that is not implemented? Isn't that how optional interfaces tend to work in general?Never heard about such a thing like "optional interface". And git grep tends to second that...
I meant any library interface that can be turned on or off
So? If I disable CRC32, can I use the either_crc()? In case of that networking header, the answer is yes. In some other piece of code the answer is no. Is that correct?
Quoted 18 lines
Quoted 18 lines
Quoted 17 lines
Quoted 4 lines
The only header requiring the crc32 and bitreverse prototypes is
include/linux/etherdevice.h. Thus, protect inclusion of corresponding headers in the etherdevice withCONFIG_CRC32, together with the only function depending on it....
include/linux/etherdevice.h line 27
Quoted 8 lines
#include <
linux/if_ether.h>
#include <linux/netdevice.h>
#include <linux/random.h>include/linux/etherdevice.h +2
-0+#ifdef CONFIG_CRC32 #include <linux/crc32.h> +#endif #include <linux/unaligned.h> #include <asm/bitsperlong.h> Don't add #ifdef blocks around headers. If the header cannot be included without side-effects, change the
linux/crc32.hfile instead of its users.
Quoted 6 lines
Quoted 2 lines
linux/acpi.hdoes that like many othes. What exactly is wrong with protecting headers inclusion?There is no "protecting" here, you just add complexity to the build when headers are sometimes included indirectly and but other times are not, depending on kernel configuration.
Sorry, don't understand... I use the 'protecting' term with the meaning:
the functionality that is explicitly disabled should be never used. Otherwise, what for we disable it?
This comes with the cost of increasing the config space, but that's how it works.
In terms of build complexity, I guess that the more you disable on preprocessing stage, the less a compiler should actually process.
Quoted 26 lines
It's unlikely to cause problems for the
crc32.hheader, but the acpi example definitely risks running into circular inclusions when you end up with some other header that depending on configuration ends up includinglinux/acpi.hwhile also bring included indirectly from that one.Quoted 20 lines
Quoted 5 lines
It looks like the problem is the check for
CONFIG_GENERIC_BITREVERSEininclude/asm-generic/bitops/__bitrev.h, which ends up hinding thegeneric___bitrev32()helper without need.Simply removing the #ifdef there should avoid the build failure.
OK, it seems like this is what I don't understand.
We've got an optional feature, like CRC32, which is enabled by
CONFIG_CRC32. The most conservative way is to declare everything CRC32-related in the corresponding header, and then protect the header withIS_ENABLED(CONFIG_CRC32).I understand that from practical perspective, we can declare some simple macros, like header size, unprotected. But what we've got now is a sort of mess: all CRC32-related functions are declared unprotected, and generic headers are good to use them. Compiler is happy while those functions are actually unused. Next, CRC32 depends on BITREVERSE, which is again unprotected, and it may optionally have an arch implementation.
Quoted 6 lines
Quoted 4 lines
So if arch
bitrev()is implemented, you can use part of bitreverse and crc32 APIs despite that they are explicitly disabled - just because they are implemented as macros in unprotected headers. And you cannot use some others - because they are implemented differently, as a real functions.I think you trying to solve a non-problem here.
This was reported by Nathan for tinyconfig. At least x86 and s390 are affected.
https://lore.kernel.org/all/20260429202922.GA3575295@ax162/
Is tinyconfig important?
Quoted 9 lines
It is extremely rare
for any kernel to be built without crc32 or bitrev. Even in randconfig builds, both are usually selected by some driver. The behavior we've always had here is that in the rare one randconfig in 10000 that hits the one driver missing 'select BITREVERSE', we get a link failure with an obvious fix. If you add the extra #ifdef, it gets a little more likely to actually run into the bug, and it shows up a little earlier as a compile failure instead of a link failure, but otherwise, nothing changes.
Except that it fixes the tinyconfig build.
The above argument sounds to me like you believe that CRC32 config is really useless because it's enabled for any practical configuration. And the same for BITREVERSE. Would it be better to get rid of them? Would simplification of config space overweights the bloat of the tinyconfig?
On the context: CONFIG_CRC32 is pre-historic, and BITREVERSE is 2006.
Right now half CRC32 is available if CONFIG_CRC32 is on, and half is not available. The bitreverse is the same. If HAVE_ARCH_BITREVERSE is enabled, one can use the API, bypassing the BITREVERSE. This doesn't sound right to me long-term.
Whatever this ends up, let's figure out a consistent solution please?
Thanks,
Yury
Arnd Bergmann
maintainerRe: [PATCH 1/6] lib: include crc32.h conditionally on CONFIG_CRC32
Quoted 10 lines · Yury Norov
Quoted 5 lines · Arnd Bergmann
Quoted 2 lines · Yury Norov
Never heard about such a thing like "optional interface". And git grep tends to second that...
I meant any library interface that can be turned on or off
So? If I disable CRC32, can I use the
either_crc()? In case of that networking header, the answer is yes. In some other piece of code the answer is no. Is that correct?
Since it's a macro defiend in terms of both bitref32 and crc32_le, you can only call it from dead code, such as an inline function that is not itself used, or from inside of a block that is protected with IS_ENABLED(CONFIG_CRC32) etc.
Quoted 8 lines
Quoted 8 lines
Quoted 7 lines
Quoted 4 lines
Don't add #ifdef blocks around headers. If the header cannot be included without side-effects, change the
linux/crc32.hfile instead of its users.
linux/acpi.hdoes that like many othes. What exactly is wrong with protecting headers inclusion?
Quoted 7 lines
Quoted 3 lines
There is no "protecting" here, you just add complexity to the build when headers are sometimes included indirectly and but other times are not, depending on kernel configuration.
Sorry, don't understand... I use the 'protecting' term with the meaning:
the functionality that is explicitly disabled should be never used. Otherwise, what for we disable it?
Arguably, both configuration symbols are at the point of not actually saving enough object code size to actually be worth the Kconfig dependencies.
As long as we have CONFIG_CRC32 and CONFIG_BITREVERSE, the point of having the Kconfig symbols is to let drivers request the inclusion of the library helpers.
Quoted 31 lines
Quoted 31 lines
It's unlikely to cause problems for the
crc32.hheader, but the acpi example definitely risks running into circular inclusions when you end up with some other header that depending on configuration ends up includinglinux/acpi.hwhile also bring included indirectly from that one.Quoted 24 lines
Quoted 5 lines
It looks like the problem is the check for
CONFIG_GENERIC_BITREVERSEininclude/asm-generic/bitops/__bitrev.h, which ends up hinding thegeneric___bitrev32()helper without need.Simply removing the #ifdef there should avoid the build failure.
OK, it seems like this is what I don't understand.
We've got an optional feature, like CRC32, which is enabled by
CONFIG_CRC32. The most conservative way is to declare everything CRC32-related in the corresponding header, and then protect the header withIS_ENABLED(CONFIG_CRC32).I understand that from practical perspective, we can declare some simple macros, like header size, unprotected. But what we've got now is a sort of mess: all CRC32-related functions are declared unprotected, and generic headers are good to use them. Compiler is happy while those functions are actually unused. Next, CRC32 depends on BITREVERSE, which is again unprotected, and it may optionally have an arch implementation.
So if arch
bitrev()is implemented, you can use part of bitreverse and crc32 APIs despite that they are explicitly disabled - just because they are implemented as macros in unprotected headers. And you cannot use some others - because they are implemented differently, as a real functions.
Quoted 8 lines
Quoted 1 line
I think you trying to solve a non-problem here.
This was reported by Nathan for tinyconfig. At least x86 and s390 are affected.
https://lore.kernel.org/all/20260429202922.GA3575295@ax162/
Is tinyconfig important?
Nathan reported a build regression caused by a small mistake in 596a9ea9015b ("bitops: Define generic __bitrev8/16/32 for reuse"), which is of course needs to be fixed.
What I meant is that there is no reason to not use the obvious fix and do
include/asm-generic/bitops/__bitrev.h +0 -2
@@ -2,7 +2,6 @@#ifndef _ASM_GENERIC_BITOPS___BITREV_H_#define _ASM_GENERIC_BITOPS___BITREV_H_ -#ifdef CONFIG_GENERIC_BITREVERSE #include <asm/types.h> extern u8 const byte_rev_table[256 ];
@@ -20,6 +19,5 @@ static __always_inline __attribute_const__ u32 generic___bitrev32(u32 x) {return (generic___bitrev16 (x &0xffff ) <<16 ) |generic___bitrev16 (x >>16 ); } -#endif /* CONFIG_GENERIC_BITREVERSE */ #endif /* _ASM_GENERIC_BITOPS___BITREV_H_ */
Quoted 6 lines
Right now half CRC32 is available if
CONFIG_CRC32is on, and half is not available. The bitreverse is the same. IfHAVE_ARCH_BITREVERSEis enabled, one can use the API, bypassing the BITREVERSE. This doesn't sound right to me long-term.Whatever this ends up, let's figure out a consistent solution please?
I really don't think we need any sort of solution here, aside from the trivial regression fix that returns it to the previous working state.
Arnd
Re: [PATCH 1/6] lib: include crc32.h conditionally on CONFIG_CRC32
Quoted 111 lines · Arnd Bergmann
Quoted 10 lines · Yury Norov
Quoted 5 lines · Arnd Bergmann
Quoted 2 lines · Yury Norov
Never heard about such a thing like "optional interface". And git grep tends to second that...
I meant any library interface that can be turned on or off
So? If I disable CRC32, can I use the
either_crc()? In case of that networking header, the answer is yes. In some other piece of code the answer is no. Is that correct?Since it's a macro defiend in terms of both bitref32 and
crc32_le, you can only call it from dead code, such as an inline function that is not itself used, or from inside of a block that is protected withIS_ENABLED(CONFIG_CRC32)etc.Quoted 8 lines
Quoted 8 lines
Quoted 7 lines
Quoted 4 lines
Don't add #ifdef blocks around headers. If the header cannot be included without side-effects, change the
linux/crc32.hfile instead of its users.
linux/acpi.hdoes that like many othes. What exactly is wrong with protecting headers inclusion?Quoted 7 lines
Quoted 3 lines
There is no "protecting" here, you just add complexity to the build when headers are sometimes included indirectly and but other times are not, depending on kernel configuration.
Sorry, don't understand... I use the 'protecting' term with the meaning:
the functionality that is explicitly disabled should be never used. Otherwise, what for we disable it?Arguably, both configuration symbols are at the point of not actually saving enough object code size to actually be worth the
Kconfigdependencies.As long as we have
CONFIG_CRC32andCONFIG_BITREVERSE, the point of having theKconfigsymbols is to let drivers request the inclusion of the library helpers.Quoted 31 lines
Quoted 31 lines
It's unlikely to cause problems for the
crc32.hheader, but the acpi example definitely risks running into circular inclusions when you end up with some other header that depending on configuration ends up includinglinux/acpi.hwhile also bring included indirectly from that one.Quoted 24 lines
Quoted 5 lines
It looks like the problem is the check for
CONFIG_GENERIC_BITREVERSEininclude/asm-generic/bitops/__bitrev.h, which ends up hinding thegeneric___bitrev32()helper without need.Simply removing the #ifdef there should avoid the build failure.
OK, it seems like this is what I don't understand.
We've got an optional feature, like CRC32, which is enabled by
CONFIG_CRC32. The most conservative way is to declare everything CRC32-related in the corresponding header, and then protect the header withIS_ENABLED(CONFIG_CRC32).I understand that from practical perspective, we can declare some simple macros, like header size, unprotected. But what we've got now is a sort of mess: all CRC32-related functions are declared unprotected, and generic headers are good to use them. Compiler is happy while those functions are actually unused. Next, CRC32 depends on BITREVERSE, which is again unprotected, and it may optionally have an arch implementation.
So if arch
bitrev()is implemented, you can use part of bitreverse and crc32 APIs despite that they are explicitly disabled - just because they are implemented as macros in unprotected headers. And you cannot use some others - because they are implemented differently, as a real functions.Quoted 8 lines
Quoted 1 line
I think you trying to solve a non-problem here.
This was reported by Nathan for tinyconfig. At least x86 and s390 are affected.
https://lore.kernel.org/all/20260429202922.GA3575295@ax162/
Is tinyconfig important?
Nathan reported a build regression caused by a small mistake in
596a9ea9015b("bitops: Define generic __bitrev8/16/32 for reuse"), which is of course needs to be fixed.What I meant is that there is no reason to not use the obvious fix and do
include/asm-generic/bitops/__bitrev.h +0
-2@@ -2,7 +2,6 @@#ifndef _ASM_GENERIC_BITOPS___BITREV_H_#define _ASM_GENERIC_BITOPS___BITREV_H_ -#ifdef CONFIG_GENERIC_BITREVERSE #include <asm/types.h> extern u8 const byte_rev_table[256 ];@@ -20,6 +19,5 @@ static __always_inline __attribute_const__ u32 generic___bitrev32(u32 x) {return (generic___bitrev16 (x &0xffff ) <<16 ) |generic___bitrev16 (x >>16 ); } -#endif /* CONFIG_GENERIC_BITREVERSE */ #endif /* _ASM_GENERIC_BITOPS___BITREV_H_ */ Quoted 6 lines
Right now half CRC32 is available if
CONFIG_CRC32is on, and half is not available. The bitreverse is the same. IfHAVE_ARCH_BITREVERSEis enabled, one can use the API, bypassing the BITREVERSE. This doesn't sound right to me long-term.
Quoted 5 lines
Quoted 1 line
Whatever this ends up, let's figure out a consistent solution please?
I really don't think we need any sort of solution here, aside from the trivial regression fix that returns it to the previous working state.
OK. Not that I've got the answers to my questions, but I trust your expertise, so will do as suggested.
Thanks,
Yury
Re: [PATCH 1/6] lib: include crc32.h conditionally on CONFIG_CRC32
In reply to Arnd BergmannQuoted 111 lines · Arnd Bergmann
Quoted 10 lines · Yury Norov
Quoted 5 lines · Arnd Bergmann
Quoted 2 lines · Yury Norov
Never heard about such a thing like "optional interface". And git grep tends to second that...
I meant any library interface that can be turned on or off
So? If I disable CRC32, can I use the
either_crc()? In case of that networking header, the answer is yes. In some other piece of code the answer is no. Is that correct?Since it's a macro defiend in terms of both bitref32 and
crc32_le, you can only call it from dead code, such as an inline function that is not itself used, or from inside of a block that is protected withIS_ENABLED(CONFIG_CRC32)etc.Quoted 8 lines
Quoted 8 lines
Quoted 7 lines
Quoted 4 lines
Don't add #ifdef blocks around headers. If the header cannot be included without side-effects, change the
linux/crc32.hfile instead of its users.
linux/acpi.hdoes that like many othes. What exactly is wrong with protecting headers inclusion?Quoted 7 lines
Quoted 3 lines
There is no "protecting" here, you just add complexity to the build when headers are sometimes included indirectly and but other times are not, depending on kernel configuration.
Sorry, don't understand... I use the 'protecting' term with the meaning:
the functionality that is explicitly disabled should be never used. Otherwise, what for we disable it?Arguably, both configuration symbols are at the point of not actually saving enough object code size to actually be worth the
Kconfigdependencies.As long as we have
CONFIG_CRC32andCONFIG_BITREVERSE, the point of having theKconfigsymbols is to let drivers request the inclusion of the library helpers.Quoted 31 lines
Quoted 31 lines
It's unlikely to cause problems for the
crc32.hheader, but the acpi example definitely risks running into circular inclusions when you end up with some other header that depending on configuration ends up includinglinux/acpi.hwhile also bring included indirectly from that one.Quoted 24 lines
Quoted 5 lines
It looks like the problem is the check for
CONFIG_GENERIC_BITREVERSEininclude/asm-generic/bitops/__bitrev.h, which ends up hinding thegeneric___bitrev32()helper without need.Simply removing the #ifdef there should avoid the build failure.
OK, it seems like this is what I don't understand.
We've got an optional feature, like CRC32, which is enabled by
CONFIG_CRC32. The most conservative way is to declare everything CRC32-related in the corresponding header, and then protect the header withIS_ENABLED(CONFIG_CRC32).I understand that from practical perspective, we can declare some simple macros, like header size, unprotected. But what we've got now is a sort of mess: all CRC32-related functions are declared unprotected, and generic headers are good to use them. Compiler is happy while those functions are actually unused. Next, CRC32 depends on BITREVERSE, which is again unprotected, and it may optionally have an arch implementation.
So if arch
bitrev()is implemented, you can use part of bitreverse and crc32 APIs despite that they are explicitly disabled - just because they are implemented as macros in unprotected headers. And you cannot use some others - because they are implemented differently, as a real functions.Quoted 8 lines
Quoted 1 line
I think you trying to solve a non-problem here.
This was reported by Nathan for tinyconfig. At least x86 and s390 are affected.
https://lore.kernel.org/all/20260429202922.GA3575295@ax162/
Is tinyconfig important?
Nathan reported a build regression caused by a small mistake in
596a9ea9015b("bitops: Define generic __bitrev8/16/32 for reuse"), which is of course needs to be fixed.What I meant is that there is no reason to not use the obvious fix and do
include/asm-generic/bitops/__bitrev.h +0
-2@@ -2,7 +2,6 @@#ifndef _ASM_GENERIC_BITOPS___BITREV_H_#define _ASM_GENERIC_BITOPS___BITREV_H_ -#ifdef CONFIG_GENERIC_BITREVERSE #include <asm/types.h> extern u8 const byte_rev_table[256 ];@@ -20,6 +19,5 @@ static __always_inline __attribute_const__ u32 generic___bitrev32(u32 x) {return (generic___bitrev16 (x &0xffff ) <<16 ) |generic___bitrev16 (x >>16 ); } -#endif /* CONFIG_GENERIC_BITREVERSE */ #endif /* _ASM_GENERIC_BITOPS___BITREV_H_ */ Quoted 6 lines
Right now half CRC32 is available if
CONFIG_CRC32is on, and half is not available. The bitreverse is the same. IfHAVE_ARCH_BITREVERSEis enabled, one can use the API, bypassing the BITREVERSE. This doesn't sound right to me long-term.
Quoted 5 lines
Quoted 1 line
Whatever this ends up, let's figure out a consistent solution please?
I really don't think we need any sort of solution here, aside from the trivial regression fix that returns it to the previous working state.
I agree with Arnd. The standard practice in the kernel is for headers to be includeable even when the associated kconfig (if any) isn't enabled. In such a case usually the inline functions, macros, and the declarations of out-of-line functions and global variables are still made available in the header.
The point of the kconfig symbol is just to build the file containing the definitions of out-of-line functions and global variables.
Yes, when those definitions are needed and calling code forgets to select that kconfig symbol, a build error results. But it's "only" a build error; those get found and fixed relatively easily. The alternative of hiding everything would mean a lot more #ifdefs in other code, which ends up being very messy.
- Eric
This series was applied to riscv/linux.git (fixes) by Yury Norov <[email protected]>:
Hello:
This series was applied to riscv/linux.git (fixes) by Yury Norov <[email protected]>:
Quoted 9 lines · On Thu, 30 Apr 2026 17:13:44 -0400 you wrote:
This series is a resend for Jinjie Ruan's "
arch/riscv: Addbitrev.hfile to support rev8 and brev8" [1], my follow-up "lib: compile generic bitrev based onGENERIC_BITREVERSE" [2], and the fix for a build error reported by Nathan Chancellor [3].No changes, except for combining pieces together and rebasing on top of the tree.
[...]
Here is the summary with links:
- [1/6] lib: include crc32.h conditionally on CONFIG_CRC32 (no matching commit)
- [2/6] lib/bitrev: Introduce GENERIC_BITREVERSE and cleanup Kconfig (no matching commit)
- [3/6] bitops: Define generic __bitrev8/16/32 for reuse
(no matching commit)
- [4/6] arch/riscv: Add bitrev.h file to support rev8 and brev8 (no matching commit)
- [5/6] lib: compile generic bitrev.c conditionally on GENERIC_BITREVERSE (no matching commit)
- [6/6] MAINTAINERS: BITOPS: include bitrev.[ch]
https://git.kernel.org/riscv/c/7b2c5b4e43aa
You are awesome, thank you!
Signature · 2 lines
Deet-doot-dot, I am a bot. https://korg.docs.kernel.org/patchwork/pwbot.html