From: Helge Deller Date: Fri, 10 Apr 2026 14:12:31 +0000 (+0200) Subject: parisc: Drop ip_fast_csum() inline assembly implementation X-Git-Url: http://git.ipfire.org/gitweb.cgi?a=commitdiff_plain;h=3dd31a370c1dccb580f729af7c580ccb1ae3c0c9;p=thirdparty%2Fkernel%2Flinux.git parisc: Drop ip_fast_csum() inline assembly implementation The assembly code of ip_fast_csum() triggers unaligned access warnings if the IP header isn't correctly aligned: Kernel: unaligned access to 0x173d22e76 in inet_gro_receive+0xbc/0x2e8 (iir 0x0e8810b6) Kernel: unaligned access to 0x173d22e7e in inet_gro_receive+0xc4/0x2e8 (iir 0x0e88109a) Kernel: unaligned access to 0x173d22e82 in inet_gro_receive+0xc8/0x2e8 (iir 0x0e90109d) Kernel: unaligned access to 0x173d22e7a in inet_gro_receive+0xd0/0x2e8 (iir 0x0e9810b8) Kernel: unaligned access to 0x173d22e86 in inet_gro_receive+0xdc/0x2e8 (iir 0x0e8810b8) We have the option to a) ignore the warnings, b) work around it by adding more code to check for alignment, or c) to switch to the generic implementation and rely on the compiler to optimize the code. Let's go with c), because a) isn't nice, and b) would effectively lead to an implementation which is basically equal to c). Signed-off-by: Helge Deller Cc: stable@vger.kernel.org # v7.0+ --- diff --git a/arch/parisc/Kconfig b/arch/parisc/Kconfig index 62d5a89d5c7bc..d7ee2f18bccde 100644 --- a/arch/parisc/Kconfig +++ b/arch/parisc/Kconfig @@ -130,6 +130,9 @@ config GENERIC_BUG config GENERIC_BUG_RELATIVE_POINTERS bool +config GENERIC_CSUM + def_bool y + config GENERIC_HWEIGHT bool default y diff --git a/arch/parisc/include/asm/checksum.h b/arch/parisc/include/asm/checksum.h index 2aceebcd695c8..382758808726a 100644 --- a/arch/parisc/include/asm/checksum.h +++ b/arch/parisc/include/asm/checksum.h @@ -4,73 +4,7 @@ #include -/* - * computes the checksum of a memory block at buff, length len, - * and adds in "sum" (32-bit) - * - * returns a 32-bit number suitable for feeding into itself - * or csum_tcpudp_magic - * - * this function must be called with even lengths, except - * for the last fragment, which may be odd - * - * it's best to have buff aligned on a 32-bit boundary - */ -extern __wsum csum_partial(const void *, int, __wsum); - -/* - * Optimized for IP headers, which always checksum on 4 octet boundaries. - * - * Written by Randolph Chung , and then mucked with by - * LaMont Jones - */ -static inline __sum16 ip_fast_csum(const void *iph, unsigned int ihl) -{ - unsigned int sum; - unsigned long t0, t1, t2; - - __asm__ __volatile__ ( -" ldws,ma 4(%1), %0\n" -" addib,<= -4, %2, 2f\n" -"\n" -" ldws 4(%1), %4\n" -" ldws 8(%1), %5\n" -" add %0, %4, %0\n" -" ldws,ma 12(%1), %3\n" -" addc %0, %5, %0\n" -" addc %0, %3, %0\n" -"1: ldws,ma 4(%1), %3\n" -" addib,> -1, %2, 1b\n" -" addc %0, %3, %0\n" -"\n" -" extru %0, 31, 16, %4\n" -" extru %0, 15, 16, %5\n" -" addc %4, %5, %0\n" -" extru %0, 15, 16, %5\n" -" add %0, %5, %0\n" -" subi -1, %0, %0\n" -"2:\n" - : "=r" (sum), "=r" (iph), "=r" (ihl), "=r" (t0), "=r" (t1), "=r" (t2) - : "1" (iph), "2" (ihl) - : "memory"); - - return (__force __sum16)sum; -} - -/* - * Fold a partial checksum - */ -static inline __sum16 csum_fold(__wsum csum) -{ - u32 sum = (__force u32)csum; - /* add the swapped two 16-bit halves of sum, - a possible carry from adding the two 16-bit halves, - will carry from the lower half into the upper half, - giving us the correct sum in the upper half. */ - sum += (sum << 16) + (sum >> 16); - return (__force __sum16)(~sum >> 16); -} - +#define csum_tcpudp_nofold csum_tcpudp_nofold static inline __wsum csum_tcpudp_nofold(__be32 saddr, __be32 daddr, __u32 len, __u8 proto, __wsum sum) @@ -85,26 +19,7 @@ static inline __wsum csum_tcpudp_nofold(__be32 saddr, __be32 daddr, return sum; } -/* - * computes the checksum of the TCP/UDP pseudo-header - * returns a 16-bit checksum, already complemented - */ -static inline __sum16 csum_tcpudp_magic(__be32 saddr, __be32 daddr, - __u32 len, __u8 proto, - __wsum sum) -{ - return csum_fold(csum_tcpudp_nofold(saddr,daddr,len,proto,sum)); -} - -/* - * this routine is used for miscellaneous IP-like checksums, mainly - * in icmp.c - */ -static inline __sum16 ip_compute_csum(const void *buf, int len) -{ - return csum_fold (csum_partial(buf, len, 0)); -} - +#include #define _HAVE_ARCH_IPV6_CSUM static __inline__ __sum16 csum_ipv6_magic(const struct in6_addr *saddr, diff --git a/arch/parisc/lib/Makefile b/arch/parisc/lib/Makefile index 7b197667faf6c..d5975d1fb4068 100644 --- a/arch/parisc/lib/Makefile +++ b/arch/parisc/lib/Makefile @@ -3,7 +3,7 @@ # Makefile for parisc-specific library files # -lib-y := lusercopy.o bitops.o checksum.o io.o memset.o memcpy.o \ +lib-y := lusercopy.o bitops.o io.o memset.o memcpy.o \ ucmpdi2.o delay.o obj-y := iomap.o diff --git a/arch/parisc/lib/checksum.c b/arch/parisc/lib/checksum.c deleted file mode 100644 index 59d8c15d81bd0..0000000000000 --- a/arch/parisc/lib/checksum.c +++ /dev/null @@ -1,99 +0,0 @@ -// SPDX-License-Identifier: GPL-2.0-or-later -/* - * INET An implementation of the TCP/IP protocol suite for the LINUX - * operating system. INET is implemented using the BSD Socket - * interface as the means of communication with the user level. - * - * MIPS specific IP/TCP/UDP checksumming routines - * - * Authors: Ralf Baechle, - * Lots of code moved from tcp.c and ip.c; see those files - * for more names. - */ -#include -#include - -#include -#include -#include -#include - -#define addc(_t,_r) \ - __asm__ __volatile__ ( \ -" add %0, %1, %0\n" \ -" addc %0, %%r0, %0\n" \ - : "=r"(_t) \ - : "r"(_r), "0"(_t)); - -static inline unsigned int do_csum(const unsigned char * buff, int len) -{ - int odd, count; - unsigned int result = 0; - - if (len <= 0) - goto out; - odd = 1 & (unsigned long) buff; - if (odd) { - result = be16_to_cpu(*buff); - len--; - buff++; - } - count = len >> 1; /* nr of 16-bit words.. */ - if (count) { - if (2 & (unsigned long) buff) { - result += *(unsigned short *) buff; - count--; - len -= 2; - buff += 2; - } - count >>= 1; /* nr of 32-bit words.. */ - if (count) { - while (count >= 4) { - unsigned int r1, r2, r3, r4; - r1 = *(unsigned int *)(buff + 0); - r2 = *(unsigned int *)(buff + 4); - r3 = *(unsigned int *)(buff + 8); - r4 = *(unsigned int *)(buff + 12); - addc(result, r1); - addc(result, r2); - addc(result, r3); - addc(result, r4); - count -= 4; - buff += 16; - } - while (count) { - unsigned int w = *(unsigned int *) buff; - count--; - buff += 4; - addc(result, w); - } - result = (result & 0xffff) + (result >> 16); - } - if (len & 2) { - result += *(unsigned short *) buff; - buff += 2; - } - } - if (len & 1) - result += le16_to_cpu(*buff); - result = csum_from32to16(result); - if (odd) - result = swab16(result); -out: - return result; -} - -/* - * computes a partial checksum, e.g. for TCP/UDP fragments - */ -/* - * why bother folding? - */ -__wsum csum_partial(const void *buff, int len, __wsum sum) -{ - unsigned int result = do_csum(buff, len); - addc(result, sum); - return (__force __wsum)csum_from32to16(result); -} - -EXPORT_SYMBOL(csum_partial);