linux/arch/parisc/lib, branch master

parisc: Drop ip_fast_csum() inline assembly implementation

2026-04-17T09:32:46Z

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+

parisc: Remove spurious if statement from raw_copy_from_user()

2025-10-07T14:44:29Z

Accidently introduced in commit 91428ca9320e. Signed-off-by: John David Anglin Signed-off-by: Helge Deller Fixes: 91428ca9320e ("parisc: Check region is readable by user in raw_copy_from_user()") Cc: stable@vger.kernel.org # v5.12+

parisc: Check region is readable by user in raw_copy_from_user()

2025-07-25T20:45:23Z

Because of the way the _PAGE_READ is handled in the parisc PTE, an access interruption is not generated when the kernel reads from a region where the _PAGE_READ is zero. The current code was written assuming read access faults would also occur in the kernel. This change adds user access checks to raw_copy_from_user(). The prober_user() define checks whether user code has read access to a virtual address. Note that page faults are not handled in the exception support for the probe instruction. For this reason, we precede the probe by a ldb access check. Signed-off-by: John David Anglin Signed-off-by: Helge Deller Cc: stable@vger.kernel.org # v5.12+

parisc: Remove memcpy_fromio

2025-02-03T18:27:00Z

Fully migrate parisc to the IO functions from lib/iomem_copy.c. In a recent patch the functions memset_io and memcpy_toio were removed, but the memcpy_fromio was kept, because for very short sequences it does half word accesses, whereas the functions in lib/iomem_copy.c do byte accesses until the memory is naturally aligned and then do machine word accesses. But I don't think the single half-word access merits keeping the arch specific implementation, so, remove it as well. Signed-off-by: Julian Vetter Acked-by: Arnd Bergmann Signed-off-by: Helge Deller

parisc: Fix formatting errors in io.c

2025-02-03T18:26:52Z

Mutliple lines in the file contain tabs in lines where there is no code. Just remove them. Signed-off-by: Julian Vetter Signed-off-by: Helge Deller

Merge tag 'parisc-for-6.14-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux

2025-01-24T19:52:59Z

Pull parisc architecture updates from Helge Deller: - Temporarily disable jump label support to avoid kernel crash with 32-bit kernel - Add vdso linker script to 'targets' instead of extra-y - Remove parisc versions of memcpy_toio and memset_io * tag 'parisc-for-6.14-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux: parisc: Temporarily disable jump label support parisc: add vdso linker script to 'targets' instead of extra-y parisc: Remove memcpy_toio and memset_io

parisc: Remove memcpy_toio and memset_io

2024-12-07T16:46:09Z

Recently new functions for IO memcpy and IO memset were added in libs/iomem_copy.c. So, remove the arch specific implementations, to fall back to the generic ones which do exactly the same. Keep memcpy_fromio for now, because it's slight more optimized by doing 'u16' accesses if the buffer is aligned this way. Signed-off-by: Julian Vetter Reviewed-by: Arnd Bergmann Signed-off-by: Helge Deller

net: checksum: Move from32to16() to generic header

2024-10-30T14:29:59Z

from32to16() is used by lib/checksum.c and also by arch/parisc/lib/checksum.c. The next patch will use it in the bpf_csum_diff helper. Move from32to16() to the include/net/checksum.h as csum_from32to16() and remove other implementations. Signed-off-by: Puranjay Mohan Signed-off-by: Daniel Borkmann Reviewed-by: Toke Høiland-Jørgensen Acked-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20241026125339.26459-2-puranjay@kernel.org

parisc: add u16 support to cmpxchg()

2024-04-10T05:06:00Z

Add (and export) __cmpxchg_u16(), teach __cmpxchg() to use it. And get rid of manual truncation down to u8, etc. in there - the only reason for those is to avoid bogus warnings about constant truncation from sparse, and those are easy to avoid by turning that switch into conditional expression. Signed-off-by: Al Viro Signed-off-by: Paul E. McKenney

parisc: unify implementations of __cmpxchg_u{8,32,64}

2024-04-10T05:06:00Z

identical except for type name involved Signed-off-by: Al Viro Signed-off-by: Paul E. McKenney