riscv: Allow PROT_WRITE-only mmap()
authorAndrew Bresticker <abrestic@rivosinc.com>
Thu, 15 Sep 2022 19:37:02 +0000 (15:37 -0400)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 26 Oct 2022 10:34:22 +0000 (12:34 +0200)
commit3c3c4fa118a44d554c2eafd56bb99a529720234c
tree6cf1ee989ce28bf8643d98d99929d9d34fc46007
parentaf3aaee08df8127422e03e5e1291f91cbb5d4f09
riscv: Allow PROT_WRITE-only mmap()

commit 9e2e6042a7ec6504fe8e366717afa2f40cf16488 upstream.

Commit 2139619bcad7 ("riscv: mmap with PROT_WRITE but no PROT_READ is
invalid") made mmap() return EINVAL if PROT_WRITE was set wihtout
PROT_READ with the justification that a write-only PTE is considered a
reserved PTE permission bit pattern in the privileged spec. This check
is unnecessary since we let VM_WRITE imply VM_READ on RISC-V, and it is
inconsistent with other architectures that don't support write-only PTEs,
creating a potential software portability issue. Just remove the check
altogether and let PROT_WRITE imply PROT_READ as is the case on other
architectures.

Note that this also allows PROT_WRITE|PROT_EXEC mappings which were
disallowed prior to the aforementioned commit; PROT_READ is implied in
such mappings as well.

Fixes: 2139619bcad7 ("riscv: mmap with PROT_WRITE but no PROT_READ is invalid")
Reviewed-by: Atish Patra <atishp@rivosinc.com>
Signed-off-by: Andrew Bresticker <abrestic@rivosinc.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20220915193702.2201018-3-abrestic@rivosinc.com/
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
arch/riscv/kernel/sys_riscv.c