[<prev] [next>] [day] [month] [year] [list]
Message-ID: <PH7PR14MB55947E9E82EFE9A2817CD692CEB49@PH7PR14MB5594.namprd14.prod.outlook.com>
Date: Fri, 24 Jun 2022 18:14:28 +0800
From: Celeste Liu <coelacanthus@...look.com>
To: Palmer Dabbelt <palmer@...osinc.com>,
Paul Walmsley <paul.walmsley@...ive.com>,
Albert Ou <aou@...s.berkeley.edu>, Guo Ren <guoren@...nel.org>,
Atish Patra <atishp@...osinc.com>,
Jisheng Zhang <jszhang@...nel.org>,
linux-riscv@...ts.infradead.org
Cc: Celeste Liu <coelacanthus@...look.com>,
Xiongchuan Tan <xc-tan@...look.com>,
Wang Ruikang <dramforever@...e.com>,
Ruizhe Pan <c141028@...il.com>, linux-kernel@...r.kernel.org,
Yash Shah <yash.shah@...ive.com>
Subject: [PATCH v4] riscv: don't allow write but not read page mapping request in mmap
When Xiongchuan Tan tries to run oe of libaio's tests[1], it encounters
a strange behavior: for the same PROT_WRITE only mapping, there was a
discrepancy in whether it could be read before and after writing
(readable before writing, unreadable after writing).
After some investigation, I found that mmap allows write only mapping,
an undefined behavior, on RISC-V.
As mentioned in Table 4.5 in RISC-V spec Volume 2 Section 4.3 version
"20211203 Privileged Architecture v1.12, Ratified"[2], the PTE permission
bit combination of "write+!read" is "Reserved for future use.". Hence,
don't allow such mapping request in mmap call.
In the current code[3], write+exec only is marked as invalid,
but write only is not marked as invalid.
This patch refines that judgment.
[1]: https://pagure.io/libaio/blob/1b18bfafc6a2f7b9fa2c6be77a95afed8b7be448/f/harness/cases/5.t
[2]: https://github.com/riscv/riscv-isa-manual/releases/download/Priv-v1.12/riscv-privileged-20211203.pdf
[3]: commit e0d17c842c0f ("RISC-V: Don't allow write+exec only page mapping request in mmap")
Reported-by: Xiongchuan Tan <xc-tan@...look.com>
Co-developed-by: Wang Ruikang <dramforever@...e.com>
Signed-off-by: Wang Ruikang <dramforever@...e.com>
Co-developed-by: Ruizhe Pan <c141028@...il.com>
Signed-off-by: Ruizhe Pan <c141028@...il.com>
Signed-off-by: Celeste Liu <coelacanthus@...look.com>
Reviewed-by: Guo Ren <guoren@...nel.org>
Cc: linux-riscv@...ts.infradead.org
Cc: linux-kernel@...r.kernel.org
Cc: Guo Ren <guoren@...nel.org>
Cc: Yash Shah <yash.shah@...ive.com>
---
v2: This version adds a link to the referenced spec, and reference of the
previous related modification.
v3: fix DCO signoff name and add comment for reason
v4: fix typo in changelog and comment
arch/riscv/kernel/sys_riscv.c | 11 ++++++++---
1 file changed, 8 insertions(+), 3 deletions(-)
diff --git a/arch/riscv/kernel/sys_riscv.c b/arch/riscv/kernel/sys_riscv.c
index 9c0194f176fc..f29456cfe68a 100644
--- a/arch/riscv/kernel/sys_riscv.c
+++ b/arch/riscv/kernel/sys_riscv.c
@@ -18,9 +18,14 @@ static long riscv_sys_mmap(unsigned long addr, unsigned long len,
if (unlikely(offset & (~PAGE_MASK >> page_shift_offset)))
return -EINVAL;
- if ((prot & PROT_WRITE) && (prot & PROT_EXEC))
- if (unlikely(!(prot & PROT_READ)))
- return -EINVAL;
+ /*
+ * As mentioned in Table 4.5 in RISC-V spec (20211203) Volume 2
+ * Section 4.3 the PTE permission bit combination of "write+!read"
+ * is "Reserved for future use.". Hence, don't allow such mapping
+ * request in mmap call.
+ */
+ if (unlikely((prot & PROT_WRITE) && !(prot & PROT_READ)))
+ return -EINVAL;
return ksys_mmap_pgoff(addr, len, prot, flags, fd,
offset >> (PAGE_SHIFT - page_shift_offset));
--
2.36.1
Powered by blists - more mailing lists