From: Ard Biesheuvel <ardb+git@google.com>
To: linux-arm-kernel@lists.infradead.org
Cc: linux-kernel@vger.kernel.org, will@kernel.org,
catalin.marinas@arm.com, mark.rutland@arm.com,
Ard Biesheuvel <ardb@kernel.org>,
Ryan Roberts <ryan.roberts@arm.com>,
Anshuman Khandual <anshuman.khandual@arm.com>,
Liz Prucka <lizprucka@google.com>,
Seth Jenkins <sethjenkins@google.com>,
Kees Cook <kees@kernel.org>, Mike Rapoport <rppt@kernel.org>,
David Hildenbrand <david@kernel.org>,
Andrew Morton <akpm@linux-foundation.org>,
linux-mm@kvack.org, linux-hardening@vger.kernel.org
Subject: [PATCH v4 15/15] arm64: mm: Remap linear aliases of the fixmap page tables read-only
Date: Mon, 27 Apr 2026 17:34:32 +0200 [thread overview]
Message-ID: <20260427153416.2103979-32-ardb+git@google.com> (raw)
In-Reply-To: <20260427153416.2103979-17-ardb+git@google.com>
From: Ard Biesheuvel <ardb@kernel.org>
The fixmap page tables are statically allocated, and are currently
mapped read-write both in the kernel mapping as well as its linear
alias. Due to lack of randomization of the linear map, these tables will
appear at a priori known offsets in the virtual address space when
booting without physical randomization, which means that a single kernel
write primitive is sufficient for an attacker to map memory of their own
choosing with any permissions at a known virtual address in the kernel's
address space.
To harden against this, move the fixmap PUD and PMD tables to
.pgdir_rodata, so that both their kernel mappings as well as their
linear aliases are mapped read-only during ordinary execution.
The PTE table needs to remain read-write accessible via the kernel
mapping, but its linear alias can be remapped read-only as well.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
---
arch/arm64/include/asm/pgtable.h | 6 ++++--
arch/arm64/kernel/vmlinux.lds.S | 1 +
arch/arm64/mm/fixmap.c | 5 +++--
arch/arm64/mm/mmu.c | 5 +++++
4 files changed, 13 insertions(+), 4 deletions(-)
diff --git a/arch/arm64/include/asm/pgtable.h b/arch/arm64/include/asm/pgtable.h
index 94235dd428be..21afe923cd71 100644
--- a/arch/arm64/include/asm/pgtable.h
+++ b/arch/arm64/include/asm/pgtable.h
@@ -822,8 +822,10 @@ extern void set_rodata_pte(pte_t *ptep, pte_t pte);
static inline bool in_pgdir_rodata(void *addr)
{
- return addr >= (void *)__pgdir_rodata_start &&
- addr < (void *)__pgdir_rodata_end;
+ phys_addr_t pa = __pa_nodebug(addr);
+
+ return pa >= __pa_symbol_nodebug(__pgdir_rodata_start) &&
+ pa < __pa_symbol_nodebug(__pgdir_rodata_end);
}
static inline void set_pmd(pmd_t *pmdp, pmd_t pmd)
diff --git a/arch/arm64/kernel/vmlinux.lds.S b/arch/arm64/kernel/vmlinux.lds.S
index e5e1d0fd7f27..9b346dd24d1c 100644
--- a/arch/arm64/kernel/vmlinux.lds.S
+++ b/arch/arm64/kernel/vmlinux.lds.S
@@ -247,6 +247,7 @@ SECTIONS
__pgdir_rodata_start = .;
swapper_pg_dir = .;
. += PAGE_SIZE;
+ *(.fixmap_rodata)
__pgdir_rodata_end = .;
}
diff --git a/arch/arm64/mm/fixmap.c b/arch/arm64/mm/fixmap.c
index b649ea1a46e4..ad6d46e5c23e 100644
--- a/arch/arm64/mm/fixmap.c
+++ b/arch/arm64/mm/fixmap.c
@@ -32,9 +32,10 @@ static_assert(NR_BM_PMD_TABLES == 1);
#define BM_PTE_TABLE_IDX(addr) __BM_TABLE_IDX(addr, PMD_SHIFT)
#define __fixmap_bss __section(".fixmap_bss") __aligned(PAGE_SIZE)
+#define __fixmap_rodata __section(".fixmap_rodata") __aligned(PAGE_SIZE)
static pte_t bm_pte[NR_BM_PTE_TABLES][PTRS_PER_PTE] __fixmap_bss;
-static pmd_t bm_pmd[PTRS_PER_PMD] __fixmap_bss __maybe_unused;
-static pud_t bm_pud[PTRS_PER_PUD] __fixmap_bss __maybe_unused;
+static pmd_t bm_pmd[PTRS_PER_PMD] __fixmap_rodata __maybe_unused;
+static pud_t bm_pud[PTRS_PER_PUD] __fixmap_rodata __maybe_unused;
static inline pte_t *fixmap_pte(unsigned long addr)
{
diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c
index 84d81bae07a7..e76fe5b0c5fe 100644
--- a/arch/arm64/mm/mmu.c
+++ b/arch/arm64/mm/mmu.c
@@ -1076,6 +1076,11 @@ void __init mark_linear_text_alias_ro(void)
(unsigned long)__init_begin - (unsigned long)_text,
pgprot_tagged(PAGE_KERNEL_RO));
+ /* Map the fixmap PTE table at __fixmap_pgdir_start R/O in linear map too */
+ update_mapping_prot(__pa_symbol(__fixmap_pgdir_start),
+ (unsigned long)lm_alias(__fixmap_pgdir_start),
+ PAGE_SIZE, pgprot_tagged(PAGE_KERNEL_RO));
+
remap_linear_data_alias(true);
if (IS_ENABLED(CONFIG_HIBERNATION)) {
--
2.54.0.rc2.544.gc7ae2d5bb8-goog
next prev parent reply other threads:[~2026-04-27 15:36 UTC|newest]
Thread overview: 49+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-04-27 15:34 [PATCH v4 00/15] arm64: Unmap linear alias of kernel data/bss Ard Biesheuvel
2026-04-27 15:34 ` [PATCH v4 01/15] arm64: mm: Map the linear alias of text/rodata as tagged Ard Biesheuvel
2026-04-28 14:16 ` Kevin Brodsky
2026-04-28 16:23 ` Ard Biesheuvel
2026-04-29 7:57 ` Kevin Brodsky
2026-04-29 7:58 ` Ard Biesheuvel
2026-04-27 15:34 ` [PATCH v4 02/15] mm: Make empty_zero_page __ro_after_init Ard Biesheuvel
2026-04-28 12:27 ` Mike Rapoport
2026-04-28 14:16 ` Kevin Brodsky
2026-04-28 19:51 ` David Hildenbrand (Arm)
2026-05-09 11:04 ` Kiryl Shutsemau
2026-05-08 17:02 ` Jann Horn
2026-05-11 8:59 ` Ard Biesheuvel
2026-05-11 14:40 ` Jann Horn
2026-05-12 12:56 ` Ard Biesheuvel
2026-05-13 8:50 ` Mike Rapoport
2026-05-13 8:53 ` Ard Biesheuvel
2026-05-13 10:28 ` Mike Rapoport
2026-05-11 18:45 ` Kees Cook
2026-05-11 19:01 ` Jann Horn
2026-05-11 2:55 ` Feng Tang
2026-04-27 15:34 ` [PATCH v4 03/15] arm64: mm: Preserve existing table mappings when mapping DRAM Ard Biesheuvel
2026-04-27 15:34 ` [PATCH v4 04/15] arm64: mm: Preserve non-contiguous descriptors " Ard Biesheuvel
2026-04-27 15:34 ` [PATCH v4 05/15] arm64: mm: Remove bogus stop condition from map_mem() loop Ard Biesheuvel
2026-04-28 14:33 ` Kevin Brodsky
2026-04-27 15:34 ` [PATCH v4 06/15] arm64: mm: Drop redundant pgd_t* argument from map_mem() Ard Biesheuvel
2026-04-28 14:33 ` Kevin Brodsky
2026-04-27 15:34 ` [PATCH v4 07/15] arm64: mm: Permit contiguous descriptors to be rewritten Ard Biesheuvel
2026-04-27 15:34 ` [PATCH v4 08/15] arm64: kfence: Avoid NOMAP tricks when mapping the early pool Ard Biesheuvel
2026-04-27 15:34 ` [PATCH v4 09/15] arm64: mm: Permit contiguous attribute for preliminary mappings Ard Biesheuvel
2026-04-27 15:34 ` [PATCH v4 10/15] arm64: Move fixmap page tables to end of kernel image Ard Biesheuvel
2026-04-29 13:52 ` Kevin Brodsky
2026-04-27 15:34 ` [PATCH v4 11/15] arm64: mm: Don't abuse memblock NOMAP to check for overlaps Ard Biesheuvel
2026-04-29 10:54 ` Kevin Brodsky
2026-04-29 14:23 ` Ard Biesheuvel
2026-04-29 14:30 ` Kevin Brodsky
2026-04-27 15:34 ` [PATCH v4 12/15] arm64: mm: Map the kernel data/bss read-only in the linear map Ard Biesheuvel
2026-04-29 13:54 ` Kevin Brodsky
2026-04-29 14:46 ` Ard Biesheuvel
2026-05-04 8:50 ` Kevin Brodsky
2026-04-27 15:34 ` [PATCH v4 13/15] arm64: mm: Unmap kernel data/bss entirely from " Ard Biesheuvel
2026-04-29 13:55 ` Kevin Brodsky
2026-04-29 17:37 ` Ard Biesheuvel
2026-05-04 8:52 ` Kevin Brodsky
2026-04-27 15:34 ` [PATCH v4 14/15] arm64: mm: Generalize manipulation code of read-only descriptors Ard Biesheuvel
2026-04-29 13:57 ` Kevin Brodsky
2026-04-27 15:34 ` Ard Biesheuvel [this message]
2026-04-29 13:57 ` [PATCH v4 15/15] arm64: mm: Remap linear aliases of the fixmap page tables read-only Kevin Brodsky
2026-04-29 14:08 ` Ard Biesheuvel
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20260427153416.2103979-32-ardb+git@google.com \
--to=ardb+git@google.com \
--cc=akpm@linux-foundation.org \
--cc=anshuman.khandual@arm.com \
--cc=ardb@kernel.org \
--cc=catalin.marinas@arm.com \
--cc=david@kernel.org \
--cc=kees@kernel.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-hardening@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=lizprucka@google.com \
--cc=mark.rutland@arm.com \
--cc=rppt@kernel.org \
--cc=ryan.roberts@arm.com \
--cc=sethjenkins@google.com \
--cc=will@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.