On architectures with CONFIG_HAVE_ARCH_HUGE_VMAP set, ioremap() may
create pud/pmd mappings. A kernel panic was observed on arm64 systems
with Cortex-A75 in the following steps as described by Hanjun Guo.
1. ioremap a 4K size, valid page table will build,
2. iounmap it, pte0 will set to 0;
3. ioremap the same address with 2M size, pgd/pmd is unchanged,
then set the a new value for pmd;
4. pte0 is leaked;
5. CPU may meet exception because the old pmd is still in TLB,
which will lead to kernel panic.
This panic is not reproducible on x86. INVLPG, called from iounmap,
purges all levels of entries associated with purged address on x86. x86
still has memory leak.
The patch changes the ioremap path to free unmapped page table(s) since
doing so in the unmap path has the following issues:
- The iounmap() path is shared with vunmap(). Since vmap() only
supports pte mappings, making vunmap() to free a pte page is an
overhead for regular vmap users as they do not need a pte page freed
up.
- Checking if all entries in a pte page are cleared in the unmap path
is racy, and serializing this check is expensive.
- The unmap path calls free_vmap_area_noflush() to do lazy TLB purges.
Clearing a pud/pmd entry before the lazy TLB purges needs extra TLB
purge.
Add two interfaces, pud_free_pmd_page() and pmd_free_pte_page(), which
clear a given pud/pmd entry and free up a page for the lower level
entries.
This patch implements their stub functions on x86 and arm64, which work
as workaround.
[akpm@linux-foundation.org: fix typo in pmd_free_pte_page() stub]
Link: http://lkml.kernel.org/r/20180314180155.19492-2-toshi.kani@hpe.com
Fixes: e61ce6ade4 ("mm: change ioremap to set up huge I/O mappings")
Reported-by: Lei Li <lious.lilei@hisilicon.com>
Signed-off-by: Toshi Kani <toshi.kani@hpe.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Wang Xuefeng <wxf.wang@hisilicon.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Hanjun Guo <guohanjun@huawei.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chintan Pandya <cpandya@codeaurora.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The routine pgattr_change_is_safe() was extended in commit 4e60205655
("arm64: mm: Permit transitioning from Global to Non-Global without BBM")
to permit changing the nG attribute from not set to set, but did so in a
way that inadvertently disallows such changes if other permitted attribute
changes take place at the same time. So update the code to take this into
account.
Fixes: 4e60205655 ("arm64: mm: Permit transitioning from Global to ...")
Cc: <stable@vger.kernel.org> # 4.14.x-
Acked-by: Mark Rutland <mark.rutland@arm.com>
Reviewed-by: Marc Zyngier <marc.zyngier@arm.com>
Acked-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
ioremap_page_range doesn't honour break-before-make and attempts to put
down huge mappings (using p*d_set_huge) over the top of pre-existing
table entries. This leads to us leaking page table memory and also gives
rise to TLB conflicts and spurious aborts, which have been seen in
practice on Cortex-A75.
Until this has been resolved, refuse to put block mappings when the
existing entry is found to be present.
Fixes: 324420bf91 ("arm64: add support for ioremap() block mappings")
Reported-by: Hanjun Guo <hanjun.guo@linaro.org>
Reported-by: Lei Li <lious.lilei@hisilicon.com>
Acked-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
In many cases, page tables can be accessed concurrently by either another
CPU (due to things like fast gup) or by the hardware page table walker
itself, which may set access/dirty bits. In such cases, it is important
to use READ_ONCE/WRITE_ONCE when accessing page table entries so that
entries cannot be torn, merged or subject to apparent loss of coherence
due to compiler transformations.
Whilst there are some scenarios where this cannot happen (e.g. pinned
kernel mappings for the linear region), the overhead of using READ_ONCE
/WRITE_ONCE everywhere is minimal and makes the code an awful lot easier
to reason about. This patch consistently uses these macros in the arch
code, as well as explicitly namespacing pointers to page table entries
from the entries themselves by using adopting a 'p' suffix for the former
(as is sometimes used elsewhere in the kernel source).
Tested-by: Yury Norov <ynorov@caviumnetworks.com>
Tested-by: Richard Ruigrok <rruigrok@codeaurora.org>
Reviewed-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Spectre v1 mitigation:
- back-end version of array_index_mask_nospec()
- masking of the syscall number to restrict speculation through the
syscall table
- masking of __user pointers prior to deference in uaccess routines
Spectre v2 mitigation update:
- using the new firmware SMC calling convention specification update
- removing the current PSCI GET_VERSION firmware call mitigation as
vendors are deploying new SMCCC-capable firmware
- additional branch predictor hardening for synchronous exceptions and
interrupts while in user mode
Meltdown v3 mitigation update for Cavium Thunder X: unaffected but
hardware erratum gets in the way. The kernel now starts with the page
tables mapped as global and switches to non-global if kpti needs to be
enabled.
Other:
- Theoretical trylock bug fixed
-----BEGIN PGP SIGNATURE-----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=Y0TT
-----END PGP SIGNATURE-----
Merge tag 'arm64-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Pull more arm64 updates from Catalin Marinas:
"As I mentioned in the last pull request, there's a second batch of
security updates for arm64 with mitigations for Spectre/v1 and an
improved one for Spectre/v2 (via a newly defined firmware interface
API).
Spectre v1 mitigation:
- back-end version of array_index_mask_nospec()
- masking of the syscall number to restrict speculation through the
syscall table
- masking of __user pointers prior to deference in uaccess routines
Spectre v2 mitigation update:
- using the new firmware SMC calling convention specification update
- removing the current PSCI GET_VERSION firmware call mitigation as
vendors are deploying new SMCCC-capable firmware
- additional branch predictor hardening for synchronous exceptions
and interrupts while in user mode
Meltdown v3 mitigation update:
- Cavium Thunder X is unaffected but a hardware erratum gets in the
way. The kernel now starts with the page tables mapped as global
and switches to non-global if kpti needs to be enabled.
Other:
- Theoretical trylock bug fixed"
* tag 'arm64-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux: (38 commits)
arm64: Kill PSCI_GET_VERSION as a variant-2 workaround
arm64: Add ARM_SMCCC_ARCH_WORKAROUND_1 BP hardening support
arm/arm64: smccc: Implement SMCCC v1.1 inline primitive
arm/arm64: smccc: Make function identifiers an unsigned quantity
firmware/psci: Expose SMCCC version through psci_ops
firmware/psci: Expose PSCI conduit
arm64: KVM: Add SMCCC_ARCH_WORKAROUND_1 fast handling
arm64: KVM: Report SMCCC_ARCH_WORKAROUND_1 BP hardening support
arm/arm64: KVM: Turn kvm_psci_version into a static inline
arm/arm64: KVM: Advertise SMCCC v1.1
arm/arm64: KVM: Implement PSCI 1.0 support
arm/arm64: KVM: Add smccc accessors to PSCI code
arm/arm64: KVM: Add PSCI_VERSION helper
arm/arm64: KVM: Consolidate the PSCI include files
arm64: KVM: Increment PC after handling an SMC trap
arm: KVM: Fix SMCCC handling of unimplemented SMC/HVC calls
arm64: KVM: Fix SMCCC handling of unimplemented SMC/HVC calls
arm64: entry: Apply BP hardening for suspicious interrupts from EL0
arm64: entry: Apply BP hardening for high-priority synchronous exceptions
arm64: futex: Mask __user pointers prior to dereference
...
Break-before-make is not needed when transitioning from Global to
Non-Global mappings, provided that the contiguous hint is not being used.
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
* Require struct page by default for filesystem DAX to remove a number of
surprising failure cases. This includes failures with direct I/O, gdb and
fork(2).
* Add support for the new Platform Capabilities Structure added to the NFIT in
ACPI 6.2a. This new table tells us whether the platform supports flushing
of CPU and memory controller caches on unexpected power loss events.
* Revamp vmem_altmap and dev_pagemap handling to clean up code and better
support future future PCI P2P uses.
* Deprecate the ND_IOCTL_SMART_THRESHOLD command whose payload has become
out-of-sync with recent versions of the NVDIMM_FAMILY_INTEL spec, and
instead rely on the generic ND_CMD_CALL approach used by the two other IOCTL
families, NVDIMM_FAMILY_{HPE,MSFT}.
* Enhance nfit_test so we can test some of the new things added in version 1.6
of the DSM specification. This includes testing firmware download and
simulating the Last Shutdown State (LSS) status.
-----BEGIN PGP SIGNATURE-----
iQIcBAABAgAGBQJaeOg0AAoJEJ/BjXdf9fLBAFoQAI/IgcgJ2h9lfEpgjBRTC44t
2p8dxwT1Ofw3Y1aR/tI8nYRXjRtAGuP4UIeRVnb1CL/N7PagJyoMGU+6hmzg+ptY
c7cEDvw6nZOhrFwXx/xn7R53sYG8zH+UE6+jTR/PP/G4mQJfFCg4iF9R72Y7z0n7
aurf82Kz137NPUy6dNr4V9bmPMJWAaOci9WOj5SKddR5ZSNbjoxylTwQRvre5y4r
7HQTScEkirABOdSf1JoXTSUXCH/RC9UFFXR03ScHstGb1HjCj3KdcicVc50Q++Ub
qsEudhE6i44PEW1Hh4Qkg6hjHMEa8qHP+ShBuRuVaUmlghYTQn66niJAYLZilwdz
EVjE7vR+toHA5g3YCalEmYVutUEhIDkh/xfpd7vM6ZorUGJy95a2elEJs2fHBffC
gEhnCip7FROPcK5RDNUM8hBgnG/q5wwWPQMKY+6rKDZQx3mXssCrKp2Vlx7kBwMG
rpblkEpYjPonbLEHxsSU8yTg9Uq55ciIWgnOToffcjZvjbihi8WUVlHcwHUMPf/o
DWElg+4qmG0Sdd4S2NeAGwTl1Ewrf2RrtUGMjHtH4OUFs1wo6ZmfrxFzzMfoZ1Od
ko/s65v4uwtTzECh2o+XQaNsReR5YETXxmA40N/Jpo7/7twABIoZ/ASvj/3ZBYj+
sie+u2rTod8/gQWSfHpJ
=MIMX
-----END PGP SIGNATURE-----
Merge tag 'libnvdimm-for-4.16' of git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm
Pull libnvdimm updates from Ross Zwisler:
- Require struct page by default for filesystem DAX to remove a number
of surprising failure cases. This includes failures with direct I/O,
gdb and fork(2).
- Add support for the new Platform Capabilities Structure added to the
NFIT in ACPI 6.2a. This new table tells us whether the platform
supports flushing of CPU and memory controller caches on unexpected
power loss events.
- Revamp vmem_altmap and dev_pagemap handling to clean up code and
better support future future PCI P2P uses.
- Deprecate the ND_IOCTL_SMART_THRESHOLD command whose payload has
become out-of-sync with recent versions of the NVDIMM_FAMILY_INTEL
spec, and instead rely on the generic ND_CMD_CALL approach used by
the two other IOCTL families, NVDIMM_FAMILY_{HPE,MSFT}.
- Enhance nfit_test so we can test some of the new things added in
version 1.6 of the DSM specification. This includes testing firmware
download and simulating the Last Shutdown State (LSS) status.
* tag 'libnvdimm-for-4.16' of git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm: (37 commits)
libnvdimm, namespace: remove redundant initialization of 'nd_mapping'
acpi, nfit: fix register dimm error handling
libnvdimm, namespace: make min namespace size 4K
tools/testing/nvdimm: force nfit_test to depend on instrumented modules
libnvdimm/nfit_test: adding support for unit testing enable LSS status
libnvdimm/nfit_test: add firmware download emulation
nfit-test: Add platform cap support from ACPI 6.2a to test
libnvdimm: expose platform persistence attribute for nd_region
acpi: nfit: add persistent memory control flag for nd_region
acpi: nfit: Add support for detect platform CPU cache flush on power loss
device-dax: Fix trailing semicolon
libnvdimm, btt: fix uninitialized err_lock
dax: require 'struct page' by default for filesystem dax
ext2: auto disable dax instead of failing mount
ext4: auto disable dax instead of failing mount
mm, dax: introduce pfn_t_special()
mm: Fix devm_memremap_pages() collision handling
mm: Fix memory size alignment in devm_memremap_pages_release()
memremap: merge find_dev_pagemap into get_dev_pagemap
memremap: change devm_memremap_pages interface to use struct dev_pagemap
...
Currently the early assembler page table code assumes that precisely
1xpgd, 1xpud, 1xpmd are sufficient to represent the early kernel text
mappings.
Unfortunately this is rarely the case when running with a 16KB granule,
and we also run into limits with 4KB granule when building much larger
kernels.
This patch re-writes the early page table logic to compute indices of
mappings for each level of page table, and if multiple indices are
required, the next-level page table is scaled up accordingly.
Also the required size of the swapper_pg_dir is computed at link time
to cover the mapping [KIMAGE_ADDR + VOFFSET, _end]. When KASLR is
enabled, an extra page is set aside for each level that may require extra
entries at runtime.
Tested-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Steve Capper <steve.capper@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
SDEI needs to calculate an offset in the trampoline page too. Move
the extern char[] to sections.h.
This patch just moves code around.
Signed-off-by: James Morse <james.morse@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
We can just pass this on instead of having to do a radix tree lookup
without proper locking a few levels into the callchain.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
We can just pass this on instead of having to do a radix tree lookup
without proper locking a few levels into the callchain.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
* for-next/52-bit-pa:
arm64: enable 52-bit physical address support
arm64: allow ID map to be extended to 52 bits
arm64: handle 52-bit physical addresses in page table entries
arm64: don't open code page table entry creation
arm64: head.S: handle 52-bit PAs in PTEs in early page table setup
arm64: handle 52-bit addresses in TTBR
arm64: limit PA size to supported range
arm64: add kconfig symbol to configure physical address size
Currently, when using VA_BITS < 48, if the ID map text happens to be
placed in physical memory above VA_BITS, we increase the VA size (up to
48) and create a new table level, in order to map in the ID map text.
This is okay because the system always supports 48 bits of VA.
This patch extends the code such that if the system supports 52 bits of
VA, and the ID map text is placed that high up, then we increase the VA
size accordingly, up to 52.
One difference from the current implementation is that so far the
condition of VA_BITS < 48 has meant that the top level table is always
"full", with the maximum number of entries, and an extra table level is
always needed. Now, when VA_BITS = 48 (and using 64k pages), the top
level table is not full, and we simply need to increase the number of
entries in it, instead of creating a new table level.
Tested-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Reviewed-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Reviewed-by: Marc Zyngier <marc.zyngier@arm.com>
Tested-by: Bob Picco <bob.picco@oracle.com>
Reviewed-by: Bob Picco <bob.picco@oracle.com>
Signed-off-by: Kristina Martsenko <kristina.martsenko@arm.com>
[catalin.marinas@arm.com: reduce arguments to __create_hyp_mappings()]
[catalin.marinas@arm.com: reworked/renamed __cpu_uses_extended_idmap_level()]
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Instead of open coding the generation of page table entries, use the
macros/functions that exist for this - pfn_p*d and p*d_populate. Most
code in the kernel already uses these macros, this patch tries to fix
up the few places that don't. This is useful for the next patch in this
series, which needs to change the page table entry logic, and it's
better to have that logic in one place.
The KVM extended ID map is special, since we're creating a level above
CONFIG_PGTABLE_LEVELS and the required function isn't available. Leave
it as is and add a comment to explain it. (The normal kernel ID map code
doesn't need this change because its page tables are created in assembly
(__create_page_tables)).
Tested-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Reviewed-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Reviewed-by: Marc Zyngier <marc.zyngier@arm.com>
Tested-by: Bob Picco <bob.picco@oracle.com>
Reviewed-by: Bob Picco <bob.picco@oracle.com>
Signed-off-by: Kristina Martsenko <kristina.martsenko@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
The literal pool entry for identifying the vectors base is the only piece
of information in the trampoline page that identifies the true location
of the kernel.
This patch moves it into a page-aligned region of the .rodata section
and maps this adjacent to the trampoline text via an additional fixmap
entry, which protects against any accidental leakage of the trampoline
contents.
Suggested-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Tested-by: Laura Abbott <labbott@redhat.com>
Tested-by: Shanker Donthineni <shankerd@codeaurora.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
The exception entry trampoline needs to be mapped at the same virtual
address in both the trampoline page table (which maps nothing else)
and also the kernel page table, so that we can swizzle TTBR1_EL1 on
exceptions from and return to EL0.
This patch maps the trampoline at a fixed virtual address in the fixmap
area of the kernel virtual address space, which allows the kernel proper
to be randomized with respect to the trampoline when KASLR is enabled.
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Laura Abbott <labbott@redhat.com>
Tested-by: Shanker Donthineni <shankerd@codeaurora.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Nothing calls arch_apei_flush_tlb_one() anymore, instead relying on
__set_fixmap() to do the invalidation. Remove it.
Move the IPI-considered-harmful comment to __set_fixmap().
Signed-off-by: James Morse <james.morse@arm.com>
Acked-by: Will Deacon <will.deacon@arm.com>
Tested-by: Tyler Baicar <tbaicar@codeaurora.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: All applicable <stable@vger.kernel.org>
The vast majority of virtual allocations in the vmalloc region are followed
by a guard page, which can help to avoid overruning on vma into another,
which may map a read-sensitive device.
This patch adds a guard page to the end of the kernel image mapping (i.e.
following the data/bss segments).
Cc: Mark Rutland <mark.rutland@arm.com>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
arm64's mm/mmu.c uses vm_area_add_early, struct vm_area and other
definitions but relies on implict inclusion of linux/vmalloc.h which
means that changes in other headers could break the build. Thus, add an
explicit include.
Acked-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Tobias Klauser <tklauser@distanz.ch>
Signed-off-by: Will Deacon <will.deacon@arm.com>
arch_kexec_protect_crashkres() and arch_kexec_unprotect_crashkres()
are meant to be called by kexec_load() in order to protect the memory
allocated for crash dump kernel once the image is loaded.
The protection is implemented by unmapping the relevant segments in crash
dump kernel memory, rather than making it read-only as other archs do,
to prevent coherency issues due to potential cache aliasing (with
mismatched attributes).
Page-level mappings are consistently used here so that we can change
the attributes of segments in page granularity as well as shrink the region
also in page granularity through /sys/kernel/kexec_crash_size, putting
the freed memory back to buddy system.
Signed-off-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
This is the third attempt at enabling the use of contiguous hints for
kernel mappings. The most recent attempt 0bfc445dec was reverted after
it turned out that updating permission attributes on live contiguous ranges
may result in TLB conflicts. So this time, the contiguous hint is not set
for .rodata or for the linear alias of .text/.rodata, both of which are
mapped read-write initially, and remapped read-only at a later stage.
(Note that the latter region could also be unmapped and remapped again
with updated permission attributes, given that the region, while live, is
only mapped for the convenience of the hibernation code, but that also
means the TLB footprint is negligible anyway, so why bother)
This enables the following contiguous range sizes for the virtual mapping
of the kernel image, and for the linear mapping:
granule size | cont PTE | cont PMD |
-------------+------------+------------+
4 KB | 64 KB | 32 MB |
16 KB | 2 MB | 1 GB* |
64 KB | 2 MB | 16 GB* |
* Only when built for 3 or more levels of translation. This is due to the
fact that a 2 level configuration only consists of PGDs and PTEs, and the
added complexity of dealing with folded PMDs is not justified considering
that 16 GB contiguous ranges are likely to be ignored by the hardware (and
16k/2 levels is a niche configuration)
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
The routines __pud_populate and __pmd_populate only create a table
entry at their respective level which refers to the next level page
by its physical address, so there is no reason to map this page and
then unmap it immediately after.
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
In preparation of extending the policy for manipulating kernel mappings
with whether or not contiguous hints may be used in the page tables,
replace the bool 'page_mappings_only' with a flags field and a flag
NO_BLOCK_MAPPINGS.
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
A mapping with the contiguous bit cannot be safely manipulated while
live, regardless of whether the bit changes between the old and new
mapping. So take this into account when deciding whether the change
is safe.
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
The debug_pagealloc facility manipulates kernel mappings in the linear
region at page granularity to detect out of bounds or use-after-free
accesses. Since the kernel segments are not allocated dynamically,
there is no point in taking the debug_pagealloc_enabled flag into
account for them, and we can use block mappings unconditionally.
Note that this applies equally to the linear alias of text/rodata:
we will never have dynamic allocations there given that the same
memory is statically in use by the kernel image.
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Align the function prototype of alloc_init_pte() with its pmd and pud
counterparts by replacing the pfn parameter with the equivalent physical
address.
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
To avoid having mappings that are writable and executable at the same
time, split the init region into a .init.text region that is mapped
read-only, and a .init.data region that is mapped non-executable.
This is possible now that the alternative patching occurs via the linear
mapping, and the linear alias of the init region is always mapped writable
(but never executable).
Since the alternatives descriptions themselves are read-only data, move
those into the .init.text region.
Reviewed-by: Laura Abbott <labbott@redhat.com>
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Now that alternatives patching code no longer relies on the primary
mapping of .text being writable, we can remove the code that removes
the writable permissions post-init time, and map it read-only from
the outset.
To preserve the existing behavior under rodata=off, which is relied
upon by external debuggers to manage software breakpoints (as pointed
out by Mark), add an early_param() check for rodata=, and use RWX
permissions if it set to 'off'.
Reviewed-by: Laura Abbott <labbott@redhat.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
One important rule of thumb when desiging a secure software system is
that memory should never be writable and executable at the same time.
We mostly adhere to this rule in the kernel, except at boot time, when
regions may be mapped RWX until after we are done applying alternatives
or making other one-off changes.
For the alternative patching, we can improve the situation by applying
the fixups via the linear mapping, which is never mapped with executable
permissions. So map the linear alias of .text with RW- permissions
initially, and remove the write permissions as soon as alternative
patching has completed.
Reviewed-by: Laura Abbott <labbott@redhat.com>
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
In preparation of refactoring the kernel mapping logic so that text regions
are never mapped writable, which would require adding explicit TLB
maintenance to new call sites of create_mapping_late() (which is currently
invoked twice from the same function), move the TLB maintenance from the
call site into create_mapping_late() itself, and change it from a full
TLB flush into a flush by VA, which is more appropriate here.
Also, given that create_mapping_late() has evolved into a routine that only
updates protection bits on existing mappings, rename it to
update_mapping_prot()
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
This reverts commit 0bfc445dec.
When we change the permissions of regions mapped using contiguous
entries, the architecture requires us to follow a Break-Before-Make
strategy, breaking *all* associated entries before we can change any of
the following properties from the entries:
- presence of the contiguous bit
- output address
- attributes
- permissiones
Failure to do so can result in a number of problems (e.g. TLB conflict
aborts and/or erroneous results from TLB lookups).
See ARM DDI 0487A.k_iss10775, "Misprogramming of the Contiguous bit",
page D4-1762.
We do not take this into account when altering the permissions of kernel
segments in mark_rodata_ro(), where we change the permissions of live
contiguous entires one-by-one, leaving them transiently inconsistent.
This has been observed to result in failures on some fast model
configurations.
Unfortunately, we cannot follow Break-Before-Make here as we'd have to
unmap kernel text and data used to perform the sequence.
For the timebeing, revert commit 0bfc445dec so as to avoid issues
resulting from this misuse of the contiguous bit.
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reported-by: Jean-Philippe Brucker <jean-philippe.brucker@arm.com>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <Will.Deacon@arm.com>
Cc: stable@vger.kernel.org # v4.10
Signed-off-by: Will Deacon <will.deacon@arm.com>
With 4 levels of 16KB pages, we get this warning about the fact that we are
copying a whole page into an array that is declared as having only two pointers
for the top level of the page table:
arch/arm64/mm/mmu.c: In function 'paging_init':
arch/arm64/mm/mmu.c:528:2: error: 'memcpy' writing 16384 bytes into a region of size 16 overflows the destination [-Werror=stringop-overflow=]
This is harmless since we actually reserve a whole page in the definition of the
array that comes from, and just the extern declaration is short. The pgdir
is initialized to zero either way, so copying the actual entries here seems
like the best solution.
Acked-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Cosmetic change to use phys_addr_t instead of unsigned long for the
return value of __pa_symbol().
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Miles Chen <miles.chen@mediatek.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
__pa_symbol is technically the marcro that should be used for kernel
symbols. Switch to this as a pre-requisite for DEBUG_VIRTUAL which
will do bounds checking.
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Laura Abbott <labbott@redhat.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Page mappings with full RWX permissions are a security risk. x86
has an option to walk the page tables and dump any bad pages.
(See e1a58320a3 ("x86/mm: Warn on W^X mappings")). Add a similar
implementation for arm64.
Reviewed-by: Kees Cook <keescook@chromium.org>
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Laura Abbott <labbott@redhat.com>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
[catalin.marinas@arm.com: folded fix for KASan out of bounds from Mark Rutland]
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Now that we no longer allow live kernel PMDs to be split, it is safe to
start using the contiguous bit for kernel mappings. So set the contiguous
bit in the kernel page mappings for regions whose size and alignment are
suitable for this.
This enables the following contiguous range sizes for the virtual mapping
of the kernel image, and for the linear mapping:
granule size | cont PTE | cont PMD |
-------------+------------+------------+
4 KB | 64 KB | 32 MB |
16 KB | 2 MB | 1 GB* |
64 KB | 2 MB | 16 GB* |
* Only when built for 3 or more levels of translation. This is due to the
fact that a 2 level configuration only consists of PGDs and PTEs, and the
added complexity of dealing with folded PMDs is not justified considering
that 16 GB contiguous ranges are likely to be ignored by the hardware (and
16k/2 levels is a niche configuration)
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
In preparation of adding support for contiguous PTE and PMD mappings,
let's replace 'block_mappings_allowed' with 'page_mappings_only', which
will be a more accurate description of the nature of the setting once we
add such contiguous mappings into the mix.
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Now that we take care not manipulate the live kernel page tables in a
way that may lead to TLB conflicts, the case where a table mapping is
replaced by a block mapping can no longer occur. So remove the handling
of this at the PUD and PMD levels, and instead, BUG() on any occurrence
of live kernel page table manipulations that modify anything other than
the permission bits.
Since mark_rodata_ro() is the only caller where the kernel mappings that
are being manipulated are actually live, drop the various conditional
flush_tlb_all() invocations, and add a single call to mark_rodata_ro()
instead.
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
There is only fixup_init() in mm.h , and it is only called
in free_initmem(), so move the codes from fixup_init() into
free_initmem(), then drop fixup_init() and mm.h.
Acked-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
These objects are set during initialization, thereafter are read only.
Previously I only want to mark vdso_pages, vdso_spec, vectors_page and
cpu_ops as __read_mostly from performance point of view. Then inspired
by Kees's patch[1] to apply more __ro_after_init for arm, I think it's
better to mark them as __ro_after_init. What's more, I find some more
objects are also read only after init. So apply __ro_after_init to all
of them.
This patch also removes global vdso_pagelist and tries to clean up
vdso_spec[] assignment code.
[1] http://www.spinics.net/lists/arm-kernel/msg523188.html
Acked-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Jisheng Zhang <jszhang@marvell.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
As reported by Zijun, the fdt_check_header() call in __fixmap_remap_fdt()
is not safe since it is not guaranteed that the FDT header is mapped
completely. Due to the minimum alignment of 8 bytes, the only fields we
can assume to be mapped are 'magic' and 'totalsize'.
Since the OF layer is in charge of validating the FDT image, and we are
only interested in making reasonably sure that the size field contains
a meaningful value, replace the fdt_check_header() call with an explicit
comparison of the magic field's value against the expected value.
Cc: <stable@vger.kernel.org>
Reported-by: Zijun Hu <zijun_hu@htc.com>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
The kernel page table creation routines are accessible to other subsystems
(e.g., EFI) via the create_pgd_mapping() entry point, which allows mappings
to be created that are not covered by init_mm.
Since generic code such as apply_to_page_range() may expect translation
table pages that are not associated with init_mm to be covered by fully
constructed struct pages, add a call to pgtable_page_ctor() in the alloc
function used by create_pgd_mapping. Since it is no longer used by
create_mapping_late(), also update the name of this function to better
reflect its purpose.
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Laura Abbott <labbott@redhat.com>
Tested-by: Sudeep Holla <sudeep.holla@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
The only purpose served by create_mapping_late() is to remap the already
mapped .text and .rodata kernel segments with read-only permissions. Since
we no longer allow block mappings to be split or merged,
create_mapping_late() should not pass an allocation function pointer into
__create_pgd_mapping(). So pass NULL instead.
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Laura Abbott <labbott@redhat.com>
Tested-by: Sudeep Holla <sudeep.holla@arm.com>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
The routine __create_pgd_mapping() does nothing except calling init_pgd(),
which has no other callers. So fold the latter into the former. Also, drop
a comment that has gone stale.
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Since the efi_create_mapping() no longer generates block mappings
and being the last user of the split_p*d code, remove these functions
and the corresponding TLBI.
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
[ardb: replace 'overlapping regions' with 'block mappings' in commit log]
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Add a bool parameter 'allow_block_mappings' to create_pgd_mapping() and
the various helper functions that it descends into, to give the caller
control over whether block entries may be used to create the mapping.
The UEFI runtime mapping routines will use this to avoid creating block
entries that would need to split up into page entries when applying the
permissions listed in the Memory Attributes firmware table.
This also replaces the block_mappings_allowed() helper function that was
added for DEBUG_PAGEALLOC functionality, but the resulting code is
functionally equivalent (given that debug_page_alloc does not operate on
EFI page table entries anyway)
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
The memblock_alloc() and memblock_alloc_base() will panic on their own
if no free memory, remove pointless BUG_ON.
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
As Kees Cook notes in the ARM counterpart of this patch [0]:
The _etext position is defined to be the end of the kernel text code,
and should not include any part of the data segments. This interferes
with things that might check memory ranges and expect executable code
up to _etext.
In particular, Kees is referring to the HARDENED_USERCOPY patch set [1],
which rejects attempts to call copy_to_user() on kernel ranges containing
executable code, but does allow access to the .rodata segment. Regardless
of whether one may or may not agree with the distinction, it makes sense
for _etext to have the same meaning across architectures.
So let's put _etext where it belongs, between .text and .rodata, and fix
up existing references to use __init_begin instead, which unlike _end_rodata
includes the exception and notes sections as well.
The _etext references in kaslr.c are left untouched, since its references
to [_stext, _etext) are meant to capture potential jump instruction targets,
and so disregarding .rodata is actually an improvement here.
[0] http://article.gmane.org/gmane.linux.kernel/2245084
[1] http://thread.gmane.org/gmane.linux.kernel.hardened.devel/2502
Reported-by: Kees Cook <keescook@chromium.org>
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
In order to extract NUMA information from the device tree, we need to
have the tree in its unflattened form.
Move the call to bootmem_init() in the tail of paging_init() into
setup_arch, and adjust header files so that its declaration is
visible.
Move the unflatten_device_tree() call between the calls to
paging_init() and bootmem_init(). Follow on patches add NUMA handling
to bootmem_init().
Signed-off-by: David Daney <david.daney@cavium.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Keeping .head.text out of the .text mapping buys us very little: its actual
payload is only 4 KB, most of which is padding, but the page alignment may
add up to 2 MB (in case of CONFIG_DEBUG_ALIGN_RODATA=y) of additional
padding to the uncompressed kernel Image.
Also, on 4 KB granule kernels, the 4 KB misalignment of .text forces us to
map the adjacent 56 KB of code without the PTE_CONT attribute, and since
this region contains things like the vector table and the GIC interrupt
handling entry point, this region is likely to benefit from the reduced TLB
pressure that results from PTE_CONT mappings.
So remove the alignment between the .head.text and .text sections, and use
the [_text, _etext) rather than the [_stext, _etext) interval for mapping
the .text segment.
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
