When a kernel is built without CONFIG_ARM64_MODULE_PLTS, we don't
generate the expected branch instruction in ftrace_make_nop(). This
means we pass zero (rather than a valid branch) to ftrace_modify_code()
as the expected instruction to validate. This causes us to return
-EINVAL to the core ftrace code for a valid case, resulting in a splat
at boot time.
This was an unintended effect of commit:
687644209a6e9557 ("arm64: ftrace: fix building without CONFIG_MODULES")
... which incorrectly moved the generation of the branch instruction
into the ifdef for CONFIG_ARM64_MODULE_PLTS.
This patch fixes the issue by moving the ifdef inside of the relevant
if-else case, and always checking that the branch is in range,
regardless of CONFIG_ARM64_MODULE_PLTS. This ensures that we generate
the expected branch instruction, and also improves our sanity checks.
For consistency, both ftrace_make_nop() and ftrace_make_call() are
updated with this pattern.
Fixes: 687644209a6e9557 ("arm64: ftrace: fix building without CONFIG_MODULES") Signed-off-by: Mark Rutland <mark.rutland@arm.com> Reported-by: Marc Zyngier <marc.zyngier@arm.com> Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Will Deacon <will.deacon@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
Dave Martin [Tue, 20 Jun 2017 17:23:39 +0000 (18:23 +0100)]
arm64: signal: Allow expansion of the signal frame
This patch defines an extra_context signal frame record that can be
used to describe an expanded signal frame, and modifies the context
block allocator and signal frame setup and parsing code to create,
populate, parse and decode this block as necessary.
To avoid abuse by userspace, parse_user_sigframe() attempts to
ensure that:
* no more than one extra_context is accepted;
* the extra context data is a sensible size, and properly placed
and aligned.
The extra_context data is required to start at the first 16-byte
aligned address immediately after the dummy terminator record
following extra_context in rt_sigframe.__reserved[] (as ensured
during signal delivery). This serves as a sanity-check that the
signal frame has not been moved or copied without taking the extra
data into account.
Signed-off-by: Dave Martin <Dave.Martin@arm.com> Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
[will: add __force annotation when casting extra_datap to __user pointer] Signed-off-by: Will Deacon <will.deacon@arm.com>
Mark Rutland [Wed, 21 Jun 2017 17:11:23 +0000 (18:11 +0100)]
arm64: dump cpu_hwcaps at panic time
When debugging a kernel panic(), it can be useful to know which CPU
features have been detected by the kernel, as some code paths can depend
on these (and may have been patched at runtime).
This patch adds a notifier to dump the detected CPU caps (as a hex
string) at panic(), when we log other information useful for debugging.
On a Juno R1 system running v4.12-rc5, this looks like:
A developer can decode this by looking at the corresponding
<asm/cpucaps.h> bits. For example, the above decodes as:
* bit 1: ARM64_WORKAROUND_DEVICE_LOAD_ACQUIRE
* bit 2: ARM64_WORKAROUND_845719
* bit 7: ARM64_WORKAROUND_834220
* bit 13: ARM64_HAS_32BIT_EL0
Signed-off-by: Mark Rutland <mark.rutland@arm.com> Acked-by: Steve Capper <steve.capper@arm.com> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Suzuki K Poulose <suzuki.poulose@arm.com> Cc: Will Deacon <will.deacon@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
Dave Martin [Wed, 21 Jun 2017 15:00:44 +0000 (16:00 +0100)]
arm64: ptrace: Flush user-RW TLS reg to thread_struct before reading
When reading current's user-writable TLS register (which occurs
when dumping core for native tasks), it is possible that userspace
has modified it since the time the task was last scheduled out.
The new TLS register value is not guaranteed to have been written
immediately back to thread_struct in this case.
As a result, a coredump can capture stale data for this register.
Reading the register for a stopped task via ptrace is unaffected.
For native tasks, this patch explicitly flushes the TPIDR_EL0
register back to thread_struct before dumping when operating on
current, thus ensuring that coredump contents are up to date. For
compat tasks, the TLS register is not user-writable and so cannot
be out of sync, so no flush is required in compat_tls_get().
Signed-off-by: Dave Martin <Dave.Martin@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
Dave Martin [Wed, 21 Jun 2017 15:00:43 +0000 (16:00 +0100)]
arm64: ptrace: Flush FPSIMD regs back to thread_struct before reading
When reading the FPSIMD state of current (which occurs when dumping
core), it is possible that userspace has modified the FPSIMD
registers since the time the task was last scheduled out. Such
changes are not guaranteed to be reflected immedately in
thread_struct.
As a result, a coredump can contain stale values for these
registers. Reading the registers of a stopped task via ptrace is
unaffected.
This patch explicitly flushes the CPU state back to thread_struct
before dumping when operating on current, thus ensuring that
coredump contents are up to date.
Signed-off-by: Dave Martin <Dave.Martin@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
Dave Martin [Wed, 21 Jun 2017 15:00:42 +0000 (16:00 +0100)]
arm64: ptrace: Fix VFP register dumping in compat coredumps
Currently, VFP registers are omitted from coredumps for compat
processes, due to a bug in the REGSET_COMPAT_VFP regset
implementation.
compat_vfp_get() needs to transfer non-contiguous data from
thread_struct.fpsimd_state, and uses put_user() to handle the
offending trailing word (FPSCR). This fails when copying to a
kernel address (i.e., kbuf && !ubuf), which is what happens when
dumping core. As a result, the ELF coredump core code silently
omits the NT_ARM_VFP note from the dump.
It would be possible to work around this with additional special
case code for the put_user(), but since user_regset_copyout() is
explicitly designed to handle this scenario it is cleaner to port
the put_user() to a user_regset_copyout() call, which this patch
does.
Signed-off-by: Dave Martin <Dave.Martin@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
When using sparse on the arm64 tree we get many thousands of
warnings like 'constant ... is so big it is unsigned long long'
or 'shift too big (32) for type unsigned long'. This happens
because by default sparse considers the machine as 32bit and
defines the size of the types accordingly.
Fix this by passing the '-m64' flag to sparse so that
sparse can correctly define longs as being 64bit.
CC: Catalin Marinas <catalin.marinas@arm.com> CC: Will Deacon <will.deacon@arm.com> CC: linux-arm-kernel@lists.infradead.org Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
Dave Martin [Thu, 15 Jun 2017 14:03:41 +0000 (15:03 +0100)]
arm64: signal: factor out signal frame record allocation
This patch factors out the allocator for signal frame optional
records into a separate function, to ensure consistency and
facilitate later expansion.
No overrun checking is currently done, because the allocation is in
user memory and anyway the kernel never tries to allocate enough
space in the signal frame yet for an overrun to occur. This
behaviour will be refined in future patches.
The approach taken in this patch to allocation of the terminator
record is not very clean: this will also be replaced in subsequent
patches.
For future extension, a comment is added in sigcontext.h
documenting the current static allocations in __reserved[]. This
will be important for determining under what circumstances
userspace may or may not see an expanded signal frame.
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by: Dave Martin <Dave.Martin@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
Dave Martin [Thu, 15 Jun 2017 14:03:40 +0000 (15:03 +0100)]
arm64: signal: factor frame layout and population into separate passes
In preparation for expanding the signal frame, this patch refactors
the signal frame setup code in setup_sigframe() into two separate
passes.
The first pass, setup_sigframe_layout(), determines the size of the
signal frame and its internal layout, including the presence and
location of optional records. The resulting knowledge is used to
allocate and locate the user stack space required for the signal
frame and to determine which optional records to include.
The second pass, setup_sigframe(), is called once the stack frame
is allocated in order to populate it with the necessary context
information.
As a result of these changes, it becomes more natural to represent
locations in the signal frame by a base pointer and an offset,
since the absolute address of each location is not known during the
layout pass. To be more consistent with this logic,
parse_user_sigframe() is refactored to describe signal frame
locations in a similar way.
This change has no effect on the signal ABI, but will make it
easier to expand the signal frame in future patches.
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by: Dave Martin <Dave.Martin@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
Dave Martin [Thu, 15 Jun 2017 14:03:39 +0000 (15:03 +0100)]
arm64: signal: Refactor sigcontext parsing in rt_sigreturn
Currently, rt_sigreturn does very limited checking on the
sigcontext coming from userspace.
Future additions to the sigcontext data will increase the potential
for surprises. Also, it is not clear whether the sigcontext
extension records are supposed to occur in a particular order.
To allow the parsing code to be extended more easily, this patch
factors out the sigcontext parsing into a separate function, and
adds extra checks to validate the well-formedness of the sigcontext
structure.
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by: Dave Martin <Dave.Martin@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
Dave Martin [Thu, 15 Jun 2017 14:03:38 +0000 (15:03 +0100)]
arm64: signal: split frame link record from sigcontext structure
In order to be able to increase the amount of the data currently
written to the __reserved[] array in the signal frame, it is
necessary to overwrite the locations currently occupied by the
{fp,lr} frame link record pushed at the top of the signal stack.
In order for this to work, this patch detaches the frame link
record from struct rt_sigframe and places it separately at the top
of the signal stack. This will allow subsequent patches to insert
data between it and __reserved[].
This change relies on the non-ABI status of the placement of the
frame record with respect to struct sigframe: this status is
undocumented, but the placement is not declared or described in the
user headers, and known unwinder implementations (libgcc,
libunwind, gdb) appear not to rely on it.
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by: Dave Martin <Dave.Martin@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
Ard Biesheuvel [Wed, 14 Jun 2017 10:43:55 +0000 (12:43 +0200)]
arm64: mm: select CONFIG_ARCH_PROC_KCORE_TEXT
To avoid issues with the /proc/kcore code getting confused about the
kernels block mappings in the VMALLOC region, enable the existing
facility that describes the [_text, _end) interval as a separate
KCORE_TEXT region, which supersedes the KCORE_VMALLOC region that
it intersects with on arm64.
Reported-by: Tan Xiaojun <tanxiaojun@huawei.com> Tested-by: Tan Xiaojun <tanxiaojun@huawei.com> Tested-by: Mark Rutland <mark.rutland@arm.com> Acked-by: Mark Rutland <mark.rutland@arm.com> Reviewed-by: Laura Abbott <labbott@redhat.com> Reviewed-by: Jiri Olsa <jolsa@kernel.org> Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Will Deacon <will.deacon@arm.com>
Ard Biesheuvel [Wed, 14 Jun 2017 10:43:54 +0000 (12:43 +0200)]
fs/proc: kcore: use kcore_list type to check for vmalloc/module address
Instead of passing each start address into is_vmalloc_or_module_addr()
to decide whether it falls into either the VMALLOC or the MODULES region,
we can simply check the type field of the current kcore_list entry, since
it will be set to KCORE_VMALLOC based on exactly the same conditions.
As a bonus, when reading the KCORE_TEXT region on architectures that have
one, this will avoid using vread() on the region if it happens to intersect
with a KCORE_VMALLOC region. This is due the fact that the KCORE_TEXT
region is the first one to be added to the kcore region list.
Reported-by: Tan Xiaojun <tanxiaojun@huawei.com> Tested-by: Tan Xiaojun <tanxiaojun@huawei.com> Tested-by: Mark Rutland <mark.rutland@arm.com> Acked-by: Mark Rutland <mark.rutland@arm.com> Reviewed-by: Laura Abbott <labbott@redhat.com> Reviewed-by: Jiri Olsa <jolsa@kernel.org> Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Will Deacon <will.deacon@arm.com>
Ard Biesheuvel [Tue, 20 Jun 2017 06:59:00 +0000 (08:59 +0200)]
drivers/char: kmem: disable on arm64
As it turns out, arm64 deviates from other architectures in the way it
maps the VMALLOC region: on most (all?) other architectures, it resides
strictly above the kernel's direct mapping of DRAM, but on arm64, this
is the other way around. For instance, for a 48-bit VA configuration,
we have
This has mostly gone unnoticed until now, but it does appear that it
breaks an assumption in the kmem read/write code, which does something
like
if (p < (unsigned long) high_memory) {
... use straight copy_[to|from]_user() using p as virtual address ...
}
...
if (count > 0) {
... use vread/vwrite for accesses past high_memory ...
}
The first condition will inadvertently hold for the VMALLOC region if
VMALLOC_START < PAGE_OFFSET [which is the case on arm64], but the read
or write will subsequently fail the virt_addr_valid() check, resulting
in a -ENXIO return value.
Given how kmem seems to be living in borrowed time anyway, and given
the fact that nobody noticed that the read/write interface is broken
on arm64 in the first place, let's not bother trying to fix it, but
simply disable the /dev/kmem interface entirely for arm64.
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>
Dustin Brown [Tue, 13 Jun 2017 18:40:56 +0000 (11:40 -0700)]
arm64: Export save_stack_trace_tsk()
The kernel watchdog is a great debugging tool for finding tasks that
consume a disproportionate amount of CPU time in contiguous chunks. One
can imagine building a similar watchdog for arbitrary driver threads
using save_stack_trace_tsk() and print_stack_trace(). However, this is
not viable for dynamically loaded driver modules on ARM platforms
because save_stack_trace_tsk() is not exported for those architectures.
Export save_stack_trace_tsk() for the ARM64 architecture to align with
x86 and support various debugging use cases such as arbitrary driver
thread watchdog timers.
Signed-off-by: Dustin Brown <dustinb@codeaurora.org> Signed-off-by: Will Deacon <will.deacon@arm.com>
ARM64/irqchip: Update ACPI_IORT symbol selection logic
ACPI IORT is an ACPI addendum to describe the connection topology of
devices with IOMMUs and interrupt controllers on ARM64 ACPI systems.
Currently the ACPI IORT Kbuild symbol is selected whenever the Kbuild
symbol ARM_GIC_V3_ITS is enabled, which in turn is selected by ARM64
Kbuild defaults. This makes the logic behind ACPI_IORT selection a bit
twisted and not easy to follow. On ARM64 systems enabling ACPI the
kbuild symbol ACPI_IORT should always be selected in that it is a kernel
layer provided to the ARM64 arch code to parse and enable ACPI firmware
bindings.
Make the ACPI_IORT selection explicit in ARM64 Kbuild and remove the
selection from ARM_GIC_V3_ITS entry, making the ACPI_IORT selection
logic clearer to follow.
Acked-by: Hanjun Guo <hanjun.guo@linaro.org> Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> Cc: Will Deacon <will.deacon@arm.com> Cc: Hanjun Guo <hanjun.guo@linaro.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
The current null-pointer check in __dma_alloc_coherent and
__dma_free_coherent is not needed anymore since the
__dma_alloc/__dma_free functions won't be called if !dev (dummy ops will
be called instead).
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by: Olav Haugan <ohaugan@codeaurora.org> Signed-off-by: Will Deacon <will.deacon@arm.com>
Arvind Yadav [Thu, 15 Jun 2017 09:25:37 +0000 (14:55 +0530)]
arm: perf: make of_device_ids const
of_device_ids are not supposed to change at runtime. All functions
working with of_device_ids provided by <linux/of.h> work with const
of_device_ids. So mark the non-const structs as const.
Signed-off-by: Arvind Yadav <arvind.yadav.cs@gmail.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
Punit Agrawal [Thu, 8 Jun 2017 17:25:28 +0000 (18:25 +0100)]
arm64: mm: Update perf accounting to handle poison faults
Re-organise the perf accounting for fault handling in preparation for
enabling handling of hardware poison faults in subsequent commits. The
change updates perf accounting to be inline with the behaviour on
x86.
With this update, the perf fault accounting -
* Always report PERF_COUNT_SW_PAGE_FAULTS
* Doesn't report anything else for VM_FAULT_ERROR (which includes
hwpoison faults)
* Reports PERF_COUNT_SW_PAGE_FAULTS_MAJ if it's a major
fault (indicated by VM_FAULT_MAJOR)
Add VM_FAULT_HWPOISON[_LARGE] handling to the arm64 page fault
handler. Handling of VM_FAULT_HWPOISON[_LARGE] is very similar
to VM_FAULT_OOM, the only difference is that a different si_code
(BUS_MCEERR_AR) is passed to user space and si_addr_lsb field is
initialized.
Signed-off-by: Jonathan (Zhixiong) Zhang <zjzhang@codeaurora.org> Signed-off-by: Tyler Baicar <tbaicar@codeaurora.org>
(fix new __do_user_fault call-site) Signed-off-by: Punit Agrawal <punit.agrawal@arm.com> Acked-by: Steve Capper <steve.capper@arm.com> Tested-by: Manoj Iyer <manoj.iyer@canonical.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
Punit Agrawal [Thu, 8 Jun 2017 17:25:26 +0000 (18:25 +0100)]
arm64: hugetlb: Fix huge_pte_offset to return poisoned page table entries
When memory failure is enabled, a poisoned hugepage pte is marked as a
swap entry. huge_pte_offset() does not return the poisoned page table
entries when it encounters PUD/PMD hugepages.
This behaviour of huge_pte_offset() leads to error such as below when
munmap is called on poisoned hugepages.
[ 344.165544] mm/pgtable-generic.c:33: bad pmd 000000083af00074.
Fix huge_pte_offset() to return the poisoned pte which is then
appropriately handled by the generic layer code.
Signed-off-by: Punit Agrawal <punit.agrawal@arm.com> Acked-by: Steve Capper <steve.capper@arm.com> Reviewed-by: Catalin Marinas <catalin.marinas@arm.com> Cc: David Woods <dwoods@mellanox.com> Tested-by: Manoj Iyer <manoj.iyer@canonical.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
Will Deacon [Mon, 12 Jun 2017 13:43:25 +0000 (14:43 +0100)]
arm64: ftrace: fix building without CONFIG_MODULES
When CONFIG_MODULES is disabled, we cannot dereference a module pointer:
arch/arm64/kernel/ftrace.c: In function 'ftrace_make_call':
arch/arm64/kernel/ftrace.c:107:36: error: dereferencing pointer to incomplete type 'struct module'
trampoline = (unsigned long *)mod->arch.ftrace_trampoline;
Also, the within_module() function is not defined:
arch/arm64/kernel/ftrace.c: In function 'ftrace_make_nop':
arch/arm64/kernel/ftrace.c:171:8: error: implicit declaration of function 'within_module'; did you mean 'init_module'? [-Werror=implicit-function-declaration]
This addresses both by adding replacing the IS_ENABLED(CONFIG_ARM64_MODULE_PLTS)
checks with #ifdef versions.
Fixes: e71a4e1bebaf ("arm64: ftrace: add support for far branches to dynamic ftrace") Reported-by: Arnd Bergmann <arnd@arndb.de> Acked-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Will Deacon <will.deacon@arm.com>
Will Deacon [Mon, 15 May 2017 14:23:58 +0000 (15:23 +0100)]
arm64: fault: Print info about page table structure when dumping pte
Whilst debugging a remote crash, I noticed that show_pte is unhelpful
when it comes to describing the structure of the page table being walked.
This is easily fixed by printing out the page table (swapper vs user),
page size and virtual address size when displaying the PGD address.
Acked-by: Mark Rutland <mark.rutland@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
Print out the name of the file associated with the vma that faulted.
This is usually the executable or shared library name. We already print
out the task name, but also printing the library name is useful for
pinpointing bugs to libraries.
Also print the base address and size of the vma, which together with the
PC (printed by __show_regs) gives the offset into the library.
Fault prints now look like:
test[2361]: unhandled level 2 translation fault (11) at 0x00000012, esr 0x92000006, in libfoo.so[ffffa0145000+1000]
This is already done on x86, for more details see commit 03252919b798
("x86: print which shared library/executable faulted in segfault etc.
messages v3").
Acked-by: Mark Rutland <mark.rutland@arm.com> Signed-off-by: Kristina Martsenko <kristina.martsenko@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
arm64: mm: don't print out page table entries on EL0 faults
When we take a fault from EL0 that can't be handled, we print out the
page table entries associated with the faulting address. This allows
userspace to print out any current page table entries, including kernel
(TTBR1) entries. Exposing kernel mappings like this could pose a
security risk, so don't print out page table information on EL0 faults.
(But still print it out for EL1 faults.) This also follows the same
behaviour as x86, printing out page table entries on kernel mode faults
but not user mode faults.
Acked-by: Mark Rutland <mark.rutland@arm.com> Signed-off-by: Kristina Martsenko <kristina.martsenko@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
When we take a fault that can't be handled, we print out the page table
entries associated with the faulting address. In some cases we currently
print out the wrong entries. For a faulting TTBR1 address, we sometimes
print out TTBR0 table entries instead, and for a faulting TTBR0 address
we sometimes print out TTBR1 table entries. Fix this by choosing the
tables based on the faulting address.
Acked-by: Mark Rutland <mark.rutland@arm.com> Signed-off-by: Kristina Martsenko <kristina.martsenko@arm.com>
[will: zero-extend addrs to 64-bit, don't walk swapper w/ TTBR0 addr] Signed-off-by: Will Deacon <will.deacon@arm.com>
Ard Biesheuvel [Tue, 6 Jun 2017 17:00:22 +0000 (17:00 +0000)]
arm64: ftrace: add support for far branches to dynamic ftrace
Currently, dynamic ftrace support in the arm64 kernel assumes that all
core kernel code is within range of ordinary branch instructions that
occur in module code, which is usually the case, but is no longer
guaranteed now that we have support for module PLTs and address space
randomization.
Since on arm64, all patching of branch instructions involves function
calls to the same entry point [ftrace_caller()], we can emit the modules
with a trampoline that has unlimited range, and patch both the trampoline
itself and the branch instruction to redirect the call via the trampoline.
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
[will: minor clarification to smp_wmb() comment] Signed-off-by: Will Deacon <will.deacon@arm.com>
Ard Biesheuvel [Tue, 6 Jun 2017 17:00:21 +0000 (17:00 +0000)]
arm64: ftrace: don't validate branch via PLT in ftrace_make_nop()
When turning branch instructions into NOPs, we attempt to validate the
action by comparing the old value at the call site with the opcode of
a direct relative branch instruction pointing at the old target.
However, these call sites are statically initialized to call _mcount(),
and may be redirected via a PLT entry if the module is loaded far away
from the kernel text, leading to false negatives and spurious errors.
So skip the validation if CONFIG_ARM64_MODULE_PLTS is configured.
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Will Deacon <will.deacon@arm.com>
Will Deacon [Mon, 5 Jun 2017 10:40:23 +0000 (11:40 +0100)]
arm64: cpufeature: Fix CPU_OUT_OF_SPEC taint for uniform systems
Commit 3fde2999fac5 ("arm64: cpufeature: Don't dump useless backtrace on
CPU_OUT_OF_SPEC") changed the cpufeature detection code to use add_taint
instead of WARN_TAINT_ONCE when detecting a heterogeneous system with
mismatched feature support. Unfortunately, this resulted in all systems
getting the taint, regardless of any feature mismatch.
This patch fixes the problem by conditionalising the taint on detecting
a feature mismatch.
Acked-by: Mark Rutland <mark.rutland@arm.com> Reported-by: Heiner Kallweit <hkallweit1@gmail.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
Ard Biesheuvel [Fri, 19 May 2017 15:42:00 +0000 (16:42 +0100)]
arm64: kernel: restrict /dev/mem read() calls to linear region
When running lscpu on an AArch64 system that has SMBIOS version 2.0
tables, it will segfault in the following way:
Unable to handle kernel paging request at virtual address ffff8000bfff0000
pgd = ffff8000f9615000
[ffff8000bfff0000] *pgd=0000000000000000
Internal error: Oops: 96000007 [#1] PREEMPT SMP
Modules linked in:
CPU: 0 PID: 1284 Comm: lscpu Not tainted 4.11.0-rc3+ #103
Hardware name: QEMU QEMU Virtual Machine, BIOS 0.0.0 02/06/2015
task: ffff8000fa78e800 task.stack: ffff8000f9780000
PC is at __arch_copy_to_user+0x90/0x220
LR is at read_mem+0xcc/0x140
This is caused by the fact that lspci issues a read() on /dev/mem at the
offset where it expects to find the SMBIOS structure array. However, this
region is classified as EFI_RUNTIME_SERVICE_DATA (as per the UEFI spec),
and so it is omitted from the linear mapping.
So let's restrict /dev/mem read/write access to those areas that are
covered by the linear region.
Reported-by: Alexander Graf <agraf@suse.de> Fixes: 4dffbfc48d65 ("arm64/efi: mark UEFI reserved regions as MEMBLOCK_NOMAP") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Will Deacon <will.deacon@arm.com>
Shaokun Zhang [Wed, 24 May 2017 07:43:18 +0000 (15:43 +0800)]
arm64: perf: Extend event config for ARMv8.1
Perf has supported ARMv8.1 feature with 16-bit evtCount filed [see c210ae8
arm64: perf: Extend event mask for ARMv8.1], event config should be
extended to 16-bit too, otherwise, if use -e event_name whose event_code
is more than 0x3ff, pmu_config_term will return -EINVAL because function
pmu_format_max_value depends on event config.
This patch extends event config to 16-bit.
Signed-off-by: Shaokun Zhang <zhangshaokun@hisilicon.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
PCI core requires the NUMA node for the struct pci_host_bridge.dev to
be set by using the pcibus_to_node(struct pci_bus*) API, that on ARM64
systems relies on the struct pci_host_bridge->bus.dev NUMA node.
The struct pci_host_bridge.dev NUMA node is then propagated through
the PCI device hierarchy as PCI devices (and bridges) are enumerated
under it.
Therefore, in order to set-up the PCI NUMA hierarchy appropriately, the
struct pci_host_bridge->bus.dev NUMA node must be set before core
code calls pcibus_to_node(struct pci_bus*) on it so that PCI core can
retrieve the NUMA node for the struct pci_host_bridge.dev device and can
propagate it through the PCI bus tree.
On ARM64 ACPI based systems the struct pci_host_bridge->bus.dev NUMA
node can be set-up in pcibios_root_bridge_prepare() by parsing the root
bridge ACPI device firmware binding.
Add code to the pcibios_root_bridge_prepare() that, when booting with
ACPI, parse the root bridge ACPI device companion NUMA binding and set
the corresponding struct pci_host_bridge->bus.dev NUMA node
appropriately.
Cc: Bjorn Helgaas <bhelgaas@google.com> Cc: Catalin Marinas <catalin.marinas@arm.com> Reviewed-by: Robert Richter <rrichter@cavium.com> Tested-by: Robert Richter <rrichter@cavium.com> Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
Will Deacon [Wed, 5 Apr 2017 10:14:05 +0000 (11:14 +0100)]
arm64: futex: Fix undefined behaviour with FUTEX_OP_OPARG_SHIFT usage
FUTEX_OP_OPARG_SHIFT instructs the futex code to treat the 12-bit oparg
field as a shift value, potentially leading to a left shift value that
is negative or with an absolute value that is significantly larger then
the size of the type. UBSAN chokes with:
* Making encoded_op an unsigned type, so we can shift it left even if
the top bit is set.
* Casting to signed prior to shifting right when extracting oparg
and cmparg
* Consider only the bottom 5 bits of oparg when using it as a left-shift
value.
Whilst I think this catches all of the issues, I'd much prefer to remove
this stuff, as I think it's unused and the bugs are copy-pasted between
a bunch of architectures.
Reviewed-by: Robin Murphy <robin.murphy@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
Kefeng Wang [Tue, 16 May 2017 07:36:16 +0000 (15:36 +0800)]
arm64: check return value of of_flat_dt_get_machine_name
It's useless to print machine name and setup arch-specific system
identifiers if of_flat_dt_get_machine_name() return NULL, especially
when ACPI-based boot.
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be> Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
Will Deacon [Thu, 18 May 2017 15:10:34 +0000 (16:10 +0100)]
arm64: cpufeature: Don't dump useless backtrace on CPU_OUT_OF_SPEC
Unfortunately, it turns out that mismatched CPU features in big.LITTLE
systems are starting to appear in the wild. Whilst we should continue to
taint the kernel with CPU_OUT_OF_SPEC for features that differ in ways
that we can't fix up, dumping a useless backtrace out of the cpufeature
code is pointless and irritating.
This patch removes the backtrace from the taint.
Signed-off-by: Mark Rutland <mark.rutland@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
Tobias Klauser [Mon, 15 May 2017 11:40:20 +0000 (13:40 +0200)]
arm64: mm: explicity include linux/vmalloc.h
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.
Kefeng Wang [Tue, 9 May 2017 01:53:37 +0000 (09:53 +0800)]
arm64: Add dump_backtrace() in show_regs
Generic code expects show_regs() to dump the stack, but arm64's
show_regs() does not. This makes it hard to debug softlockups and
other issues that result in show_regs() being called.
This patch updates arm64's show_regs() to dump the stack, as common
code expects.
Acked-by: Mark Rutland <mark.rutland@arm.com> Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
[will: folded in bug_handler fix from mrutland] Signed-off-by: Will Deacon <will.deacon@arm.com>
Kefeng Wang [Tue, 9 May 2017 01:53:36 +0000 (09:53 +0800)]
arm64: Call __show_regs directly
Generic code expects show_regs() to also dump the stack, but arm64's
show_reg() does not do this. Some arm64 callers of show_regs() *only*
want the registers dumped, without the stack.
To enable generic code to work as expected, we need to make
show_regs() dump the stack. Where we only want the registers dumped,
we must use __show_regs().
This patch updates code to use __show_regs() where only registers are
desired. A subsequent patch will modify show_regs().
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>
Dong Bo [Tue, 25 Apr 2017 06:11:29 +0000 (14:11 +0800)]
arm64: Preventing READ_IMPLIES_EXEC propagation
Like arch/arm/, we inherit the READ_IMPLIES_EXEC personality flag across
fork(). This is undesirable for a number of reasons:
* ELF files that don't require executable stack can end up with it
anyway
* We end up performing un-necessary I-cache maintenance when mapping
what should be non-executable pages
* Restricting what is executable is generally desirable when defending
against overflow attacks
This patch clears the personality flag when setting up the personality for
newly spwaned native tasks. Given that semi-recent AArch64 toolchains emit
a non-executable PT_GNU_STACK header, userspace applications can already
not rely on READ_IMPLIES_EXEC so shouldn't be adversely affected by this
change.
Cc: <stable@vger.kernel.org> Reported-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Dong Bo <dongbo4@huawei.com>
[will: added comment to compat code, rewrote commit message] Signed-off-by: Will Deacon <will.deacon@arm.com>
Linus Torvalds [Sun, 28 May 2017 23:18:27 +0000 (16:18 -0700)]
Merge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/evalenti/linux-soc-thermal
Pull thermal SoC management fixes from Eduardo Valentin:
- fixes to TI SoC driver, Broadcom, qoriq
- small sparse warning fix on thermal core
* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/evalenti/linux-soc-thermal:
thermal: broadcom: ns-thermal: default on iProc SoCs
ti-soc-thermal: Fix a typo in a comment line
ti-soc-thermal: Delete error messages for failed memory allocations in ti_bandgap_build()
ti-soc-thermal: Use devm_kcalloc() in ti_bandgap_build()
thermal: core: make thermal_emergency_poweroff static
thermal: qoriq: remove useless call for of_thermal_get_trip_points()
Linus Torvalds [Sat, 27 May 2017 16:39:09 +0000 (09:39 -0700)]
Merge tag 'tty-4.12-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty
Pull tty/serial fixes from Greg KH:
"Here are some serial and tty fixes for 4.12-rc3. They are a bit bigger
than normal, which is why I had them bake in linux-next for a few
weeks and didn't send them to you for -rc2.
They revert a few of the serdev patches from 4.12-rc1, and bring
things back to how they were in 4.11, to try to make things a bit more
stable there. Rob and Johan both agree that this is the way forward,
so this isn't people squabbling over semantics. Other than that, just
a few minor serial driver fixes that people have had problems with.
All of these have been in linux-next for a few weeks with no reported
issues"
* tag 'tty-4.12-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty:
serial: altera_uart: call iounmap() at driver remove
serial: imx: ensure UCR3 and UFCR are setup correctly
MAINTAINERS/serial: Change maintainer of jsm driver
serial: enable serdev support
tty/serdev: add serdev registration interface
serdev: Restore serdev_device_write_buf for atomic context
serial: core: fix crash in uart_suspend_port
tty: fix port buffer locking
tty: ehv_bytechan: clean up init error handling
serial: ifx6x60: fix use-after-free on module unload
serial: altera_jtaguart: adding iounmap()
serial: exar: Fix stuck MSIs
serial: efm32: Fix parity management in 'efm32_uart_console_get_options()'
serdev: fix tty-port client deregistration
Revert "tty_port: register tty ports with serdev bus"
drivers/tty: 8250: only call fintek_8250_probe when doing port I/O
Linus Torvalds [Sat, 27 May 2017 16:28:34 +0000 (09:28 -0700)]
Merge tag 'powerpc-4.12-4' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux
Pull powerpc fixes from Michael Ellerman:
"Fix running SPU programs on Cell, and a few other minor fixes.
Thanks to Alistair Popple, Jeremy Kerr, Michael Neuling, Nicholas
Piggin"
* tag 'powerpc-4.12-4' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
powerpc: Add PPC_FEATURE userspace bits for SCV and DARN instructions
powerpc/spufs: Fix hash faults for kernel regions
powerpc: Fix booting P9 hash with CONFIG_PPC_RADIX_MMU=N
powerpc/powernv/npu-dma.c: Fix opal_npu_destroy_context() call
selftests/powerpc: Fix TM resched DSCR test with some compilers
Linus Torvalds [Sat, 27 May 2017 16:17:58 +0000 (09:17 -0700)]
Merge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 fixes from Thomas Gleixner:
"A series of fixes for X86:
- The final fix for the end-of-stack issue in the unwinder
- Handle non PAT systems gracefully
- Prevent access to uninitiliazed memory
- Move early delay calaibration after basic init
- Fix Kconfig help text
- Fix a cross compile issue
- Unbreak older make versions"
* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/timers: Move simple_udelay_calibration past init_hypervisor_platform
x86/alternatives: Prevent uninitialized stack byte read in apply_alternatives()
x86/PAT: Fix Xorg regression on CPUs that don't support PAT
x86/watchdog: Fix Kconfig help text file path reference to lockup watchdog documentation
x86/build: Permit building with old make versions
x86/unwind: Add end-of-stack check for ftrace handlers
Revert "x86/entry: Fix the end of the stack for newly forked tasks"
x86/boot: Use CROSS_COMPILE prefix for readelf
Linus Torvalds [Sat, 27 May 2017 16:02:41 +0000 (09:02 -0700)]
Merge branch 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull perf tooling fixes from Thomas Gleixner:
- Synchronization of tools and kernel headers
- A series of fixes for perf report addressing various failures:
* Handle invalid maps proper
* Plug a memory leak
* Handle frames and callchain order correctly
- Fixes for handling inlines and children mode
* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
tools/include: Sync kernel ABI headers with tooling headers
perf tools: Put caller above callee in --children mode
perf report: Do not drop last inlined frame
perf report: Always honor callchain order for inlined nodes
perf script: Add --inline option for debugging
perf report: Fix off-by-one for non-activation frames
perf report: Fix memory leak in addr2line when called by addr2inlines
perf report: Don't crash on invalid maps in `-g srcline` mode
Linus Torvalds [Sat, 27 May 2017 15:30:30 +0000 (08:30 -0700)]
Merge tag 'trace-v4.12-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace
Pull ftrace fixes from Steven Rostedt:
"There's been a few memory issues found with ftrace.
One was simply a memory leak where not all was being freed that should
have been in releasing a file pointer on set_graph_function.
Then Thomas found that the ftrace trampolines were marked for
read/write as well as execute. To shrink the possible attack surface,
he added calls to set them to ro. Which also uncovered some other
issues with freeing module allocated memory that had its permissions
changed.
Kprobes had a similar issue which is fixed and a selftest was added to
trigger that issue again"
* tag 'trace-v4.12-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace:
x86/ftrace: Make sure that ftrace trampolines are not RWX
x86/mm/ftrace: Do not bug in early boot on irqs_disabled in cpu_flush_range()
selftests/ftrace: Add a testcase for many kprobe events
kprobes/x86: Fix to set RWX bits correctly before releasing trampoline
ftrace: Fix memory leak in ftrace_graph_release()
Thomas Gleixner [Thu, 25 May 2017 08:57:51 +0000 (10:57 +0200)]
x86/ftrace: Make sure that ftrace trampolines are not RWX
ftrace use module_alloc() to allocate trampoline pages. The mapping of
module_alloc() is RWX, which makes sense as the memory is written to right
after allocation. But nothing makes these pages RO after writing to them.
Add proper set_memory_rw/ro() calls to protect the trampolines after
modification.
Interrupts should not be enabled at this early in the boot process. It is
also fine to leave interrupts enabled during this time as there's only one
CPU running, and on_each_cpu() means to only run on the current CPU.
If early_boot_irqs_disabled is set, it is safe to run cpu_flush_range() with
interrupts disabled. Don't trigger a BUG_ON() in that case.
Masami Hiramatsu [Fri, 26 May 2017 04:44:54 +0000 (13:44 +0900)]
selftests/ftrace: Add a testcase for many kprobe events
Add a testcase to test kprobes via ftrace interface
with many concurrent kprobe events.
This tries to add many kprobe events (up to 256) on
kernel functions. To avoid making ftrace-based
kprobes (kprobes on fentry), it skips first N bytes
(on x86 N=5, on ppc or arm N=4) of function entry.
After that, it enables all those events, disable it,
and remove it.
Since the unoptimization buffer reclaiming will
be delayed, after removing events, it will wait
enough time.
Masami Hiramatsu [Thu, 25 May 2017 10:38:17 +0000 (19:38 +0900)]
kprobes/x86: Fix to set RWX bits correctly before releasing trampoline
Fix kprobes to set(recover) RWX bits correctly on trampoline
buffer before releasing it. Releasing readonly page to
module_memfree() crash the kernel.
Without this fix, if kprobes user register a bunch of kprobes
in function body (since kprobes on function entry usually
use ftrace) and unregister it, kernel hits a BUG and crash.
Luis Henriques [Thu, 25 May 2017 15:20:38 +0000 (16:20 +0100)]
ftrace: Fix memory leak in ftrace_graph_release()
ftrace_hash is being kfree'ed in ftrace_graph_release(), however the
->buckets field is not. This results in a memory leak that is easily
captured by kmemleak:
Linus Torvalds [Fri, 26 May 2017 23:45:13 +0000 (16:45 -0700)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
Pull input layer fixes from Dmitry Torokhov:
"Just a few fixups to a couple of drivers"
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
Input: elan_i2c - ignore signals when finishing updating firmware
Input: elan_i2c - clear INT before resetting controller
Input: atmel_mxt_ts - add T100 as a readable object
Input: edt-ft5x06 - increase allowed data range for threshold parameter
1) Fix state pruning in bpf verifier wrt. alignment, from Daniel
Borkmann.
2) Handle non-linear SKBs properly in SCTP ICMP parsing, from Davide
Caratti.
3) Fix bit field definitions for rss_hash_type of descriptors in mlx5
driver, from Jesper Brouer.
4) Defer slave->link updates until bonding is ready to do a full commit
to the new settings, from Nithin Sujir.
5) Properly reference count ipv4 FIB metrics to avoid use after free
situations, from Eric Dumazet and several others including Cong Wang
and Julian Anastasov.
6) Fix races in llc_ui_bind(), from Lin Zhang.
7) Fix regression of ESP UDP encapsulation for TCP packets, from
Steffen Klassert.
8) Fix mdio-octeon driver Kconfig deps, from Randy Dunlap.
9) Fix regression in setting DSCP on ipv6/GRE encapsulation, from Peter
Dawson.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (43 commits)
ipv4: add reference counting to metrics
net: ethernet: ax88796: don't call free_irq without request_irq first
ip6_tunnel, ip6_gre: fix setting of DSCP on encapsulated packets
sctp: fix ICMP processing if skb is non-linear
net: llc: add lock_sock in llc_ui_bind to avoid a race condition
bonding: Don't update slave->link until ready to commit
test_bpf: Add a couple of tests for BPF_JSGE.
bpf: add various verifier test cases
bpf: fix wrong exposure of map_flags into fdinfo for lpm
bpf: add bpf_clone_redirect to bpf_helper_changes_pkt_data
bpf: properly reset caller saved regs after helper call and ld_abs/ind
bpf: fix incorrect pruning decision when alignment must be tracked
arp: fixed -Wuninitialized compiler warning
tcp: avoid fastopen API to be used on AF_UNSPEC
net: move somaxconn init from sysctl code
net: fix potential null pointer dereference
geneve: fix fill_info when using collect_metadata
virtio-net: enable TSO/checksum offloads for Q-in-Q vlans
be2net: Fix offload features for Q-in-Q packets
vlan: Fix tcp checksum offloads in Q-in-Q vlans
...
- Fix warnings about unused variables that appeared in -rc1.
- Don't spew errors when bmapping a local format directory
- Fix an off-by-one error in a delalloc eof assertion
- Make fsmap only return inode information for CAP_SYS_ADMIN
- Fix a potential mount time deadlock recovering cow extents
- Fix unaligned memory access in _btree_visit_blocks
- Fix various SEEK_HOLE/SEEK_DATA bugs"
* tag 'xfs-4.12-fixes-2' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux:
xfs: Move handling of missing page into one place in xfs_find_get_desired_pgoff()
xfs: Fix off-by-in in loop termination in xfs_find_get_desired_pgoff()
xfs: Fix missed holes in SEEK_HOLE implementation
xfs: fix off-by-one on max nr_pages in xfs_find_get_desired_pgoff()
xfs: fix unaligned access in xfs_btree_visit_blocks
xfs: avoid mount-time deadlock in CoW extent recovery
xfs: only return detailed fsmap info if the caller has CAP_SYS_ADMIN
xfs: bad assertion for delalloc an extent that start at i_size
xfs: fix warnings about unused stack variables
xfs: BMAPX shouldn't barf on inline-format directories
xfs: fix indlen accounting error on partial delalloc conversion
2) At the same time run following loop :
while :
do
ip ro add 10.246.7.152 dev eth0 src 10.246.7.151 mtu 1500
ip ro del 10.246.7.152 dev eth0 src 10.246.7.151 mtu 1500
done
Cong Wang attempted to add back rt->fi in commit 82486aa6f1b9 ("ipv4: restore rt->fi for reference counting")
but this proved to add some issues that were complex to solve.
Instead, I suggested to add a refcount to the metrics themselves,
being a standalone object (in particular, no reference to other objects)
I tried to make this patch as small as possible to ease its backport,
instead of being super clean. Note that we believe that only ipv4 dst
need to take care of the metric refcount. But if this is wrong,
this patch adds the basic infrastructure to extend this to other
families.
Many thanks to Julian Anastasov for reviewing this patch, and Cong Wang
for his efforts on this problem.
Fixes: 2860583fe840 ("ipv4: Kill rt->fi") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: Andrey Konovalov <andreyknvl@google.com> Reviewed-by: Julian Anastasov <ja@ssi.bg> Acked-by: Cong Wang <xiyou.wangcong@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
net: ethernet: ax88796: don't call free_irq without request_irq first
The function ax_init_dev (which is called only from the driver's .probe
function) calls free_irq in the error path without having requested the
irq in the first place. So drop the free_irq call in the error path.
Fixes: 825a2ff1896e ("AX88796 network driver") Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de> Signed-off-by: David S. Miller <davem@davemloft.net>
Peter Dawson [Thu, 25 May 2017 20:35:18 +0000 (06:35 +1000)]
ip6_tunnel, ip6_gre: fix setting of DSCP on encapsulated packets
This fix addresses two problems in the way the DSCP field is formulated
on the encapsulating header of IPv6 tunnels.
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=195661
1) The IPv6 tunneling code was manipulating the DSCP field of the
encapsulating packet using the 32b flowlabel. Since the flowlabel is
only the lower 20b it was incorrect to assume that the upper 12b
containing the DSCP and ECN fields would remain intact when formulating
the encapsulating header. This fix handles the 'inherit' and
'fixed-value' DSCP cases explicitly using the extant dsfield u8 variable.
2) The use of INET_ECN_encapsulate(0, dsfield) in ip6_tnl_xmit was
incorrect and resulted in the DSCP value always being set to 0.
Commit 90427ef5d2a4 ("ipv6: fix flow labels when the traffic class
is non-0") caused the regression by masking out the flowlabel
which exposed the incorrect handling of the DSCP portion of the
flowlabel in ip6_tunnel and ip6_gre.
Fixes: 90427ef5d2a4 ("ipv6: fix flow labels when the traffic class is non-0") Signed-off-by: Peter Dawson <peter.a.dawson@boeing.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Davide Caratti [Thu, 25 May 2017 17:14:56 +0000 (19:14 +0200)]
sctp: fix ICMP processing if skb is non-linear
sometimes ICMP replies to INIT chunks are ignored by the client, even if
the encapsulated SCTP headers match an open socket. This happens when the
ICMP packet is carried by a paged skb: use skb_header_pointer() to read
packet contents beyond the SCTP header, so that chunk header and initiate
tag are validated correctly.
v2:
- don't use skb_header_pointer() to read the transport header, since
icmp_socket_deliver() already puts these 8 bytes in the linear area.
- change commit message to make specific reference to INIT chunks.
Signed-off-by: Davide Caratti <dcaratti@redhat.com> Acked-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com> Acked-by: Vlad Yasevich <vyasevich@gmail.com> Reviewed-by: Xin Long <lucien.xin@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
linzhang [Thu, 25 May 2017 06:07:18 +0000 (14:07 +0800)]
net: llc: add lock_sock in llc_ui_bind to avoid a race condition
There is a race condition in llc_ui_bind if two or more processes/threads
try to bind a same socket.
If more processes/threads bind a same socket success that will lead to
two problems, one is this action is not what we expected, another is
will lead to kernel in unstable status or oops(in my simple test case,
cause llc2.ko can't unload).
The current code is test SOCK_ZAPPED bit to avoid a process to
bind a same socket twice but that is can't avoid more processes/threads
try to bind a same socket at the same time.
So, add lock_sock in llc_ui_bind like others, such as llc_ui_connect.
Signed-off-by: Lin Zhang <xiaolou4617@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Linus Torvalds [Fri, 26 May 2017 18:05:22 +0000 (11:05 -0700)]
Merge branch 'for-linus' of git://git.kernel.dk/linux-block
Pull block fixes from Jens Axboe:
"A collection of fixes that should go into this series. This contains:
- A set of NVMe fixes, pulled from Christoph. This includes a set of
fixes for the fiber channel bits from James Smart, rdma queue depth
fix from Marta, controller removal fixes from Ming, and some more
APST quirk updates from Andy.
- A blk-mq debugfs fix from Bart, fixing a problem with the
untangling of the sysfs and debugfs blk-mq bits that was added in
this series.
- Error code fix in add_partition() from Dan.
- A small series of fixes for the new blk-throttle code from Shaohua"
* 'for-linus' of git://git.kernel.dk/linux-block: (21 commits)
blk-mq: Only register debugfs attributes for blk-mq queues
nvme: Quirk APST on Intel 600P/P3100 devices
nvme: only setup block integrity if supported by the driver
nvme: replace is_flags field in nvme_ctrl_ops with a flags field
nvme-pci: consistencly use ctrl->device for logging
partitions/msdos: FreeBSD UFS2 file systems are not recognized
block: fix an error code in add_partition()
blk-throttle: force user to configure all settings for io.low
blk-throttle: respect 0 bps/iops settings for io.low
blk-throttle: output some debug info in trace
blk-throttle: add hierarchy support for latency target and idle time
nvme_fc: remove extra controller reference taken on reconnect
nvme_fc: correct nvme status set on abort
nvme_fc: set logging level on resets/deletes
nvme_fc: revise comment on teardown
nvme_fc: Support ctrl_loss_tmo
nvme_fc: get rid of local reconnect_delay
blk-mq: remove blk_mq_abort_requeue_list()
nvme: avoid to use blk_mq_abort_requeue_list()
nvme: use blk_mq_start_hw_queues() in nvme_kill_queues()
...
Linus Torvalds [Fri, 26 May 2017 17:51:18 +0000 (10:51 -0700)]
Merge tag 'pci-v4.12-fixes-1' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci
Pull PCI fixes from Bjorn Helgaas:
- fix PCI_ENDPOINT build error (merged for v4.12)
- fix Switchtec driver (merged for v4.12)
- fix imx6 config read timeouts, fallout from changing to non-postable
reads
- add PM "needs_resume" flag for i915 suspend issue
* tag 'pci-v4.12-fixes-1' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci:
PCI/PM: Add needs_resume flag to avoid suspend complete optimization
PCI: imx6: Fix config read timeout handling
switchtec: Fix minor bug with partition ID register
switchtec: Use new cdev_device_add() helper function
PCI: endpoint: Make PCI_ENDPOINT depend on HAS_DMA
Linus Torvalds [Fri, 26 May 2017 16:35:22 +0000 (09:35 -0700)]
Merge tag 'ceph-for-4.12-rc3' of git://github.com/ceph/ceph-client
Pul ceph fixes from Ilya Dryomov:
"A bunch of make W=1 and static checker fixups, a RECONNECT_SEQ
messenger patch from Zheng and Luis' fallocate fix"
* tag 'ceph-for-4.12-rc3' of git://github.com/ceph/ceph-client:
ceph: check that the new inode size is within limits in ceph_fallocate()
libceph: cleanup old messages according to reconnect seq
libceph: NULL deref on crush_decode() error path
libceph: fix error handling in process_one_ticket()
libceph: validate blob_struct_v in process_one_ticket()
libceph: drop version variable from ceph_monmap_decode()
libceph: make ceph_msg_data_advance() return void
libceph: use kbasename() and kill ceph_file_part()
Linus Torvalds [Fri, 26 May 2017 16:05:35 +0000 (09:05 -0700)]
Merge tag 'mmc-v4.12-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/ulfh/mmc
Pull MMC fixes from Ulf Hansson:
"This contains fixes to make the WiFi work again for the ARM64 Hikey
board.
Together with a couple of DTS updates for the Hikey board we have also
extended the mmc pwrseq_simple, to support a new power-off-delay-us DT
property, as that was required to enable a graceful power off sequence
for the WiFi chip"
* tag 'mmc-v4.12-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/ulfh/mmc:
arm64: dts: hikey: Fix WiFi support
arm64: dts: hi6220: Move board data from the dwmmc nodes to hikey dts
arm64: dts: hikey: Add the SYS_5V and the VDD_3V3 regulators
arm64: dts: hi6220: Move the fixed_5v_hub regulator to the hikey dts
arm64: dts: hikey: Add clock for the pmic mfd
mfd: dts: hi655x: Add clock binding for the pmic
mmc: pwrseq_simple: Parse DTS for the power-off-delay-us property
mmc: dt: pwrseq-simple: Invent power-off-delay-us
Linus Torvalds [Fri, 26 May 2017 16:03:09 +0000 (09:03 -0700)]
Merge tag 'sound-4.12-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound
Pull sound fixes from Takashi Iwai:
"This contains a few HD-audio device-specific quirks and an endianess
fix for USB-audio, as well as the update of quirk model list document.
All fixes are small and trivial.
The document update could have been postponed, but it's a good thing
for user and has absolutely zero risk of breakage, so included here"
* tag 'sound-4.12-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
ALSA: hda - apply STAC_9200_DELL_M22 quirk for Dell Latitude D430
ALSA: hda - Update the list of quirk models
ALSA: hda - Provide dual-codecs model option for a few Realtek codecs
ALSA: hda - Apply dual-codec quirk for MSI Z270-Gaming mobo
ALSA: hda - No loopback on ALC299 codec
ALSA: usb-audio: fix Amanero Combo384 quirk on big-endian hosts
Linus Torvalds [Fri, 26 May 2017 15:54:06 +0000 (08:54 -0700)]
Merge tag 'drm-fixes-for-v4.12-rc3' of git://people.freedesktop.org/~airlied/linux
Pull drm fixes from Dave Airlie:
"Not a whole lot happening here, a set of amdgpu fixes and one core
deadlock fix, and some misc drivers fixes"
* tag 'drm-fixes-for-v4.12-rc3' of git://people.freedesktop.org/~airlied/linux:
drm/amdgpu: fix null point error when rmmod amdgpu.
drm/amd/powerplay: fix a signedness bugs
drm/amdgpu: fix NULL pointer panic of emit_gds_switch
drm/radeon: Unbreak HPD handling for r600+
drm/amd/powerplay/smu7: disable mclk switching for high refresh rates
drm/amd/powerplay/smu7: add vblank check for mclk switching (v2)
drm/radeon/ci: disable mclk switching for high refresh rates (v2)
drm/amdgpu/ci: disable mclk switching for high refresh rates (v2)
drm/amdgpu: fix fundamental suspend/resume issue
drm/gma500/psb: Actually use VBT mode when it is found
drm: Fix deadlock retry loop in page_flip_ioctl
drm: qxl: Delay entering atomic context during cursor update
drm/radeon: Fix oops upon driver load on PowerXpress laptops
Jens Axboe [Fri, 26 May 2017 15:11:19 +0000 (09:11 -0600)]
Merge branch 'nvme-4.12' of git://git.infradead.org/nvme into for-linus
Christoph writes:
"A couple of fixes for the next rc on the nvme front. Various FC fixes
from James, controller removal fixes from Ming (including a block layer
patch), a APST related device quirk from Andy, a RDMA fix for small
queue depth device from Marta, as well as fixes for the lack of
metadata support in non-PCIe drivers and the printk logging format from
me."
Bart Van Assche [Thu, 25 May 2017 23:38:06 +0000 (16:38 -0700)]
blk-mq: Only register debugfs attributes for blk-mq queues
The code in blk-mq-debugfs.c assumes that it is working on a blk-mq
queue and is not intended to work on a blk-sq queue. Hence only
register blk-mq debugfs attributes for blk-mq queues.
Fixes: commit 9c1051aacde8 ("blk-mq: untangle debugfs and sysfs") Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Ming Lei <ming.lei@redhat.com> Reviewed-by: Omar Sandoval <osandov@fb.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Signed-off-by: Jens Axboe <axboe@fb.com>
Andy Lutomirski [Wed, 24 May 2017 22:06:31 +0000 (15:06 -0700)]
nvme: Quirk APST on Intel 600P/P3100 devices
They have known firmware bugs. A fix is apparently in the works --
once fixed firmware is available, someone from Intel (Hi, Keith!)
can adjust the quirk accordingly.
Cc: stable@vger.kernel.org # v4.11 Cc: Kai-Heng Feng <kai.heng.feng@canonical.com> Cc: Mario Limonciello <mario_limonciello@dell.com> Signed-off-by: Andy Lutomirski <luto@kernel.org> Signed-off-by: Christoph Hellwig <hch@lst.de>
nvme: only setup block integrity if supported by the driver
Currently only the PCIe driver supports metadata, so we should not claim
integrity support for the other drivers. This prevents nasty crashes
with targets that advertise metadata support on fabrics.
Also use the opportunity to factor out some code into a separate helper
that isn't even compiled if CONFIG_BLK_DEV_INTEGRITY is disabled.
Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Keith Busch <keith.busch@intel.com>
Dave Airlie [Fri, 26 May 2017 01:51:55 +0000 (11:51 +1000)]
Merge branch 'drm-fixes-4.12' of git://people.freedesktop.org/~agd5f/linux into drm-fixes
A bunch of bug fixes:
- Fix display flickering on some chips at high refresh rates
- suspend/resume fix
- hotplug fix
- a couple of segfault fixes for certain cases
* 'drm-fixes-4.12' of git://people.freedesktop.org/~agd5f/linux:
drm/amdgpu: fix null point error when rmmod amdgpu.
drm/amd/powerplay: fix a signedness bugs
drm/amdgpu: fix NULL pointer panic of emit_gds_switch
drm/radeon: Unbreak HPD handling for r600+
drm/amd/powerplay/smu7: disable mclk switching for high refresh rates
drm/amd/powerplay/smu7: add vblank check for mclk switching (v2)
drm/radeon/ci: disable mclk switching for high refresh rates (v2)
drm/amdgpu/ci: disable mclk switching for high refresh rates (v2)
drm/amdgpu: fix fundamental suspend/resume issue
Dave Airlie [Fri, 26 May 2017 01:51:28 +0000 (11:51 +1000)]
Merge tag 'drm-misc-fixes-2017-05-25' of git://anongit.freedesktop.org/git/drm-misc into drm-fixes
Core Changes:
- Don't drop vblank reference more than once in cases of ww retry (Daniel)
Driver Changes:
- radeon: Fix oops during radeon probe trying to reference wrong device (Lukas)
- qxl: Avoid sleeping while in atomic context on cursor update (Gabriel)
- gma500: Use VBT mode instead of pre-programmed mode for LVDS (Patrik)
Cc: Lukas Wunner <lukas@wunner.de> Cc: Gabriel Krisman Bertazi <krisman@collabora.co.uk> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
* tag 'drm-misc-fixes-2017-05-25' of git://anongit.freedesktop.org/git/drm-misc:
drm/gma500/psb: Actually use VBT mode when it is found
drm: Fix deadlock retry loop in page_flip_ioctl
drm: qxl: Delay entering atomic context during cursor update
drm/radeon: Fix oops upon driver load on PowerXpress laptops
Nithin Sujir [Thu, 25 May 2017 02:45:17 +0000 (19:45 -0700)]
bonding: Don't update slave->link until ready to commit
In the loadbalance arp monitoring scheme, when a slave link change is
detected, the slave->link is immediately updated and slave_state_changed
is set. Later down the function, the rtnl_lock is acquired and the
changes are committed, updating the bond link state.
However, the acquisition of the rtnl_lock can fail. The next time the
monitor runs, since slave->link is already updated, it determines that
link is unchanged. This results in the bond link state permanently out
of sync with the slave link.
This patch modifies bond_loadbalance_arp_mon() to handle link changes
identical to bond_ab_arp_{inspect/commit}(). The new link state is
maintained in slave->new_link until we're ready to commit at which point
it's copied into slave->link.
NOTE: miimon_{inspect/commit}() has a more complex state machine
requiring the use of the bond_{propose,commit}_link_state() functions
which maintains the intermediate state in slave->link_new_state. The arp
monitors don't require that.
Testing: This bug is very easy to reproduce with the following steps.
1. In a loop, toggle a slave link of a bond slave interface.
2. In a separate loop, do ifconfig up/down of an unrelated interface to
create contention for rtnl_lock.
Within a few iterations, the bond link goes out of sync with the slave
link.
Signed-off-by: Nithin Nayak Sujir <nsujir@tintri.com> Cc: Mahesh Bandewar <maheshb@google.com> Cc: Jay Vosburgh <jay.vosburgh@canonical.com> Acked-by: Mahesh Bandewar <maheshb@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
David Daney [Wed, 24 May 2017 23:35:49 +0000 (16:35 -0700)]
test_bpf: Add a couple of tests for BPF_JSGE.
Some JITs can optimize comparisons with zero. Add a couple of
BPF_JSGE tests against immediate zero.
Signed-off-by: David Daney <david.daney@cavium.com> Acked-by: Daniel Borkmann <daniel@iogearbox.net> Signed-off-by: David S. Miller <davem@davemloft.net>
David S. Miller [Thu, 25 May 2017 17:44:29 +0000 (13:44 -0400)]
Merge branch 'bpf-fixes'
Daniel Borkmann says:
====================
Various BPF fixes
Follow-up to fix incorrect pruning when alignment tracking is
in use and to properly clear regs after call to not leave stale
data behind, also a fix that adds bpf_clone_redirect to the
bpf_helper_changes_pkt_data helper and exposes correct map_flags
for lpm map into fdinfo. For details, please see individual
patches.
v1 -> v2:
- Reworked first patch so that env->strict_alignment is the
final indicator on whether we have to deal with strict
alignment rather than having CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS
checks on various locations, so only checking env->strict_alignment
is sufficient after that. Thanks for spotting, Dave!
- Added patch 3 and 4.
- Rest as is.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Borkmann [Wed, 24 May 2017 23:05:09 +0000 (01:05 +0200)]
bpf: add various verifier test cases
This patch adds various verifier test cases:
1) A test case for the pruning issue when tracking alignment
is used.
2) Various PTR_TO_MAP_VALUE_OR_NULL tests to make sure pointer
arithmetic turns such register into UNKNOWN_VALUE type.
3) Test cases for the special treatment of LD_ABS/LD_IND to
make sure verifier doesn't break calling convention here.
Latter is needed, since f.e. arm64 JIT uses r1 - r5 for
storing temporary data, so they really must be marked as
NOT_INIT.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Acked-by: Alexei Starovoitov <ast@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Borkmann [Wed, 24 May 2017 23:05:08 +0000 (01:05 +0200)]
bpf: fix wrong exposure of map_flags into fdinfo for lpm
trie_alloc() always needs to have BPF_F_NO_PREALLOC passed in via
attr->map_flags, since it does not support preallocation yet. We
check the flag, but we never copy the flag into trie->map.map_flags,
which is later on exposed into fdinfo and used by loaders such as
iproute2. Latter uses this in bpf_map_selfcheck_pinned() to test
whether a pinned map has the same spec as the one from the BPF obj
file and if not, bails out, which is currently the case for lpm
since it exposes always 0 as flags.
Also copy over flags in array_map_alloc() and stack_map_alloc().
They always have to be 0 right now, but we should make sure to not
miss to copy them over at a later point in time when we add actual
flags for them to use.
Fixes: b95a5c4db09b ("bpf: add a longest prefix match trie map implementation") Reported-by: Jarno Rajahalme <jarno@covalent.io> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Acked-by: Alexei Starovoitov <ast@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Borkmann [Wed, 24 May 2017 23:05:07 +0000 (01:05 +0200)]
bpf: add bpf_clone_redirect to bpf_helper_changes_pkt_data
The bpf_clone_redirect() still needs to be listed in
bpf_helper_changes_pkt_data() since we call into
bpf_try_make_head_writable() from there, thus we need
to invalidate prior pkt regs as well.
Fixes: 36bbef52c7eb ("bpf: direct packet write and access for helpers for clsact progs") Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Acked-by: Alexei Starovoitov <ast@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Borkmann [Wed, 24 May 2017 23:05:06 +0000 (01:05 +0200)]
bpf: properly reset caller saved regs after helper call and ld_abs/ind
Currently, after performing helper calls, we clear all caller saved
registers, that is r0 - r5 and fill r0 depending on struct bpf_func_proto
specification. The way we reset these regs can affect pruning decisions
in later paths, since we only reset register's imm to 0 and type to
NOT_INIT. However, we leave out clearing of other variables such as id,
min_value, max_value, etc, which can later on lead to pruning mismatches
due to stale data.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Acked-by: Alexei Starovoitov <ast@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Borkmann [Wed, 24 May 2017 23:05:05 +0000 (01:05 +0200)]
bpf: fix incorrect pruning decision when alignment must be tracked
Currently, when we enforce alignment tracking on direct packet access,
the verifier lets the following program pass despite doing a packet
write with unaligned access:
from 6 to 8:
R0=pkt(id=0,off=14,r=0) R1=ctx R2=pkt(id=0,off=0,r=0)
R3=pkt_end R7=inv,min_value=0,max_value=1 R10=fp
8: (b7) r0 = 0
9: (95) exit
from 5 to 10:
R0=pkt(id=0,off=14,r=0) R1=ctx R2=pkt(id=0,off=0,r=0)
R3=pkt_end R7=inv,min_value=2 R10=fp
10: (07) r0 += 1
11: (05) goto pc-6
6: safe <----- here, wrongly found safe
processed 15 insns
However, if we enforce a pruning mismatch by adding state into r8
which is then being mismatched in states_equal(), we find that for
the otherwise same program, the verifier detects a misaligned packet
access when actually walking that path:
... will let the above pass. The situation we run into is that
old->off <= cur->off (14 <= 15), meaning that prior walked paths
went with smaller offset, which was later used in the packet
access after successful packet range check and found to be safe
already.
For example: Given is R0=pkt(id=0,off=0,r=0). Adding offset 14
as in above program to it, results in R0=pkt(id=0,off=14,r=0)
before the packet range test. Now, testing this against R3=pkt_end
with 'if r0 > r3 goto out' will transform R0 into R0=pkt(id=0,off=14,r=14)
for the case when we're within bounds. A write into the packet
at offset *(u32 *)(r0 -4), that is, 2 + 14 -4, is valid and
aligned (2 is for NET_IP_ALIGN). After processing this with
all fall-through paths, we later on check paths from branches.
When the above skb->mark test is true, then we jump near the
end of the program, perform r0 += 1, and jump back to the
'if r0 > r3 goto out' test we've visited earlier already. This
time, R0 is of type R0=pkt(id=0,off=15,r=0), and we'll prune
that part because this time we'll have a larger safe packet
range, and we already found that with off=14 all further insn
were already safe, so it's safe as well with a larger off.
However, the problem is that the subsequent write into the packet
with 2 + 15 -4 is then unaligned, and not caught by the alignment
tracking. Note that min_align, aux_off, and aux_off_align were
all 0 in this example.
Since we cannot tell at this time what kind of packet access was
performed in the prior walk and what minimal requirements it has
(we might do so in the future, but that requires more complexity),
fix it to disable this pruning case for strict alignment for now,
and let the verifier do check such paths instead. With that applied,
the test cases pass and reject the program due to misalignment.
Fixes: d1174416747d ("bpf: Track alignment of register values in the verifier.")
Reference: http://patchwork.ozlabs.org/patch/761909/ Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Acked-by: Alexei Starovoitov <ast@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
net/ipv4/arp.c:880:35: warning: 'addr_type' may be used uninitialized in
this function [-Wmaybe-uninitialized]
While the code logic seems to be correct and doesn't allow the variable
to be used uninitialized, and the warning is not consistently
reproducible, it's still worth fixing it for other people not to waste
time looking at the warning in case it pops up in the build environment.
Yes, compiler is probably at fault, but we will need to accommodate.
Fixes: 7d472a59c0e5 ("arp: always override existing neigh entries with gratuitous ARP") Signed-off-by: Ihar Hrachyshka <ihrachys@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Wei Wang [Wed, 24 May 2017 16:59:31 +0000 (09:59 -0700)]
tcp: avoid fastopen API to be used on AF_UNSPEC
Fastopen API should be used to perform fastopen operations on the TCP
socket. It does not make sense to use fastopen API to perform disconnect
by calling it with AF_UNSPEC. The fastopen data path is also prone to
race conditions and bugs when using with AF_UNSPEC.
One issue reported and analyzed by Vegard Nossum is as follows:
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Thread A: Thread B:
------------------------------------------------------------------------
sendto()
- tcp_sendmsg()
- sk_stream_memory_free() = 0
- goto wait_for_sndbuf
- sk_stream_wait_memory()
- sk_wait_event() // sleep
| sendto(flags=MSG_FASTOPEN, dest_addr=AF_UNSPEC)
| - tcp_sendmsg()
| - tcp_sendmsg_fastopen()
| - __inet_stream_connect()
| - tcp_disconnect() //because of AF_UNSPEC
| - tcp_transmit_skb()// send RST
| - return 0; // no reconnect!
| - sk_stream_wait_connect()
| - sock_error()
| - xchg(&sk->sk_err, 0)
| - return -ECONNRESET
- ... // wake up, see sk->sk_err == 0
- skb_entail() on TCP_CLOSE socket
If the connection is reopened then we will send a brand new SYN packet
after thread A has already queued a buffer. At this point I think the
socket internal state (sequence numbers etc.) becomes messed up.
When the new connection is closed, the FIN-ACK is rejected because the
sequence number is outside the window. The other side tries to
retransmit,
but __tcp_retransmit_skb() calls tcp_trim_head() on an empty skb which
corrupts the skb data length and hits a BUG() in copy_and_csum_bits().
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Hence, this patch adds a check for AF_UNSPEC in the fastopen data path
and return EOPNOTSUPP to user if such case happens.
Fixes: cf60af03ca4e7 ("tcp: Fast Open client - sendmsg(MSG_FASTOPEN)") Reported-by: Vegard Nossum <vegard.nossum@oracle.com> Signed-off-by: Wei Wang <weiwan@google.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Roman Kapl [Wed, 24 May 2017 08:22:22 +0000 (10:22 +0200)]
net: move somaxconn init from sysctl code
The default value for somaxconn is set in sysctl_core_net_init(), but this
function is not called when kernel is configured without CONFIG_SYSCTL.
This results in the kernel not being able to accept TCP connections,
because the backlog has zero size. Usually, the user ends up with:
"TCP: request_sock_TCP: Possible SYN flooding on port 7. Dropping request. Check SNMP counters."
If SYN cookies are not enabled the connection is rejected.
Before ef547f2ac16 (tcp: remove max_qlen_log), the effects were less
severe, because the backlog was always at least eight slots long.
Signed-off-by: Roman Kapl <roman.kapl@sysgo.com> Signed-off-by: David S. Miller <davem@davemloft.net>
projects / karo-tx-linux.git / log