Ever since commit 091d42e43d ("iommu/vt-d: Copy translation tables from
old kernel") the kdump kernel copies the IOMMU context tables from the
previous kernel. Each device mappings will be destroyed once the driver
for the respective device takes over.
This unfortunately breaks the workflow of mapping and unmapping a new
context to the IOMMU. The mapping function assumes that either:
1) Unmapping did the proper IOMMU flushing and it only ever flush if the
IOMMU unit supports caching invalid entries.
2) The system just booted and the initialization code took care of
flushing all IOMMU caches.
This assumption is not true for the kdump kernel since the context
tables have been copied from the previous kernel and translations could
have been cached ever since. So make sure to flush the IOTLB as well
when we destroy these old copied mappings.
Cc: Joerg Roedel <joro@8bytes.org> Cc: David Woodhouse <dwmw2@infradead.org> Cc: David Woodhouse <dwmw@amazon.co.uk> Cc: Anthony Liguori <aliguori@amazon.com> Signed-off-by: KarimAllah Ahmed <karahmed@amazon.de> Acked-by: David Woodhouse <dwmw@amazon.co.uk> Fixes: 091d42e43d ("iommu/vt-d: Copy translation tables from old kernel") Signed-off-by: Joerg Roedel <jroedel@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
vchiq_arm supports transfers less than one page and at arbitrary
alignment, using the dma-mapping API to perform its cache maintenance
(even though the VPU drives the DMA hardware). Read (DMA_FROM_DEVICE)
operations use cache invalidation for speed, falling back to
clean+invalidate on partial cache lines, with writes (DMA_TO_DEVICE)
using flushes.
If a read transfer has ends which aren't page-aligned, performing cache
maintenance as if they were whole pages can lead to memory corruption
since the partial cache lines at the ends (and any cache lines before or
after the transfer area) will be invalidated. This bug was masked until
the disabling of the cache flush in flush_dcache_page().
Honouring the requested transfer start- and end-points prevents the
corruption.
Fixes: cf9caf192988 ("staging: vc04_services: Replace dmac_map_area with dmac_map_sg") Signed-off-by: Phil Elwell <phil@raspberrypi.org> Reported-by: Stefan Wahren <stefan.wahren@i2se.com> Tested-by: Stefan Wahren <stefan.wahren@i2se.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Some kernel features don't currently work if a task puts a non-zero
address tag in its stack pointer, frame pointer, or frame record entries
(FP, LR).
For example, with a tagged stack pointer, the kernel can't deliver
signals to the process, and the task is killed instead. As another
example, with a tagged frame pointer or frame records, perf fails to
generate call graphs or resolve symbols.
For now, just document these limitations, instead of finding and fixing
everything that doesn't work, as it's not known if anyone needs to use
tags in these places anyway.
In addition, as requested by Dave Martin, generalize the limitations
into a general kernel address tag policy, and refactor
tagged-pointers.txt to include it.
Fixes: d50240a5f6ce ("arm64: mm: permit use of tagged pointers at EL0") Reviewed-by: Dave Martin <Dave.Martin@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Kristina Martsenko <kristina.martsenko@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When handling a data abort from EL0, we currently zero the top byte of
the faulting address, as we assume the address is a TTBR0 address, which
may contain a non-zero address tag. However, the address may be a TTBR1
address, in which case we should not zero the top byte. This patch fixes
that. The effect is that the full TTBR1 address is passed to the task's
signal handler (or printed out in the kernel log).
When handling a data abort from EL1, we leave the faulting address
intact, as we assume it's either a TTBR1 address or a TTBR0 address with
tag 0x00. This is true as far as I'm aware, we don't seem to access a
tagged TTBR0 address anywhere in the kernel. Regardless, it's easy to
forget about address tags, and code added in the future may not always
remember to remove tags from addresses before accessing them. So add tag
handling to the EL1 data abort handler as well. This also makes it
consistent with the EL0 data abort handler.
Fixes: d50240a5f6ce ("arm64: mm: permit use of tagged pointers at EL0") Reviewed-by: Dave Martin <Dave.Martin@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Kristina Martsenko <kristina.martsenko@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When we take a watchpoint exception, the address that triggered the
watchpoint is found in FAR_EL1. We compare it to the address of each
configured watchpoint to see which one was hit.
The configured watchpoint addresses are untagged, while the address in
FAR_EL1 will have an address tag if the data access was done using a
tagged address. The tag needs to be removed to compare the address to
the watchpoints.
Currently we don't remove it, and as a result can report the wrong
watchpoint as being hit (specifically, always either the highest TTBR0
watchpoint or lowest TTBR1 watchpoint). This patch removes the tag.
Fixes: d50240a5f6ce ("arm64: mm: permit use of tagged pointers at EL0") Acked-by: Mark Rutland <mark.rutland@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Kristina Martsenko <kristina.martsenko@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When we emulate userspace cache maintenance in the kernel, we can
currently send the task a SIGSEGV even though the maintenance was done
on a valid address. This happens if the address has a non-zero address
tag, and happens to not be mapped in.
When we get the address from a user register, we don't currently remove
the address tag before performing cache maintenance on it. If the
maintenance faults, we end up in either __do_page_fault, where find_vma
can't find the VMA if the address has a tag, or in do_translation_fault,
where the tagged address will appear to be above TASK_SIZE. In both
cases, the address is not mapped in, and the task is sent a SIGSEGV.
This patch removes the tag from the address before using it. With this
patch, the fault is handled correctly, the address gets mapped in, and
the cache maintenance succeeds.
As a second bug, if cache maintenance (correctly) fails on an invalid
tagged address, the address gets passed into arm64_notify_segfault,
where find_vma fails to find the VMA due to the tag, and the wrong
si_code may be sent as part of the siginfo_t of the segfault. With this
patch, the correct si_code is sent.
Our access_ok() simply hands its arguments over to __range_ok(), which
implicitly assummes that the addr parameter is 64 bits wide. This isn't
necessarily true for compat code, which might pass down a 32-bit address
parameter.
In these cases, we don't have a guarantee that the address has been zero
extended to 64 bits, and the upper bits of the register may contain
unknown values, potentially resulting in a suprious failure.
Avoid this by explicitly casting the addr parameter to an unsigned long
(as is done on other architectures), ensuring that the parameter is
widened appropriately.
Fixes: 0aea86a2176c ("arm64: User access library functions") Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Mark Rutland <mark.rutland@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Our compat swp emulation holds the compat user address in an unsigned
int, which it passes to __user_swpX_asm(). When a 32-bit value is passed
in a register, the upper 32 bits of the register are unknown, and we
must extend the value to 64 bits before we can use it as a base address.
This patch casts the address to unsigned long to ensure it has been
suitably extended, avoiding the potential issue, and silencing a related
warning from clang.
Fixes: bd35a4adc413 ("arm64: Port SWP/SWPB emulation support from arm") Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Mark Rutland <mark.rutland@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When an inline assembly operand's type is narrower than the register it
is allocated to, the least significant bits of the register (up to the
operand type's width) are valid, and any other bits are permitted to
contain any arbitrary value. This aligns with the AAPCS64 parameter
passing rules.
Our __smp_store_release() implementation does not account for this, and
implicitly assumes that operands have been zero-extended to the width of
the type being stored to. Thus, we may store unknown values to memory
when the value type is narrower than the pointer type (e.g. when storing
a char to a long).
This patch fixes the issue by casting the value operand to the same
width as the pointer operand in all cases, which ensures that the value
is zero-extended as we expect. We use the same union trickery as
__smp_load_acquire and {READ,WRITE}_ONCE() to avoid GCC complaining that
pointers are potentially cast to narrower width integers in unreachable
paths.
A whitespace issue at the top of __smp_store_release() is also
corrected.
No changes are necessary for __smp_load_acquire(). Load instructions
implicitly clear any upper bits of the register, and the compiler will
only consider the least significant bits of the register as valid
regardless.
Fixes: 47933ad41a86 ("arch: Introduce smp_load_acquire(), smp_store_release()") Fixes: 878a84d5a8a1 ("arm64: add missing data types in smp_load_acquire/smp_store_release") Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Mark Rutland <mark.rutland@arm.com> Cc: Matthias Kaehlcke <mka@chromium.org> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The inline assembly in __XCHG_CASE() uses a +Q constraint to hazard
against other accesses to the memory location being exchanged. However,
the pointer passed to the constraint is a u8 pointer, and thus the
hazard only applies to the first byte of the location.
GCC can take advantage of this, assuming that other portions of the
location are unchanged, as demonstrated with the following test case:
union u {
unsigned long l;
unsigned int i[2];
};
unsigned long update_char_hazard(union u *u)
{
unsigned int a, b;
a = u->i[1];
asm ("str %1, %0" : "+Q" (*(char *)&u->l) : "r" (0UL));
b = u->i[1];
return a ^ b;
}
unsigned long update_long_hazard(union u *u)
{
unsigned int a, b;
a = u->i[1];
asm ("str %1, %0" : "+Q" (*(long *)&u->l) : "r" (0UL));
b = u->i[1];
return a ^ b;
}
The linaro 15.08 GCC 5.1.1 toolchain compiles the above as follows when
using -O2 or above:
This patch fixes the issue by passing an unsigned long pointer into the
+Q constraint, as we do for our cmpxchg code. This may hazard against
more than is necessary, but this is better than missing a necessary
hazard.
The MMC hosts could be left in an unconsistent or uninitialized state from
the firmware. Instead of assuming, the firmware did the right things, let's
reset the host controllers.
This change fixes a bug when the mmc2/sdio is initialized leading to a hung
task:
The board file for imx6sx-sdb overrides cpufreq operating points to use
higher voltages. This is done because the board has a shared rail for
VDD_ARM_IN and VDD_SOC_IN and when using LDO bypass the shared voltage
needs to be a value suitable for both ARM and SOC.
This only applies to LDO bypass mode, a feature not present in upstream.
When LDOs are enabled the effect is to use higher voltages than necessary
for no good reason.
Setting these higher voltages can make some boards fail to boot with ugly
semi-random crashes reminiscent of memory corruption. These failures only
happen on board rev. C, rev. B is reported to still work.
Signed-off-by: Leonard Crestez <leonard.crestez@nxp.com> Fixes: 54183bd7f766 ("ARM: imx6sx-sdb: add revb board and make it default") Signed-off-by: Shawn Guo <shawnguo@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We save/restore registers around v7m_invalidate_l1 to address pointed
by r12, which is vector table, so the first eight entries are
overwritten with a garbage. We already have stack setup at that stage,
so use it to save/restore register.
Fixes: 6a8146f420be ("ARM: 8609/1: V7M: Add support for the Cortex-M7 processor") Signed-off-by: Vladimir Murzin <vladimir.murzin@arm.com> Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To cope with the variety in ARM architectures and configurations, the
pagetable attributes for kernel memory are generated at runtime to match
the system the kernel finds itself on. This calculated value is stored
in pgprot_kernel.
However, when early fixmap support was added for ARM (commit a5f4c561b3b1) the attributes used for mappings were hard coded because
pgprot_kernel is not set up early enough. Unfortunately, when fixmap is
used after early boot this means the memory being mapped can have
different attributes to existing mappings, potentially leading to
unpredictable behaviour. A specific problem also exists due to the hard
coded values not include the 'shareable' attribute which means on
systems where this matters (e.g. those with multiple CPU clusters) the
cache contents for a memory location can become inconsistent between
CPUs.
To resolve these issues we change fixmap to use the same memory
attributes (from pgprot_kernel) that the rest of the kernel uses. To
enable this we need to refactor the initialisation code so
build_mem_type_table() is called early enough. Note, that relies on early
param parsing for memory type overrides passed via the kernel command
line, so we need to make sure this call is still after
parse_early_params().
[ardb: keep early_fixmap_init() before param parsing, for earlycon]
Fixes: a5f4c561b3b1 ("ARM: 8415/1: early fixmap support for earlycon") Tested-by: afzal mohammed <afzal.mohd.ma@gmail.com> Signed-off-by: Jon Medhurst <tixy@linaro.org> Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Since commit 35fa91eed817 ("ARM: kernel: merge core and init PLTs"),
the ARM module PLT code allocates all PLT entries in a single core
section, since the overhead of having a separate init PLT section is
not justified by the small number of PLT entries usually required for
init code.
However, the core and init module regions are allocated independently,
and there is a corner case where the core region may be allocated from
the VMALLOC region if the dedicated module region is exhausted, but the
init region, being much smaller, can still be allocated from the module
region. This puts the PLT entries out of reach of the relocated branch
instructions, defeating the whole purpose of PLTs.
So split the core and init PLT regions, and name the latter ".init.plt"
so it gets allocated along with (and sufficiently close to) the .init
sections that it serves. Also, given that init PLT entries may need to
be emitted for branches that target the core module, modify the logic
that disregards defined symbols to only disregard symbols that are
defined in the same section.
Fixes: 35fa91eed817 ("ARM: kernel: merge core and init PLTs") Reported-by: Angus Clark <angus@angusclark.org> Tested-by: Angus Clark <angus@angusclark.org> Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Hardware debugging in guests is not intercepted currently, it means
that a malicious guest can bring down the entire machine by writing
to the debug registers.
This patch enable trapping of all debug registers, preventing the
guests to access the debug registers. This includes access to the
debug mode(DBGDSCR) in the guest world all the time which could
otherwise mess with the host state. Reads return 0 and writes are
ignored (RAZ_WI).
The result is the guest cannot detect any working hardware based debug
support. As debug exceptions are still routed to the guest normal
debug using software based breakpoints still works.
To support debugging using hardware registers we need to implement a
debug register aware world switch as well as special trapping for
registers that may affect the host state.
Signed-off-by: Zhichao Huang <zhichao.huang@linaro.org> Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Christoffer Dall <cdall@linaro.org> Signed-off-by: Christoffer Dall <cdall@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When an interrupt is injected with the HW bit set (indicating that
deactivation should be propagated to the physical distributor),
special care must be taken so that we never mark the corresponding
LR with the Active+Pending state (as the pending state is kept in
the physycal distributor).
Fixes: 59529f69f504 ("KVM: arm/arm64: vgic-new: Add GICv3 world switch backend") Signed-off-by: Marc Zyngier <marc.zyngier@arm.com> Reviewed-by: Christoffer Dall <cdall@linaro.org> Signed-off-by: Christoffer Dall <cdall@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When an interrupt is injected with the HW bit set (indicating that
deactivation should be propagated to the physical distributor),
special care must be taken so that we never mark the corresponding
LR with the Active+Pending state (as the pending state is kept in
the physycal distributor).
Fixes: 140b086dd197 ("KVM: arm/arm64: vgic-new: Add GICv2 world switch backend") Signed-off-by: Marc Zyngier <marc.zyngier@arm.com> Reviewed-by: Christoffer Dall <cdall@linaro.org> Signed-off-by: Christoffer Dall <cdall@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We like living dangerously. Nothing explicitely forbids stack-protector
to be used in the HYP code, while distributions routinely compile their
kernel with it. We're just lucky that no code actually triggers the
instrumentation.
Let's not try our luck for much longer, and disable stack-protector
for code living at HYP.
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com> Acked-by: Christoffer Dall <cdall@linaro.org> Signed-off-by: Christoffer Dall <cdall@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We like living dangerously. Nothing explicitely forbids stack-protector
to be used in the EL2 code, while distributions routinely compile their
kernel with it. We're just lucky that no code actually triggers the
instrumentation.
Let's not try our luck for much longer, and disable stack-protector
for code living at EL2.
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com> Acked-by: Christoffer Dall <cdall@linaro.org> Signed-off-by: Christoffer Dall <cdall@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In commit dc3106690b20 ("powerpc: tm: Always use fp_state and vr_state
to store live registers"), a section of code was removed that copied
the current state to checkpointed state. That code should not have been
removed.
When an FP (Floating Point) unavailable is taken inside a transaction,
we need to abort the transaction. This is because at the time of the
tbegin, the FP state is bogus so the state stored in the checkpointed
registers is incorrect. To fix this, we treclaim (to get the
checkpointed GPRs) and then copy the thread_struct FP live state into
the checkpointed state. We then trecheckpoint so that the FP state is
correctly restored into the CPU.
The copying of the FP registers from live to checkpointed is what was
missing.
This simplifies the logic slightly from the original patch.
tm_reclaim_thread() will now always write the checkpointed FP
state. Either the checkpointed FP state will be written as part of
the actual treclaim (in tm.S), or it'll be a copy of the live
state. Which one we use is based on MSR[FP] from userspace.
Similarly for VMX.
Fixes: dc3106690b20 ("powerpc: tm: Always use fp_state and vr_state to store live registers") Signed-off-by: Michael Neuling <mikey@neuling.org> Reviewed-by: cyrilbur@gmail.com Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The page table dump code doesn't know about huge pages, so currently
it crashes (or walks random memory, usually leading to a crash), if it
finds a huge page. On Book3S we only see huge pages in the Linux page
tables when we're using the P9 Radix MMU.
Teaching the code to properly handle huge pages is a bit more involved,
so for now just prevent the crash.
Fixes: 8eb07b187000 ("powerpc/mm: Dump linux pagetables") Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Debug interrupts can be taken during interrupt entry, since interrupt
entry does not automatically turn them off. The kernel will check
whether the faulting instruction is between [interrupt_base_book3e,
__end_interrupts], and if so clear MSR[DE] and return.
However, when the kernel is built with CONFIG_RELOCATABLE, it can't use
LOAD_REG_IMMEDIATE(r14,interrupt_base_book3e) and
LOAD_REG_IMMEDIATE(r15,__end_interrupts), as they ignore relocation.
Thus, if the kernel is actually running at a different address than it
was built at, the address comparison will fail, and the exception entry
code will hang at kernel_dbg_exc.
r2(toc) is also not usable here, as r2 still holds data from the
interrupted context, so LOAD_REG_ADDR() doesn't work either. So we use
the *name@got* to get the EV of two labels directly.
Test programs test.c shows as follows:
int main(int argc, char *argv[])
{
if (access("/proc/sys/kernel/perf_event_paranoid", F_OK) == -1)
printf("Kernel doesn't have perf_event support\n");
}
Steps to reproduce the bug, for example:
1) ./gdb ./test
2) (gdb) b access
3) (gdb) r
4) (gdb) s
Signed-off-by: Liu Hailong <liu.hailong6@zte.com.cn> Signed-off-by: Jiang Xuexin <jiang.xuexin@zte.com.cn> Reviewed-by: Jiang Biao <jiang.biao2@zte.com.cn> Reviewed-by: Liu Song <liu.song11@zte.com.cn> Reviewed-by: Huang Jian <huang.jian@zte.com.cn>
[scottwood: cleaned up commit message, and specified bad behavior
as a hang rather than an oops to correspond to mainline kernel behavior] Fixes: 1cb6e0649248 ("powerpc/book3e: support CONFIG_RELOCATABLE") Signed-off-by: Scott Wood <oss@buserror.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 616badd2fb49 ("powerpc/powernv: Use OPAL call for TCE kill on
NVLink2") forced all TCE kills to go via the OPAL call for
NVLink2. However the PHB3 implementation of TCE kill was still being
called directly from some functions which in some circumstances caused
a machine check.
This patch adds an equivalent IODA2 version of the function which uses
the correct invalidation method depending on PHB model and changes all
external callers to use it instead.
Fixes: 616badd2fb49 ("powerpc/powernv: Use OPAL call for TCE kill on NVLink2") Signed-off-by: Alistair Popple <alistair@popple.id.au> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The CMA pages migration code does not support compound pages at
the moment so it performs few tests before proceeding to actual page
migration.
One of the tests - PageTransHuge() - has VM_BUG_ON_PAGE(PageTail()) as
it is designed to be called on head pages only. Since we also test for
PageCompound(), and it contains PageTail() and PageHead(), we can
simplify the check by leaving just PageCompound() and therefore avoid
possible VM_BUG_ON_PAGE.
Fixes: 2e5bbb5461f1 ("KVM: PPC: Book3S HV: Migrate pinned pages out of CMA") Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru> Acked-by: Balbir Singh <bsingharora@gmail.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
For CPUs present at boot each logical CPU acquires a reference to the
associated device node of the core. This happens in register_cpu() which
is called by topology_init(). The result of this is that we end up with
a reference held by each thread of the core. However, these references
are never freed if the CPU core is DLPAR removed.
This patch fixes the reference leaks by acquiring and releasing the references
in the CPU hotplug callbacks un/register_cpu_online(). With this patch symmetric
reference counting is observed with both CPUs present at boot, and those DLPAR
added after boot.
Historically struct device_node references were tracked using a kref embedded as
a struct field. Commit 75b57ecf9d1d ("of: Make device nodes kobjects so they
show up in sysfs") (Mar 2014) refactored device_nodes to be kobjects such that
the device tree could by more simply exposed to userspace using sysfs.
Commit 0829f6d1f69e ("of: device_node kobject lifecycle fixes") (Mar 2014)
followed up these changes to better control the kobject lifecycle and in
particular the referecne counting via of_node_get(), of_node_put(), and
of_node_init().
A result of this second commit was that it introduced an of_node_put() call when
a dynamic node is detached, in of_node_remove(), that removes the initial kobj
reference created by of_node_init().
Traditionally as the original dynamic device node user the pseries code had
assumed responsibilty for releasing this final reference in its platform
specific DLPAR detach code.
This patch fixes a refcount underflow introduced by commit 0829f6d1f6, and
recently exposed by the upstreaming of the recount API.
Messages like the following are no longer seen in the kernel log with this
patch following DLPAR remove operations of cpus and pci devices.
rpadlpar_io: slot PHB 72 removed
refcount_t: underflow; use-after-free.
------------[ cut here ]------------
WARNING: CPU: 5 PID: 3335 at lib/refcount.c:128 refcount_sub_and_test+0xf4/0x110
Fixes: 0829f6d1f69e ("of: device_node kobject lifecycle fixes") Signed-off-by: Tyrel Datwyler <tyreld@linux.vnet.ibm.com>
[mpe: Make change log commit references more verbose] Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
machine_check_early() gets called in real mode. The very first time when
add_taint() is called, it prints a warning which ends up calling opal
call (that uses OPAL_CALL wrapper) for writing it to console. If we get a
very first machine check while we are in opal we are doomed. OPAL_CALL
overwrites the PACASAVEDMSR in r13 and in this case when we are done with
MCE handling the original opal call will use this new MSR on it's way
back to opal_return. This usually leads to unexpected behaviour or the
kernel to panic. Instead move the add_taint() call later in the virtual
mode where it is safe to call.
This is broken with current FW level. We got lucky so far for not getting
very first MCE hit while in OPAL. But easily reproducible on Mambo.
Fixes: 27ea2c420cad ("powerpc: Set the correct kernel taint on machine check errors.") Signed-off-by: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
eeh_handle_special_event() is called when an EEH event is detected but
can't be narrowed down to a specific PE. This function looks through
every PE to find one in an erroneous state, then calls the regular event
handler eeh_handle_normal_event() once it knows which PE has an error.
However, if eeh_handle_normal_event() found that the PE cannot possibly
be recovered, it will free it, rendering the passed PE stale.
This leads to a use after free in eeh_handle_special_event() as it attempts to
clear the "recovering" state on the PE after eeh_handle_normal_event() returns.
Thus, make sure the PE is valid when attempting to clear state in
eeh_handle_special_event().
Fixes: 8a6b1bc70dbb ("powerpc/eeh: EEH core to handle special event") Reported-by: Alexey Kardashevskiy <aik@ozlabs.ru> Signed-off-by: Russell Currey <ruscur@russell.cc> Reviewed-by: Gavin Shan <gwshan@linux.vnet.ibm.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
powerpc expects IRQs to already be (soft) disabled when switch_mm() is
called, as made clear in the commit message of 9c1e105238c4 ("powerpc: Allow
perf_counters to access user memory at interrupt time").
Aside from any race conditions that might exist between switch_mm() and an IRQ,
there is also an unconditional hard_irq_disable() in switch_slb(). If that isn't
followed at some point by an IRQ enable then interrupts will remain disabled
until we return to userspace.
It is true that when switch_mm() is called from the scheduler IRQs are off, but
not when it's called by use_mm(). Looking closer we see that last year in commit f98db6013c55 ("sched/core: Add switch_mm_irqs_off() and use it in the scheduler")
this was made more explicit by the addition of switch_mm_irqs_off() which is now
called by the scheduler, vs switch_mm() which is used by use_mm().
Arguably it is a bug in use_mm() to call switch_mm() in a different context than
it expects, but fixing that will take time.
This was discovered recently when vhost started throwing warnings such as:
BUG: sleeping function called from invalid context at kernel/mutex.c:578
in_atomic(): 0, irqs_disabled(): 1, pid: 10768, name: vhost-10760
no locks held by vhost-10760/10768.
irq event stamp: 10
hardirqs last enabled at (9): _raw_spin_unlock_irq+0x40/0x80
hardirqs last disabled at (10): switch_slb+0x2e4/0x490
softirqs last enabled at (0): copy_process+0x5e8/0x1260
softirqs last disabled at (0): (null)
Call Trace:
show_stack+0x88/0x390 (unreliable)
dump_stack+0x30/0x44
__might_sleep+0x1c4/0x2d0
mutex_lock_nested+0x74/0x5c0
cgroup_attach_task_all+0x5c/0x180
vhost_attach_cgroups_work+0x58/0x80 [vhost]
vhost_worker+0x24c/0x3d0 [vhost]
kthread+0xec/0x100
ret_from_kernel_thread+0x5c/0xd4
Prior to commit 04b96e5528ca ("vhost: lockless enqueuing") (Aug 2016) the
vhost_worker() would do a spin_unlock_irq() not long after calling use_mm(),
which had the effect of reenabling IRQs. Since that commit removed the locking
in vhost_worker() the body of the vhost_worker() loop now runs with interrupts
off causing the warnings.
This patch addresses the problem by making the powerpc code mirror the x86 code,
ie. we disable interrupts in switch_mm(), and optimise the scheduler case by
defining switch_mm_irqs_off().
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
[mpe: Flesh out/rewrite change log, add stable] Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Make sure to check the number of endpoints to avoid dereferencing a
NULL-pointer or accessing memory beyond the endpoint array should a
malicious device lack the expected endpoints.
Fixes: e0d3bafd0258 ("V4L/DVB (10954): Add cx231xx USB driver") Cc: Sri Deevi <Srinivasa.Deevi@conexant.com> Signed-off-by: Johan Hovold <johan@kernel.org> Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Make sure to check the number of endpoints to avoid dereferencing a
NULL-pointer or accessing memory beyond the endpoint array should a
malicious device lack the expected endpoints.
Fixes: e0d3bafd0258 ("V4L/DVB (10954): Add cx231xx USB driver") Cc: Sri Deevi <Srinivasa.Deevi@conexant.com> Signed-off-by: Johan Hovold <johan@kernel.org> Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Clock should be turned off after calling s5p_mfc_init_hw() from the
watchdog worker, like it is already done in the s5p_mfc_open() which also
calls this function.
Make sure to check the number of endpoints to avoid accessing memory
beyond the endpoint array should a device lack the expected endpoints.
Note that, as far as I can tell, the gspca framework has already made
sure there is at least one endpoint in the current alternate setting so
there should be no risk for a NULL-pointer dereference here.
Fixes: b517af722860 ("V4L/DVB: gspca_konica: New gspca subdriver for
konica chipset using cams")
Cc: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Johan Hovold <johan@kernel.org> Signed-off-by: Hans Verkuil <hansverk@cisco.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Interrupt routine must wake process waiting for given interrupt AFTER
updating driver's internal structures and contexts. Doing it in-between
is a serious bug. This patch moves all calls to the wake() function to
the end of the interrupt processing block to avoid potential and real
races, especially on multi-core platforms. This also fixes following issue
reported from clock core (clocks were disabled in interrupt after being
unprepared from the other place in the driver, the stack trace however
points to the different place than s5p_mfc driver because of the race):
WARNING: CPU: 1 PID: 18 at drivers/clk/clk.c:544 clk_core_unprepare+0xc8/0x108
Modules linked in:
CPU: 1 PID: 18 Comm: kworker/1:0 Not tainted 4.10.0-next-20170223-00070-g04e18bc99ab9-dirty #2154
Hardware name: SAMSUNG EXYNOS (Flattened Device Tree)
Workqueue: pm pm_runtime_work
[<c010d8b0>] (unwind_backtrace) from [<c010a534>] (show_stack+0x10/0x14)
[<c010a534>] (show_stack) from [<c033292c>] (dump_stack+0x74/0x94)
[<c033292c>] (dump_stack) from [<c011cef4>] (__warn+0xd4/0x100)
[<c011cef4>] (__warn) from [<c011cf40>] (warn_slowpath_null+0x20/0x28)
[<c011cf40>] (warn_slowpath_null) from [<c0387a84>] (clk_core_unprepare+0xc8/0x108)
[<c0387a84>] (clk_core_unprepare) from [<c0389d84>] (clk_unprepare+0x24/0x2c)
[<c0389d84>] (clk_unprepare) from [<c03d4660>] (exynos_sysmmu_suspend+0x48/0x60)
[<c03d4660>] (exynos_sysmmu_suspend) from [<c042b9b0>] (pm_generic_runtime_suspend+0x2c/0x38)
[<c042b9b0>] (pm_generic_runtime_suspend) from [<c0437580>] (genpd_runtime_suspend+0x94/0x220)
[<c0437580>] (genpd_runtime_suspend) from [<c042e240>] (__rpm_callback+0x134/0x208)
[<c042e240>] (__rpm_callback) from [<c042e334>] (rpm_callback+0x20/0x80)
[<c042e334>] (rpm_callback) from [<c042d3b8>] (rpm_suspend+0xdc/0x458)
[<c042d3b8>] (rpm_suspend) from [<c042ea24>] (pm_runtime_work+0x80/0x90)
[<c042ea24>] (pm_runtime_work) from [<c01322c4>] (process_one_work+0x120/0x318)
[<c01322c4>] (process_one_work) from [<c0132520>] (worker_thread+0x2c/0x4ac)
[<c0132520>] (worker_thread) from [<c0137ab0>] (kthread+0xfc/0x134)
[<c0137ab0>] (kthread) from [<c0107978>] (ret_from_fork+0x14/0x3c)
---[ end trace 1ead49a7bb83f0d8 ]---
Currently when the RC Core is enabled (reachable) core code located
in cec_register_adapter() attempts to populate the RC structure with
a pointer to the 'parent' passed in by the caller.
Unfortunately if the caller did not specify RC capability when calling
cec_allocate_adapter(), then there will be no RC structure to populate.
This causes a "NULL pointer dereference" error.
Fixes: f51e80804f0 ("[media] cec: pass parent device in register(), not allocate()") Signed-off-by: Lee Jones <lee.jones@linaro.org> Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This change undo the change done by 'commit 3bec24747446
("iio: hid-sensor-trigger: Change get poll value function order to avoid
sensor properties losing after resume from S3")' as this breaks some
USB/i2c sensor hubs.
Instead of relying on HW for restoring poll and hysteresis, driver stores
and restores on resume (S3). In this way user space modified settings are
not lost for any kind of sensor hub behavior.
In this change, whenever user space modifies sampling frequency or
hysteresis driver will get the feature value from the hub and store in the
per device hid_sensor_common data structure. On resume callback from S3,
system will set the feature to sensor hub, if user space ever modified the
feature value.
Fixes: 3bec24747446 ("iio: hid-sensor-trigger: Change get poll value function order to avoid sensor properties losing after resume from S3") Reported-by: Ritesh Raj Sarraf <rrs@researchut.com> Tested-by: Ritesh Raj Sarraf <rrs@researchut.com> Tested-by: Song, Hongyan <hongyan.song@intel.com> Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Jonathan Cameron <jic23@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We should call ipxitf_put() if the copy_to_user() fails.
Reported-by: 李强 <liqiang6-s@360.cn> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Add missing sanity check on the non-SuperSpeed hub-descriptor length in
order to avoid parsing and leaking two bytes of uninitialised slab data
through sysfs removable-attributes (or a compound-device debug
statement).
Note that we only make sure that the DeviceRemovable field is always
present (and specifically ignore the unused PortPwrCtrlMask field) in
order to continue support any hubs with non-compliant descriptors. As a
further safeguard, the descriptor buffer is also cleared.
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: Johan Hovold <johan@kernel.org> Acked-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Drop erroneous cpu_to_le32 when setting the baud rate, something which
corrupted the divisor on big-endian hosts.
Found using sparse:
warning: incorrect type in argument 1 (different base types)
expected unsigned int [unsigned] [usertype] val
got restricted __le32 [usertype] <noident>
Fixes: af2ac1a091bc ("USB: serial mct_usb232: move DMA buffers to heap") Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Acked-By: Pete Zaitcev <zaitcev@yahoo.com> Signed-off-by: Johan Hovold <johan@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In their infinite wisdom, and never ending quest for end user frustration,
Lenovo has decided to use new USB device IDs for the wwan modules in
their 2017 laptops. The actual hardware is still the Sierra Wireless
EM7455 or EM7430, depending on region.
Commit d8e5f0eca1e8 ("usb: musb: Fix hardirq-safe hardirq-unsafe
lock order error") caused a regression where musb keeps trying to
enable host mode with no cable connected. This seems to be caused
by the fact that now phy is enabled earlier, and we are wrongly
trying to force USB host mode on an OTG port. The errors we are
getting are "trying to suspend as a_idle while active".
For ports configured as OTG, we should not need to do anything
to try to force USB host mode on it's OTG port. Trying to force host
mode in this case just seems to completely confuse the musb state
machine.
Let's fix the issue by making musb_host_setup() attempt to force the
mode only if port_mode is configured for host mode.
Fixes: d8e5f0eca1e8 ("usb: musb: Fix hardirq-safe hardirq-unsafe lock order error") Cc: Johan Hovold <johan@kernel.org> Reported-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Reported-by: Peter Ujfalusi <peter.ujfalusi@ti.com> Tested-by: Peter Ujfalusi <peter.ujfalusi@ti.com> Signed-off-by: Tony Lindgren <tony@atomide.com> Signed-off-by: Bin Liu <b-liu@ti.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We have one register for each EP to set the maximum packet size for both
TX and RX.
If for example an RX programming would happen before the previous TX
transfer finishes we would reset the TX packet side.
To fix this issue, only modify the TX or RX part of the register.
Fixes: 550a7375fe72 ("USB: Add MUSB and TUSB support") Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com> Tested-by: Tony Lindgren <tony@atomide.com> Signed-off-by: Bin Liu <b-liu@ti.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The dwc3 driver can overwite its previous events if its top-half IRQ
handler (TH) gets invoked again before processing the events in the
cache. We see this as a hang in the file transfer and the host will
attempt to reset the device. TH gets the event count and deasserts the
interrupt line by writing DWC3_GEVNTSIZ_INTMASK to DWC3_GEVNTSIZ. If
there's a new event coming between reading the event count and interrupt
deassertion, dwc3 will lose previous pending events. More generally, we
will see 0 event count, which should not affect anything.
This shouldn't be possible in the current dwc3 implementation. However,
through testing and reading the PCIe trace, the TH occasionally still
gets invoked one more time after HW interrupt deassertion. (With PCIe
legacy interrupts, TH is called repeatedly as long as the interrupt line
is asserted). We suspect that there is a small detection delay in the
SW.
To avoid this issue, Check DWC3_EVENT_PENDING flag to determine if the
events are processed in the bottom-half IRQ handler. If not, return
IRQ_HANDLED and don't process new event.
Make sure to check for the required out endpoint to avoid dereferencing
a NULL-pointer in mce_request_packet should a malicious device lack such
an endpoint. Note that this path is hit during probe.
Fixes: 66e89522aff7 ("V4L/DVB: IR: add mceusb IR receiver driver") Signed-off-by: Johan Hovold <johan@kernel.org> Signed-off-by: Sean Young <sean@mess.org> Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Make sure to check the number of endpoints to avoid dereferencing a
NULL-pointer or accessing memory beyond the endpoint array should a
malicious device lack the expected endpoints.
Fixes: 2a9f8b5d25be ("V4L/DVB (5206): Usbvision: set alternate interface
modification")
Add missing endianness conversion when using the USB device-descriptor
bcdDevice field to construct a firmware file name.
Fixes: 8ef80aef118e ("[IRDA]: irda-usb.c: STIR421x cleanups") Cc: Nick Fedchik <nfedchik@atlantic-link.com.ua> Signed-off-by: Johan Hovold <johan@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
With threaded interrupts, bottom-half handlers are called with
interrupts enabled. Therefore they can't safely use spin_lock(); they
have to use spin_lock_irqsave(). Lockdep warns about a violation
occurring in xhci_irq():
=========================================================
[ INFO: possible irq lock inversion dependency detected ]
4.11.0-rc8-dbg+ #1 Not tainted
---------------------------------------------------------
swapper/7/0 just changed the state of lock:
(&(&ehci->lock)->rlock){-.-...}, at: [<ffffffffa0130a69>]
ehci_hrtimer_func+0x29/0xc0 [ehci_hcd]
but this lock took another, HARDIRQ-unsafe lock in the past:
(hcd_urb_list_lock){+.....}
and interrupts could create inverse lock ordering between them.
other info that might help us debug this:
Possible interrupt unsafe locking scenario:
platform_get_irq() returns an error code, but the xhci-plat driver
ignores it and always returns -ENODEV. This is not correct, and
prevents -EPROBE_DEFER from being propagated properly.
The wrong index into the csbases/csmasks arrays was being passed to
the function to compute the chip select sizes, which resulted in the
wrong size being computed. Address that so that the correct values are
computed and printed.
Also, redo how we calculate the number of pages in a CS row.
Reported-by: Benjamin Bennett <benbennett@gmail.com> Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com> Cc: linux-edac <linux-edac@vger.kernel.org> Link: http://lkml.kernel.org/r/1493313114-11260-1-git-send-email-Yazen.Ghannam@amd.com
[ Remove unneeded integer math comment, minor cleanups. ] Signed-off-by: Borislav Petkov <bp@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently DAX read fault can race with write(2) in the following way:
CPU1 - write(2) CPU2 - read fault
dax_iomap_pte_fault()
->iomap_begin() - sees hole
dax_iomap_rw()
iomap_apply()
->iomap_begin - allocates blocks
dax_iomap_actor()
invalidate_inode_pages2_range()
- there's nothing to invalidate
grab_mapping_entry()
- we add zero page in the radix tree
and map it to page tables
The result is that hole page is mapped into page tables (and thus zeros
are seen in mmap) while file has data written in that place.
Fix the problem by locking exception entry before mapping blocks for the
fault. That way we are sure invalidate_inode_pages2_range() call for
racing write will either block on entry lock waiting for the fault to
finish (and unmap stale page tables after that) or read fault will see
already allocated blocks by write(2).
Fixes: 9f141d6ef6258a3a37a045842d9ba7e68f368956 Link: http://lkml.kernel.org/r/20170510085419.27601-5-jack@suse.cz Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Ross Zwisler <ross.zwisler@linux.intel.com> Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com> Cc: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
stat/lstat/fstatat need to pass AT_NO_AUTOMOUNT to vfs_statx() as the
pre-statx code didn't set LOOKUP_AUTOMOUNT, even though fstatat() accepted
the AT_NO_AUTOMOUNT flag.
Fixes: a528d35e8bfc ("statx: Add a system call to make enhanced file info available") Reported-by: Ian Kent <raven@themaw.net> Signed-off-by: David Howells <dhowells@redhat.com> Tested-by: Ian Kent <raven@themaw.net> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Add missing endianness conversion when applying the Alea timeout quirk.
Found using sparse:
warning: restricted __le16 degrades to integer
Fixes: e4a886e811cd ("hwrng: chaoskey - Fix URB warning due to timeout on Alea") Cc: Bob Ham <bob.ham@collabora.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: Keith Packard <keithp@keithp.com> Signed-off-by: Johan Hovold <johan@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch adds support for recognition of ARM-USB-TINY(H) devices which
are almost identical to ARM-USB-OCD(H) but lacking separate barrel jack
and serial console.
By suggestion from Johan Hovold it is possible to replace
ftdi_jtag_quirk with a bit more generic construction. Since all
Olimex-ARM debuggers has exactly two ports, we could safely always use
only second port within the debugger family.
Commit 557aaa7ffab6 ("ft232: support the ASYNC_LOW_LATENCY
flag") enables unprivileged users to set the FTDI latency timer,
but there was a logic flaw that skipped sending the corresponding
USB control message to the device.
Specifically, the device latency timer would not be updated until next
open, something which was later also inadvertently broken by commit c19db4c9e49a ("USB: ftdi_sio: set device latency timeout at port
probe").
A recent commit c6dce2626606 ("USB: serial: ftdi_sio: fix extreme
low-latency setting") disabled the low-latency mode by default so we now
need this fix to allow unprivileged users to again enable it.
Signed-off-by: Anthony Mallet <anthony.mallet@laas.fr>
[johan: amend commit message] Fixes: 557aaa7ffab6 ("ft232: support the ASYNC_LOW_LATENCY flag") Fixes: c19db4c9e49a ("USB: ftdi_sio: set device latency timeout at port probe"). Signed-off-by: Johan Hovold <johan@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Imagine we have a pid namespace and a task from its parent's pid_ns,
which made setns() to the pid namespace. The task is doing fork(),
while the pid namespace's child reaper is dying. We have the race
between them:
Task from parent pid_ns Child reaper
copy_process() ..
alloc_pid() ..
.. zap_pid_ns_processes()
.. disable_pid_allocation()
.. read_lock(&tasklist_lock)
.. iterate over pids in pid_ns
.. kill tasks linked to pids
.. read_unlock(&tasklist_lock)
write_lock_irq(&tasklist_lock); ..
attach_pid(p, PIDTYPE_PID); ..
.. ..
So, just created task p won't receive SIGKILL signal,
and the pid namespace will be in contradictory state.
Only manual kill will help there, but does the userspace
care about this? I suppose, the most users just inject
a task into a pid namespace and wait a SIGCHLD from it.
The patch fixes the problem. It simply checks for
(pid_ns->nr_hashed & PIDNS_HASH_ADDING) in copy_process().
We do it under the tasklist_lock, and can't skip
PIDNS_HASH_ADDING as noted by Oleg:
"zap_pid_ns_processes() does disable_pid_allocation()
and then takes tasklist_lock to kill the whole namespace.
Given that copy_process() checks PIDNS_HASH_ADDING
under write_lock(tasklist) they can't race;
if copy_process() takes this lock first, the new child will
be killed, otherwise copy_process() can't miss
the change in ->nr_hashed."
If allocation is disabled, we just return -ENOMEM
like it's made for such cases in alloc_pid().
v2: Do not move disable_pid_allocation(), do not
introduce a new variable in copy_process() and simplify
the patch as suggested by Oleg Nesterov.
Account the problem with double irq enabling
found by Eric W. Biederman.
Fixes: c876ad768215 ("pidns: Stop pid allocation when init dies") Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com> CC: Andrew Morton <akpm@linux-foundation.org> CC: Ingo Molnar <mingo@kernel.org> CC: Peter Zijlstra <peterz@infradead.org> CC: Oleg Nesterov <oleg@redhat.com> CC: Mike Rapoport <rppt@linux.vnet.ibm.com> CC: Michal Hocko <mhocko@suse.com> CC: Andy Lutomirski <luto@kernel.org> CC: "Eric W. Biederman" <ebiederm@xmission.com> CC: Andrei Vagin <avagin@openvz.org> CC: Cyrill Gorcunov <gorcunov@openvz.org> CC: Serge Hallyn <serge@hallyn.com> Acked-by: Oleg Nesterov <oleg@redhat.com> Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The code can potentially sleep for an indefinite amount of time in
zap_pid_ns_processes triggering the hung task timeout, and increasing
the system average. This is undesirable. Sleep with a task state of
TASK_INTERRUPTIBLE instead of TASK_UNINTERRUPTIBLE to remove these
undesirable side effects.
Apparently under heavy load this has been allowing Chrome to trigger
the hung time task timeout error and cause ChromeOS to reboot.
Reported-by: Vovo Yang <vovoy@google.com> Reported-by: Guenter Roeck <linux@roeck-us.net> Tested-by: Guenter Roeck <linux@roeck-us.net> Fixes: 6347e9009104 ("pidns: guarantee that the pidns init will be the last pidns process reaped") Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The only context that frees user_exp_rcv data structures is the last
context closed (from a sub-context set). This leaks the allocations
from the other sub-contexts. Separate the common frees from the
specific frees and call them at the appropriate time.
Using KEDR to check for memory leaks we get:
Before test:
[leak_check] Possible leaks: 25
After test:
[leak_check] Possible leaks: 31 (6 leaked data structures)
After patch applied (before and after test have the same value)
[leak_check] Possible leaks: 25
Each leak is 192 + 13440 + 6720 = 20352 bytes per sub-context.
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When prescaler (PSC) is 0, it means div factor is 1: counter clock
frequency is equal to input clk / (PSC + 1).
When reload value is 8 for example, counter counts 9 cycles, from 0 to 8.
This is handled in frequency write routine, by writing respectively:
- prescaler - 1 to PSC
- reload value - 1 to ARR
This fix does the opposite when reading the frequency from PSC and ARR:
- prescaler is PSC + 1
- reload value is ARR + 1
Thus, PSC may be 0, depending on requested sampling frequency (div 1).
In this case, reading freq wrongly reports 0, instead of computing and
reporting correct value.
Remove test on !psc and !arr.
Small test on stm32f4 (example on tim1_trgo), before this fix:
$ cd /sys/bus/iio/devices/triggerX
$ echo 10000 > sampling_frequency
$ cat sampling_frequency
0
After this fix:
$ echo 10000 > sampling_frequency
$ cat sampling_frequency
10000
While calculating the compensation of the humidity there are negative values
interpreted as unsigned because of unsigned variables used. These values as
well as the constants need to be casted to signed as indicated by the
documentation of the sensor.
Signed-off-by: Andreas Klinger <ak@it-klinger.de> Acked-by: Linus Walleij <linus.walleij@linaro.org> Reviewed-by: Matt Ranostay <matt.ranostay@konsulko.com> Signed-off-by: Jonathan Cameron <jic23@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To select the appropriate shost template, the driver is issuing a
mailbox command to retrieve the wwn. Turns out the sending of the
command precedes the reset of the function. On SLI-4 adapters, this is
inconsequential as the mailbox command location is specified by dma via
the BMBX register. However, on SLI-3 adapters, the location of the
mailbox command submission area changes. When the function is first
powered on or reset, the cmd is submitted via PCI bar memory. Later the
driver changes the function config to use host memory and DMA. The
request to start a mailbox command is the same, a simple doorbell write,
regardless of submission area. So.. if there has not been a boot driver
run against the adapter, the mailbox command works as defaults are
ok. But, if the boot driver has configured the card and, and if no
platform pci function/slot reset occurs as the os starts, the mailbox
command will fail. The SLI-3 device will use the stale boot driver dma
location. This can cause PCI eeh errors.
Fix is to reset the sli-3 function before sending the mailbox command,
thus synchronizing the function/driver on mailbox location.
Note: The fix uses routines that are typically invoked later in the call
flow to reset the sli-3 device. The issue in using those routines is
that the normal (non-fix) flow does additional initialization, namely
the allocation of the pport structure. So, rather than significantly
reworking the initialization flow so that the pport is alloc'd first,
pointer checks are added to work around it. Checks are limited to the
routines invoked by a sli-3 adapter (s3 routines) as this fix/early call
is only invoked on a sli3 adapter. Nothing changes post the
fix. Subsequent initialization, and another adapter reset, still occur -
both on sli-3 and sli-4 adapters.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com> Signed-off-by: James Smart <james.smart@broadcom.com> Fixes: 96418b5e2c88 ("scsi: lpfc: Fix eh_deadline setting for sli3 adapters.") Reviewed-by: Ewan D. Milne <emilne@redhat.com> Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The driver is sending a response to the actual scsi op that was
aborted by an abort task TM, while LIO is sending a response to
the abort task TM.
ibmvscsis_tgt does not send the response to the client until
release_cmd time. The reason for this was because if we did it
at queue_status time, then the client would be free to reuse the
tag for that command, but we're still using the tag until the
command is released at release_cmd time, so we chose to delay
sending the response until then. That then caused this issue, because
release_cmd is always called, even if queue_status is not.
SCSI spec says that the initiator that sends the abort task
TM NEVER gets a response to the aborted op and with the current
code it will send a response. Thus this fix will remove that response
if the CMD_T_ABORTED && !CMD_T_TAS.
Another case with a small timing window is the case where if LIO sends a
TMR_DOES_NOT_EXIST, and the release_cmd callback is called for the TMR Abort
cmd before the release_cmd for the (attemped) aborted cmd, then we need to
ensure that we send the response for the (attempted) abort cmd to the client
before we send the response for the TMR Abort cmd.
Signed-off-by: Bryant G. Ly <bryantly@linux.vnet.ibm.com> Signed-off-by: Michael Cyr <mikecyr@linux.vnet.ibm.com> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The memory allocator passed to __unflatten_device_tree() (e.g. a wrapped
kzalloc) can fail so add the missing sanity check to avoid dereferencing
a NULL pointer.
Fixes: fe14042358fa ("of/flattree: Refactor unflatten_device_tree and add fdt_unflatten_tree") Signed-off-by: Johan Hovold <johan@kernel.org> Signed-off-by: Rob Herring <robh@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The call to of_find_node_by_path("/cpus") returns the cpus device_node
with its reference count incremented. There is no matching of_node_put()
call in of_numa_parse_cpu_nodes() which results in a leaked reference
to the "/cpus" node.
This patch adds an of_node_put() to release the reference.
fixes: 298535c00a2c ("of, numa: Add NUMA of binding implementation.") Signed-off-by: Tyrel Datwyler <tyreld@linux.vnet.ibm.com> Acked-by: David Daney <david.daney@cavium.com> Signed-off-by: Rob Herring <robh@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
proc_create_mount_point() forgot to increase the parent's nlink, and
it resulted in unbalanced hard link numbers, e.g. /proc/fs shows one
less than expected.
Fixes: eb6d38d5427b ("proc: Allow creating permanently empty directories...") Reported-by: Tristan Ye <tristan.ye@suse.com> Signed-off-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fix a boundary condition where in some cases an eeh event that results
in card reset isn't passed on to a driver attached to the virtual PCI
device associated with a slice. This will happen in case when a slice
attached device driver returns a value other than
PCI_ERS_RESULT_NEED_RESET from the eeh error_detected() callback. This
would result in an early return from cxl_pci_error_detected() and
other drivers attached to other AFUs on the card wont be notified.
The patch fixes this by making sure that all slice attached
device-drivers are notified and the return values from
error_detected() callback are aggregated in a scheme where request for
'disconnect' trumps all and 'none' trumps 'need_reset'.
During an eeh event when the cxl card is fenced and card sysfs attr
perst_reloads_same_image is set following warning message is seen in the
kernel logs:
Adapter context unlocked with 0 active contexts
------------[ cut here ]------------
WARNING: CPU: 12 PID: 627 at
../drivers/misc/cxl/main.c:325 cxl_adapter_context_unlock+0x60/0x80 [cxl]
Even though this warning is harmless, it clutters the kernel log
during an eeh event. This warning is triggered as the EEH callback
cxl_pci_error_detected doesn't obtain a context-lock before forcibly
detaching all active context and when context-lock is released during
call to cxl_configure_adapter from cxl_pci_slot_reset, a warning in
cxl_adapter_context_unlock is triggered.
To fix this warning, we acquire the adapter context-lock via
cxl_adapter_context_lock() in the eeh callback
cxl_pci_error_detected() once all the virtual AFU PHBs are notified
and their contexts detached. The context-lock is released in
cxl_pci_slot_reset() after the adapter is successfully reconfigured
and before the we call the slot_reset callback on slice attached
device-drivers.
Fixes: 70b565bbdb91 ("cxl: Prevent adapter reset if an active context exists") Reported-by: Andrew Donnellan <andrew.donnellan@au1.ibm.com> Signed-off-by: Vaibhav Jain <vaibhav@linux.vnet.ibm.com> Acked-by: Frederic Barrat <fbarrat@linux.vnet.ibm.com> Reviewed-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com> Tested-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
On a loaded virtualization host (dozen guests booting at the same time)
it may happen that the ohci controller emulation doesn't manage to do
timely frame processing, with the result that the io watchdog fires and
considers the controller being dead, even though it's only the emulation
being unusual slow due to the load peak.
So, add a quirk for qemu and don't use the watchdog in case we figure we
are running on emulated ohci. The virtual ohci controller masquerades
as apple ohci controller, but we can identify it by subsystem id.
Notifications may only be 8 bytes long. Accessing the 9th and
10th byte of unimplemented/unknown notifications may be insecure.
Also check the length of known notifications before accessing anything
behind the 8th byte.
omap_gpio_debounce() does not validate that the requested debounce
is within a range it can handle. Instead it lets the register value
wrap silently, and always returns success.
This can lead to all sorts of unexpected behavior, such as gpio_keys
asking for a too-long debounce, but getting a very short debounce in
practice.
Fix this by returning -EINVAL if the requested value does not fit into
the register field. If there is no debounce clock available at all,
return -ENOTSUPP.
The idea here was to avoid having to "manually" program the HW if there's
a new earliest alarm. This was lazy and bad, as it leads to loads of fun
races between inter-related callers (ie. therm).
Turns out, it's not so difficult after all. Go figure ;)
Signed-off-by: Ben Skeggs <bskeggs@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
projects / thirdparty / kernel / stable.git / log
