GitHub/LineageOS/G12/android_kernel_amlogic_linux-4.9.git
8 years agokvm: x86: Convert ioapic->rtc_status.dest_map to a struct
Joerg Roedel [Mon, 29 Feb 2016 15:04:43 +0000 (16:04 +0100)]
kvm: x86: Convert ioapic->rtc_status.dest_map to a struct

Currently this is a bitmap which tracks which CPUs we expect
an EOI from. Move this bitmap to a struct so that we can
track additional information there.

Signed-off-by: Joerg Roedel <jroedel@suse.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMerge branch 'kvm-ppc-next' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus...
Paolo Bonzini [Thu, 3 Mar 2016 13:35:44 +0000 (14:35 +0100)]
Merge branch 'kvm-ppc-next' of git://git./linux/kernel/git/paulus/powerpc into HEAD

The highlights are:

* Enable VFIO device on PowerPC, from David Gibson
* Optimizations to speed up IPIs between vcpus in HV KVM,
  from Suresh Warrier (who is also Suresh E. Warrier)
* In-kernel handling of IOMMU hypercalls, and support for dynamic DMA
  windows (DDW), from Alexey Kardashevskiy.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: PPC: Add support for 64bit TCE windows
Alexey Kardashevskiy [Tue, 1 Mar 2016 06:54:40 +0000 (17:54 +1100)]
KVM: PPC: Add support for 64bit TCE windows

The existing KVM_CREATE_SPAPR_TCE only supports 32bit windows which is not
enough for directly mapped windows as the guest can get more than 4GB.

This adds KVM_CREATE_SPAPR_TCE_64 ioctl and advertises it
via KVM_CAP_SPAPR_TCE_64 capability. The table size is checked against
the locked memory limit.

Since 64bit windows are to support Dynamic DMA windows (DDW), let's add
@bus_offset and @page_shift which are also required by DDW.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Paul Mackerras <paulus@samba.org>
8 years agoKVM: PPC: Add @offset to kvmppc_spapr_tce_table
Alexey Kardashevskiy [Tue, 1 Mar 2016 06:54:39 +0000 (17:54 +1100)]
KVM: PPC: Add @offset to kvmppc_spapr_tce_table

This enables userspace view of TCE tables to start from non-zero offset
on a bus. This will be used for huge DMA windows.

This only changes the internal structure, the user interface needs to
change in order to use an offset.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
8 years agoKVM: PPC: Add @page_shift to kvmppc_spapr_tce_table
Alexey Kardashevskiy [Tue, 1 Mar 2016 06:54:38 +0000 (17:54 +1100)]
KVM: PPC: Add @page_shift to kvmppc_spapr_tce_table

At the moment the kvmppc_spapr_tce_table struct can only describe
4GB windows and handle fixed size (4K) pages. Dynamic DMA windows
support more so these limits need to be extended.

This replaces window_size (in bytes, 4GB max) with page_shift (32bit)
and size (64bit, in pages).

This should cause no behavioural change as this is changing
the internal structures only - the user interface still only
allows one to create a 32-bit table with 4KiB pages at this stage.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
8 years agoKVM: PPC: Reserve KVM_CAP_SPAPR_TCE_64 capability number
Alexey Kardashevskiy [Tue, 1 Mar 2016 06:54:37 +0000 (17:54 +1100)]
KVM: PPC: Reserve KVM_CAP_SPAPR_TCE_64 capability number

This adds a capability number for 64-bit TCE tables support.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Paul Mackerras <paulus@samba.org>
8 years agoKVM: PPC: Book3S HV: Add tunable to control H_IPI redirection
Suresh E. Warrier [Mon, 21 Dec 2015 22:33:57 +0000 (16:33 -0600)]
KVM: PPC: Book3S HV: Add tunable to control H_IPI redirection

Redirecting the wakeup of a VCPU from the H_IPI hypercall to
a core running in the host is usually a good idea, most workloads
seemed to benefit. However, in one heavily interrupt-driven SMT1
workload, some regression was observed. This patch adds a kvm_hv
module parameter called h_ipi_redirect to control this feature.

The default value for this tunable is 1 - that is enable the feature.

Signed-off-by: Suresh Warrier <warrier@linux.vnet.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
8 years agoKVM: PPC: Book3S HV: Send IPI to host core to wake VCPU
Suresh E. Warrier [Mon, 21 Dec 2015 22:22:51 +0000 (16:22 -0600)]
KVM: PPC: Book3S HV: Send IPI to host core to wake VCPU

This patch adds support to real-mode KVM to search for a core
running in the host partition and send it an IPI message with
VCPU to be woken. This avoids having to switch to the host
partition to complete an H_IPI hypercall when the VCPU which
is the target of the the H_IPI is not loaded (is not running
in the guest).

The patch also includes the support in the IPI handler running
in the host to do the wakeup by calling kvmppc_xics_ipi_action
for the PPC_MSG_RM_HOST_ACTION message.

When a guest is being destroyed, we need to ensure that there
are no pending IPIs waiting to wake up a VCPU before we free
the VCPUs of the guest. This is accomplished by:
- Forces a PPC_MSG_CALL_FUNCTION IPI to be completed by all CPUs
  before freeing any VCPUs in kvm_arch_destroy_vm().
- Any PPC_MSG_RM_HOST_ACTION messages must be executed first
  before any other PPC_MSG_CALL_FUNCTION messages.

Signed-off-by: Suresh Warrier <warrier@linux.vnet.ibm.com>
Acked-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
8 years agoKVM: PPC: Book3S HV: Host side kick VCPU when poked by real-mode KVM
Suresh Warrier [Thu, 17 Dec 2015 20:59:09 +0000 (14:59 -0600)]
KVM: PPC: Book3S HV: Host side kick VCPU when poked by real-mode KVM

This patch adds the support for the kick VCPU operation for
kvmppc_host_rm_ops. The kvmppc_xics_ipi_action() function
provides the function to be invoked for a host side operation
when poked by the real mode KVM. This is initiated by KVM by
sending an IPI to any free host core.

KVM real mode must set the rm_action to XICS_RM_KICK_VCPU and
rm_data to point to the VCPU to be woken up before sending the IPI.
Note that we have allocated one kvmppc_host_rm_core structure
per core. The above values need to be set in the structure
corresponding to the core to which the IPI will be sent.

Signed-off-by: Suresh Warrier <warrier@linux.vnet.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
8 years agoKVM: PPC: Book3S HV: kvmppc_host_rm_ops - handle offlining CPUs
Suresh Warrier [Thu, 17 Dec 2015 20:59:08 +0000 (14:59 -0600)]
KVM: PPC: Book3S HV: kvmppc_host_rm_ops - handle offlining CPUs

The kvmppc_host_rm_ops structure keeps track of which cores are
are in the host by maintaining a bitmask of active/runnable
online CPUs that have not entered the guest. This patch adds
support to manage the bitmask when a CPU is offlined or onlined
in the host.

Signed-off-by: Suresh Warrier <warrier@linux.vnet.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
8 years agoKVM: PPC: Book3S HV: Manage core host state
Suresh Warrier [Thu, 17 Dec 2015 20:59:07 +0000 (14:59 -0600)]
KVM: PPC: Book3S HV: Manage core host state

Update the core host state in kvmppc_host_rm_ops whenever
the primary thread of the core enters the guest or returns
back.

Signed-off-by: Suresh Warrier <warrier@linux.vnet.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
8 years agoKVM: PPC: Book3S HV: Host-side RM data structures
Suresh Warrier [Thu, 17 Dec 2015 20:59:06 +0000 (14:59 -0600)]
KVM: PPC: Book3S HV: Host-side RM data structures

This patch defines the data structures to support the setting up
of host side operations while running in real mode in the guest,
and also the functions to allocate and free it.

The operations are for now limited to virtual XICS operations.
Currently, we have only defined one operation in the data
structure:
         - Wake up a VCPU sleeping in the host when it
           receives a virtual interrupt

The operations are assigned at the core level because PowerKVM
requires that the host run in SMT off mode. For each core,
we will need to manage its state atomically - where the state
is defined by:
1. Is the core running in the host?
2. Is there a Real Mode (RM) operation pending on the host?

Currently, core state is only managed at the whole-core level
even when the system is in split-core mode. This just limits
the number of free or "available" cores in the host to perform
any host-side operations.

The kvmppc_host_rm_core.rm_data allows any data to be passed by
KVM in real mode to the host core along with the operation to
be performed.

The kvmppc_host_rm_ops structure is allocated the very first time
a guest VM is started. Initial core state is also set - all online
cores are in the host. This structure is never deleted, not even
when there are no active guests. However, it needs to be freed
when the module is unloaded because the kvmppc_host_rm_ops_hv
can contain function pointers to kvm-hv.ko functions for the
different supported host operations.

Signed-off-by: Suresh Warrier <warrier@linux.vnet.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
8 years agopowerpc/xics: Add icp_native_cause_ipi_rm
Suresh Warrier [Thu, 17 Dec 2015 20:59:05 +0000 (14:59 -0600)]
powerpc/xics: Add icp_native_cause_ipi_rm

Function to cause an IPI by directly updating the MFFR register
in the XICS. The function is meant for real-mode callers since
they cannot use the smp_ops->cause_ipi function which uses an
ioremapped address.

Normal usage is for the the KVM real mode code to set the IPI message
using smp_muxed_ipi_message_pass and then invoke icp_native_cause_ipi_rm
to cause the actual IPI.

The function requires kvm_hstate.xics_phys to have been initialized
with the physical address of XICS.

Signed-off-by: Suresh Warrier <warrier@linux.vnet.ibm.com>
Acked-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
8 years agopowerpc/smp: Add smp_muxed_ipi_set_message
Suresh Warrier [Thu, 17 Dec 2015 20:59:04 +0000 (14:59 -0600)]
powerpc/smp: Add smp_muxed_ipi_set_message

smp_muxed_ipi_message_pass() invokes smp_ops->cause_ipi, which
uses an ioremapped address to access registers on the XICS
interrupt controller to cause the IPI. Because of this real
mode callers cannot call smp_muxed_ipi_message_pass() for IPI
messaging.

This patch creates a separate function smp_muxed_ipi_set_message
just to set the IPI message without the cause_ipi routine.
After calling this function to set the IPI message, real
mode callers must cause the IPI by writing to the XICS registers
directly.

As part of this, we also change smp_muxed_ipi_message_pass
to call smp_muxed_ipi_set_message to set the message instead
of doing it directly inside the routine.

Signed-off-by: Suresh Warrier <warrier@linux.vnet.ibm.com>
Acked-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
8 years agopowerpc/smp: Support more IPI messages
Suresh Warrier [Thu, 17 Dec 2015 20:59:03 +0000 (14:59 -0600)]
powerpc/smp: Support more IPI messages

This patch increases the number of demuxed messages for a
controller with a single ipi to 8 for 64-bit systems.

This is required because we want to use the IPI mechanism
to send messages from a CPU running in KVM real mode in a
guest to a CPU in the host to take some action. Currently,
we only support 4 messages and all 4 are already taken.

Define a fifth message PPC_MSG_RM_HOST_ACTION for this
purpose.

Signed-off-by: Suresh Warrier <warrier@linux.vnet.ibm.com>
Acked-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
8 years agoKVM: async_pf: use list_first_entry
Geliang Tang [Fri, 1 Jan 2016 11:47:15 +0000 (19:47 +0800)]
KVM: async_pf: use list_first_entry

To make the intention clearer, use list_first_entry instead of
list_entry.

Signed-off-by: Geliang Tang <geliangtang@163.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: x86: use list_last_entry
Geliang Tang [Fri, 1 Jan 2016 11:47:14 +0000 (19:47 +0800)]
KVM: x86: use list_last_entry

To make the intention clearer, use list_last_entry instead of
list_entry.

Signed-off-by: Geliang Tang <geliangtang@163.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: x86: use list_for_each_entry*
Geliang Tang [Fri, 1 Jan 2016 11:47:13 +0000 (19:47 +0800)]
KVM: x86: use list_for_each_entry*

Use list_for_each_entry*() instead of list_for_each*() to simplify
the code.

Signed-off-by: Geliang Tang <geliangtang@163.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: use list_for_each_entry_safe
Geliang Tang [Fri, 1 Jan 2016 11:47:12 +0000 (19:47 +0800)]
KVM: use list_for_each_entry_safe

Use list_for_each_entry_safe() instead of list_for_each_safe() to
simplify the code.

Signed-off-by: Geliang Tang <geliangtang@163.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: x86: MMU: Move handle_mmio_page_fault() call to kvm_mmu_page_fault()
Takuya Yoshikawa [Mon, 22 Feb 2016 08:23:41 +0000 (17:23 +0900)]
KVM: x86: MMU: Move handle_mmio_page_fault() call to kvm_mmu_page_fault()

Rather than placing a handle_mmio_page_fault() call in each
vcpu->arch.mmu.page_fault() handler, moving it up to
kvm_mmu_page_fault() makes the code better:

 - avoids code duplication
 - for kvm_arch_async_page_ready(), which is the other caller of
   vcpu->arch.mmu.page_fault(), removes an extra error_code check
 - avoids returning both RET_MMIO_PF_* values and raw integer values
   from vcpu->arch.mmu.page_fault()

Signed-off-by: Takuya Yoshikawa <yoshikawa_takuya_b1@lab.ntt.co.jp>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: x86: MMU: Consolidate quickly_check_mmio_pf() and is_mmio_page_fault()
Takuya Yoshikawa [Mon, 22 Feb 2016 08:23:40 +0000 (17:23 +0900)]
KVM: x86: MMU: Consolidate quickly_check_mmio_pf() and is_mmio_page_fault()

These two have only slight differences:
 - whether 'addr' is of type u64 or of type gva_t
 - whether they have 'direct' parameter or not

Concerning the former, quickly_check_mmio_pf()'s u64 is better because
'addr' needs to be able to have both a guest physical address and a
guest virtual address.

The latter is just a stylistic issue as we can always calculate the mode
from the 'vcpu' as is_mmio_page_fault() does.  This patch keeps the
parameter to make the following patch cleaner.

In addition, the patch renames the function to mmio_info_in_cache() to
make it clear what it actually checks for.

Signed-off-by: Takuya Yoshikawa <yoshikawa_takuya_b1@lab.ntt.co.jp>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: x86: pass kvm_get_time_scale arguments in hertz
Paolo Bonzini [Mon, 8 Feb 2016 14:11:15 +0000 (15:11 +0100)]
KVM: x86: pass kvm_get_time_scale arguments in hertz

Prepare for improving the precision in the next patch.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agokvm/x86: Hyper-V VMBus hypercall userspace exit
Andrey Smetanin [Thu, 11 Feb 2016 13:45:01 +0000 (16:45 +0300)]
kvm/x86: Hyper-V VMBus hypercall userspace exit

The patch implements KVM_EXIT_HYPERV userspace exit
functionality for Hyper-V VMBus hypercalls:
HV_X64_HCALL_POST_MESSAGE, HV_X64_HCALL_SIGNAL_EVENT.

Changes v3:
* use vcpu->arch.complete_userspace_io to setup hypercall
result

Changes v2:
* use KVM_EXIT_HYPERV for hypercalls

Signed-off-by: Andrey Smetanin <asmetanin@virtuozzo.com>
Reviewed-by: Roman Kagan <rkagan@virtuozzo.com>
CC: Gleb Natapov <gleb@kernel.org>
CC: Paolo Bonzini <pbonzini@redhat.com>
CC: Joerg Roedel <joro@8bytes.org>
CC: "K. Y. Srinivasan" <kys@microsoft.com>
CC: Haiyang Zhang <haiyangz@microsoft.com>
CC: Roman Kagan <rkagan@virtuozzo.com>
CC: Denis V. Lunev <den@openvz.org>
CC: qemu-devel@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agokvm/x86: Reject Hyper-V hypercall continuation
Andrey Smetanin [Thu, 11 Feb 2016 13:45:00 +0000 (16:45 +0300)]
kvm/x86: Reject Hyper-V hypercall continuation

Currently we do not support Hyper-V hypercall continuation
so reject it.

Signed-off-by: Andrey Smetanin <asmetanin@virtuozzo.com>
Reviewed-by: Roman Kagan <rkagan@virtuozzo.com>
CC: Gleb Natapov <gleb@kernel.org>
CC: Paolo Bonzini <pbonzini@redhat.com>
CC: Joerg Roedel <joro@8bytes.org>
CC: "K. Y. Srinivasan" <kys@microsoft.com>
CC: Haiyang Zhang <haiyangz@microsoft.com>
CC: Roman Kagan <rkagan@virtuozzo.com>
CC: Denis V. Lunev <den@openvz.org>
CC: qemu-devel@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agokvm/x86: Pass return code of kvm_emulate_hypercall
Andrey Smetanin [Thu, 11 Feb 2016 13:44:59 +0000 (16:44 +0300)]
kvm/x86: Pass return code of kvm_emulate_hypercall

Pass the return code from kvm_emulate_hypercall on to the caller,
in order to allow it to indicate to the userspace that
the hypercall has to be handled there.

Also adjust all the existing code paths to return 1 to make sure the
hypercall isn't passed to the userspace without setting kvm_run
appropriately.

Signed-off-by: Andrey Smetanin <asmetanin@virtuozzo.com>
Reviewed-by: Roman Kagan <rkagan@virtuozzo.com>
CC: Gleb Natapov <gleb@kernel.org>
CC: Paolo Bonzini <pbonzini@redhat.com>
CC: Joerg Roedel <joro@8bytes.org>
CC: "K. Y. Srinivasan" <kys@microsoft.com>
CC: Haiyang Zhang <haiyangz@microsoft.com>
CC: Roman Kagan <rkagan@virtuozzo.com>
CC: Denis V. Lunev <den@openvz.org>
CC: qemu-devel@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agodrivers/hv: Move VMBus hypercall codes into Hyper-V UAPI header
Andrey Smetanin [Thu, 11 Feb 2016 13:44:58 +0000 (16:44 +0300)]
drivers/hv: Move VMBus hypercall codes into Hyper-V UAPI header

VMBus hypercall codes inside Hyper-V UAPI header will
be used by QEMU to implement VMBus host devices support.

Signed-off-by: Andrey Smetanin <asmetanin@virtuozzo.com>
Acked-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Roman Kagan <rkagan@virtuozzo.com>
CC: Gleb Natapov <gleb@kernel.org>
CC: Paolo Bonzini <pbonzini@redhat.com>
CC: Joerg Roedel <joro@8bytes.org>
CC: "K. Y. Srinivasan" <kys@microsoft.com>
CC: Haiyang Zhang <haiyangz@microsoft.com>
CC: Roman Kagan <rkagan@virtuozzo.com>
CC: Denis V. Lunev <den@openvz.org>
CC: qemu-devel@nongnu.org
[Do not rename the constant at the same time as moving it, as that
 would cause semantic conflicts with the Hyper-V tree. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agokvm/x86: Rename Hyper-V long spin wait hypercall
Andrey Smetanin [Thu, 11 Feb 2016 13:44:57 +0000 (16:44 +0300)]
kvm/x86: Rename Hyper-V long spin wait hypercall

Rename HV_X64_HV_NOTIFY_LONG_SPIN_WAIT by HVCALL_NOTIFY_LONG_SPIN_WAIT,
so the name is more consistent with the other hypercalls.

Signed-off-by: Andrey Smetanin <asmetanin@virtuozzo.com>
Reviewed-by: Roman Kagan <rkagan@virtuozzo.com>
CC: Gleb Natapov <gleb@kernel.org>
CC: Paolo Bonzini <pbonzini@redhat.com>
CC: Joerg Roedel <joro@8bytes.org>
CC: "K. Y. Srinivasan" <kys@microsoft.com>
CC: Haiyang Zhang <haiyangz@microsoft.com>
CC: Roman Kagan <rkagan@virtuozzo.com>
CC: Denis V. Lunev <den@openvz.org>
CC: qemu-devel@nongnu.org
[Change name, Andrey used HV_X64_HCALL_NOTIFY_LONG_SPIN_WAIT. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: x86: fix missed hardware breakpoints
Paolo Bonzini [Wed, 10 Feb 2016 16:50:23 +0000 (17:50 +0100)]
KVM: x86: fix missed hardware breakpoints

Sometimes when setting a breakpoint a process doesn't stop on it.
This is because the debug registers are not loaded correctly on
VCPU load.

The following simple reproducer from Oleg Nesterov tries using debug
registers in both the host and the guest, for example by running "./bp
0 1" on the host and "./bp 14 15" under QEMU.

    #include <unistd.h>
    #include <signal.h>
    #include <stdlib.h>
    #include <stdio.h>
    #include <sys/wait.h>
    #include <sys/ptrace.h>
    #include <sys/user.h>
    #include <asm/debugreg.h>
    #include <assert.h>

    #define offsetof(TYPE, MEMBER) ((size_t) &((TYPE *)0)->MEMBER)

    unsigned long encode_dr7(int drnum, int enable, unsigned int type, unsigned int len)
    {
        unsigned long dr7;

        dr7 = ((len | type) & 0xf)
            << (DR_CONTROL_SHIFT + drnum * DR_CONTROL_SIZE);
        if (enable)
            dr7 |= (DR_GLOBAL_ENABLE << (drnum * DR_ENABLE_SIZE));

        return dr7;
    }

    int write_dr(int pid, int dr, unsigned long val)
    {
        return ptrace(PTRACE_POKEUSER, pid,
                offsetof (struct user, u_debugreg[dr]),
                val);
    }

    void set_bp(pid_t pid, void *addr)
    {
        unsigned long dr7;
        assert(write_dr(pid, 0, (long)addr) == 0);
        dr7 = encode_dr7(0, 1, DR_RW_EXECUTE, DR_LEN_1);
        assert(write_dr(pid, 7, dr7) == 0);
    }

    void *get_rip(int pid)
    {
        return (void*)ptrace(PTRACE_PEEKUSER, pid,
                offsetof(struct user, regs.rip), 0);
    }

    void test(int nr)
    {
        void *bp_addr = &&label + nr, *bp_hit;
        int pid;

        printf("test bp %d\n", nr);
        assert(nr < 16); // see 16 asm nops below

        pid = fork();
        if (!pid) {
            assert(ptrace(PTRACE_TRACEME, 0,0,0) == 0);
            kill(getpid(), SIGSTOP);
            for (;;) {
                label: asm (
                    "nop; nop; nop; nop;"
                    "nop; nop; nop; nop;"
                    "nop; nop; nop; nop;"
                    "nop; nop; nop; nop;"
                );
            }
        }

        assert(pid == wait(NULL));
        set_bp(pid, bp_addr);

        for (;;) {
            assert(ptrace(PTRACE_CONT, pid, 0, 0) == 0);
            assert(pid == wait(NULL));

            bp_hit = get_rip(pid);
            if (bp_hit != bp_addr)
                fprintf(stderr, "ERR!! hit wrong bp %ld != %d\n",
                    bp_hit - &&label, nr);
        }
    }

    int main(int argc, const char *argv[])
    {
        while (--argc) {
            int nr = atoi(*++argv);
            if (!fork())
                test(nr);
        }

        while (wait(NULL) > 0)
            ;
        return 0;
    }

Cc: stable@vger.kernel.org
Suggested-by: Nadadv Amit <namit@cs.technion.ac.il>
Reported-by: Andrey Wagin <avagin@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: x86: fix *NULL on invalid low-prio irq
Radim Krčmář [Fri, 12 Feb 2016 14:00:15 +0000 (15:00 +0100)]
KVM: x86: fix *NULL on invalid low-prio irq

Smatch noticed a NULL dereference in kvm_intr_is_single_vcpu_fast that
happens if VM already warned about invalid lowest-priority interrupt.

Create a function for common code while fixing it.

Fixes: 6228a0da8057 ("KVM: x86: Add lowest-priority support for vt-d posted-interrupts")
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: x86: rewrite handling of scaled TSC for kvmclock
Paolo Bonzini [Mon, 8 Feb 2016 13:51:40 +0000 (14:51 +0100)]
KVM: x86: rewrite handling of scaled TSC for kvmclock

This is the same as before:

    kvm_scale_tsc(tgt_tsc_khz)
        = tgt_tsc_khz * ratio
        = tgt_tsc_khz * user_tsc_khz / tsc_khz   (see set_tsc_khz)
        = user_tsc_khz                           (see kvm_guest_time_update)
        = vcpu->arch.virtual_tsc_khz             (see kvm_set_tsc_khz)

However, computing it through kvm_scale_tsc will make it possible
to include the NTP correction in tgt_tsc_khz.

Reviewed-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: x86: rename argument to kvm_set_tsc_khz
Paolo Bonzini [Mon, 8 Feb 2016 13:51:12 +0000 (14:51 +0100)]
KVM: x86: rename argument to kvm_set_tsc_khz

This refers to the desired (scaled) frequency, which is called
user_tsc_khz in the rest of the file.

Reviewed-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: VMX: Fix guest debugging while in L2
Jan Kiszka [Tue, 9 Feb 2016 19:15:18 +0000 (20:15 +0100)]
KVM: VMX: Fix guest debugging while in L2

When we take a #DB or #BP vmexit while in guest mode, we first of all
need to check if there is ongoing guest debugging that might be
interested in the event. Currently, we unconditionally leave L2 and
inject the event into L1 if it is intercepting the exceptions. That
breaks things marvelously.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: VMX: Factor out is_exception_n helper
Jan Kiszka [Tue, 9 Feb 2016 19:14:21 +0000 (20:14 +0100)]
KVM: VMX: Factor out is_exception_n helper

There is quite some common code in all these is_<exception>() helpers.
Factor it out before adding even more of them.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: halt_polling: improve grow/shrink settings
Christian Borntraeger [Tue, 9 Feb 2016 12:47:55 +0000 (13:47 +0100)]
KVM: halt_polling: improve grow/shrink settings

Right now halt_poll_ns can be change during runtime. The
grow and shrink factors can only be set during module load.
Lets fix several aspects of grow shrink:
- make grow/shrink changeable by root
- make all variables unsigned int
- read the variables once to prevent races

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMerge tag 'kvm-s390-next-4.6-1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Paolo Bonzini [Tue, 16 Feb 2016 17:48:14 +0000 (18:48 +0100)]
Merge tag 'kvm-s390-next-4.6-1' of git://git./linux/kernel/git/kvms390/linux into HEAD

KVM: s390: Fixes and features for kvm/next (4.6)

1. also provide the floating point registers via sync regs
2. Separate out intruction vs. data accesses
3. Fix program interrupts in some cases
4. Documentation fixes
5. dirty log improvements for huge guests

8 years agoKVM: PPC: Add support for multiple-TCE hcalls
Alexey Kardashevskiy [Mon, 15 Feb 2016 01:55:09 +0000 (12:55 +1100)]
KVM: PPC: Add support for multiple-TCE hcalls

This adds real and virtual mode handlers for the H_PUT_TCE_INDIRECT and
H_STUFF_TCE hypercalls for user space emulated devices such as IBMVIO
devices or emulated PCI. These calls allow adding multiple entries
(up to 512) into the TCE table in one call which saves time on
transition between kernel and user space.

The current implementation of kvmppc_h_stuff_tce() allows it to be
executed in both real and virtual modes so there is one helper.
The kvmppc_rm_h_put_tce_indirect() needs to translate the guest address
to the host address and since the translation is different, there are
2 helpers - one for each mode.

This implements the KVM_CAP_PPC_MULTITCE capability. When present,
the kernel will try handling H_PUT_TCE_INDIRECT and H_STUFF_TCE if these
are enabled by the userspace via KVM_CAP_PPC_ENABLE_HCALL.
If they can not be handled by the kernel, they are passed on to
the user space. The user space still has to have an implementation
for these.

Both HV and PR-syle KVM are supported.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
8 years agoKVM: PPC: Move reusable bits of H_PUT_TCE handler to helpers
Alexey Kardashevskiy [Mon, 15 Feb 2016 01:55:08 +0000 (12:55 +1100)]
KVM: PPC: Move reusable bits of H_PUT_TCE handler to helpers

Upcoming multi-tce support (H_PUT_TCE_INDIRECT/H_STUFF_TCE hypercalls)
will validate TCE (not to have unexpected bits) and IO address
(to be within the DMA window boundaries).

This introduces helpers to validate TCE and IO address. The helpers are
exported as they compile into vmlinux (to work in realmode) and will be
used later by KVM kernel module in virtual mode.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
8 years agoKVM: PPC: Replace SPAPR_TCE_SHIFT with IOMMU_PAGE_SHIFT_4K
Alexey Kardashevskiy [Mon, 15 Feb 2016 01:55:07 +0000 (12:55 +1100)]
KVM: PPC: Replace SPAPR_TCE_SHIFT with IOMMU_PAGE_SHIFT_4K

SPAPR_TCE_SHIFT is used in few places only and since IOMMU_PAGE_SHIFT_4K
can be easily used instead, remove SPAPR_TCE_SHIFT.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
8 years agoKVM: PPC: Account TCE-containing pages in locked_vm
Alexey Kardashevskiy [Mon, 15 Feb 2016 01:55:06 +0000 (12:55 +1100)]
KVM: PPC: Account TCE-containing pages in locked_vm

At the moment pages used for TCE tables (in addition to pages addressed
by TCEs) are not counted in locked_vm counter so a malicious userspace
tool can call ioctl(KVM_CREATE_SPAPR_TCE) as many times as
RLIMIT_NOFILE and lock a lot of memory.

This adds counting for pages used for TCE tables.

This counts the number of pages required for a table plus pages for
the kvmppc_spapr_tce_table struct (TCE table descriptor) itself.

This changes release_spapr_tce_table() to store @npages on stack to
avoid calling kvmppc_stt_npages() in the loop (tiny optimization,
probably).

This does not change the amount of used memory.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
8 years agoKVM: PPC: Use RCU for arch.spapr_tce_tables
Alexey Kardashevskiy [Mon, 15 Feb 2016 01:55:05 +0000 (12:55 +1100)]
KVM: PPC: Use RCU for arch.spapr_tce_tables

At the moment only spapr_tce_tables updates are protected against races
but not lookups. This fixes missing protection by using RCU for the list.
As lookups also happen in real mode, this uses
list_for_each_entry_lockless() (which is expected not to access any
vmalloc'd memory).

This converts release_spapr_tce_table() to a RCU scheduled handler.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
8 years agoKVM: PPC: Rework H_PUT_TCE/H_GET_TCE handlers
Alexey Kardashevskiy [Mon, 15 Feb 2016 01:55:04 +0000 (12:55 +1100)]
KVM: PPC: Rework H_PUT_TCE/H_GET_TCE handlers

This reworks the existing H_PUT_TCE/H_GET_TCE handlers to have following
patches applied nicer.

This moves the ioba boundaries check to a helper and adds a check for
least bits which have to be zeros.

The patch is pretty mechanical (only check for least ioba bits is added)
so no change in behaviour is expected.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
8 years agopowerpc: Make vmalloc_to_phys() public
Alexey Kardashevskiy [Mon, 15 Feb 2016 01:55:03 +0000 (12:55 +1100)]
powerpc: Make vmalloc_to_phys() public

This makes vmalloc_to_phys() public as there will be another user
(KVM in-kernel VFIO acceleration) for it soon. As this new user
can be compiled as a module, this exports the symbol.

As a little optimization, this changes the helper to call
vmalloc_to_pfn() instead of vmalloc_to_page() as the size of the
struct page may not be power-of-two aligned which will make gcc use
multiply instructions instead of shifts.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Acked-by: Michael Ellerman <mpe@ellerman.id.au>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
8 years agovfio: Enable VFIO device for powerpc
David Gibson [Mon, 1 Feb 2016 00:14:15 +0000 (11:14 +1100)]
vfio: Enable VFIO device for powerpc

ec53500f "kvm: Add VFIO device" added a special KVM pseudo-device which is
used to handle any necessary interactions between KVM and VFIO.

Currently that device is built on x86 and ARM, but not powerpc, although
powerpc does support both KVM and VFIO.  This makes things awkward in
userspace

Currently qemu prints an alarming error message if you attempt to use VFIO
and it can't initialize the KVM VFIO device.  We don't want to remove the
warning, because lack of the KVM VFIO device could mean coherency problems
on x86.  On powerpc, however, the error is harmless but looks disturbing,
and a test based on host architecture in qemu would be ugly, and break if
we do need the KVM VFIO device for something important in future.

There's nothing preventing the KVM VFIO device from being built for
powerpc, so this patch turns it on.  It won't actually do anything, since
we don't define any of the arch_*() hooks, but it will make qemu happy and
we can extend it in future if we need to.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Eric Auger <eric.auger@linaro.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
8 years agoKVM: s390: bail out early on fatal signal in dirty logging
Christian Borntraeger [Wed, 3 Feb 2016 10:12:34 +0000 (11:12 +0100)]
KVM: s390: bail out early on fatal signal in dirty logging

A KVM_GET_DIRTY_LOG ioctl might take a long time.
This can result in fatal signals seemingly being ignored.
Lets bail out during the dirty bit sync, if a fatal signal
is pending.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: do not block CPU on dirty logging
Christian Borntraeger [Tue, 2 Feb 2016 14:15:56 +0000 (15:15 +0100)]
KVM: s390: do not block CPU on dirty logging

When doing dirty logging on huge guests (e.g.600GB) we sometimes
get rcu stall timeouts with backtraces like

[ 2753.194083] ([<0000000000112fb2>] show_trace+0x12a/0x130)
[ 2753.194092]  [<0000000000113024>] show_stack+0x6c/0xe8
[ 2753.194094]  [<00000000001ee6a8>] rcu_pending+0x358/0xa48
[ 2753.194099]  [<00000000001f20cc>] rcu_check_callbacks+0x84/0x168
[ 2753.194102]  [<0000000000167654>] update_process_times+0x54/0x80
[ 2753.194107]  [<00000000001bdb5c>] tick_sched_handle.isra.16+0x4c/0x60
[ 2753.194113]  [<00000000001bdbd8>] tick_sched_timer+0x68/0x90
[ 2753.194115]  [<0000000000182a88>] __run_hrtimer+0x88/0x1f8
[ 2753.194119]  [<00000000001838ba>] hrtimer_interrupt+0x122/0x2b0
[ 2753.194121]  [<000000000010d034>] do_extint+0x16c/0x170
[ 2753.194123]  [<00000000005e206e>] ext_skip+0x38/0x3e
[ 2753.194129]  [<000000000012157c>] gmap_test_and_clear_dirty+0xcc/0x118
[ 2753.194134] ([<00000000001214ea>] gmap_test_and_clear_dirty+0x3a/0x118)
[ 2753.194137]  [<0000000000132da4>] kvm_vm_ioctl_get_dirty_log+0xd4/0x1b0
[ 2753.194143]  [<000000000012ac12>] kvm_vm_ioctl+0x21a/0x548
[ 2753.194146]  [<00000000002b57f6>] do_vfs_ioctl+0x30e/0x518
[ 2753.194149]  [<00000000002b5a9c>] SyS_ioctl+0x9c/0xb0
[ 2753.194151]  [<00000000005e1ae6>] sysc_tracego+0x14/0x1a
[ 2753.194153]  [<000003ffb75f3972>] 0x3ffb75f3972

We should do a cond_resched in here.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: do not take mmap_sem on dirty log query
Christian Borntraeger [Wed, 3 Feb 2016 09:18:41 +0000 (10:18 +0100)]
KVM: s390: do not take mmap_sem on dirty log query

Dirty log query can take a long time for huge guests.
Holding the mmap_sem for very long times  can cause some unwanted
latencies.
Turns out that we do not need to hold the mmap semaphore.
We hold the slots_lock for gfn->hva translation and walk the page
tables with that address, so no need to look at the VMAs. KVM also
holds a reference to the mm, which should prevent other things
going away. During the walk we take the necessary ptl locks.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: usage hint for adapter mappings
Cornelia Huck [Tue, 26 Jan 2016 09:48:59 +0000 (10:48 +0100)]
KVM: s390: usage hint for adapter mappings

The interface for adapter mappings was designed with code in mind
that maps each address only once; let's document this.

Otherwise, duplicate mappings are added to the list, which makes
the code ineffective and uses up the limited amount of mapping
needlessly.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: add documentation of KVM_S390_VM_CRYPTO
David Hildenbrand [Thu, 21 Jan 2016 12:45:50 +0000 (13:45 +0100)]
KVM: s390: add documentation of KVM_S390_VM_CRYPTO

Let's properly document KVM_S390_VM_CRYPTO and its attributes.

Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: add documentation of KVM_S390_VM_TOD
David Hildenbrand [Thu, 21 Jan 2016 12:39:14 +0000 (13:39 +0100)]
KVM: s390: add documentation of KVM_S390_VM_TOD

Let's properly document KVM_S390_VM_TOD and its attributes.

Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: remove old fragment of vector registers
David Hildenbrand [Thu, 14 Jan 2016 21:15:08 +0000 (22:15 +0100)]
KVM: s390: remove old fragment of vector registers

Since commit 9977e886cbbc ("s390/kernel: lazy restore fpu registers"),
vregs in struct sie_page is unsed. We can safely remove the field and
the definition.

Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: instruction-fetching exceptions on SIE faults
David Hildenbrand [Tue, 12 Jan 2016 16:40:54 +0000 (17:40 +0100)]
KVM: s390: instruction-fetching exceptions on SIE faults

On instruction-fetch exceptions, we have to forward the PSW by any
valid ilc and correctly use that ilc when injecting the irq. Injection
will already take care of rewinding the PSW if we injected a nullifying
program irq, so we don't need special handling prior to injection.

Until now, autodetection would have guessed an ilc of 0.

Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: provide prog irq ilc on SIE faults
David Hildenbrand [Tue, 12 Jan 2016 16:37:58 +0000 (17:37 +0100)]
KVM: s390: provide prog irq ilc on SIE faults

On SIE faults, the ilc cannot be detected automatically, as the icptcode
is 0. The ilc indicated in the program irq will always be 0. Therefore we
have to manually specify the ilc in order to tell the guest which ilen was
used when forwarding the PSW.

Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: irq delivery should not rely on icptcode
David Hildenbrand [Wed, 4 Nov 2015 15:46:55 +0000 (16:46 +0100)]
KVM: s390: irq delivery should not rely on icptcode

Program irq injection during program irq intercepts is the last candidates
that injects nullifying irqs and relies on delivery to do the right thing.

As we should not rely on the icptcode during any delivery (because that
value will not be migrated), let's add a flag, telling prog IRQ delivery
to not rewind the PSW in case of nullifying prog IRQs.

Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: clean up prog irq injection on prog irq icpts
David Hildenbrand [Wed, 4 Nov 2015 15:22:19 +0000 (16:22 +0100)]
KVM: s390: clean up prog irq injection on prog irq icpts

__extract_prog_irq() is used only once for getting the program check data
in one place. Let's combine it with an injection function to avoid a memset
and to prevent misuse on injection by simplifying the interface to only
have the VCPU as parameter.

Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: read the correct opcode on SIE faults
David Hildenbrand [Mon, 16 Nov 2015 15:17:45 +0000 (16:17 +0100)]
KVM: s390: read the correct opcode on SIE faults

Let's use our fresh new function read_guest_instr() to access
guest storage via the correct addressing schema.

Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: gaccess: implement instruction fetching mode
David Hildenbrand [Mon, 16 Nov 2015 14:48:59 +0000 (15:48 +0100)]
KVM: s390: gaccess: implement instruction fetching mode

When an instruction is to be fetched, special handling applies to
secondary-space mode and access-register mode. The instruction is to be
fetched from primary space.

We can easily support this by selecting the right asce for translation.
Access registers will never be used during translation, so don't
include them in the interface. As we only want to read from the current
PSW address for now, let's also hide that detail.

Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: gaccess: introduce access modes
David Hildenbrand [Mon, 16 Nov 2015 14:42:11 +0000 (15:42 +0100)]
KVM: s390: gaccess: introduce access modes

We will need special handling when fetching instructions, so let's
introduce new guest access modes GACC_FETCH and GACC_STORE instead
of a write flag. An additional patch will then introduce GACC_IFETCH.

Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: migration / injection of prog irq ilc
David Hildenbrand [Wed, 4 Nov 2015 15:33:33 +0000 (16:33 +0100)]
KVM: s390: migration / injection of prog irq ilc

We have to migrate the program irq ilc and someday we will have to
specify the ilc without KVM trying to autodetect the value.

Let's reuse one of the spare fields in our program irq that should
always be set to 0 by user space. Because we also want to make use
of 0 ilcs ("not available"), we need a validity indicator.

If no valid ilc is given, we try to autodetect the ilc via the current
icptcode and icptstatus + parameter and store the valid ilc in the
irq structure.

This has a nice effect: QEMU's making use of KVM_S390_IRQ /
KVM_S390_SET_IRQ_STATE / KVM_S390_GET_IRQ_STATE for migration will
directly migrate the ilc without any changes.

Please note that we use bit 0 as validity and bit 1,2 for the ilc, so
by applying the ilc mask we directly get the ilen which is usually what
we work with.

Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: PSW forwarding / rewinding / ilc rework
David Hildenbrand [Wed, 4 Nov 2015 12:47:58 +0000 (13:47 +0100)]
KVM: s390: PSW forwarding / rewinding / ilc rework

We have some confusion about ilc vs. ilen in our current code. So let's
correctly use the term ilen when dealing with (ilc << 1).

Program irq injection didn't take care of the correct ilc in case of
irqs triggered by EXECUTE functions, let's provide one function
kvm_s390_get_ilen() to take care of all that.

Also, manually specifying in intercept handlers the size of the
instruction (and sometimes overwriting that value for EXECUTE internally)
doesn't make too much sense. So also provide the functions:
- kvm_s390_retry_instr to retry the currently intercepted instruction
- kvm_s390_rewind_psw to rewind the PSW without internal overwrites
- kvm_s390_forward_psw to forward the PSW

Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: sync of fp registers via kvm_run
David Hildenbrand [Mon, 18 Jan 2016 13:46:34 +0000 (14:46 +0100)]
KVM: s390: sync of fp registers via kvm_run

As we already store the floating point registers in the vector save area
in floating point register format when we don't have MACHINE_HAS_VX, we can
directly expose them to user space using a new sync flag.

The floating point registers will be valid when KVM_SYNC_FPRS is set. The
fpc will also be valid when KVM_SYNC_FPRS is set.

Either KVM_SYNC_FPRS or KVM_SYNC_VRS will be enabled, never both.

Let's also change two positions where we access vrs, making the code easier
to read and one comment superfluous.

Suggested-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: allow sync of fp registers via vregs
David Hildenbrand [Fri, 15 Jan 2016 13:11:46 +0000 (14:11 +0100)]
KVM: s390: allow sync of fp registers via vregs

If we have MACHINE_HAS_VX, the floating point registers are stored
in the vector register format, event if the guest isn't enabled for vector
registers. So we can allow KVM_SYNC_VRS as soon as MACHINE_HAS_VX is
available.

This can in return be used by user space to support floating point
registers via struct kvm_run when the machine has vector registers.

Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: x86: consolidate different ways to test for in-kernel LAPIC
Paolo Bonzini [Fri, 8 Jan 2016 12:48:51 +0000 (13:48 +0100)]
KVM: x86: consolidate different ways to test for in-kernel LAPIC

Different pieces of code checked for vcpu->arch.apic being (non-)NULL,
or used kvm_vcpu_has_lapic (more optimized) or lapic_in_kernel.
Replace everything with lapic_in_kernel's name and kvm_vcpu_has_lapic's
implementation.

Reviewed-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: x86: consolidate "has lapic" checks into irq.c
Paolo Bonzini [Fri, 8 Jan 2016 12:41:16 +0000 (13:41 +0100)]
KVM: x86: consolidate "has lapic" checks into irq.c

Do for kvm_cpu_has_pending_timer and kvm_inject_pending_timer_irqs
what the other irq.c routines have been doing.

Reviewed-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: APIC: remove unnecessary double checks on APIC existence
Paolo Bonzini [Fri, 8 Jan 2016 12:42:24 +0000 (13:42 +0100)]
KVM: APIC: remove unnecessary double checks on APIC existence

Usually the in-kernel APIC's existence is checked in the caller.  Do not
bother checking it again in lapic.c.

Reviewed-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM/VMX: Add host irq information in trace event when updating IRTE for posted interrupts
Feng Wu [Mon, 25 Jan 2016 08:53:35 +0000 (16:53 +0800)]
KVM/VMX: Add host irq information in trace event when updating IRTE for posted interrupts

Add host irq information in trace event, so we can better understand
which irq is in posted mode.

Signed-off-by: Feng Wu <feng.wu@intel.com>
Reviewed-by: Radim Krcmar <rkrcmar@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: x86: Add lowest-priority support for vt-d posted-interrupts
Feng Wu [Mon, 25 Jan 2016 08:53:34 +0000 (16:53 +0800)]
KVM: x86: Add lowest-priority support for vt-d posted-interrupts

Use vector-hashing to deliver lowest-priority interrupts for
VT-d posted-interrupts. This patch extends kvm_intr_is_single_vcpu()
to support lowest-priority handling.

Signed-off-by: Feng Wu <feng.wu@intel.com>
Reviewed-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: x86: Use vector-hashing to deliver lowest-priority interrupts
Feng Wu [Mon, 25 Jan 2016 08:53:33 +0000 (16:53 +0800)]
KVM: x86: Use vector-hashing to deliver lowest-priority interrupts

Use vector-hashing to deliver lowest-priority interrupts, As an
example, modern Intel CPUs in server platform use this method to
handle lowest-priority interrupts.

Signed-off-by: Feng Wu <feng.wu@intel.com>
Reviewed-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: Recover IRTE to remapped mode if the interrupt is not single-destination
Feng Wu [Mon, 25 Jan 2016 08:53:32 +0000 (16:53 +0800)]
KVM: Recover IRTE to remapped mode if the interrupt is not single-destination

When the interrupt is not single destination any more, we need
to change back IRTE to remapped mode explicitly.

Signed-off-by: Feng Wu <feng.wu@intel.com>
Reviewed-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: x86: introduce do_shl32_div32
Paolo Bonzini [Fri, 22 Jan 2016 10:39:22 +0000 (11:39 +0100)]
KVM: x86: introduce do_shl32_div32

This is similar to the existing div_frac function, but it returns the
remainder too.  Unlike div_frac, it can be used to implement long
division, e.g. (a << 64) / b for 32-bit a and b.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Mon, 8 Feb 2016 18:32:30 +0000 (10:32 -0800)]
Merge tag 'for-linus' of git://git./virt/kvm/kvm

Pull KVM fixes from Paolo Bonzini:
 "KVM-ARM fixes, mostly coming from the PMU work"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  arm64: KVM: Fix guest dead loop when register accessor returns false
  arm64: KVM: Fix comments of the CP handler
  arm64: KVM: Fix wrong use of the CPSR MODE mask for 32bit guests
  arm64: KVM: Obey RES0/1 reserved bits when setting CPTR_EL2
  arm64: KVM: Fix AArch64 guest userspace exception injection

8 years agoMerge tag 'regmap-fix-v4.5-big-endian' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 8 Feb 2016 18:20:06 +0000 (10:20 -0800)]
Merge tag 'regmap-fix-v4.5-big-endian' of git://git./linux/kernel/git/broonie/regmap

Pull regmap fix from Mark Brown:
 "A single revert back to v4.4 endianness handling.

  Commit 29bb45f25ff3 ("regmap-mmio: Use native endianness for
  read/write") attempted to fix some long standing bugs in the MMIO
  implementation for big endian systems caused by duplicate byte
  swapping in both regmap and readl()/writel().  Sadly the fix makes
  things worse rather than better, so revert it for now"

* tag 'regmap-fix-v4.5-big-endian' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap:
  regmap: mmio: Revert to v4.4 endianness handling

8 years agoscatterlist: fix a typo in comment block of sg_miter_stop()
Masahiro Yamada [Mon, 8 Feb 2016 07:09:08 +0000 (16:09 +0900)]
scatterlist: fix a typo in comment block of sg_miter_stop()

Fix the doubled "started" and tidy up the following sentences.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoMerge tag 'kvm-arm-for-4.5-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git...
Paolo Bonzini [Mon, 8 Feb 2016 15:20:51 +0000 (16:20 +0100)]
Merge tag 'kvm-arm-for-4.5-rc2' of git://git./linux/kernel/git/kvmarm/kvmarm into kvm-master

KVM/ARM fixes for v4.5-rc2

A few random fixes, mostly coming from the PMU work by Shannon:

- fix for injecting faults coming from the guest's userspace
- cleanup for our CPTR_EL2 accessors (reserved bits)
- fix for a bug impacting perf (user/kernel discrimination)
- fix for a 32bit sysreg handling bug

8 years agoLinux 4.5-rc3
Linus Torvalds [Sun, 7 Feb 2016 23:38:30 +0000 (15:38 -0800)]
Linux 4.5-rc3

8 years agoMerge tag 'armsoc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Linus Torvalds [Sun, 7 Feb 2016 23:23:20 +0000 (15:23 -0800)]
Merge tag 'armsoc-fixes' of git://git./linux/kernel/git/arm/arm-soc

Pull ARM SoC fixes from Olof Johansson:
 "The first real batch of fixes for this release cycle, so there are a
  few more than usual.

  Most of these are fixes and tweaks to board support (DT bugfixes,
  etc).  I've also picked up a couple of small cleanups that seemed
  innocent enough that there was little reason to wait (const/
  __initconst and Kconfig deps).

  Quite a bit of the changes on OMAP were due to fixes to no longer
  write to rodata from assembly when ARM_KERNMEM_PERMS was enabled, but
  there were also other fixes.

  Kirkwood had a bunch of gpio fixes for some boards.  OMAP had RTC
  fixes on OMAP5, and Nomadik had changes to MMC parameters in DT.

  All in all, mostly the usual mix of various fixes"

* tag 'armsoc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (46 commits)
  ARM: multi_v7_defconfig: enable DW_WATCHDOG
  ARM: nomadik: fix up SD/MMC DT settings
  ARM64: tegra: Add chosen node for tegra132 norrin
  ARM: realview: use "depends on" instead of "if" after prompt
  ARM: tango: use "depends on" instead of "if" after prompt
  ARM: tango: use const and __initconst for smp_operations
  ARM: realview: use const and __initconst for smp_operations
  bus: uniphier-system-bus: revive tristate prompt
  arm64: dts: Add missing DMA Abort interrupt to Juno
  bus: vexpress-config: Add missing of_node_put
  ARM: dts: am57xx: sbc-am57x: correct Eth PHY settings
  ARM: dts: am57xx: cl-som-am57x: fix CPSW EMAC pinmux
  ARM: dts: am57xx: sbc-am57x: fix UART3 pinmux
  ARM: dts: am57xx: cl-som-am57x: update SPI Flash frequency
  ARM: dts: am57xx: cl-som-am57x: set HOST mode for USB2
  ARM: dts: am57xx: sbc-am57x: fix SB-SOM EEPROM I2C address
  ARM: dts: LogicPD Torpedo: Revert Duplicative Entries
  ARM: dts: am437x: pixcir_tangoc: use correct flags for irq types
  ARM: dts: am4372: fix irq type for arm twd and global timer
  ARM: dts: at91: sama5d4 xplained: fix phy0 IRQ type
  ...

8 years agoMerge branch 'mailbox-devel' of git://git.linaro.org/landing-teams/working/fujitsu...
Linus Torvalds [Sun, 7 Feb 2016 23:17:47 +0000 (15:17 -0800)]
Merge branch 'mailbox-devel' of git://git.linaro.org/landing-teams/working/fujitsu/integration

Pull mailbox fixes from Jassi Brar:

 - fix getting element from the pcc-channels array by simply indexing
   into it

 - prevent building mailbox-test driver for archs that don't have IOMEM

* 'mailbox-devel' of git://git.linaro.org/landing-teams/working/fujitsu/integration:
  mailbox: Fix dependencies for !HAS_IOMEM archs
  mailbox: pcc: fix channel calculation in get_pcc_channel()

8 years agoMerge tag 'usb-4.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb
Linus Torvalds [Sun, 7 Feb 2016 06:14:46 +0000 (22:14 -0800)]
Merge tag 'usb-4.5-rc3' of git://git./linux/kernel/git/gregkh/usb

Pull USB fixes from Greg KH:
 "Here are some USB fixes for 4.5-rc3.

  The usual, xhci fixes for reported issues, combined with some small
  gadget driver fixes, and a MAINTAINERS file update.  All have been in
  linux-next with no reported issues"

* tag 'usb-4.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb:
  xhci: harden xhci_find_next_ext_cap against device removal
  xhci: Fix list corruption in urb dequeue at host removal
  usb: host: xhci-plat: fix NULL pointer in probe for device tree case
  usb: xhci-mtk: fix AHB bus hang up caused by roothubs polling
  usb: xhci-mtk: fix bpkts value of LS/HS periodic eps not behind TT
  usb: xhci: apply XHCI_PME_STUCK_QUIRK to Intel Broxton-M platforms
  usb: xhci: set SSIC port unused only if xhci_suspend succeeds
  usb: xhci: add a quirk bit for ssic port unused
  usb: xhci: handle both SSIC ports in PME stuck quirk
  usb: dwc3: gadget: set the OTG flag in dwc3 gadget driver.
  Revert "xhci: don't finish a TD if we get a short-transfer event mid TD"
  MAINTAINERS: fix my email address
  usb: dwc2: Fix probe problem on bcm2835
  Revert "usb: dwc2: Move reset into dwc2_get_hwparams()"
  usb: musb: ux500: Fix NULL pointer dereference at system PM
  usb: phy: mxs: declare variable with initialized value
  usb: phy: msm: fix error handling in probe.

8 years agoMerge tag 'staging-4.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Sun, 7 Feb 2016 06:13:16 +0000 (22:13 -0800)]
Merge tag 'staging-4.5-rc3' of git://git./linux/kernel/git/gregkh/staging

Pull staging and IIO driver fixes from Greg KH:
 "Here are some IIO and staging driver fixes for 4.5-rc3.

  All of them, except one, are for IIO drivers, and one is for a speakup
  driver fix caused by some earlier patches, to resolve a reported build
  failure"

* tag 'staging-4.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging:
  Staging: speakup: Fix allyesconfig build on mn10300
  iio: dht11: Use boottime
  iio: ade7753: avoid uninitialized data
  iio: pressure: mpl115: fix temperature offset sign
  iio: imu: Fix dependencies for !HAS_IOMEM archs
  staging: iio: Fix dependencies for !HAS_IOMEM archs
  iio: adc: Fix dependencies for !HAS_IOMEM archs
  iio: inkern: fix a NULL dereference on error
  iio:adc:ti_am335x_adc Fix buffered mode by identifying as software buffer.
  iio: light: acpi-als: Report data as processed
  iio: dac: mcp4725: set iio name property in sysfs
  iio: add HAS_IOMEM dependency to VF610_ADC
  iio: add IIO_TRIGGER dependency to STK8BA50
  iio: proximity: lidar: correct return value
  iio-light: Use a signed return type for ltr501_match_samp_freq()

8 years agoMerge branch 'akpm' (patches from Andrew)
Linus Torvalds [Sat, 6 Feb 2016 04:20:07 +0000 (20:20 -0800)]
Merge branch 'akpm' (patches from Andrew)

Merge fixes from Andrew Morton:
 "22 fixes"

* emailed patches from Andrew Morton <akpm@linux-foundation.org>: (22 commits)
  epoll: restrict EPOLLEXCLUSIVE to POLLIN and POLLOUT
  radix-tree: fix oops after radix_tree_iter_retry
  MAINTAINERS: trim the file triggers for ABI/API
  dax: dirty inode only if required
  thp: make deferred_split_scan() work again
  mm: replace vma_lock_anon_vma with anon_vma_lock_read/write
  ocfs2/dlm: clear refmap bit of recovery lock while doing local recovery cleanup
  um: asm/page.h: remove the pte_high member from struct pte_t
  mm, hugetlb: don't require CMA for runtime gigantic pages
  mm/hugetlb: fix gigantic page initialization/allocation
  mm: downgrade VM_BUG in isolate_lru_page() to warning
  mempolicy: do not try to queue pages from !vma_migratable()
  mm, vmstat: fix wrong WQ sleep when memory reclaim doesn't make any progress
  vmstat: make vmstat_update deferrable
  mm, vmstat: make quiet_vmstat lighter
  mm/Kconfig: correct description of DEFERRED_STRUCT_PAGE_INIT
  memblock: don't mark memblock_phys_mem_size() as __init
  dump_stack: avoid potential deadlocks
  mm: validate_mm browse_rb SMP race condition
  m32r: fix build failure due to SMP and MMU
  ...

8 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph...
Linus Torvalds [Sat, 6 Feb 2016 03:52:57 +0000 (19:52 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/sage/ceph-client

Pull Ceph fixes from Sage Weil:
 "We have a few wire protocol compatibility fixes, ports of a few recent
  CRUSH mapping changes, and a couple error path fixes"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client:
  libceph: MOSDOpReply v7 encoding
  libceph: advertise support for TUNABLES5
  crush: decode and initialize chooseleaf_stable
  crush: add chooseleaf_stable tunable
  crush: ensure take bucket value is valid
  crush: ensure bucket id is valid before indexing buckets array
  ceph: fix snap context leak in error path
  ceph: checking for IS_ERR instead of NULL

8 years agoMerge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux
Linus Torvalds [Sat, 6 Feb 2016 03:38:15 +0000 (19:38 -0800)]
Merge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux

Pull drm fixes from Dave Airlie:
 "Fixes all over the place:

   - amdkfd: two static checker fixes
   - mst: a bunch of static checker and spec/hw interaction fixes
   - amdgpu: fix Iceland hw properly, and some fiji bugs, along with
     some write-combining fixes.
   - exynos: some regression fixes
   - adv7511: fix some EDID reading issues"

* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux: (38 commits)
  drm/dp/mst: deallocate payload on port destruction
  drm/dp/mst: Reverse order of MST enable and clearing VC payload table.
  drm/dp/mst: move GUID storage from mgr, port to only mst branch
  drm/dp/mst: change MST detection scheme
  drm/dp/mst: Calculate MST PBN with 31.32 fixed point
  drm: Add drm_fixp_from_fraction and drm_fixp2int_ceil
  drm/mst: Add range check for max_payloads during init
  drm/mst: Don't ignore the MST PBN self-test result
  drm: fix missing reference counting decrease
  drm/amdgpu: disable uvd and vce clockgating on Fiji
  drm/amdgpu: remove exp hardware support from iceland
  drm/amdgpu: load MEC ucode manually on iceland
  drm/amdgpu: don't load MEC2 on topaz
  drm/amdgpu: drop topaz support from gmc8 module
  drm/amdgpu: pull topaz gmc bits into gmc_v7
  drm/amdgpu: The VI specific EXE bit should only apply to GMC v8.0 above
  drm/amdgpu: iceland use CI based MC IP
  drm/amdgpu: move gmc7 support out of CIK dependency
  drm/amdgpu/gfx7: enable cp inst/reg error interrupts
  drm/amdgpu/gfx8: enable cp inst/reg error interrupts
  ...

8 years agoMerge tag 'pm+acpi-4.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Sat, 6 Feb 2016 02:11:23 +0000 (18:11 -0800)]
Merge tag 'pm+acpi-4.5-rc3' of git://git./linux/kernel/git/rafael/linux-pm

Pull power management and ACPI fixes from Rafael Wysocki:
 "These are: a fix for a recently introduced false-positive warnings
  about PM domain pointers being changed inappropriately (harmless but
  annoying), an MCH size workaround quirk for one more platform, a
  compiler warning fix (generic power domains framework), an ACPI LPSS
  (Intel SoCs) driver fixup and a cleanup of the ACPI CPPC core code.

  Specifics:

   - PM core fix to avoid false-positive warnings generated when the
     pm_domain field is cleared for a device that appears to be bound to
     a driver (Rafael Wysocki).

   - New MCH size workaround quirk for Intel Haswell-ULT (Josh Boyer).

   - Fix for an "unused function" compiler warning in the generic power
     domains framework (Ulf Hansson).

   - Fixup for the ACPI driver for Intel SoCs (acpi-lpss) to set the PM
     domain pointer of a device properly in one place that was
     overlooked by a recent PM core update (Andy Shevchenko).

   - Removal of a redundant function declaration in the ACPI CPPC core
     code (Timur Tabi)"

* tag 'pm+acpi-4.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  PM: Avoid false-positive warnings in dev_pm_domain_set()
  PM / Domains: Silence compiler warning for an unused function
  ACPI / CPPC: remove redundant mbox_send_message() declaration
  ACPI / LPSS: set PM domain via helper setter
  PNP: Add Haswell-ULT to Intel MCH size workaround

8 years agoepoll: restrict EPOLLEXCLUSIVE to POLLIN and POLLOUT
Jason Baron [Fri, 5 Feb 2016 23:37:04 +0000 (15:37 -0800)]
epoll: restrict EPOLLEXCLUSIVE to POLLIN and POLLOUT

In the current implementation of the EPOLLEXCLUSIVE flag (added for
4.5-rc1), if epoll waiters create different POLL* sets and register them
as exclusive against the same target fd, the current implementation will
stop waking any further waiters once it finds the first idle waiter.
This means that waiters could miss wakeups in certain cases.

For example, when we wake up a pipe for reading we do:
wake_up_interruptible_sync_poll(&pipe->wait, POLLIN | POLLRDNORM); So if
one epoll set or epfd is added to pipe p with POLLIN and a second set
epfd2 is added to pipe p with POLLRDNORM, only epfd may receive the
wakeup since the current implementation will stop after it finds any
intersection of events with a waiter that is blocked in epoll_wait().

We could potentially address this by requiring all epoll waiters that
are added to p be required to pass the same set of POLL* events.  IE the
first EPOLL_CTL_ADD that passes EPOLLEXCLUSIVE establishes the set POLL*
flags to be used by any other epfds that are added as EPOLLEXCLUSIVE.
However, I think it might be somewhat confusing interface as we would
have to reference count the number of users for that set, and so
userspace would have to keep track of that count, or we would need a
more involved interface.  It also adds some shared state that we'd have
store somewhere.  I don't think anybody will want to bloat
__wait_queue_head for this.

I think what we could do instead, is to simply restrict EPOLLEXCLUSIVE
such that it can only be specified with EPOLLIN and/or EPOLLOUT.  So
that way if the wakeup includes 'POLLIN' and not 'POLLOUT', we can stop
once we hit the first idle waiter that specifies the EPOLLIN bit, since
any remaining waiters that only have 'POLLOUT' set wouldn't need to be
woken.  Likewise, we can do the same thing if 'POLLOUT' is in the wakeup
bit set and not 'POLLIN'.  If both 'POLLOUT' and 'POLLIN' are set in the
wake bit set (there is at least one example of this I saw in fs/pipe.c),
then we just wake the entire exclusive list.  Having both 'POLLOUT' and
'POLLIN' both set should not be on any performance critical path, so I
think that's ok (in fs/pipe.c its in pipe_release()).  We also continue
to include EPOLLERR and EPOLLHUP by default in any exclusive set.  Thus,
the user can specify EPOLLERR and/or EPOLLHUP but is not required to do
so.

Since epoll waiters may be interested in other events as well besides
EPOLLIN, EPOLLOUT, EPOLLERR and EPOLLHUP, these can still be added by
doing a 'dup' call on the target fd and adding that as one normally
would with EPOLL_CTL_ADD.  Since I think that the POLLIN and POLLOUT
events are what we are interest in balancing, I think that the 'dup'
thing could perhaps be added to only one of the waiter threads.
However, I think that EPOLLIN, EPOLLOUT, EPOLLERR and EPOLLHUP should be
sufficient for the majority of use-cases.

Since EPOLLEXCLUSIVE is intended to be used with a target fd shared
among multiple epfds, where between 1 and n of the epfds may receive an
event, it does not satisfy the semantics of EPOLLONESHOT where only 1
epfd would get an event.  Thus, it is not allowed to be specified in
conjunction with EPOLLEXCLUSIVE.

EPOLL_CTL_MOD is also not allowed if the fd was previously added as
EPOLLEXCLUSIVE.  It seems with the limited number of flags to not be as
interesting, but this could be relaxed at some further point.

Signed-off-by: Jason Baron <jbaron@akamai.com>
Tested-by: Madars Vitolins <m@silodev.com>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Al Viro <viro@ftp.linux.org.uk>
Cc: Eric Wong <normalperson@yhbt.net>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Hagen Paul Pfeifer <hagen@jauu.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoradix-tree: fix oops after radix_tree_iter_retry
Konstantin Khlebnikov [Fri, 5 Feb 2016 23:37:01 +0000 (15:37 -0800)]
radix-tree: fix oops after radix_tree_iter_retry

Helper radix_tree_iter_retry() resets next_index to the current index.
In following radix_tree_next_slot current chunk size becomes zero.  This
isn't checked and it tries to dereference null pointer in slot.

Tagged iterator is fine because retry happens only at slot 0 where tag
bitmask in iter->tags is filled with single bit.

Fixes: 46437f9a554f ("radix-tree: fix race in gang lookup")
Signed-off-by: Konstantin Khlebnikov <koct9i@gmail.com>
Cc: Matthew Wilcox <willy@linux.intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Ohad Ben-Cohen <ohad@wizery.com>
Cc: Jeremiah Mahler <jmmahler@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoMAINTAINERS: trim the file triggers for ABI/API
Michael Kerrisk (man-pages) [Fri, 5 Feb 2016 23:36:58 +0000 (15:36 -0800)]
MAINTAINERS: trim the file triggers for ABI/API

Commit ea8f8fc8631 ("MAINTAINERS: add linux-api for review of API/ABI
changes") added file triggers for various paths that likely indicated
API/ABI changes.  However, catching all changes in Documentation/ABI/
and include/uapi/ produces a large volume of mail to linux-api, rather
than only API/ABI changes.  Drop those two entries, but leave
include/linux/syscalls.h and kernel/sys_ni.c to catch syscall-related
changes.

[josh@joshtriplett.org: redid changelog]
Signed-off-by: Michael Kerrisk <mtk.man-pages@gmail.com>
Acked-by: Shuah khan <shuahkh@osg.samsung.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agodax: dirty inode only if required
Dmitry Monakhov [Fri, 5 Feb 2016 23:36:55 +0000 (15:36 -0800)]
dax: dirty inode only if required

Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Cc: Matthew Wilcox <matthew.r.wilcox@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agothp: make deferred_split_scan() work again
Kirill A. Shutemov [Fri, 5 Feb 2016 23:36:53 +0000 (15:36 -0800)]
thp: make deferred_split_scan() work again

We need to iterate over split_queue, not local empty list to get
anything split from the shrinker.

Fixes: e3ae19535c66 ("thp: limit number of object to scan on deferred_split_scan()")
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomm: replace vma_lock_anon_vma with anon_vma_lock_read/write
Konstantin Khlebnikov [Fri, 5 Feb 2016 23:36:50 +0000 (15:36 -0800)]
mm: replace vma_lock_anon_vma with anon_vma_lock_read/write

Sequence vma_lock_anon_vma() - vma_unlock_anon_vma() isn't safe if
anon_vma appeared between lock and unlock.  We have to check anon_vma
first or call anon_vma_prepare() to be sure that it's here.  There are
only few users of these legacy helpers.  Let's get rid of them.

This patch fixes anon_vma lock imbalance in validate_mm().  Write lock
isn't required here, read lock is enough.

And reorders expand_downwards/expand_upwards: security_mmap_addr() and
wrapping-around check don't have to be under anon vma lock.

Link: https://lkml.kernel.org/r/CACT4Y+Y908EjM2z=706dv4rV6dWtxTLK9nFg9_7DhRMLppBo2g@mail.gmail.com
Signed-off-by: Konstantin Khlebnikov <koct9i@gmail.com>
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoocfs2/dlm: clear refmap bit of recovery lock while doing local recovery cleanup
xuejiufei [Fri, 5 Feb 2016 23:36:47 +0000 (15:36 -0800)]
ocfs2/dlm: clear refmap bit of recovery lock while doing local recovery cleanup

When recovery master down, dlm_do_local_recovery_cleanup() only remove
the $RECOVERY lock owned by dead node, but do not clear the refmap bit.
Which will make umount thread falling in dead loop migrating $RECOVERY
to the dead node.

Signed-off-by: xuejiufei <xuejiufei@huawei.com>
Reviewed-by: Joseph Qi <joseph.qi@huawei.com>
Cc: Mark Fasheh <mfasheh@suse.de>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoum: asm/page.h: remove the pte_high member from struct pte_t
Nicolai Stange [Fri, 5 Feb 2016 23:36:44 +0000 (15:36 -0800)]
um: asm/page.h: remove the pte_high member from struct pte_t

Commit 16da306849d0 ("um: kill pfn_t") introduced a compile warning for
defconfig (SUBARCH=i386):

  arch/um/kernel/skas/mmu.c:38:206:
      warning: right shift count >= width of type [-Wshift-count-overflow]

Aforementioned patch changes the definition of the phys_to_pfn() macro
from

  ((pfn_t) ((p) >> PAGE_SHIFT))

to

  ((p) >> PAGE_SHIFT)

This effectively changes the phys_to_pfn() expansion's type from
unsigned long long to unsigned long.

Through the callchain init_stub_pte() => mk_pte(), the expansion of
phys_to_pfn() is (indirectly) fed into the 'phys' argument of the
pte_set_val(pte, phys, prot) macro, eventually leading to

  (pte).pte_high = (phys) >> 32;

This results in the warning from above.

Since UML only deals with 32 bit addresses, the upper 32 bits from
'phys' used to be always zero anyway.  Also, all page protection flags
defined by UML don't use any bits beyond bit 9.  Since the contents of a
PTE are defined within architecture scope only, the ->pte_high member
can be safely removed.

Remove the ->pte_high member from struct pte_t.
Rename ->pte_low to ->pte.
Adapt the pte helper macros in arch/um/include/asm/page.h.

Noteworthy is the pte_copy() macro where a smp_wmb() gets dropped.  This
write barrier doesn't seem to be paired with any read barrier though and
thus, was useless anyway.

Fixes: 16da306849d0 ("um: kill pfn_t")
Signed-off-by: Nicolai Stange <nicstange@gmail.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Richard Weinberger <richard@nod.at>
Cc: Nicolai Stange <nicstange@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomm, hugetlb: don't require CMA for runtime gigantic pages
Vlastimil Babka [Fri, 5 Feb 2016 23:36:41 +0000 (15:36 -0800)]
mm, hugetlb: don't require CMA for runtime gigantic pages

Commit 944d9fec8d7a ("hugetlb: add support for gigantic page allocation
at runtime") has added the runtime gigantic page allocation via
alloc_contig_range(), making this support available only when CONFIG_CMA
is enabled.  Because it doesn't depend on MIGRATE_CMA pageblocks and the
associated infrastructure, it is possible with few simple adjustments to
require only CONFIG_MEMORY_ISOLATION instead of full CONFIG_CMA.

After this patch, alloc_contig_range() and related functions are
available and used for gigantic pages with just CONFIG_MEMORY_ISOLATION
enabled.  Note CONFIG_CMA selects CONFIG_MEMORY_ISOLATION.  This allows
supporting runtime gigantic pages without the CMA-specific checks in
page allocator fastpaths.

Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Luiz Capitulino <lcapitulino@redhat.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Cc: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomm/hugetlb: fix gigantic page initialization/allocation
Mike Kravetz [Fri, 5 Feb 2016 23:36:38 +0000 (15:36 -0800)]
mm/hugetlb: fix gigantic page initialization/allocation

Attempting to preallocate 1G gigantic huge pages at boot time with
"hugepagesz=1G hugepages=1" on the kernel command line will prevent
booting with the following:

  kernel BUG at mm/hugetlb.c:1218!

When mapcount accounting was reworked, the setting of
compound_mapcount_ptr in prep_compound_gigantic_page was overlooked.  As
a result, the validation of mapcount in free_huge_page fails.

The "BUG_ON" checks in free_huge_page were also changed to
"VM_BUG_ON_PAGE" to assist with debugging.

Fixes: 53f9263baba69 ("mm: rework mapcount accounting to enable 4k mapping of THPs")
Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com>
Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: David Rientjes <rientjes@google.com>
Tested-by: Vlastimil Babka <vbabka@suse.cz>
Cc: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>
Cc: Jerome Marchand <jmarchan@redhat.com>
Cc: Michal Hocko <mhocko@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomm: downgrade VM_BUG in isolate_lru_page() to warning
Kirill A. Shutemov [Fri, 5 Feb 2016 23:36:36 +0000 (15:36 -0800)]
mm: downgrade VM_BUG in isolate_lru_page() to warning

Calling isolate_lru_page() is wrong and shouldn't happen, but it not
nessesary fatal: the page just will not be isolated if it's not on LRU.

Let's downgrade the VM_BUG_ON_PAGE() to WARN_RATELIMIT().

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomempolicy: do not try to queue pages from !vma_migratable()
Kirill A. Shutemov [Fri, 5 Feb 2016 23:36:33 +0000 (15:36 -0800)]
mempolicy: do not try to queue pages from !vma_migratable()

Maybe I miss some point, but I don't see a reason why we try to queue
pages from non migratable VMAs.

This testcase steps on VM_BUG_ON_PAGE() in isolate_lru_page():

    #include <fcntl.h>
    #include <unistd.h>
    #include <stdio.h>
    #include <sys/mman.h>
    #include <numaif.h>

    #define SIZE 0x2000

    int foo;

    int main()
    {
        int fd;
        char *p;
        unsigned long mask = 2;

        fd = open("/dev/sg0", O_RDWR);
        p = mmap(NULL, SIZE, PROT_READ | PROT_WRITE, MAP_PRIVATE, fd, 0);
        /* Faultin pages */
        foo = p[0] + p[0x1000];
        mbind(p, SIZE, MPOL_BIND, &mask, 4, MPOL_MF_MOVE | MPOL_MF_STRICT);
        return 0;
    }

The only case when we can queue pages from such VMA is MPOL_MF_STRICT
plus MPOL_MF_MOVE or MPOL_MF_MOVE_ALL for VMA which has pages on LRU,
but gfp mask is not sutable for migaration (see mapping_gfp_mask() check
in vma_migratable()).  That's looks like a bug to me.

Let's filter out non-migratable vma at start of queue_pages_test_walk()
and go to queue_pages_pte_range() only if MPOL_MF_MOVE or
MPOL_MF_MOVE_ALL flag is set.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Signed-off-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomm, vmstat: fix wrong WQ sleep when memory reclaim doesn't make any progress
Tetsuo Handa [Fri, 5 Feb 2016 23:36:30 +0000 (15:36 -0800)]
mm, vmstat: fix wrong WQ sleep when memory reclaim doesn't make any progress

Jan Stancek has reported that system occasionally hanging after "oom01"
testcase from LTP triggers OOM.  Guessing from a result that there is a
kworker thread doing memory allocation and the values between "Node 0
Normal free:" and "Node 0 Normal:" differs when hanging, vmstat is not
up-to-date for some reason.

According to commit 373ccbe59270 ("mm, vmstat: allow WQ concurrency to
discover memory reclaim doesn't make any progress"), it meant to force
the kworker thread to take a short sleep, but it by error used
schedule_timeout(1).  We missed that schedule_timeout() in state
TASK_RUNNING doesn't do anything.

Fix it by using schedule_timeout_uninterruptible(1) which forces the
kworker thread to take a short sleep in order to make sure that vmstat
is up-to-date.

Fixes: 373ccbe59270 ("mm, vmstat: allow WQ concurrency to discover memory reclaim doesn't make any progress")
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Reported-by: Jan Stancek <jstancek@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Cristopher Lameter <clameter@sgi.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Arkadiusz Miskiewicz <arekm@maven.pl>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agovmstat: make vmstat_update deferrable
Michal Hocko [Fri, 5 Feb 2016 23:36:27 +0000 (15:36 -0800)]
vmstat: make vmstat_update deferrable

Commit 0eb77e988032 ("vmstat: make vmstat_updater deferrable again and
shut down on idle") made vmstat_shepherd deferrable.  vmstat_update
itself is still useing standard timer which might interrupt idle task.
This is possible because "mm, vmstat: make quiet_vmstat lighter" removed
cancel_delayed_work from the quiet_vmstat.

Change vmstat_work to use DEFERRABLE_WORK to prevent from pointless
wakeups from the idle context.

Acked-by: Christoph Lameter <cl@linux.com>
Signed-off-by: Michal Hocko <mhocko@suse.com>
Cc: Mike Galbraith <umgwanakikbuti@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomm, vmstat: make quiet_vmstat lighter
Michal Hocko [Fri, 5 Feb 2016 23:36:24 +0000 (15:36 -0800)]
mm, vmstat: make quiet_vmstat lighter

Mike has reported a considerable overhead of refresh_cpu_vm_stats from
the idle entry during pipe test:

    12.89%  [kernel]       [k] refresh_cpu_vm_stats.isra.12
     4.75%  [kernel]       [k] __schedule
     4.70%  [kernel]       [k] mutex_unlock
     3.14%  [kernel]       [k] __switch_to

This is caused by commit 0eb77e988032 ("vmstat: make vmstat_updater
deferrable again and shut down on idle") which has placed quiet_vmstat
into cpu_idle_loop.  The main reason here seems to be that the idle
entry has to get over all zones and perform atomic operations for each
vmstat entry even though there might be no per cpu diffs.  This is a
pointless overhead for _each_ idle entry.

Make sure that quiet_vmstat is as light as possible.

First of all it doesn't make any sense to do any local sync if the
current cpu is already set in oncpu_stat_off because vmstat_update puts
itself there only if there is nothing to do.

Then we can check need_update which should be a cheap way to check for
potential per-cpu diffs and only then do refresh_cpu_vm_stats.

The original patch also did cancel_delayed_work which we are not doing
here.  There are two reasons for that.  Firstly cancel_delayed_work from
idle context will blow up on RT kernels (reported by Mike):

  CPU: 1 PID: 0 Comm: swapper/1 Not tainted 4.5.0-rt3 #7
  Hardware name: MEDION MS-7848/MS-7848, BIOS M7848W08.20C 09/23/2013
  Call Trace:
    dump_stack+0x49/0x67
    ___might_sleep+0xf5/0x180
    rt_spin_lock+0x20/0x50
    try_to_grab_pending+0x69/0x240
    cancel_delayed_work+0x26/0xe0
    quiet_vmstat+0x75/0xa0
    cpu_idle_loop+0x38/0x3e0
    cpu_startup_entry+0x13/0x20
    start_secondary+0x114/0x140

And secondly, even on !RT kernels it might add some non trivial overhead
which is not necessary.  Even if the vmstat worker wakes up and preempts
idle then it will be most likely a single shot noop because the stats
were already synced and so it would end up on the oncpu_stat_off anyway.
We just need to teach both vmstat_shepherd and vmstat_update to stop
scheduling the worker if there is nothing to do.

[mgalbraith@suse.de: cancel pending work of the cpu_stat_off CPU]
Signed-off-by: Michal Hocko <mhocko@suse.com>
Reported-by: Mike Galbraith <umgwanakikbuti@gmail.com>
Acked-by: Christoph Lameter <cl@linux.com>
Signed-off-by: Mike Galbraith <mgalbraith@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomm/Kconfig: correct description of DEFERRED_STRUCT_PAGE_INIT
Vlastimil Babka [Fri, 5 Feb 2016 23:36:21 +0000 (15:36 -0800)]
mm/Kconfig: correct description of DEFERRED_STRUCT_PAGE_INIT

The description mentions kswapd threads, while the deferred struct page
initialization is actually done by one-off "pgdatinitX" threads.

Fix the description so that potentially users are not confused about
pgdatinit threads using CPU after boot instead of kswapd.

Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomemblock: don't mark memblock_phys_mem_size() as __init
David Gibson [Fri, 5 Feb 2016 23:36:19 +0000 (15:36 -0800)]
memblock: don't mark memblock_phys_mem_size() as __init

At the moment memblock_phys_mem_size() is marked as __init, and so is
discarded after boot.  This is different from most of the memblock
functions which are marked __init_memblock, and are only discarded after
boot if memory hotplug is not configured.

To allow for upcoming code which will need memblock_phys_mem_size() in
the hotplug path, change it from __init to __init_memblock.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agodump_stack: avoid potential deadlocks
Eric Dumazet [Fri, 5 Feb 2016 23:36:16 +0000 (15:36 -0800)]
dump_stack: avoid potential deadlocks

Some servers experienced fatal deadlocks because of a combination of
bugs, leading to multiple cpus calling dump_stack().

The checksumming bug was fixed in commit 34ae6a1aa054 ("ipv6: update
skb->csum when CE mark is propagated").

The second problem is a faulty locking in dump_stack()

CPU1 runs in process context and calls dump_stack(), grabs dump_lock.

   CPU2 receives a TCP packet under softirq, grabs socket spinlock, and
   call dump_stack() from netdev_rx_csum_fault().

   dump_stack() spins on atomic_cmpxchg(&dump_lock, -1, 2), since
   dump_lock is owned by CPU1

While dumping its stack, CPU1 is interrupted by a softirq, and happens
to process a packet for the TCP socket locked by CPU2.

CPU1 spins forever in spin_lock() : deadlock

Stack trace on CPU1 looked like :

    NMI backtrace for cpu 1
    RIP: _raw_spin_lock+0x25/0x30
    ...
    Call Trace:
      <IRQ>
      tcp_v6_rcv+0x243/0x620
      ip6_input_finish+0x11f/0x330
      ip6_input+0x38/0x40
      ip6_rcv_finish+0x3c/0x90
      ipv6_rcv+0x2a9/0x500
      process_backlog+0x461/0xaa0
      net_rx_action+0x147/0x430
      __do_softirq+0x167/0x2d0
      call_softirq+0x1c/0x30
      do_softirq+0x3f/0x80
      irq_exit+0x6e/0xc0
      smp_call_function_single_interrupt+0x35/0x40
      call_function_single_interrupt+0x6a/0x70
      <EOI>
      printk+0x4d/0x4f
      printk_address+0x31/0x33
      print_trace_address+0x33/0x3c
      print_context_stack+0x7f/0x119
      dump_trace+0x26b/0x28e
      show_trace_log_lvl+0x4f/0x5c
      show_stack_log_lvl+0x104/0x113
      show_stack+0x42/0x44
      dump_stack+0x46/0x58
      netdev_rx_csum_fault+0x38/0x3c
      __skb_checksum_complete_head+0x6e/0x80
      __skb_checksum_complete+0x11/0x20
      tcp_rcv_established+0x2bd5/0x2fd0
      tcp_v6_do_rcv+0x13c/0x620
      sk_backlog_rcv+0x15/0x30
      release_sock+0xd2/0x150
      tcp_recvmsg+0x1c1/0xfc0
      inet_recvmsg+0x7d/0x90
      sock_recvmsg+0xaf/0xe0
      ___sys_recvmsg+0x111/0x3b0
      SyS_recvmsg+0x5c/0xb0
      system_call_fastpath+0x16/0x1b

Fixes: b58d977432c8 ("dump_stack: serialize the output from dump_stack()")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Alex Thorlton <athorlton@sgi.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>