Skip to content

R
raspberrypi-kernel

openeuler / raspberrypi-kernel

通知 13
Star 1
Fork 0
R
raspberrypi-kernel
切换分支/标签
  1. 01 10月, 2015 7 次提交
  3. 28 9月, 2015 1 次提交
  5. 25 9月, 2015 4 次提交
    • D
      KVM: disable halt_poll_ns as default for s390x · 920552b2
      由 David Hildenbrand 提交于 
      We observed some performance degradation on s390x with dynamic
halt polling. Until we can provide a proper fix, let's enable
halt_poll_ns as default only for supported architectures.

Architectures are now free to set their own halt_poll_ns
default value.
Signed-off-by: NDavid Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>
      920552b2
    • P
      KVM: x86: fix off-by-one in reserved bits check · 58c95070
      由 Paolo Bonzini 提交于 
      29ecd660 ("KVM: x86: avoid uninitialized variable warning",
2015-09-06) introduced a not-so-subtle problem, which probably
escaped review because it was not part of the patch context.

Before the patch, leaf was always equal to iterator.level.  After,
it is equal to iterator.level - 1 in the call to is_shadow_zero_bits_set,
and when is_shadow_zero_bits_set does another "-1" the check on
reserved bits becomes incorrect.  Using "iterator.level" in the call
fixes this call trace:

WARNING: CPU: 2 PID: 17000 at arch/x86/kvm/mmu.c:3385 handle_mmio_page_fault.part.93+0x1a/0x20 [kvm]()
Modules linked in: tun sha256_ssse3 sha256_generic drbg binfmt_misc ipv6 vfat fat fuse dm_crypt dm_mod kvm_amd kvm crc32_pclmul aesni_intel aes_x86_64 lrw gf128mul glue_helper ablk_helper cryptd fam15h_power amd64_edac_mod k10temp edac_core amdkfd amd_iommu_v2 radeon acpi_cpufreq
[...]
Call Trace:
  dump_stack+0x4e/0x84
  warn_slowpath_common+0x95/0xe0
  warn_slowpath_null+0x1a/0x20
  handle_mmio_page_fault.part.93+0x1a/0x20 [kvm]
  tdp_page_fault+0x231/0x290 [kvm]
  ? emulator_pio_in_out+0x6e/0xf0 [kvm]
  kvm_mmu_page_fault+0x36/0x240 [kvm]
  ? svm_set_cr0+0x95/0xc0 [kvm_amd]
  pf_interception+0xde/0x1d0 [kvm_amd]
  handle_exit+0x181/0xa70 [kvm_amd]
  ? kvm_arch_vcpu_ioctl_run+0x68b/0x1730 [kvm]
  kvm_arch_vcpu_ioctl_run+0x6f6/0x1730 [kvm]
  ? kvm_arch_vcpu_ioctl_run+0x68b/0x1730 [kvm]
  ? preempt_count_sub+0x9b/0xf0
  ? mutex_lock_killable_nested+0x26f/0x490
  ? preempt_count_sub+0x9b/0xf0
  kvm_vcpu_ioctl+0x358/0x710 [kvm]
  ? __fget+0x5/0x210
  ? __fget+0x101/0x210
  do_vfs_ioctl+0x2f4/0x560
  ? __fget_light+0x29/0x90
  SyS_ioctl+0x4c/0x90
  entry_SYSCALL_64_fastpath+0x16/0x73
---[ end trace 37901c8686d84de6 ]---
Reported-by: NBorislav Petkov <bp@alien8.de>
Tested-by: NBorislav Petkov <bp@alien8.de>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>
      58c95070
    • P
      KVM: x86: use correct page table format to check nested page table reserved bits · 6fec2144
      由 Paolo Bonzini 提交于 
      Intel CPUID on AMD host or vice versa is a weird case, but it can
happen.  Handle it by checking the host CPU vendor instead of the
guest's in reset_tdp_shadow_zero_bits_mask.  For speed, the
check uses the fact that Intel EPT has an X (executable) bit while
AMD NPT has NX.
Reported-by: NBorislav Petkov <bp@alien8.de>
Tested-by: NBorislav Petkov <bp@alien8.de>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>
      6fec2144
    • P
      KVM: svm: do not call kvm_set_cr0 from init_vmcb · 79a8059d
      由 Paolo Bonzini 提交于 
      kvm_set_cr0 may want to call kvm_zap_gfn_range and thus access the
memslots array (SRCU protected).  Using a mini SRCU critical section
is ugly, and adding it to kvm_arch_vcpu_create doesn't work because
the VMX vcpu_create callback calls synchronize_srcu.

Fixes this lockdep splat:

===============================
[ INFO: suspicious RCU usage. ]
4.3.0-rc1+ #1 Not tainted
-------------------------------
include/linux/kvm_host.h:488 suspicious rcu_dereference_check() usage!

other info that might help us debug this:
rcu_scheduler_active = 1, debug_locks = 0
1 lock held by qemu-system-i38/17000:
 #0:  (&(&kvm->mmu_lock)->rlock){+.+...}, at: kvm_zap_gfn_range+0x24/0x1a0 [kvm]

[...]
Call Trace:
 dump_stack+0x4e/0x84
 lockdep_rcu_suspicious+0xfd/0x130
 kvm_zap_gfn_range+0x188/0x1a0 [kvm]
 kvm_set_cr0+0xde/0x1e0 [kvm]
 init_vmcb+0x760/0xad0 [kvm_amd]
 svm_create_vcpu+0x197/0x250 [kvm_amd]
 kvm_arch_vcpu_create+0x47/0x70 [kvm]
 kvm_vm_ioctl+0x302/0x7e0 [kvm]
 ? __lock_is_held+0x51/0x70
 ? __fget+0x101/0x210
 do_vfs_ioctl+0x2f4/0x560
 ? __fget_light+0x29/0x90
 SyS_ioctl+0x4c/0x90
 entry_SYSCALL_64_fastpath+0x16/0x73
Reported-by: NBorislav Petkov <bp@alien8.de>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>
      79a8059d
  7. 23 9月, 2015 3 次提交
    • A
      x86, efi, kasan: #undef memset/memcpy/memmove per arch · 769a8089
      由 Andrey Ryabinin 提交于 
      In not-instrumented code KASAN replaces instrumented memset/memcpy/memmove
with not-instrumented analogues __memset/__memcpy/__memove.

However, on x86 the EFI stub is not linked with the kernel.  It uses
not-instrumented mem*() functions from arch/x86/boot/compressed/string.c

So we don't replace them with __mem*() variants in EFI stub.

On ARM64 the EFI stub is linked with the kernel, so we should replace
mem*() functions with __mem*(), because the EFI stub runs before KASAN
sets up early shadow.

So let's move these #undef mem* into arch's asm/efi.h which is also
included by the EFI stub.

Also, this will fix the warning in 32-bit build reported by kbuild test
robot:

	efi-stub-helper.c:599:2: warning: implicit declaration of function 'memcpy'

[akpm@linux-foundation.org: use 80 cols in comment]
Signed-off-by: NAndrey Ryabinin <ryabinin.a.a@gmail.com>
Reported-by: NFengguang Wu <fengguang.wu@gmail.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Matt Fleming <matt.fleming@intel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>
      769a8089
    • A
      x86/nmi/64: Fix a paravirt stack-clobbering bug in the NMI code · 83c133cf
      由 Andy Lutomirski 提交于 
      The NMI entry code that switches to the normal kernel stack needs to
be very careful not to clobber any extra stack slots on the NMI
stack.  The code is fine under the assumption that SWAPGS is just a
normal instruction, but that assumption isn't really true.  Use
SWAPGS_UNSAFE_STACK instead.

This is part of a fix for some random crashes that Sasha saw.

Fixes: 9b6e6a83 ("x86/nmi/64: Switch stacks on userspace NMI entry")
Reported-and-tested-by: NSasha Levin <sasha.levin@oracle.com>
Signed-off-by: NAndy Lutomirski <luto@kernel.org>
Cc: stable@vger.kernel.org
Link: http://lkml.kernel.org/r/974bc40edffdb5c2950a5c4977f821a446b76178.1442791737.git.luto@kernel.orgSigned-off-by: NThomas Gleixner <tglx@linutronix.de>
      83c133cf
    • A
      x86/paravirt: Replace the paravirt nop with a bona fide empty function · fc57a7c6
      由 Andy Lutomirski 提交于 
      PARAVIRT_ADJUST_EXCEPTION_FRAME generates this code (using nmi as an
example, trimmed for readability):

    ff 15 00 00 00 00       callq  *0x0(%rip)        # 2796 <nmi+0x6>
              2792: R_X86_64_PC32     pv_irq_ops+0x2c

That's a call through a function pointer to regular C function that
does nothing on native boots, but that function isn't protected
against kprobes, isn't marked notrace, and is certainly not
guaranteed to preserve any registers if the compiler is feeling
perverse.  This is bad news for a CLBR_NONE operation.

Of course, if everything works correctly, once paravirt ops are
patched, it gets nopped out, but what if we hit this code before
paravirt ops are patched in?  This can potentially cause breakage
that is very difficult to debug.

A more subtle failure is possible here, too: if _paravirt_nop uses
the stack at all (even just to push RBP), it will overwrite the "NMI
executing" variable if it's called in the NMI prologue.

The Xen case, perhaps surprisingly, is fine, because it's already
written in asm.

Fix all of the cases that default to paravirt_nop (including
adjust_exception_frame) with a big hammer: replace paravirt_nop with
an asm function that is just a ret instruction.

The Xen case may have other problems, so document them.

This is part of a fix for some random crashes that Sasha saw.
Reported-and-tested-by: NSasha Levin <sasha.levin@oracle.com>
Signed-off-by: NAndy Lutomirski <luto@kernel.org>
Cc: stable@vger.kernel.org
Link: http://lkml.kernel.org/r/8f5d2ba295f9d73751c33d97fda03e0495d9ade0.1442791737.git.luto@kernel.orgSigned-off-by: NThomas Gleixner <tglx@linutronix.de>
      fc57a7c6
  9. 21 9月, 2015 1 次提交
  11. 18 9月, 2015 1 次提交
    • I
      kvm: svm: reset mmu on VCPU reset · ebae871a
      由 Igor Mammedov 提交于 
      When INIT/SIPI sequence is sent to VCPU which before that
was in use by OS, VMRUN might fail with:

 KVM: entry failed, hardware error 0xffffffff
 EAX=00000000 EBX=00000000 ECX=00000000 EDX=000006d3
 ESI=00000000 EDI=00000000 EBP=00000000 ESP=00000000
 EIP=00000000 EFL=00000002 [-------] CPL=0 II=0 A20=1 SMM=0 HLT=0
 ES =0000 00000000 0000ffff 00009300
 CS =9a00 0009a000 0000ffff 00009a00
 [...]
 CR0=60000010 CR2=b6f3e000 CR3=01942000 CR4=000007e0
 [...]
 EFER=0000000000000000

with corresponding SVM error:
 KVM: FAILED VMRUN WITH VMCB:
 [...]
 cpl:            0                efer:         0000000000001000
 cr0:            0000000080010010 cr2:          00007fd7fe85bf90
 cr3:            0000000187d0c000 cr4:          0000000000000020
 [...]

What happens is that VCPU state right after offlinig:
CR0: 0x80050033  EFER: 0xd01  CR4: 0x7e0
  -> long mode with CR3 pointing to longmode page tables

and when VCPU gets INIT/SIPI following transition happens
CR0: 0 -> 0x60000010 EFER: 0x0  CR4: 0x7e0
  -> paging disabled with stale CR3

However SVM under the hood puts VCPU in Paged Real Mode*
which effectively translates CR0 0x60000010 -> 80010010 after

   svm_vcpu_reset()
       -> init_vmcb()
           -> kvm_set_cr0()
               -> svm_set_cr0()

but from  kvm_set_cr0() perspective CR0: 0 -> 0x60000010
only caching bits are changed and
commit d81135a5
 ("KVM: x86: do not reset mmu if CR0.CD and CR0.NW are changed")'
regressed svm_vcpu_reset() which relied on MMU being reset.

As result VMRUN after svm_vcpu_reset() tries to run
VCPU in Paged Real Mode with stale MMU context (longmode page tables),
which causes some AMD CPUs** to bail out with VMEXIT_INVALID.

Fix issue by unconditionally resetting MMU context
at init_vmcb() time.

	* AMD64 Architecture Programmer’s Manual,
	    Volume 2: System Programming, rev: 3.25
	      15.19 Paged Real Mode
	** Opteron 1216
Signed-off-by: NIgor Mammedov <imammedo@redhat.com>
Fixes: d81135a5
Cc: stable@vger.kernel.org
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>
      ebae871a
  13. 17 9月, 2015 1 次提交
  15. 16 9月, 2015 6 次提交
  17. 15 9月, 2015 2 次提交
    • S
      x86/apic: Serialize LVTT and TSC_DEADLINE writes · 5d7c631d
      由 Shaohua Li 提交于 
      The APIC LVTT register is MMIO mapped but the TSC_DEADLINE register is an
MSR. The write to the TSC_DEADLINE MSR is not serializing, so it's not
guaranteed that the write to LVTT has reached the APIC before the
TSC_DEADLINE MSR is written. In such a case the write to the MSR is
ignored and as a consequence the local timer interrupt never fires.

The SDM decribes this issue for xAPIC and x2APIC modes. The
serialization methods recommended by the SDM differ.

xAPIC:
 "1. Memory-mapped write to LVT Timer Register, setting bits 18:17 to 10b.
  2. WRMSR to the IA32_TSC_DEADLINE MSR a value much larger than current time-stamp counter.
  3. If RDMSR of the IA32_TSC_DEADLINE MSR returns zero, go to step 2.
  4. WRMSR to the IA32_TSC_DEADLINE MSR the desired deadline."

x2APIC:
 "To allow for efficient access to the APIC registers in x2APIC mode,
  the serializing semantics of WRMSR are relaxed when writing to the
  APIC registers. Thus, system software should not use 'WRMSR to APIC
  registers in x2APIC mode' as a serializing instruction. Read and write
  accesses to the APIC registers will occur in program order. A WRMSR to
  an APIC register may complete before all preceding stores are globally
  visible; software can prevent this by inserting a serializing
  instruction, an SFENCE, or an MFENCE before the WRMSR."

The xAPIC method is to just wait for the memory mapped write to hit
the LVTT by checking whether the MSR write has reached the hardware.
There is no reason why a proper MFENCE after the memory mapped write would
not do the same. Andi Kleen confirmed that MFENCE is sufficient for the
xAPIC case as well.

Issue MFENCE before writing to the TSC_DEADLINE MSR. This can be done
unconditionally as all CPUs which have TSC_DEADLINE also have MFENCE
support.

[ tglx: Massaged the changelog ]
Signed-off-by: NShaohua Li <shli@fb.com>
Reviewed-by: NIngo Molnar <mingo@kernel.org>
Cc: <Kernel-team@fb.com>
Cc: <lenb@kernel.org>
Cc: <fenghua.yu@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: stable@vger.kernel.org #v3.7+
Link: http://lkml.kernel.org/r/20150909041352.GA2059853@devbig257.prn2.facebook.comSigned-off-by: NThomas Gleixner <tglx@linutronix.de>
      5d7c631d
    • T
      x86/ioapic: Force affinity setting in setup_ioapic_dest() · 4857c91f
      由 Thomas Gleixner 提交于 
      The recent ioapic cleanups changed the affinity setting in
setup_ioapic_dest() from a direct write to the hardware to the delayed
affinity setup via irq_set_affinity().

That results in a warning from chained_irq_exit():
WARNING: CPU: 0 PID: 5 at kernel/irq/migration.c:32 irq_move_masked_irq
[<ffffffff810a0a88>] irq_move_masked_irq+0xb8/0xc0
[<ffffffff8103c161>] ioapic_ack_level+0x111/0x130
[<ffffffff812bbfe8>] intel_gpio_irq_handler+0x148/0x1c0

The reason is that irq_set_affinity() does not write directly to the
hardware. It marks the affinity setting as pending and executes it
from the next interrupt. The chained handler infrastructure does not
take the irq descriptor lock for performance reasons because such a
chained interrupt is not visible to any interfaces. So the delayed
affinity setting triggers the warning in irq_move_masked_irq().

Restore the old behaviour by calling the set_affinity function of the
ioapic chip in setup_ioapic_dest(). This is safe as none of the
interrupts can be on the fly at this point.

Fixes: aa5cb97f 'x86/irq: Remove x86_io_apic_ops.set_affinity and related interfaces'
Reported-and-tested-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: NThomas Gleixner <tglx@linutronix.de>
Cc: Jiang Liu <jiang.liu@linux.intel.com>
Cc: jarkko.nikula@linux.intel.com
      4857c91f
  19. 14 9月, 2015 3 次提交
    • J
      x86/paravirt: Remove the unused pv_time_ops::get_tsc_khz method · cda34fc7
      由 Juergen Gross 提交于 
      It's not used anywhere.
Signed-off-by: NJuergen Gross <jgross@suse.com>
Acked-by: NRusty Russell <rusty@rustcorp.com.au>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: akataria@vmware.com
Cc: chrisw@sous-sol.org
Cc: jeremy@goop.org
Cc: virtualization@lists.linux-foundation.org
Link: http://lkml.kernel.org/r/1442227343-403-1-git-send-email-jgross@suse.comSigned-off-by: NIngo Molnar <mingo@kernel.org>
      cda34fc7
    • J
      x86/ldt: Fix small LDT allocation for Xen · f454b478
      由 Jan Beulich 提交于 
      While the following commit:

  37868fe1 ("x86/ldt: Make modify_ldt synchronous")

added a nice comment explaining that Xen needs page-aligned
whole page chunks for guest descriptor tables, it then
nevertheless used kzalloc() on the small size path.

As I'm unaware of guarantees for kmalloc(PAGE_SIZE, ) to return
page-aligned memory blocks, I believe this needs to be switched
back to __get_free_page() (or better get_zeroed_page()).
Signed-off-by: NJan Beulich <jbeulich@suse.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: David Vrabel <david.vrabel@citrix.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/55E735D6020000780009F1E6@prv-mh.provo.novell.comSigned-off-by: NIngo Molnar <mingo@kernel.org>
      f454b478
    • I
      x86/vm86: Fix the misleading CONFIG_VM86 Kconfig help text · 1e642812
      由 Ingo Molnar 提交于 
      The CONFIG_VM86 Kconfig help text is actively misleading, so fix it:

  - Don't mark it 'obsolete' in the text as we'll support the ABI as long as CPUs
    support it.

  - Qualify the part about software emulation and mention that for some apps you
    want a real vm86 mode.

  - Don't scare users away from the option, instead explain what it does.
Reported-by: NStas Sergeev <stsp@list.ru>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Arjan van de Ven <arjan@linux.intel.com>
Cc: Austin S Hemmelgarn <ahferroin7@gmail.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Josh Boyer <jwboyer@fedoraproject.org>
Cc: Kees Cook <keescook@chromium.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matthew Garrett <mjg59@srcf.ucam.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: NIngo Molnar <mingo@kernel.org>
      1e642812
  21. 13 9月, 2015 2 次提交
  23. 12 9月, 2015 1 次提交
    • M
      sys_membarrier(): system-wide memory barrier (generic, x86) · 5b25b13a
      由 Mathieu Desnoyers 提交于 
      Here is an implementation of a new system call, sys_membarrier(), which
executes a memory barrier on all threads running on the system.  It is
implemented by calling synchronize_sched().  It can be used to
distribute the cost of user-space memory barriers asymmetrically by
transforming pairs of memory barriers into pairs consisting of
sys_membarrier() and a compiler barrier.  For synchronization primitives
that distinguish between read-side and write-side (e.g.  userspace RCU
[1], rwlocks), the read-side can be accelerated significantly by moving
the bulk of the memory barrier overhead to the write-side.

The existing applications of which I am aware that would be improved by
this system call are as follows:

* Through Userspace RCU library (http://urcu.so)
  - DNS server (Knot DNS) https://www.knot-dns.cz/
  - Network sniffer (http://netsniff-ng.org/)
  - Distributed object storage (https://sheepdog.github.io/sheepdog/)
  - User-space tracing (http://lttng.org)
  - Network storage system (https://www.gluster.org/)
  - Virtual routers (https://events.linuxfoundation.org/sites/events/files/slides/DPDK_RCU_0MQ.pdf)
  - Financial software (https://lkml.org/lkml/2015/3/23/189)

Those projects use RCU in userspace to increase read-side speed and
scalability compared to locking.  Especially in the case of RCU used by
libraries, sys_membarrier can speed up the read-side by moving the bulk of
the memory barrier cost to synchronize_rcu().

* Direct users of sys_membarrier
  - core dotnet garbage collector (https://github.com/dotnet/coreclr/issues/198)

Microsoft core dotnet GC developers are planning to use the mprotect()
side-effect of issuing memory barriers through IPIs as a way to implement
Windows FlushProcessWriteBuffers() on Linux.  They are referring to
sys_membarrier in their github thread, specifically stating that
sys_membarrier() is what they are looking for.

To explain the benefit of this scheme, let's introduce two example threads:

Thread A (non-frequent, e.g. executing liburcu synchronize_rcu())
Thread B (frequent, e.g. executing liburcu
rcu_read_lock()/rcu_read_unlock())

In a scheme where all smp_mb() in thread A are ordering memory accesses
with respect to smp_mb() present in Thread B, we can change each
smp_mb() within Thread A into calls to sys_membarrier() and each
smp_mb() within Thread B into compiler barriers "barrier()".

Before the change, we had, for each smp_mb() pairs:

Thread A                    Thread B
previous mem accesses       previous mem accesses
smp_mb()                    smp_mb()
following mem accesses      following mem accesses

After the change, these pairs become:

Thread A                    Thread B
prev mem accesses           prev mem accesses
sys_membarrier()            barrier()
follow mem accesses         follow mem accesses

As we can see, there are two possible scenarios: either Thread B memory
accesses do not happen concurrently with Thread A accesses (1), or they
do (2).

1) Non-concurrent Thread A vs Thread B accesses:

Thread A                    Thread B
prev mem accesses
sys_membarrier()
follow mem accesses
                            prev mem accesses
                            barrier()
                            follow mem accesses

In this case, thread B accesses will be weakly ordered. This is OK,
because at that point, thread A is not particularly interested in
ordering them with respect to its own accesses.

2) Concurrent Thread A vs Thread B accesses

Thread A                    Thread B
prev mem accesses           prev mem accesses
sys_membarrier()            barrier()
follow mem accesses         follow mem accesses

In this case, thread B accesses, which are ensured to be in program
order thanks to the compiler barrier, will be "upgraded" to full
smp_mb() by synchronize_sched().

* Benchmarks

On Intel Xeon E5405 (8 cores)
(one thread is calling sys_membarrier, the other 7 threads are busy
looping)

1000 non-expedited sys_membarrier calls in 33s =3D 33 milliseconds/call.

* User-space user of this system call: Userspace RCU library

Both the signal-based and the sys_membarrier userspace RCU schemes
permit us to remove the memory barrier from the userspace RCU
rcu_read_lock() and rcu_read_unlock() primitives, thus significantly
accelerating them. These memory barriers are replaced by compiler
barriers on the read-side, and all matching memory barriers on the
write-side are turned into an invocation of a memory barrier on all
active threads in the process. By letting the kernel perform this
synchronization rather than dumbly sending a signal to every process
threads (as we currently do), we diminish the number of unnecessary wake
ups and only issue the memory barriers on active threads. Non-running
threads do not need to execute such barrier anyway, because these are
implied by the scheduler context switches.

Results in liburcu:

Operations in 10s, 6 readers, 2 writers:

memory barriers in reader:    1701557485 reads, 2202847 writes
signal-based scheme:          9830061167 reads,    6700 writes
sys_membarrier:               9952759104 reads,     425 writes
sys_membarrier (dyn. check):  7970328887 reads,     425 writes

The dynamic sys_membarrier availability check adds some overhead to
the read-side compared to the signal-based scheme, but besides that,
sys_membarrier slightly outperforms the signal-based scheme. However,
this non-expedited sys_membarrier implementation has a much slower grace
period than signal and memory barrier schemes.

Besides diminishing the number of wake-ups, one major advantage of the
membarrier system call over the signal-based scheme is that it does not
need to reserve a signal. This plays much more nicely with libraries,
and with processes injected into for tracing purposes, for which we
cannot expect that signals will be unused by the application.

An expedited version of this system call can be added later on to speed
up the grace period. Its implementation will likely depend on reading
the cpu_curr()->mm without holding each CPU's rq lock.

This patch adds the system call to x86 and to asm-generic.

[1] http://urcu.so

membarrier(2) man page:

MEMBARRIER(2)              Linux Programmer's Manual             MEMBARRIER(2)

NAME
       membarrier - issue memory barriers on a set of threads

SYNOPSIS
       #include <linux/membarrier.h>

       int membarrier(int cmd, int flags);

DESCRIPTION
       The cmd argument is one of the following:

       MEMBARRIER_CMD_QUERY
              Query  the  set  of  supported commands. It returns a bitmask of
              supported commands.

       MEMBARRIER_CMD_SHARED
              Execute a memory barrier on all threads running on  the  system.
              Upon  return from system call, the caller thread is ensured that
              all running threads have passed through a state where all memory
              accesses  to  user-space  addresses  match program order between
              entry to and return from the system  call  (non-running  threads
              are de facto in such a state). This covers threads from all pro=E2=80=90
              cesses running on the system.  This command returns 0.

       The flags argument needs to be 0. For future extensions.

       All memory accesses performed  in  program  order  from  each  targeted
       thread is guaranteed to be ordered with respect to sys_membarrier(). If
       we use the semantic "barrier()" to represent a compiler barrier forcing
       memory  accesses  to  be performed in program order across the barrier,
       and smp_mb() to represent explicit memory barriers forcing full  memory
       ordering  across  the barrier, we have the following ordering table for
       each pair of barrier(), sys_membarrier() and smp_mb():

       The pair ordering is detailed as (O: ordered, X: not ordered):

                              barrier()   smp_mb() sys_membarrier()
              barrier()          X           X            O
              smp_mb()           X           O            O
              sys_membarrier()   O           O            O

RETURN VALUE
       On success, these system calls return zero.  On error, -1 is  returned,
       and errno is set appropriately. For a given command, with flags
       argument set to 0, this system call is guaranteed to always return the
       same value until reboot.

ERRORS
       ENOSYS System call is not implemented.

       EINVAL Invalid arguments.

Linux                             2015-04-15                     MEMBARRIER(2)
Signed-off-by: NMathieu Desnoyers <mathieu.desnoyers@efficios.com>
Reviewed-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: NJosh Triplett <josh@joshtriplett.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Nicholas Miell <nmiell@comcast.net>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Alan Cox <gnomes@lxorguk.ukuu.org.uk>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Stephen Hemminger <stephen@networkplumber.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: David Howells <dhowells@redhat.com>
Cc: Pranith Kumar <bobby.prani@gmail.com>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Cc: Shuah Khan <shuahkh@osg.samsung.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>
      5b25b13a
  25. 11 9月, 2015 8 次提交
    • A
      perf/x86/intel/bts: Set event->hw.itrace_started in pmu::start to match the new logic · d2498729
      由 Alexander Shishkin 提交于 
      Since event->hw.itrace_started is now set in pmu::start() to signal the beginning of
the trace, do so also in the intel_bts driver.
Signed-off-by: NAlexander Shishkin <alexander.shishkin@linux.intel.com>
Signed-off-by: NPeter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: acme@infradead.org
Cc: adrian.hunter@intel.com
Cc: hpa@zytor.com
Link: http://lkml.kernel.org/r/1437140050-23363-4-git-send-email-alexander.shishkin@linux.intel.comSigned-off-by: NIngo Molnar <mingo@kernel.org>
      d2498729
    • P
      locking/qspinlock/x86: Only emit the test-and-set fallback when building guest support · a6b27785
      由 Peter Zijlstra 提交于 
      Only emit the test-and-set fallback for Hypervisors lacking
PARAVIRT_SPINLOCKS support when building for guests.
Suggested-by: NLinus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: NPeter Zijlstra (Intel) <peterz@infradead.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Cc: stable@vger.kernel.org # 4.2
Signed-off-by: NIngo Molnar <mingo@kernel.org>
      a6b27785
    • P
      locking/qspinlock/x86: Fix performance regression under unaccelerated VMs · 43b3f028
      由 Peter Zijlstra 提交于 
      Dave ran into horrible performance on a VM without PARAVIRT_SPINLOCKS
set and Linus noted that the test-and-set implementation was retarded.

One should spin on the variable with a load, not a RMW.

While there, remove 'queued' from the name, as the lock isn't queued
at all, but a simple test-and-set.
Suggested-by: NLinus Torvalds <torvalds@linux-foundation.org>
Reported-by: NDave Chinner <david@fromorbit.com>
Tested-by: NDave Chinner <david@fromorbit.com>
Signed-off-by: NPeter Zijlstra (Intel) <peterz@infradead.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Waiman Long <Waiman.Long@hp.com>
Cc: stable@vger.kernel.org # v4.2+
Link: http://lkml.kernel.org/r/20150904152523.GR18673@twins.programming.kicks-ass.netSigned-off-by: NIngo Molnar <mingo@kernel.org>
      43b3f028
    • C
      dma-mapping: consolidate dma_set_mask · 452e06af
      由 Christoph Hellwig 提交于 
      Almost everyone implements dma_set_mask the same way, although some time
that's hidden in ->set_dma_mask methods.

This patch consolidates those into a common implementation that either
calls ->set_dma_mask if present or otherwise uses the default
implementation.  Some architectures used to only call ->set_dma_mask
after the initial checks, and those instance have been fixed to do the
full work.  h8300 implemented dma_set_mask bogusly as a no-ops and has
been fixed.

Unfortunately some architectures overload unrelated semantics like changing
the dma_ops into it so we still need to allow for an architecture override
for now.

[jcmvbkbc@gmail.com: fix xtensa]
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Michal Simek <monstr@monstr.eu>
Cc: Jonas Bonn <jonas@southpole.se>
Cc: Chris Metcalf <cmetcalf@ezchip.com>
Cc: Guan Xuetao <gxt@mprc.pku.edu.cn>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: NMax Filippov <jcmvbkbc@gmail.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>
      452e06af
    • C
      dma-mapping: consolidate dma_supported · ee196371
      由 Christoph Hellwig 提交于 
      Most architectures just call into ->dma_supported, but some also return 1
if the method is not present, or 0 if no dma ops are present (although
that should never happeb). Consolidate this more broad version into
common code.

Also fix h8300 which inorrectly always returned 0, which would have been
a problem if it's dma_set_mask implementation wasn't a similarly buggy
noop.

As a few architectures have much more elaborate implementations, we
still allow for arch overrides.

[jcmvbkbc@gmail.com: fix xtensa]
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Michal Simek <monstr@monstr.eu>
Cc: Jonas Bonn <jonas@southpole.se>
Cc: Chris Metcalf <cmetcalf@ezchip.com>
Cc: Guan Xuetao <gxt@mprc.pku.edu.cn>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: NMax Filippov <jcmvbkbc@gmail.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>
      ee196371
    • C
      dma-mapping: cosolidate dma_mapping_error · efa21e43
      由 Christoph Hellwig 提交于 
      Currently there are three valid implementations of dma_mapping_error:

 (1) call ->mapping_error
 (2) check for a hardcoded error code
 (3) always return 0

This patch provides a common implementation that calls ->mapping_error
if present, then checks for DMA_ERROR_CODE if defined or otherwise
returns 0.

[jcmvbkbc@gmail.com: fix xtensa]
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Michal Simek <monstr@monstr.eu>
Cc: Jonas Bonn <jonas@southpole.se>
Cc: Chris Metcalf <cmetcalf@ezchip.com>
Cc: Guan Xuetao <gxt@mprc.pku.edu.cn>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: NMax Filippov <jcmvbkbc@gmail.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>
      efa21e43
    • C
      dma-mapping: consolidate dma_{alloc,free}_noncoherent · 1e893752
      由 Christoph Hellwig 提交于 
      Most architectures do not support non-coherent allocations and either
define dma_{alloc,free}_noncoherent to their coherent versions or stub
them out.

Openrisc uses dma_{alloc,free}_attrs to implement them, and only Mips
implements them directly.

This patch moves the Openrisc version to common code, and handles the
DMA_ATTR_NON_CONSISTENT case in the mips dma_map_ops instance.

Note that actual non-coherent allocations require a dma_cache_sync
implementation, so if non-coherent allocations didn't work on
an architecture before this patch they still won't work after it.

[jcmvbkbc@gmail.com: fix xtensa]
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Michal Simek <monstr@monstr.eu>
Cc: Jonas Bonn <jonas@southpole.se>
Cc: Chris Metcalf <cmetcalf@ezchip.com>
Cc: Guan Xuetao <gxt@mprc.pku.edu.cn>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: NMax Filippov <jcmvbkbc@gmail.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>
      1e893752
    • C
      dma-mapping: consolidate dma_{alloc,free}_{attrs,coherent} · 6894258e
      由 Christoph Hellwig 提交于 
      Since 2009 we have a nice asm-generic header implementing lots of DMA API
functions for architectures using struct dma_map_ops, but unfortunately
it's still missing a lot of APIs that all architectures still have to
duplicate.

This series consolidates the remaining functions, although we still need
arch opt outs for two of them as a few architectures have very
non-standard implementations.

This patch (of 5):

The coherent DMA allocator works the same over all architectures supporting
dma_map operations.

This patch consolidates them and converges the minor differences:

 - the debug_dma helpers are now called from all architectures, including
   those that were previously missing them
 - dma_alloc_from_coherent and dma_release_from_coherent are now always
   called from the generic alloc/free routines instead of the ops
   dma-mapping-common.h always includes dma-coherent.h to get the defintions
   for them, or the stubs if the architecture doesn't support this feature
 - checks for ->alloc / ->free presence are removed.  There is only one
   magic instead of dma_map_ops without them (mic_dma_ops) and that one
   is x86 only anyway.

Besides that only x86 needs special treatment to replace a default devices
if none is passed and tweak the gfp_flags.  An optional arch hook is provided
for that.

[linux@roeck-us.net: fix build]
[jcmvbkbc@gmail.com: fix xtensa]
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Michal Simek <monstr@monstr.eu>
Cc: Jonas Bonn <jonas@southpole.se>
Cc: Chris Metcalf <cmetcalf@ezchip.com>
Cc: Guan Xuetao <gxt@mprc.pku.edu.cn>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NMax Filippov <jcmvbkbc@gmail.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>
      6894258e