linux_dsm_epyc7002/arch/x86/kernel
Yonghong Song e7ed9d9bd0 uprobes/x86: Emulate push insns for uprobe on x86
Uprobe is a tracing mechanism for userspace programs.
Typical uprobe will incur overhead of two traps.
First trap is caused by replaced trap insn, and
the second trap is to execute the original displaced
insn in user space.

To reduce the overhead, kernel provides hooks
for architectures to emulate the original insn
and skip the second trap. In x86, emulation
is done for certain branch insns.

This patch extends the emulation to "push <reg>"
insns. These insns are typical in the beginning
of the function. For example, bcc
in https://github.com/iovisor/bcc repo provides
tools to measure funclantency, detect memleak, etc.
The tools will place uprobes in the beginning of
function and possibly uretprobes at the end of function.
This patch is able to reduce the trap overhead for
uprobe from 2 to 1.

Without this patch, uretprobe will typically incur
three traps. With this patch, if the function starts
with "push" insn, the number of traps can be
reduced from 3 to 2.

An experiment was conducted on two local VMs,
fedora 26 64-bit VM and 32-bit VM, both 4 processors
and 4GB memory, booted with latest tip repo (and this patch).
The host is MacBook with intel i7 processor.

The test program looks like:

  #include <stdio.h>
  #include <stdlib.h>
  #include <time.h>
  #include <sys/time.h>

  static void test() __attribute__((noinline));
  void test() {}
  int main() {
    struct timeval start, end;

    gettimeofday(&start, NULL);
    for (int i = 0; i < 1000000; i++) {
      test();
    }
    gettimeofday(&end, NULL);

    printf("%ld\n", ((end.tv_sec * 1000000 + end.tv_usec)
                     - (start.tv_sec * 1000000 + start.tv_usec)));
    return 0;
  }

The program is compiled without optimization, and
the first insn for function "test" is "push %rbp".
The host is relatively idle.

Before the test run, the uprobe is inserted as below for uprobe:
  echo 'p <binary>:<test_func_offset>' > /sys/kernel/debug/tracing/uprobe_events
  echo 1 > /sys/kernel/debug/tracing/events/uprobes/enable
and for uretprobe:
  echo 'r <binary>:<test_func_offset>' > /sys/kernel/debug/tracing/uprobe_events
  echo 1 > /sys/kernel/debug/tracing/events/uprobes/enable

Unit: microsecond(usec) per loop iteration

x86_64          W/ this patch   W/O this patch
uprobe          1.55            3.1
uretprobe       2.0             3.6

x86_32          W/ this patch   W/O this patch
uprobe          1.41            3.5
uretprobe       1.75            4.0

You can see that this patch significantly reduced the overhead,
50% for uprobe and 44% for uretprobe on x86_64, and even more
on x86_32.

Signed-off-by: Yonghong Song <yhs@fb.com>
Reviewed-by: Oleg Nesterov <oleg@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: kernel-team@fb.com
Link: http://lkml.kernel.org/r/20171201001202.3706564-1-yhs@fb.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-12-11 18:42:11 +01:00
..
acpi Merge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip 2017-11-26 14:11:54 -08:00
apic x86/PCI: Remove unused HyperTransport interrupt support 2017-11-23 20:18:18 +01:00
cpu Merge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip 2017-11-26 14:11:54 -08:00
fpu Merge branch 'linus' into x86/asm, to pick up fixes and resolve conflicts 2017-11-07 10:53:06 +01:00
kprobes Merge branch 'linus' into perf/core, to fix conflicts 2017-11-07 10:30:18 +01:00
.gitignore
alternative.c kprobes, x86/alternatives: Use text_mutex to protect smp_alt_modules 2017-11-07 12:20:09 +01:00
amd_gart_64.c x86: remove arch specific dma_supported implementation 2017-06-28 06:54:46 -07:00
amd_nb.c x86/cpu/AMD: Apply the Erratum 688 fix when the BIOS doesn't 2017-10-22 13:06:02 +02:00
apb_timer.c Merge branch 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip 2016-12-25 14:30:04 -08:00
aperture_64.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
apm_32.c dmi: Mark all struct dmi_system_id instances const 2017-09-14 11:59:30 +02:00
asm-offsets_32.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
asm-offsets_64.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
asm-offsets.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
audit_64.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
bootflag.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
check.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
cpuid.c Merge branch 'smp-hotplug-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip 2016-12-12 19:25:04 -08:00
crash_dump_32.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
crash_dump_64.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
crash.c resource: Fix resource_size.cocci warnings 2017-11-07 20:44:56 +01:00
devicetree.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
doublefault.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
dumpstack_32.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
dumpstack_64.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
dumpstack.c x86/asm/32: Remove a bunch of '& 0xffff' from pt_regs segment reads 2017-07-30 12:04:41 +02:00
e820.c x86/boot/e820: Add support to determine the E820 type of an address 2017-07-18 11:38:01 +02:00
early_printk.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
early-quirks.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
ebda.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
eisa.c x86/eisa: Add missing include 2017-08-31 21:34:48 +02:00
espfix_64.c kmemcheck: stop using GFP_NOTRACK and SLAB_NOTRACK 2017-11-15 18:21:04 -08:00
ftrace_32.S License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
ftrace_64.S License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
ftrace.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
head32.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
head64.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
head_32.S Merge branch 'linus' into x86/asm, to pick up fixes and resolve conflicts 2017-11-07 10:53:06 +01:00
head_64.S Merge branch 'linus' into x86/asm, to pick up fixes and resolve conflicts 2017-11-07 10:53:06 +01:00
hpet.c x86/hpet: Cure interface abuse in the resume path 2017-08-01 13:02:37 +02:00
hw_breakpoint.c
i8237.c
i8253.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
i8259.c Merge branch 'linus' into x86/apic, to resolve conflicts 2017-11-07 10:51:10 +01:00
idt.c Merge branch 'x86-apic-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip 2017-11-13 18:29:23 -08:00
io_delay.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
ioport.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
irq_32.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
irq_64.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
irq_work.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
irq.c x86/irq: Simplify hotplug vector accounting 2017-09-25 20:52:02 +02:00
irqinit.c Merge branch 'linus' into x86/apic, to resolve conflicts 2017-11-07 10:51:10 +01:00
itmt.c sched/x86: Remove unnecessary TBM3 check to update topology 2017-01-19 08:42:37 +01:00
jump_label.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
kdebugfs.c x86, mpparse, x86/acpi, x86/PCI, x86/dmi, SFI: Use memremap() for RAM mappings 2017-07-18 11:37:58 +02:00
kexec-bzimage64.c x86/boot/e820: Rename the e820_table_firmware to e820_table_kexec 2017-07-05 10:09:02 +02:00
kgdb.c
ksysfs.c x86/sysfs: Fix off-by-one error in loop termination 2017-09-25 09:36:16 +02:00
kvm.c Merge branch 'x86-platform-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip 2017-11-13 17:04:36 -08:00
kvmclock.c xen: features and fixes for v4.15-rc1 2017-11-16 13:06:27 -08:00
ldt.c Merge branch 'x86-asm-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip 2017-11-13 14:13:48 -08:00
livepatch.c
machine_kexec_32.c x86/idt: Consolidate IDT invalidation 2017-08-29 12:07:26 +02:00
machine_kexec_64.c x86/mm, kexec: Fix memory corruption with SME on successive kexecs 2017-07-30 12:09:12 +02:00
Makefile x86/umip: Add emulation code for UMIP instructions 2017-11-08 11:16:22 +01:00
mmconf-fam10h_64.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
module.c x86/module: Detect and skip invalid relocations 2017-11-05 09:52:16 +01:00
mpparse.c x86/boot: Fix boot failure when SMP MP-table is based at 0 2017-11-17 15:30:33 +01:00
msr.c x86/msr: Remove bogus cleanup from the error path 2016-12-25 10:47:41 +01:00
nmi_selftest.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
nmi.c locking/atomics: COCCINELLE/treewide: Convert trivial ACCESS_ONCE() patterns to READ_ONCE()/WRITE_ONCE() 2017-10-25 11:01:08 +02:00
paravirt_patch_32.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
paravirt_patch_64.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
paravirt-spinlocks.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
paravirt.c locking/paravirt: Use new static key for controlling call of virt_spin_lock() 2017-10-10 11:50:12 +02:00
pci-calgary_64.c x86, calgary: Convert timers to use timer_setup() 2017-11-02 15:50:32 -07:00
pci-dma.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
pci-iommu_table.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
pci-nommu.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
pci-swiotlb.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
pcspeaker.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
perf_regs.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
platform-quirks.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
pmem.c resource: Provide resource struct in resource walk callback 2017-11-07 15:35:57 +01:00
probe_roms.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
process_32.c x86/entry/64: Remove all remaining direct thread_struct::sp0 reads 2017-11-02 11:04:47 +01:00
process_64.c x86/entry/64: Remove thread_struct::sp0 2017-11-02 11:04:48 +01:00
process.c Merge branch 'linus' into x86/asm, to pick up fixes and resolve conflicts 2017-11-07 10:53:06 +01:00
ptrace.c x86/arch_prctl/64: Rename do_arch_prctl() to do_arch_prctl_64() 2017-03-20 16:10:32 +01:00
pvclock.c x86/pvclock: add setter for pvclock_pvti_cpu0_va 2017-11-08 16:33:14 -05:00
quirks.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
reboot_fixups_32.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
reboot.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
relocate_kernel_32.S
relocate_kernel_64.S x86/mm, kexec: Fix memory corruption with SME on successive kexecs 2017-07-30 12:09:12 +02:00
resource.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
rtc.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
setup_percpu.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
setup.c Merge branch 'x86-timers-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip 2017-11-13 19:07:38 -08:00
signal_compat.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
signal.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
smp.c x86/tracing: Disentangle pagefault and resched IPI tracing key 2017-08-29 11:42:29 +02:00
smpboot.c x86/smpboot: Fix __max_logical_packages estimate 2017-11-17 16:22:31 +01:00
stacktrace.c x86/stacktrace: Avoid recording save_stack_trace() wrappers 2017-09-29 19:44:03 +02:00
step.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
sys_x86_64.c x86/mm: Prevent non-MAP_FIXED mapping across DEFAULT_MAP_WINDOW border 2017-11-16 11:43:11 +01:00
sysfb_efi.c
sysfb_simplefb.c x86/sysfb: Fix lfb_size calculation 2016-11-16 09:38:23 +01:00
sysfb.c
tboot.c iommu/vt-d: Correctly disable Intel IOMMU force on 2017-06-15 16:41:10 +02:00
tce_64.c
time.c Merge branch 'linus' into x86/apic, to resolve conflicts 2017-11-07 10:51:10 +01:00
tls.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
tls.h
topology.c
trace_clock.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
tracepoint.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
traps.c kmemcheck: remove annotations 2017-11-15 18:21:04 -08:00
tsc_msr.c x86/tsc: Set TSC_KNOWN_FREQ and TSC_RELIABLE flags on Intel Atom SoCs 2016-11-18 10:58:31 +01:00
tsc_sync.c Merge branch 'x86-timers-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip 2017-11-13 19:07:38 -08:00
tsc.c Merge branch 'x86-timers-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip 2017-11-13 19:07:38 -08:00
umip.c x86/umip: Fix insn_get_code_seg_params()'s return value 2017-11-23 20:17:59 +01:00
unwind_frame.c x86/unwind: Disable unwinder warnings on 32-bit 2017-10-10 12:49:49 +02:00
unwind_guess.c x86/unwind: Add the ORC unwinder 2017-07-26 13:18:20 +02:00
unwind_orc.c x86/unwind: Disable KASAN checking in the ORC unwinder 2017-11-08 10:21:49 +01:00
uprobes.c uprobes/x86: Emulate push insns for uprobe on x86 2017-12-11 18:42:11 +01:00
verify_cpu.S x86/boot: Annotate verify_cpu() as a callable function 2017-09-28 09:39:03 +02:00
vm86_32.c Merge branch 'linus' into x86/asm, to pick up fixes and resolve conflicts 2017-11-07 10:53:06 +01:00
vmlinux.lds.S License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
vsmp_64.c x86/apic: Remove unused callbacks 2017-09-25 20:51:58 +02:00
x86_init.c Merge branch 'x86-apic-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip 2017-11-13 18:29:23 -08:00