lkml.org 
[lkml]   [2019]   [Sep]   [23]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 15/17] KVM: retpolines: x86: eliminate retpoline from vmx.c exit handlers
On Mon, Sep 23, 2019 at 11:57:57AM +0200, Paolo Bonzini wrote:
> On 23/09/19 11:31, Vitaly Kuznetsov wrote:
> > +#ifdef CONFIG_RETPOLINE
> > + if (exit_reason == EXIT_REASON_MSR_WRITE)
> > + return handle_wrmsr(vcpu);
> > + else if (exit_reason == EXIT_REASON_PREEMPTION_TIMER)
> > + return handle_preemption_timer(vcpu);
> > + else if (exit_reason == EXIT_REASON_PENDING_INTERRUPT)
> > + return handle_interrupt_window(vcpu);
> > + else if (exit_reason == EXIT_REASON_EXTERNAL_INTERRUPT)
> > + return handle_external_interrupt(vcpu);
> > + else if (exit_reason == EXIT_REASON_HLT)
> > + return handle_halt(vcpu);
> > + else if (exit_reason == EXIT_REASON_PAUSE_INSTRUCTION)
> > + return handle_pause(vcpu);
> > + else if (exit_reason == EXIT_REASON_MSR_READ)
> > + return handle_rdmsr(vcpu);
> > + else if (exit_reason == EXIT_REASON_CPUID)
> > + return handle_cpuid(vcpu);
> > + else if (exit_reason == EXIT_REASON_EPT_MISCONFIG)
> > + return handle_ept_misconfig(vcpu);
> > +#endif
> > return kvm_vmx_exit_handlers[exit_reason](vcpu);
>
> Most of these, while frequent, are already part of slow paths.
>
> I would keep only EXIT_REASON_MSR_WRITE, EXIT_REASON_PREEMPTION_TIMER,
> EXIT_REASON_EPT_MISCONFIG and add EXIT_REASON_IO_INSTRUCTION.

Intuition doesn't work great when it comes to CPU speculative
execution runtime. I can however run additional benchmarks to verify
your theory that keeping around frequent retpolines will still perform
ok.

> If you make kvm_vmx_exit_handlers const, can the compiler substitute for
> instance kvm_vmx_exit_handlers[EXIT_REASON_MSR_WRITE] with handle_wrmsr?
> Just thinking out loud, not sure if it's an improvement code-wise.

gcc gets right if you make it const, it calls kvm_emulate_wrmsr in
fact. However I don't think const will fly
with_vmx_hardware_setup()... in fact at runtime testing nested I just
got:

BUG: unable to handle page fault for address: ffffffffa00751e0
#PF: supervisor write access in kernel mode
#PF: error_code(0x0003) - permissions violation
PGD 2424067 P4D 2424067 PUD 2425063 PMD 7cc09067 PTE 80000000741cb161
Oops: 0003 [#1] SMP NOPTI
CPU: 1 PID: 4458 Comm: insmod Not tainted 5.3.0+ #1
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS ?-20190711_202441-buildvm-armv7-10.arm.fedoraproject.or4
RIP: 0010:nested_vmx_hardware_setup+0x29a/0x37a [kvm_intel]
Code: 41 ff c5 66 89 2c 85 20 92 0b a0 66 44 89 34 85 22 92 0b a0 49 ff c7 e9 e6 fe ff ff 44 89 2d 28 24 fc ff 48
RSP: 0018:ffffc90000257c18 EFLAGS: 00010246
RAX: ffffffffa001e0b0 RBX: ffffffffa0075140 RCX: 0000000000000000
RDX: ffff888078f60000 RSI: 0000000000002401 RDI: 0000000000000018
RBP: 0000000000006c08 R08: 0000000000001000 R09: 000000000007ffdc
R10: 0000000000000000 R11: 0000000000000001 R12: 0000000000006c08
R13: 0000000000000017 R14: 0000000000000268 R15: 0000000000000018
FS: 00007f7fb7ef0b80(0000) GS:ffff88807da40000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: ffffffffa00751e0 CR3: 0000000079620001 CR4: 0000000000160ee0
Call Trace:
hardware_setup+0x4df/0x5b2 [kvm_intel]
kvm_arch_hardware_setup+0x2f/0x27b [kvm_intel]
kvm_init+0x5d/0x26d [kvm_intel]

\
 
 \ /
  Last update: 2019-09-23 21:05    [W:0.114 / U:22.508 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site