Linux kernel mirror (for testing) git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel os linux
1
fork

Configure Feed

Select the types of activity you want to include in your feed.

x86/xen/time: prefer tsc as clocksource when it is invariant

Kvm elects to use tsc instead of kvm-clock when it can detect that the
TSC is invariant.

(As of commit 7539b174aef4 ("x86: kvmguest: use TSC clocksource if
invariant TSC is exposed")).

Notable cloud vendors[1] and performance engineers[2] recommend that Xen
users preferentially select tsc over xen-clocksource due the performance
penalty incurred by the latter. These articles are persuasive and
tailored to specific use cases. In order to understand the tradeoffs
around this choice more fully, this author had to reference the
documented[3] complexities around the Xen configuration, as well as the
kernel's clocksource selection algorithm. Many users may not attempt
this to correctly configure the right clock source in their guest.

The approach taken in the kvm-clock module spares users this confusion,
where possible.

Both the Intel SDM[4] and the Xen tsc documentation explain that marking
a tsc as invariant means that it should be considered stable by the OS
and is elibile to be used as a wall clock source.

In order to obtain better out-of-the-box performance, and reduce the
need for user tuning, follow kvm's approach and decrease the xen clock
rating so that tsc is preferable, if it is invariant, stable, and the
tsc will never be emulated.

[1] https://aws.amazon.com/premiumsupport/knowledge-center/manage-ec2-linux-clock-source/
[2] https://www.brendangregg.com/blog/2021-09-26/the-speed-of-time.html
[3] https://xenbits.xen.org/docs/unstable/man/xen-tscmode.7.html
[4] Intel 64 and IA-32 Architectures Sofware Developer's Manual Volume
3b: System Programming Guide, Part 2, Section 17.17.1, Invariant TSC

Signed-off-by: Krister Johansen <kjlx@templeofstupid.com>
Code-reviewed-by: David Reaver <me@davidreaver.com>
Reviewed-by: Juergen Gross <jgross@suse.com>
Link: https://lore.kernel.org/r/20221216162118.GB2633@templeofstupid.com
Signed-off-by: Juergen Gross <jgross@suse.com>

authored by

Krister Johansen and committed by
Juergen Gross
caea091e f697cb00

+37 -1
+37 -1
arch/x86/xen/time.c
··· 474 474 xen_clocksource.vdso_clock_mode = VDSO_CLOCKMODE_PVCLOCK; 475 475 } 476 476 477 + /* 478 + * Check if it is possible to safely use the tsc as a clocksource. This is 479 + * only true if the hypervisor notifies the guest that its tsc is invariant, 480 + * the tsc is stable, and the tsc instruction will never be emulated. 481 + */ 482 + static int __init xen_tsc_safe_clocksource(void) 483 + { 484 + u32 eax, ebx, ecx, edx; 485 + 486 + if (!(boot_cpu_has(X86_FEATURE_CONSTANT_TSC))) 487 + return 0; 488 + 489 + if (!(boot_cpu_has(X86_FEATURE_NONSTOP_TSC))) 490 + return 0; 491 + 492 + if (check_tsc_unstable()) 493 + return 0; 494 + 495 + /* Leaf 4, sub-leaf 0 (0x40000x03) */ 496 + cpuid_count(xen_cpuid_base() + 3, 0, &eax, &ebx, &ecx, &edx); 497 + 498 + /* tsc_mode = no_emulate (2) */ 499 + if (ebx != 2) 500 + return 0; 501 + 502 + return 1; 503 + } 504 + 477 505 static void __init xen_time_init(void) 478 506 { 479 507 struct pvclock_vcpu_time_info *pvti; 480 508 int cpu = smp_processor_id(); 481 509 struct timespec64 tp; 482 510 483 - /* As Dom0 is never moved, no penalty on using TSC there */ 511 + /* 512 + * As Dom0 is never moved, no penalty on using TSC there. 513 + * 514 + * If it is possible for the guest to determine that the tsc is a safe 515 + * clocksource, then set xen_clocksource rating below that of the tsc 516 + * so that the system prefers tsc instead. 517 + */ 484 518 if (xen_initial_domain()) 485 519 xen_clocksource.rating = 275; 520 + else if (xen_tsc_safe_clocksource()) 521 + xen_clocksource.rating = 299; 486 522 487 523 clocksource_register_hz(&xen_clocksource, NSEC_PER_SEC); 488 524