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

x86, vmi: Mark VMI deprecated and schedule it for removal

Add text in feature-removal.txt indicating that VMI will be removed in
the 2.6.37 timeframe.

Signed-off-by: Alok N Kataria <akataria@vmware.com>
Acked-by: Chris Wright <chrisw@sous-sol.org>
LKML-Reference: <1254193238.13456.48.camel@ank32.eng.vmware.com>
[ removed a bogus Kconfig change, marked (DEPRECATED) in Kconfig ]
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

authored by

Alok Kataria and committed by
Ingo Molnar
d0153ca3 f579bbcd

+41 -2
+30
Documentation/feature-removal-schedule.txt
··· 451 451 will also allow making ALSA OSS emulation independent of 452 452 sound_core. The dependency will be broken then too. 453 453 Who: Tejun Heo <tj@kernel.org> 454 + 455 + ---------------------------- 456 + 457 + What: Support for VMware's guest paravirtuliazation technique [VMI] will be 458 + dropped. 459 + When: 2.6.37 or earlier. 460 + Why: With the recent innovations in CPU hardware acceleration technologies 461 + from Intel and AMD, VMware ran a few experiments to compare these 462 + techniques to guest paravirtualization technique on VMware's platform. 463 + These hardware assisted virtualization techniques have outperformed the 464 + performance benefits provided by VMI in most of the workloads. VMware 465 + expects that these hardware features will be ubiquitous in a couple of 466 + years, as a result, VMware has started a phased retirement of this 467 + feature from the hypervisor. We will be removing this feature from the 468 + Kernel too. Right now we are targeting 2.6.37 but can retire earlier if 469 + technical reasons (read opportunity to remove major chunk of pvops) 470 + arise. 471 + 472 + Please note that VMI has always been an optimization and non-VMI kernels 473 + still work fine on VMware's platform. 474 + Latest versions of VMware's product which support VMI are, 475 + Workstation 7.0 and VSphere 4.0 on ESX side, future maintainence 476 + releases for these products will continue supporting VMI. 477 + 478 + For more details about VMI retirement take a look at this, 479 + http://blogs.vmware.com/guestosguide/2009/09/vmi-retirement.html 480 + 481 + Who: Alok N Kataria <akataria@vmware.com> 482 + 483 + ----------------------------
+10 -1
arch/x86/Kconfig
··· 491 491 source "arch/x86/xen/Kconfig" 492 492 493 493 config VMI 494 - bool "VMI Guest support" 494 + bool "VMI Guest support (DEPRECATED)" 495 495 select PARAVIRT 496 496 depends on X86_32 497 497 ---help--- ··· 499 499 (it could be used by other hypervisors in theory too, but is not 500 500 at the moment), by linking the kernel to a GPL-ed ROM module 501 501 provided by the hypervisor. 502 + 503 + As of September 2009, VMware has started a phased retirement 504 + of this feature from VMware's products. Please see 505 + feature-removal-schedule.txt for details. If you are 506 + planning to enable this option, please note that you cannot 507 + live migrate a VMI enabled VM to a future VMware product, 508 + which doesn't support VMI. So if you expect your kernel to 509 + seamlessly migrate to newer VMware products, keep this 510 + disabled. 502 511 503 512 config KVM_CLOCK 504 513 bool "KVM paravirtualized clock"
+1 -1
arch/x86/kernel/vmi_32.c
··· 648 648 649 649 pv_info.paravirt_enabled = 1; 650 650 pv_info.kernel_rpl = kernel_cs & SEGMENT_RPL_MASK; 651 - pv_info.name = "vmi"; 651 + pv_info.name = "vmi [deprecated]"; 652 652 653 653 pv_init_ops.patch = vmi_patch; 654 654