Red Hat Bugzilla – Bug 667359
forward port xen pvops changes for evtchn
Last modified: 2011-08-30 03:26:00 EDT
The following commits will fix various bugs in event channels. They could also fix problems in migration since many of them affect the suspend/resume path. 06b0362 (xen: Fix find_unbound_irq in presence of ioapic irqs., 2010-05-14) 9fa90aa (xen: use dynamic_irq_init_keep_chip_data, 2010-08-19) aaca496 (xen: use percpu interrupts for IPIs and VIRQs, 2010-08-20) dffe2e1 (xen: handle events as edge-triggered, 2010-08-20) e041956 (xen/events: use locked set|clear_bit() for cpu_evtchn_mask, 2010-11-16) 1c6969e (xen/evtchn: clear secondary CPUs' cpu_evtchn_mask[] after restore, 2010-11-16) 6903591 (xen: events: do not unmask event channels on resume, 2010-11-01) a52521f (xen: set up IRQ before binding virq to evtchn, 2010-09-22) b0097ad (xen: ensure that all event channels start off bound to VCPU 0, 2010-10-08) 3b32f57 (xen: statically initialize cpu_evtchn_mask_p, 2009-08-13) 77dff1c (x86: xen: Sanitise sparse_irq handling, 2010-09-29)
This request was evaluated by Red Hat Product Management for inclusion in a Red Hat Enterprise Linux maintenance release. Product Management has requested further review of this request by Red Hat Engineering, for potential inclusion in a Red Hat Enterprise Linux Update release for currently deployed products. This request is not yet committed for inclusion in an Update release.
Backporting these patches unfortunately don't fix the migration issues, in fact they got worse. On two systems where I could migrate one direction, but not the other before the backports, with them I wasn't able to migrate at all. I have found that using 2.6.37-2.fc15.x86_64 I'm able to migrate in both directions. So it looks like there are some other patches necessary that need to be found.
This bug can be considered synching specifically drivers/xen/events.c Already have: aaca496 (xen: use percpu interrupts for IPIs and VIRQs, 2010-08-20) dffe2e1 (xen: handle events as edge-triggered, 2010-08-20) 06b0362 (xen: Fix find_unbound_irq in presence of ioapic irqs., 2010-05-14) Skipped: 77dff1c (x86: xen: Sanitise sparse_irq handling, 2010-09-29) * would require general kernel backports outside of xen files 6903591 (xen: events: do not unmask event channels on resume, 2010-11-01) * only first 2 hunks would be applicable, but those are for performance, so they're not worth the risk cherry-picked the remaining cleanly: 3b32f57 (xen: statically initialize cpu_evtchn_mask_p, 2009-08-13) 9fa90aa (xen: use dynamic_irq_init_keep_chip_data, 2010-08-19) a52521f (xen: set up IRQ before binding virq to evtchn, 2010-09-22) b0097ad (xen: ensure that all event channels start off bound to VCPU 0, 2010-10-08) e041956 (xen/events: use locked set|clear_bit() for cpu_evtchn_mask, 2010-11-16) 1c6969e (xen/evtchn: clear secondary CPUs' cpu_evtchn_mask[] after restore, 2010-11-16) Also cherry-picked these, which weren't originally listed b21ddbf5 (xen: dynamically allocate irq & event structures, 2010-06-07) cb52e6d9 (xen: improvements to VIRQ_DEBUG output, 2010-10-15) So far only compile tested.
Also added cherry-pick 3588fe2e (xen/events: change to using fasteoi, 2010-08-27) and synched-up with missing pieces of 38e20b07 (x86/xen: event channels delivery on HVM, 2010-05-14)
Patch(es) available on kernel-2.6.32-112.el6
An advisory has been issued which should help the problem described in this bug report. This report is therefore being closed with a resolution of ERRATA. For more information on therefore solution and/or where to find the updated files, please follow the link below. You may reopen this bug report if the solution does not work for you. http://rhn.redhat.com/errata/RHSA-2011-0542.html