kernel/git/stable/linux.git/drivers/xen/events.c, branch master

xen/events: move drivers/xen/events.c into drivers/xen/events/

2014-01-06T15:07:38+00:00

events.c will be split into multiple files so move it into its own directory. Signed-off-by: David Vrabel Reviewed-by: Konrad Rzeszutek Wilk Reviewed-by: Boris Ostrovsky

xen/events: replace raw bit ops with functions

2014-01-06T15:07:36+00:00

In preparation for adding event channel port ops, use set_evtchn() instead of sync_set_bit(). Signed-off-by: Wei Liu Signed-off-by: David Vrabel Reviewed-by: Konrad Rzeszutek Wilk Reviewed-by: Boris Ostrovsky

xen/events: introduce test_and_set_mask()

2014-01-06T15:07:35+00:00

In preparation for adding event channel port ops, add test_and_set_mask(). Signed-off-by: Wei Liu Signed-off-by: David Vrabel Reviewed-by: Konrad Rzeszutek Wilk Reviewed-by: Boris Ostrovsky

xen/events: remove unnecessary init_evtchn_cpu_bindings()

2014-01-06T15:07:33+00:00

Because the guest-side binding of an event to a VCPU (i.e., setting the local per-cpu masks) is always explicitly done after an event channel is bound to a port, there is no need to initialize all possible events as bound to VCPU 0 at start of day or after a resume. Signed-off-by: David Vrabel Reviewed-by: Konrad Rzeszutek Wilk Reviewed-by: Boris Ostrovsky

xen/events: refactor retrigger_dynirq() and resend_irq_on_evtchn()

2014-01-06T15:07:32+00:00

These two function did the same thing with different parameters, put the common bits in retrigger_evtchn(). This changes the return value of resend_irq_on_evtchn() but the only caller (in arch/ia64/xen/irq_xen.c) ignored the return value so this is fine. Signed-off-by: David Vrabel Reviewed-by: Konrad Rzeszutek Wilk Reviewed-by: Boris Ostrovsky

Merge tag 'stable/for-linus-3.12-rc0-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip

2013-09-05T00:45:39+00:00

Pull Xen updates from Konrad Rzeszutek Wilk: "A couple of features and a ton of bug-fixes. There is also some maintership changes. Jeremy is enjoying the full-time work at the startup and as much as he would love to help - he can't find the time. I have a bunch of other things that I promised to work on - paravirt diet, get SWIOTLB working everywhere, etc, but haven't been able to find the time. As such both David Vrabel and Boris Ostrovsky have graciously volunteered to help with the maintership role. They will keep the lid on regressions, bug-fixes, etc. I will be in the background to help - but eventually there will be less of me doing the Xen GIT pulls and more of them. Stefano is still doing the ARM/ARM64 and will continue on doing so. Features: - Xen Trusted Platform Module (TPM) frontend driver - with the backend in MiniOS. - Scalability improvements in event channel. - Two extra Xen co-maintainers (David, Boris) and one going away (Jeremy) Bug-fixes: - Make the 1:1 mapping work during early bootup on selective regions. - Add scratch page to balloon driver to deal with unexpected code still holding on stale pages. - Allow NMIs on PV guests (64-bit only) - Remove unnecessary TLB flush in M2P code. - Fixes duplicate callbacks in Xen granttable code. - Fixes in PRIVCMD_MMAPBATCH ioctls to allow retries - Fix for events being lost due to rescheduling on different VCPUs. - More documentation" * tag 'stable/for-linus-3.12-rc0-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip: (23 commits) hvc_xen: Remove unnecessary __GFP_ZERO from kzalloc drivers/xen-tpmfront: Fix compile issue with missing option. xen/balloon: don't set P2M entry for auto translated guest xen/evtchn: double free on error Xen: Fix retry calls into PRIVCMD_MMAPBATCH*. xen/pvhvm: Initialize xen panic handler for PVHVM guests xen/m2p: use GNTTABOP_unmap_and_replace to reinstate the original mapping xen: fix ARM build after 6efa20e4 MAINTAINERS: Remove Jeremy from the Xen subsystem. xen/events: document behaviour when scanning the start word for events x86/xen: during early setup, only 1:1 map the ISA region x86/xen: disable premption when enabling local irqs swiotlb-xen: replace dma_length with sg_dma_len() macro swiotlb: replace dma_length with sg_dma_len() macro xen/balloon: set a mapping for ballooned out pages xen/evtchn: improve scalability by using per-user locks xen/p2m: avoid unneccesary TLB flush in m2p_remove_override() MAINTAINERS: Add in two extra co-maintainers of the Xen tree. MAINTAINERS: Update the Xen subsystem's with proper mailing list. xen: replace strict_strtoul() with kstrtoul() ...

xen: fix ARM build after 6efa20e4

2013-08-20T14:23:09+00:00

The following commit: commit 6efa20e49b9cb1db1ab66870cc37323474a75a13 Author: Konrad Rzeszutek Wilk Date: Fri Jul 19 11:51:31 2013 -0400 xen: Support 64-bit PV guest receiving NMIs breaks the Xen ARM build: CC drivers/xen/events.o drivers/xen/events.c: In function 'xen_send_IPI_one': drivers/xen/events.c:1218:6: error: 'XEN_NMI_VECTOR' undeclared (first use in this function) Simply ifdef the undeclared symbol in the code. Signed-off-by: Stefano Stabellini

xen/events: document behaviour when scanning the start word for events

2013-08-20T14:15:28+00:00

The original comment on the scanning of the start word on the 2nd pass did not reflect the actual behaviour (the code was incorrectly masking bit_idx instead of the pending word itself). The documented behaviour is not actually required since if event were pending in the MSBs, they would be immediately scanned anyway as we go through the loop again. Update the documentation to reflect this (instead of trying to change the behaviour). Signed-off-by: David Vrabel Signed-off-by: Konrad Rzeszutek Wilk Reviewed-by: Boris Ostrovsky

xen/events: mask events when changing their VCPU binding

2013-08-20T14:13:04+00:00

When a event is being bound to a VCPU there is a window between the EVTCHNOP_bind_vpcu call and the adjustment of the local per-cpu masks where an event may be lost. The hypervisor upcalls the new VCPU but the kernel thinks that event is still bound to the old VCPU and ignores it. There is even a problem when the event is being bound to the same VCPU as there is a small window beween the clear_bit() and set_bit() calls in bind_evtchn_to_cpu(). When scanning for pending events, the kernel may read the bit when it is momentarily clear and ignore the event. Avoid this by masking the event during the whole bind operation. Signed-off-by: David Vrabel Signed-off-by: Konrad Rzeszutek Wilk Reviewed-by: Jan Beulich CC: stable@vger.kernel.org

xen/events: initialize local per-cpu mask for all possible events

2013-08-20T14:13:02+00:00

The sizeof() argument in init_evtchn_cpu_bindings() is incorrect resulting in only the first 64 (or 32 in 32-bit guests) ports having their bindings being initialized to VCPU 0. In most cases this does not cause a problem as request_irq() will set the irq affinity which will set the correct local per-cpu mask. However, if the request_irq() is called on a VCPU other than 0, there is a window between the unmasking of the event and the affinity being set were an event may be lost because it is not locally unmasked on any VCPU. If request_irq() is called on VCPU 0 then local irqs are disabled during the window and the race does not occur. Fix this by initializing all NR_EVENT_CHANNEL bits in the local per-cpu masks. Signed-off-by: David Vrabel Signed-off-by: Konrad Rzeszutek Wilk CC: stable@vger.kernel.org