[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[win-pv-devel] [PATCH v2 0/4] Use per-vcpu event channel upcalls if available



My recent patch posted to xen-devel provides a new hypercall to register an
upcall vector per-vcpu for event channels. This series supplies the in-guest
code to make use of that.

Patch #1 adds code to re-populate the hypercall page on resume. This needs to
be done because the guest currently relies on a set effect of setting the
callback via to latch the shared info width in Xen. Re-populating the
hypercall page does the same thing so if use of the callback via is
dropped in future, in favour of just using the per-vcpu upcall, then the
shared info width will still get latched.

Patch #2 modifies the FDO handling code to intercept
IRP_MN_FILTER_RESOURCE_REQUIREMENTS so that extra interrupts can be
allocated, one per CPU.

Patch #3 updates to the latest Xen headers, including the new HVM op.

Patch #4 switches over to using the new per-vcpu upcalls, if the hypervisor
supports the new hypercall, and allows event channels to be bound to a vcpu
other than 0 if there is an upcall registered for that vcpu. The STORE
event channel is bound to the highest numbered vcpu as an example.

v2:
 - Added missing file to patch #1
 - Removed spurious INF file change from patch #2


_______________________________________________
win-pv-devel mailing list
win-pv-devel@xxxxxxxxxxxxxxxxxxxx
http://lists.xenproject.org/cgi-bin/mailman/listinfo/win-pv-devel


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.