Clean up the domain_page.h interfaces. One common header file
<xen/domain_page.h> and map_domain_mem() -> map_domain_page(), takes
a pfn rather than a paddr.
Signed-off-by: Keir Fraser <keir@xensource.com>
Give the map_dom_mem_cache a better name, and fix a bug in
unmap_domain_mem_with_cache() [it shouldn't actually unmap the page!].
The bug was spotted by George Dunlap.
Signed-off-by: Keir Fraser <keir@xensource.com>
[PATCH] vmx-pit-reset-vector.patch
If the guest resets irq_base in the PIC, we should reset the vector
in the hypervisor as well.
Signed-off-by: Arun Sharma <arun.sharma@intel.com>
cpu_gdt_init() could allocate a frame array one element too small. Now
it is fixed size and guaranteed big enough. Spotted by George Dunlap.
Signed-off-by: Keir Fraser <keir@xensource.com>
Remove strrchr/strnlen/memscan from x86-specific string.h. They are not
used so not worth optimising (strnlen is used only in vsprintf which is
not performance critical).
Signed-off-by: Keir Fraser <keir@xensource.com>
xc_private.h, xc_linux_build.c, Makefile:
Add support for loading ``bin'' format images, as used by ReactOS.
Move image probing/parsing/loading code out of domain builder to allow
multiple image formats without having to duplicate the domain building
code.
xc_load_elf.c, xc_load_bin.c:
new file
xc_vmx_build.c:
Cleanup.
Signed-Off-By: Ge van Geldorp <gvg@reactos.com>
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
Attached patch fixes FP initialization bug with x86-64 xenlinux.=20
With this patch, # of testcase failures with LTP is < 5 (out of 800),
almost same results as the native x86-64 Linux. The other two patches
for LTP are coming soon; with these, all the testcases will pass (except
the ones that fail on the native x86-64 Linux).
Signed-off-by: Jun Nakajima <jun.nakajima@intel.com>
More console cleanups. The console ring now overwrites its tail when
it fills, rather than stop recording output. The guest serial output
is only allowed to half-fill the serial output buffer, and now
respects hypercall preemption requests.
Signed-off-by: Keir Fraser <keir@xensource.com>
ns16550 interrupt handler should not return until the IIR indicates
no pending interrupts. Also a few generic serial driver cleanups.
Signed-off-by: Keir Fraser <keir@xensource.com>
Add a manual 'sync_console' boot parameter that forces synchronous
console output. May be useful for systems that crash unexpectedly
and lose the last chunk of console output.
Signed-off-by: Keir Fraser <keir@xensource.com>
Rationalise debugger stuff in the tree.
Within Xen itself, remove PDB references and fold into domu_debug.
When domu_debug hits a breakpoint, it pauses all vcpus of the domain
and raises VIRQ_DEBUGGER (not VIRQ_PDB). The debugger in domain0 can
either poll pause status or bind to VIRQ_DEBUGGER. Before debugging
a breakpointed domain the debugger should issue PAUSEDOMAIN itself, or
it is not guaranteed to see paused state of the domain!
Outside Xen there's a new dir tools/debugger, containing gdb, pdb and
a new library libxendebug (used by pdb). I also think the ptrace
functions in libxc also ought to be moved out: either to libxendebug
or to a new library libxenptrace.
Signed-off-by: Keir Fraser <keir@xensource.com>
[patch] libxc x86-64
I've redone the patch to add domU launching support to libxc for x86-64.
Signed-off-by: Jerone Young <jyoung5@us.ibm.com>
Signed-off-by: ian@xensource.com
[PATCH] missing put_domain in DOM0_IOPORT_PERMISSION
There appears to be a missing put_domain() in the failure case. The code
doesn't seem to be used by anyone, but anyway...
Signed-off-by: John Levon <levon@movementarian.org>
Signed-off-by: ian@xensource.com
Event-channel CPU affinity. Currently all event channels still bind to
VCPU#0 at start of day, and have their binding automatically changed
when bound to a VIRQ or IPI source. XenLinux maintains a per-cpu
evtchn mask denoting which event channels are bound to each cpu.
Todo: Allow guests to change binding of of non-ipi and non-virq evtchns.
Signed-off-by: Keir Fraser <keir@xensource.com>
Ensure that _PAGE_GLOBAL bit is never set in any pagetable if the CPU
does not support PGE (Page Global Extension).
Signed-off-by: Keir Fraser <keir@xensource.com>
sched.h:
Bring back includes of xen/time.h and xen/ac_timer.h since we use types
defined in those.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
