[PATCH] xen,tools: pincpu use vcpu and cpumap_t
The following patch updates the dom0 pincpu operation to read the VCPU
value from the xend interface rather than hard-coding the exec_domain to
0. This prevented pinning VCPUS other than 0 to a particular cpu. The
pincpu cpu argument is replaced with cpumap, a u64 bitmap representing
which CPUs a VCPU can use.
I added the number of VCPUS to the main xm list output and also included
a new sub-option to xm list to display the VCPU to CPU mapping. While
working on the pincpu code, I fixed an out-of-bounds indexing for the
pincpu operation that wasn't previously exposed since the
vcpu/exec_domain value was hard-coded to 0.
Signed-off-by: Ryan Harper <ryanh@us.ibm.com>
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
[PATCH] [PATCH] Increase size of sxpr parser input buffer
Increase the size of the sxpr parser input buffer from 1024 characters, which
isn't long enough to hold larger domain configurations (xfrd transmits the
entire domain config sxpr as a single atom inside of another sxpr).
Signed-off-by: Jody Belka <knew (dash) xen (at) pimb (dot) org>
Clean up IO-APIC handling in domain0, and the hypercall interface
exported by Xen (inc. removal of PCI-related calls). This hopefully
fixes acpi=off.
Signed-off-by: Keir Fraser <keir@xensource.com>
Avoid field duplication between vcpu_guest_context and arch_exec_domain
structures. The latter now includes the former as a sub-field.
Signed-off-by: Keir Fraser <keir@xensource.com>
[PATCH] [PATCH] Assorted VMX patches
This patch contains the following VMX patches:
- ioemu: Update the address when doing a stosb/stosw/stosl. Without this patch
stosb writes req->count times at the same location req->addr. Not exactly
the intended semantics.
- xc_ptrace.c: Use the page_array to refer to partition memory. This allows
you to use gdbserver on a VMX partition (and presumably partitions that use
shadow page tables).
- dom0_ops.c: Preserve ECF_VMX_GUEST flag for gdbserver. Without it you
cannot (re)set the VMX domain state.
- vmx.c: Added support for lmsw. Unmodified FreeBSD uses this when running
inside a VMX partition.
Signed-Off-By: Leendert van Doorn <leendert@watson.ibm.com>
===== tools/ioemu/iodev/cpu.cc 1.8 vs edited =====
Remove memory and cpu parameters from DOM0_CREATEDOMAIN, and remove
DOM0_SETINITIALMEM. You can get the same effect via PINCPU,
SETMAXMEM, and do_mem_op(increase_reservation).
Signed-off-by: Keir Fraser <keir@xensource.com>
Merge freefall.cl.cam.ac.uk:/auto/anfs/scratch/font/sd386/xen-2.0-testing.bk
into freefall.cl.cam.ac.uk:/auto/groups/xeno/users/iap10/xeno-clone/xen-2.0-testing.bk
Merge freefall.cl.cam.ac.uk:/auto/groups/xeno/users/iap10/xeno-clone/xen-2.0-testing.bk
into freefall.cl.cam.ac.uk:/auto/groups/xeno/users/iap10/xeno-clone/xen-unstable.bk
Merge freefall.cl.cam.ac.uk:/auto/groups/xeno/users/iap10/xeno-clone/xen-2.0-testing.bk
into freefall.cl.cam.ac.uk:/auto/groups/xeno/users/iap10/xeno-clone/xen-unstable.bk
Merge freefall.cl.cam.ac.uk:/auto/groups/xeno/users/iap10/xeno-clone/xen-2.0-testing.bk
into freefall.cl.cam.ac.uk:/auto/groups/xeno/users/iap10/xeno-clone/xen-unstable.bk
Fix curldebug function in xendomain.c of xfrd. Function takes argument
"size_t buf_num" and uses it as the field width to print, that has to be
an int. While size_t on 32 bit systems is 4bytes (or size of an int) it
is 8bytes on 64bit systems (not size of an int). So this function should
just use "int buf_num". This function is not used in the code anywhere.
This will fix compilation on x86-64.
Signed-off-by: Jerone Young <jyoung5@us.ibm.com>
With -D_FORTIFY_SOURCE=2 (which is used in the Fedora buildroot), gcc has certain defines for functions like read() to check that things are done right.
This trips up these function pointers, and gcc becomes unhappy.
Adding parens around the function prevents gcc from expanding read to the macro it is defined to internally, and makes things compile again.
Signed-off-by: Rik van Riel <riel@redhat.com>
[PATCH] Install libraries into libdir
Attached makes it so that libxc and libxutil get installed
into /usr/lib64 instead of /usr/lib on x86_64.
Signed-off-by: Jeremy Katz <katzj@redhat.com>
===== Config.mk 1.3 vs edited =====
[PATCH] Make non-native python path handling a
> > Although I don't really agree with the installing of the
> > python bits to the non-native paths, I can see where it's
> > useful as long as the module ABI doesn't change if you're
> > trying to distribute binary tarballs. But with the current
> > handling it adds module lookup costs for the native path case
> > (which would be the norm) as well as prevents using
> > PYTHONPATH to look to a different location for the modules.
> >
> > The attached patch changes things so that at first, the xen
> > module is attempted to be imported from the stock paths. If
> > we get an ImportError exception, then it adds /usr/lib/python
> > to the end of the python path and imports again.
>
> Do you prefer this to just unconditionally addding /usr/lib/python to
> the end of the path rather than the front?
Slightly, but just because having /usr/lib/python in my sys.path is
ugly :-) I can look over it, though... patch doing that attached
Signed-off-by: Jeremy Katz <katzj@redhat.com>
===== tools/misc/netfix 1.13 vs edited =====
[PATCH] [PATCH] Man page path
The FHS has said that manpages go in /usr/share/man instead of /usr/man
since FHS 2.0 which dates back to 1997 or so, if I remember correctly.
Having the makefiles drop things in the right place makes the packages a
little cleaner.
Signed-off-by: Jeremy Katz <katzj@redhat.com>
===== tools/xentrace/Makefile 1.24 vs edited =====
[PATCH] stream fixes for migration
I've attached a patch for libxutil/libxc. This fixes one of the hangs =
I've seen during migrations. It applies against 2.0 and 2.0-testing.
Changes:
* Encountering EOF or error when xfrd reads from stream could cause an =
infinite loop.
* Cleaned up the closing of streams.
* Fixed several memory leaks.
Signed-off-by: Charles Coffing <ccoffing@novell.com>
