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

Re: [Xen-devel] Xen 4.2 TODO List Update

2012/1/25 Ian Campbell <Ian.Campbell@xxxxxxxxxx>:
> On Wed, 2012-01-25 at 14:33 +0000, Roger Pau Monnà wrote:
>> 2012/1/23 Ian Campbell <Ian.Campbell@xxxxxxxxxx>:
>> > Newly updated list follows. Please send me corrections (especially
>> > "done"). I've stopped CCing everyone, since I guess it is mostly spam to
>> > the majority.
>> >
>> > hypervisor, blockers:
>> >
>> > Â Â Â* round-up of the closing of the security hole in MSI-X
>> > Â Â Â Âpassthrough (uniformly - i.e. even for Dom0 - disallowing write
>> > Â Â Â Âaccess to MSI-X table pages). (Jan Beulich -- more fixes
>> > Â Â Â Ârequired than first thought, patches posted)
>> > Â Â Â* domctls / sysctls set up to modify scheduler parameters, like
>> > Â Â Â Âthe credit1 timeslice and schedule rate. (George Dunlap)
>> > Â Â Â* get the interface changes for sharing/paging/mem-events done and
>> > Â Â Â Âdusted so that 4.2 is a stable API that we hold to. (Tim Deegan,
>> > Â Â Â ÂAndres Lagar-Cavilla et al)
>> > Â Â Â Â Â Â Â* mem event ring management posted, seems close to going
>> > Â Â Â Â Â Â Â Âin.
>> > Â Â Â Â Â Â Â* sharing patches posted
>> >
>> > tools, blockers:
>> >
>> > Â Â Â* libxl stable API -- we would like 4.2 to define a stable API
>> > Â Â Â Âwhich downstream's can start to rely on not changing. Aspects of
>> > Â Â Â Âthis are:
>> > Â Â Â Â Â Â Â* event handling (Ian Jackson, posted several rounds,
>> > Â Â Â Â Â Â Â Ânearing completion?)
>> > Â Â Â Â Â Â Â* drop libxl_device_model_info (move bits to build_info or
>> > Â Â Â Â Â Â Â Âelsewhere as appropriate) (Ian Campbell, first RFC sent)
>> > Â Â Â Â Â Â Â* add libxl_defbool and generally try and arrange that
>> > Â Â Â Â Â Â Â Âmemset(foo,0,...) requests the defaults (Ian Campbell,
>> > Â Â Â Â Â Â Â Âfirst RFC sent)
>> > Â Â Â Â Â Â Â* topologyinfo datastructure should be a list of tuples,
>> > Â Â Â Â Â Â Â Ânot a tuple of lists. (nobody currently looking at this,
>> > Â Â Â Â Â Â Â Ânot 100% sure this makes sense, could possibly defer and
>> > Â Â Â Â Â Â Â Âchange after 4.2 in a compatible way)
>> > Â Â Â* xl to use json for machine readable output instead of sexp by
>> > Â Â Â Âdefault (Ian Campbell to revisit existing patch)
>> > Â Â Â* xl support for vcpu pinning (Dario Faggioli)
>> > Â Â Â* xl feature parity with xend wrt driver domain support (George
>> > Â Â Â ÂDunlap)
>> > Â Â Â* Integrate qemu+seabios upstream into the build (patches
>> > Â Â Â Âreposted, pending). No change in default qemu for 4.2.
>> > Â Â Â* More formally deprecate xm/xend. Manpage patches already in
>> > Â Â Â Âtree. Needs release noting and communication around -rc1 to
>> > Â Â Â Âremind people to test xl.
>> >
>> > hypervisor, nice to have:
>> >
>> > Â Â Â* solid implementation of sharing/paging/mem-events (using work
>> > Â Â Â Âqueues) (Tim Deegan, Olaf Herring et al)
>> > Â Â Â* A long standing issue is a fully synchronized p2m (locking
>> > Â Â Â Âlookups) (Andres Lagar-Cavilla)
>> > Â Â Â* NUMA improvement: domain affinity consistent with cpupool
>> > Â Â Â Âmembership (Dario Faggioli, Jeurgen Gross -- patch posted)
>> >
>> > tools, nice to have:
>> >
>> > Â Â Â* Hotplug script stuff -- internal to libxl (I think, therefore I
>> > Â Â Â Âdidn't put this under stable API above) but still good to have
>> > Â Â Â Âfor 4.2? Roger Pau Monet was looking at this but its looking
>> > Â Â Â Âlike a big can-o-worms. (discussion on-going)
>> > Â Â Â* Block script support -- follows on from hotplug script (Roger
>> > Â Â Â ÂPau Monet)
>> > Â Â Â* libyajl v2 support (patch posted by Roger Pau Monet, blocked on
>> > Â Â Â Âautoconf?)
>> > Â Â Â* Configure/control paging via xl/libxl (Olaf Herring)
>> > Â Â Â* Upstream qemu feature patches:
>> > Â Â Â Â Â Â Â* Upstream qemu PCI passthrough support (Anthony Perard)
>> > Â Â Â Â Â Â Â* Upstream qemu save restore (Anthony Perard)
>> > Â Â Â* Nested-virtualisation (currently should be marked
>> > Â Â Â Âexperimental,likely to release that way? Consider nested-svm
>> > Â Â Â Âseparate to nested-vmx. Nested-svm is in better shape)
>> Just a random thought, but I find it quite annoying to need latex to
>> compile the documentation. I think it will be wise to create a new
>> Makefile target, like man-pages and install-man-pages to allow the
>> user to build and install man pages only (or don't abort docs build if
>> latex is not found). Personally I don't have latex on my server, and I
>> don't want to install it, but I would like to have the man pages when
>> building Xen from source.
> "make -C docs man-pages" will do what you want.

Thanks, I've already done this to build them on my system, but I think
we should provide some easy way for users without latex to build and
install the man pages.

> I must admit I thought latex was optional and that "make docs" would
> simply skip those docs if it wasn't installed -- that's what the "if
> which $(TOOL)" construct used in there is (supposed to be) doing.

If latex is not found, the compilation is aborted:

# make docs
sh ./docs/check_pkgs && make -C docs install || true
= WARNING: Package 'latex' is required
=          to build Xen documentation

I think make docs should just skip latex if not found.

> Ian.
>> > Tools, need to decide if pre- or post-4.2 feature:
>> >
>> > Â Â Â* Autoconf (Roger Pau Monet posted a patch)
>> >
>> >
>> > _______________________________________________
>> > Xen-devel mailing list
>> > Xen-devel@xxxxxxxxxxxxxxxxxxx
>> > http://lists.xensource.com/xen-devel

Xen-devel mailing list



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