Keir Fraser [Wed, 28 May 2008 08:30:48 +0000 (09:30 +0100)]
stubdom: make the build more generic by moving as many stubdom parts
into stubdom/ as possible. That also permits to build all of
ioemu, c and caml stubdoms at the same time.
Signed-off-by: Samuel Thibault <samuel.thibault@eu.citrix.com>
Keir Fraser [Thu, 8 May 2008 12:14:27 +0000 (13:14 +0100)]
minios: fix and enforce block_domain atomicity
To ensure that the timer event is not lost, block_domain requires that
event delivery is disabled. SCHEDOP_block re-enables them, so for
coherency (and fixing a bug actually), we should re-disable them
after. Also, make sure that the caller disabled them.
Signed-off-by: Samuel Thibault <samuel.thibault@eu.citrix.com>
Keir Fraser [Thu, 8 May 2008 10:53:39 +0000 (11:53 +0100)]
minios: clear the event before calling the handler since (especially
in the SMP case) the handler may make another domain send an event
again, and that must not be lost.
Signed-off-by: Samuel Thibault <samuel.thibault@eu.citrix.com>
Keir Fraser [Tue, 6 May 2008 15:35:44 +0000 (16:35 +0100)]
minios: drop volatile qualifier from some status variables
since we already use memory barriers as approriate to prevent
concurrent access with event handlers.
Signed-off-by: Samuel Thibault <samuel.thibault@eu.citrix.com>
Keir Fraser [Tue, 6 May 2008 12:34:52 +0000 (13:34 +0100)]
minios: fix thread safety of xenbus watches by requiring callers to
provide their own queue of events, because else we can not dispatch to
watchers running in parallel.
Signed-off-by: Samuel Thibault <samuel.thibault@eu.citrix.com>
stubdom: make use of PVFB resize event
which with the offset support also permits to expose the VGA vram and
non-shared vram throught PVFB at the same time, switching between both
as appropriate.
Signed-off-by: Samuel Thibault <samuel.thibault@eu.citrix.com>
minios: simplify schedule()
- Merge the search, wakeup, and timeout loops.
- Avoid unnecessarily switching to the idle thread.
- Perform stack release _after_ we get out of it.
Signed-off-by: Samuel Thibault <samuel.thibault@eu.citrix.com>
stubdom: sparse application's BSS by linking it separately first, put
markers at its beginning and end, and then link with mini-os.
That permits to stick a bit more to upstream qemu.
Signed-off-by: Samuel Thibault <samuel.thibault@eu.citrix.com>
It has to check first if the memory to free is so big as to be freed
directly by free_pages. mini-os domains crash without this patch if
vfb is misconfigured.
Keir Fraser [Wed, 26 Mar 2008 13:15:00 +0000 (13:15 +0000)]
minios: blkfront_aio_poll may reenter
if the callback calls blkfront_sync for instance.
In such a case, we would see responses and hence release grants
several times. We need to be more synchronous and stop when we detect
that we have re-entered.
This fixes HVM restore with stubdomains.
Signed-off-by: Samuel Thibault <samuel.thibault@eu.citrix.com>
Keir Fraser [Wed, 26 Mar 2008 13:13:50 +0000 (13:13 +0000)]
minios: more assertions
- assert that we never allocate or free the same grant twice
- assert that network packets do not exceed a page
- assert that incoming network event IDs make sense
Signed-off-by: Samuel Thibault <samuel.thibault@eu.citrix.com>
Keir Fraser [Tue, 18 Mar 2008 11:29:18 +0000 (11:29 +0000)]
minios: Fix lost events
evtchn_bind_interdomain used to clear any already pending event before
binding a handler, because else the handler may be called before it is
ready. That however leads to missed events, which I had to workaround
for the HVM case.
This changes the semantics of bind_evtchn, and thus of all the
event channel binding functions (bind_virq, evtchn_alloc_unbound,
evtchn_bind_interdomain) into not unmasking the event itself, hence
letting the caller initialize properly before unmasking the port (e.g.
record the port number in an appropriate place).
Signed-off-by: Samuel Thibault <samuel.thibault@eu.citrix.com>
Keir Fraser [Tue, 12 Feb 2008 14:35:39 +0000 (14:35 +0000)]
Add stubdomain support. See stubdom/README for usage details.
- Move PAGE_SIZE and STACK_SIZE into __PAGE_SIZE and __STACK_SIZE in
arch_limits.h so as to permit getting them from there without
pulling all the internal Mini-OS defines.
- Setup a xen-elf cross-compilation environment in stubdom/cross-root
- Add a POSIX layer on top of Mini-OS by linking against the newlib C
library and lwIP, and implementing the Unixish part in mini-os/lib/sys.c
- Cross-compile zlib and libpci too.
- Add an xs.h-compatible layer on top of Mini-OS' xenbus.
- Cross-compile libxc with an additional xc_minios.c and a few things
disabled.
- Cross-compile ioemu with an additional block-vbd, but without sound,
tpm and other details. A few hacks are needed:
- Align ide and scsi buffers at least on sector size to permit
direct transmission to the block backend. While we are at it, just
page-align it to possibly save a segment. Also, limit the scsi
buffer size because of limitations of the block paravirtualization
protocol.
- Allocate big tables dynamically rather that letting them go to
bss: when Mini-OS gets installed in memory, bss is not lazily
allocated, and doing so during Mini-OS is unnecessarily trick while
we can simply use malloc.
- Had to change the Mini-OS compilation somehow, so as to export
Mini-OS compilation flags to the Makefiles of libxc and ioemu.
Signed-off-by: Samuel Thibault <samuel.thibault@eu.citrix.com>
Keir Fraser [Tue, 29 Jan 2008 15:18:27 +0000 (15:18 +0000)]
minios: Fix bug when blkfront reading into zero-mapped buffer
by just poking the page.
No need to use virtual_to_mfn() for the ring since that is a real page.
Signed-off-by: Samuel Thibault <samuel.thibault@eu.citrix.com>
Keir Fraser [Fri, 18 Jan 2008 15:22:42 +0000 (15:22 +0000)]
minios: extend map_frames into being able to map a series of
contiguous frames, or the same frame several times, aligned, from
another domain, with specific protection, and with potential
failures.
Signed-off-by: Samuel Thibault <samuel.thibault@eu.citrix.com>
Keir Fraser [Fri, 28 Dec 2007 15:23:28 +0000 (15:23 +0000)]
In C99 (7.19.6.2.10), the behavior of scanf("stuff... %n", ..., &n) is
said to be undefined if the value to be put is n can't fit. I guess
we can safely consider that the same applies to printf.
Signed-off-by: Samuel Thibault <samuel.thibault@citrix.com>
Keir Fraser [Sat, 24 Nov 2007 13:31:39 +0000 (13:31 +0000)]
[Mini-OS] Fix domain blocking race
A callback which wakes a thread may happen between the moment
schedule() gives hand to the idle thread and the latter blocks the
domain. Idle hence needs to atomically check that no thread is
running and block, else awoken threads may have to wait up to 10
seconds.
Signed-off-by: Samuel Thibault <samuel.thibault@citrix.com>
projects / mini-os.git / log