By default, require 4k of consecutive zero bytes for qemu-img to make the
output file sparse by not issuing a write request for the zeroed parts. Add an
-S option to allow users to tune this setting.

This helps to avoid situations where a lot of zero sectors and data sectors are
mixed and qemu-img tended to issue many tiny 512 byte writes.
Signed-off-by: NKevin Wolf <kwolf@redhat.com>

When adding a rom_device in I/O mode, we incorrectly masked off the low
bits, resulting in a pure RAM map.  Fix my masking off the high bits and
IO_MEM_ROMD, yielding a pure I/O map.
Signed-off-by: NAvi Kivity <avi@redhat.com>

This commit implements the prefetch engine feature of the GPMC
which can be used for NAND devices. This includes both interrupt
driven and DMA-filling modes.
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

Refactor the gpmc state structure so items relating to
the prefetch engine are in their own sub-struct and have
more useful names.
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

OMAP3630 adds an extra bit of address masking, so a mask of
0xb1111 is valid. Unfortunately the GPMC_REVISION is the same as
on the OMAP3430 which only has three bits of address masking, so
we have to derive this feature directly from the OMAP revision
rather than from the GPMC revision.
Signed-off-by: NJuha Riihimäki <juha.riihimaki@nokia.com>
[Riku Voipio: Fixes and restructuring patchset]
Signed-off-by: NRiku Voipio <riku.voipio@iki.fi>
[Peter Maydell: More fixes and cleanups for upstream submission]
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

Add the OMAP 3630 to the omap_mpu_model enumeration, and add the
corresponding cpu_is_omap3630() function.

(OMAP3 isn't supported yet but this is useful in upgrading common
components to be "OMAP3 ready". We already have this for OMAP3430.)
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

Support accesses to NAND devices, both by mapping them into
the GPMC address space, and via the NAND_COMMAND, NAND_ADDRESS
and NAND_DATA GPMC registers.
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

Whitespace-only change fixing indentation.
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

Signed-off-by: NJuha Riihimäki <juha.riihimaki@nokia.com>
[Riku Voipio: Fixes and restructuring patchset]
Signed-off-by: NRiku Voipio <riku.voipio@iki.fi>
[Peter Maydell: More fixes and cleanups for upstream submission]
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

Take a pointer to the omap mpu state struct in omap_gpmc_init.
Some details of GPMC behaviour depend on the OMAP version we
are a part of.
Signed-off-by: NJuha Riihimäki <juha.riihimaki@nokia.com>
[Riku Voipio: Fixes and restructuring patchset]
Signed-off-by: NRiku Voipio <riku.voipio@iki.fi>
[Peter Maydell: More fixes and cleanups for upstream submission]
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

The OMAP3 TRM is inconsistent about whether the GPMC FIFOTHRESHOLDSTATUS
bit should be set when FIFOPOINTER > FIFOTHRESHOLD or when it is >=
FIFOTHRESHOLD. Apparently the underlying functional spec from which
the TRM was created states that the behaviour is ">=", and this also
makes more conceptual sense.
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

The omap_gpmc wasn't actually wiring up its IRQ, so
anything that provoked an interrupt would be using
uninitialised data for its IRQ number.
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

Fix a bug in the handling of writes to GPMC_IRQSTATUS:
it behaves as "write one to clear, writing zero is ignored".
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

Refactor the omap_gpmc_cs_map/unmap functions:
 * take the omap_gpmc_s* and a chipselect id rather than the
   omap_gpmc_cs_file_s*, so they have access to the general gpmc
   member fields
 * extract the base and mask from the config registers in the functions
   rather than at every callsite
 * check for CSVALID in the functions rather than at every callsite
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

Now that all callers of omap_gpmc_attach pass in a MemoryRegion*,
we can remove the base_update and unmap function pointer arguments,
and the opaque pointer that was passed into these callbacks.

We can also remove the base and size fields from omap_gpmc_cs_file_s
as these are no longer necessary (you don't need the base/size
to unmap a MemoryRegion the way you did to undo a mapping made
with cpu_register_physical_memory()).
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

Minor whitespace-only cleanup (separated out from the qdevifying
patch for clarity).
Signed-off-by: NJuha Riihimäki <juha.riihimaki@nokia.com>
[Riku Voipio: Fixes and restructuring patchset]
Signed-off-by: NRiku Voipio <riku.voipio@iki.fi>
[Peter Maydell: More fixes and cleanups for upstream submission]
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

Qdevify the ONENAND device.
Signed-off-by: NJuha Riihimäki <juha.riihimaki@nokia.com>
[Riku Voipio: Fixes and restructuring patchset]
Signed-off-by: NRiku Voipio <riku.voipio@iki.fi>
[Peter Maydell: More fixes and cleanups for upstream submission]
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

Refactor onenand_command() -- since it is essentially a method of
the device object, it doesn't make sense to pass in something as
an argument which is one of the object's own member fields.
Signed-off-by: NJuha Riihimäki <juha.riihimaki@nokia.com>
[Riku Voipio: Fixes and restructuring patchset]
Signed-off-by: NRiku Voipio <riku.voipio@iki.fi>
[Peter Maydell: More fixes and cleanups for upstream submission]
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

Add a sysbus_mmio_get_region() which allows users of sysbus
devices to turn a (SysBusDevice*, mmioidx) tuple into a
MemoryRegion*. This enables some useful simplifications of
devices which pass through another device's mmio region
(either directly or by implementing some kind of memory
controller device).
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

The a7 area was set up as an alias of itself, rather than the p4 area.  This
sent the memory core into infinite recursion.

Fix by aliasing the a7 area to the p4 area.
Signed-off-by: NAvi Kivity <avi@redhat.com>

clear interrupt request if the interrupt priority < CPU pil
clear hardware interrupt request if interrupts are disabled
Signed-off-by: NArtyom Tarasenko <atar4qemu@gmail.com>
[blauwirbel@gmail.com: added a comment about magic 2]
Signed-off-by: NBlue Swirl <blauwirbel@gmail.com>

This converts ethlite, intc, timer and uartlite to use
MemoryRegions.
Signed-off-by: NEdgar E. Iglesias <edgar.iglesias@gmail.com>

Use enum TCGOpcode instead of plain old int so that the name of
current op can be seen in GDB. Add a default case to switch
so that GCC does not complain about unhandled enum cases.
Signed-off-by: NBlue Swirl <blauwirbel@gmail.com>

Remove unused or otherwise available stuff.
Signed-off-by: NBlue Swirl <blauwirbel@gmail.com>

Extract G364 ROM contents from device emulation to machine emulation,
so device emulation can be reused in other machines (Commodore Amiga)
Signed-off-by: NHervé Poussineau <hpoussin@reactos.org>
Signed-off-by: NBlue Swirl <blauwirbel@gmail.com>

Signed-off-by: NHervé Poussineau <hpoussin@reactos.org>
Signed-off-by: NBlue Swirl <blauwirbel@gmail.com>

checkpatch.pl doesn't report warning for if/else statements with missing
'else' braces:

if (something) {
    foo;
} else
    bar;

The patch has been tested using the last 100 commits.
Signed-off-by: NPavel Borzenkov <pavel.borzenkov@gmail.com>
Signed-off-by: NBlue Swirl <blauwirbel@gmail.com>

Remove PROVIDE_HIDDEN and ONLY_IF_{RO,RW} from linker scripts to make
them work with older binutils versions.  Fixes *-bsd-user build on
OpenBSD 4.9 which ships binutils 2.15.
Signed-off-by: NGerd Hoffmann <kraxel@redhat.com>
Signed-off-by: NBlue Swirl <blauwirbel@gmail.com>

The first issue is the hard coded POSIX Real Time extensions library in the
libcacard/Makefile. From looking at the code it doesn't seem this is necessary
anyway. Robert Relyea seems to think it most likely isn't necessary.

The second issue was the missing exclusion of the BSD userland binary
builds from the addition of this Makefile target for the smartcard NSS
code which breaks the builds if smartcard NSS support is enabled.

pastebin clip of the build failure..

http://pastebin.com/raw.php?i=BLCKd3s6Signed-off-by: NBrad Smith <brad@comstyle.com>
Signed-off-by: NBlue Swirl <blauwirbel@gmail.com>

Account the total latency for read/write/flush requests.  This allows
management tools to average it based on a snapshot of the nr ops
counters and allow checking for SLAs or provide statistics.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NKevin Wolf <kwolf@redhat.com>

When the vhost notifier is disabled, the userspace handler runs
immediately: virtio_pci_set_host_notifier_internal might
call virtio_queue_notify_vq.
Since the VQ state and the tap backend state aren't
recovered yet, this causes
"Guest moved used index from XXX to YYY" assertions.

The solution is to split out host notifier handling
from vhost VQ setup and disable notifiers as our last step
when we stop vhost-net. For symmetry enable them first thing
on start.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

Fix a few cases where we were passing host pointers to the
guest.
Signed-off-by: NEdgar E. Iglesias <edgar.iglesias@gmail.com>

The MemoryRegionOps callbacks expect it.
Signed-off-by: NAvi Kivity <avi@redhat.com>