Isaku Yamahata [Fri, 30 Oct 2009 12:21:09 +0000 (21:21 +0900)]
pci: introduce FMT_PCIBUS for printf format for pcibus_t.
This patch is preliminary for 64bit BAR.
Later pcibus_t will be changed from uint32_t to uint64_t.
Introduce FMT_PCIBUS for printf format for pcibus_t.
Isaku Yamahata [Fri, 30 Oct 2009 12:21:08 +0000 (21:21 +0900)]
pci: introduce pcibus_t to represent pci bus address/size instead of uint32_t
This patch is preliminary for 64 bit BAR support.
Introduce dedicated type, pcibus_t, to represent pci bus address/size
instead of uint32_t.
Later this type will be changed to uint64_t.
Isaku Yamahata [Fri, 30 Oct 2009 12:21:06 +0000 (21:21 +0900)]
pci_host.h: move functions in pci_host.h into .c file.
split static functions in pci_host.h into pci_host.c and
pci_host_template.h.
Later a structures declared in pci_host.h, PCIHostState, will be used.
However pci_host.h doesn't allow to include itself easily. This patches
addresses it.
pci_host.h includes functions which are instantiated in .c by including
pci_host.h with typedefing pci_addr_t.
pci_addr_t is per pci host bridge and is typedef'ed to uint32_t for ioio
or target_phys_addr_t for mmio in .c file.
That prevents from including pci_host.h to use PCIHostState because of
requiring type, pci_addr_t.
Its purpose to include is to instantiate io function for mmio or ioio
depending on which pci host bridge requires ioio or mmio.
To avoid including code, we always instantiate both version.
Isaku Yamahata [Fri, 30 Oct 2009 12:21:01 +0000 (21:21 +0900)]
pci/bridge: clean up of pci_bridge_initfn()
- use symbolic constant
- use helper function pci_set_xxx()
- removed lines which initializes to 0.
It is unnecessary because it is already zeroed.
- add some comments on command registers.
Some initial values are suspicious because they seems to
be specific to apb_pci.c which is the only user of pci bridge right now.
For now don't touch those values to avoid breakage.
Uri Lublin [Thu, 29 Oct 2009 18:40:01 +0000 (14:40 -0400)]
Makefile: make qemu-img dependant on config-host.h
Since config-host.h is generated by the Makefile (1215c6e76),
building (only) qemu-img fails:
[user@f12-uri qemu]$ make distclean (or git clone qemu)
[user@f12-uri qemu]$ ./configure ...
[user@f12-uri qemu]$ make qemu-img
GEN config-all-devices.mak
GEN qemu-img-cmds.h
CC qemu-img.o
In file included from qemu-img.c:24:
qemu-common.h:32:25: error: config-host.h: No such file or directory
Dustin Kirkland [Thu, 29 Oct 2009 15:34:15 +0000 (10:34 -0500)]
whitelist host virtio networking features
This patch is a followup to 8eca6b1bc770982595db2f7207c65051572436cb,
fixing crashes when guests with 2.6.25 virtio drivers have saturated
virtio network connections.
That patch should have been whitelisting *_HOST_* rather than the the
*_GUEST_* features.
I tested this by running an Ubuntu 8.04 Hardy guest (2.6.24 kernel +
2.6.25-virtio driver). I saturated both the incoming, and outgoing
network connection with nc, seeing sustained 6MB/s up and 6MB/s down
bitrates for ~20 minutes. Previously, this crashed immediately. Now,
the guest does not crash and maintains network connectivity throughout
the test.
Mark McLoughlin [Tue, 27 Oct 2009 18:16:39 +0000 (18:16 +0000)]
tap: drain queue in tap_send()
Okay, let's try re-enabling the drain-entire-queue behaviour, with a
difference - before each subsequent packet, use qemu_can_send_packet()
to check that we can send it. This is similar to how we check before
polling the tap fd and avoids having to drop a packet if the receiver
cannot handle it.
This patch should be a performance improvement since we no longer have
to go through the mainloop for each packet.
Mark McLoughlin [Tue, 27 Oct 2009 18:16:38 +0000 (18:16 +0000)]
virtio-net: split the has_buffers() logic from can_receive()
We should only return zero from receive() for a condition which we'll
get notification of when it changes. Currently, we're returning zero
if the guest driver is not ready, but we won't ever flush our queue
when that status changes.
Also, don't check buffer space in can_receive(), but instead just allow
receive() to return zero when this condition occurs and have the caller
handle queueing the packet.
Mark McLoughlin [Tue, 27 Oct 2009 18:16:35 +0000 (18:16 +0000)]
tap: disable draining queue in one go
If qemu_send_packet_async() returns zero, it means the packet has been
queued and the sent callback will be invoked once it has been flushed.
This is only possible where the NIC's receive() handler returns zero
and promises to notify the networking core that room is available in its
queue again.
In the case where the receive handler does not have this capability
(and its queue fills up) it returns -1 and the networking core does not
queue up the packet. This condition is indicated by a -1 return from
qemu_send_packet_async().
Currently, tap handles this condition simply by dropping the packet. It
should do its best to avoid getting into this situation by checking such
NIC's have room for a packet before copying the packet from the tap
interface.
tap_send() used to achieve this by only reading a single packet before
returning to the mainloop. That way, tap_can_send() is called before
reading each packet.
tap_send() was changed to completely drain the tap interface queue
without taking into account the situation where the NIC returns an
error and the packet is not queued. Let's start fixing this by
reverting to the previous behaviour of reading one packet at a time.
We have code for a quite a few block formats. While I trust that all
of these formats are useful at least for some people in some
circumstances, some of them are of a kind that friends don't let
friends use in production.
This patch provides an optional block format whitelist, default off.
If a whitelist is configured with --block-drv-whitelist, QEMU proper
can use only whitelisted formats. Other programs, like qemu-img, are
not affected.
Drivers for formats off the whitelist still participate in format
probing, to ensure all programs probe exactly the same. Without that,
QEMU proper would be prone to treat images with a format off the
whitelist as raw when the image's format is probed.
qdev: Tag isa-fdc, PIIX3 IDE and PIIX4 IDE as no-user
These devices are created automatically, and attempting to create
another one with -device fails with "qemu: hardware error:
register_ioport_write: invalid opaque".
Naphtali Sprei [Mon, 26 Oct 2009 14:25:16 +0000 (16:25 +0200)]
Added readonly flag to -drive command
This is a slightly revised patch for adding readonly flag to the -drive command.
Even though this patch is "stand-alone", it assumes a previous related patch (in Anthony staging tree), that passes
the readonly attribute of the drive to the guest OS, applied first.
This enables sharing same image between guests, with readonly access.
Implementaion mark the drive as read_only and changes the flags when actually opening the file.
The readonly attribute of a qcow also passed to it's base file.
For ide that cannot pass the readonly attribute to the guest OS, disallow the readonly flag.
Also, return error code from bdrv_truncate for readonly drive.
Blue Swirl [Sat, 7 Nov 2009 16:34:12 +0000 (16:34 +0000)]
(x86/Sparc/PPC)-user: fix cpu_copy
b55a37c981914aa8ecd21b9a2a2fb37f39b917c5 moved the call to cpu_reset
to user emulators. But cpu_copy also initializes a CPU structure, so add the
call also there.
Glauber Costa [Thu, 5 Nov 2009 18:05:15 +0000 (16:05 -0200)]
v3: don't call reset functions on cpu initialization
There is absolutely no need to call reset functions when initializing
devices. Since we are already registering them, calling qemu_system_reset()
should suffice. Actually, it is what happens when we reboot the machine,
and using the same process instead of a special case semantics will even
allow us to find bugs easier.
Furthermore, the fact that we initialize things like the cpu quite early,
leads to the need to introduce synchronization stuff like qemu_system_cond.
This patch removes it entirely. All we need to do is call qemu_system_reset()
only when we're already sure the system is up and running
I tested it with qemu (with and without io-thread) and qemu-kvm, and it
seems to be doing okay - although qemu-kvm uses a slightly different patch.
[ v2: user mode still needs cpu_reset, so put it in ifdef. ]
[ v3: leave qemu_system_cond for now. ]
Sparc: fix carry flag handling (Solaris bootblk fix)
The page 108 of the SPARC Version 8 Architecture Manual describes
that addcc and addxcc shall compute carry flag the same way.
The page 110 claims the same about subcc and subxcc instructions.
This patch fixes carry computation in corner cases and removes redundant code.
The most visible effect of the patch is enabling Solaris boot when using OBP.
Anthony Liguori [Mon, 2 Nov 2009 21:50:27 +0000 (15:50 -0600)]
Fix the build when srcdir == objdir
You would only see this error on a fresh clone when srcdir == objdir. configure
will fail because roms/pcbios doesn't exist.
git submodule integration doesn't cleanup very well when switching between
branches so you'll get an roms/pcbios directory from normal operations if you
switch between old branches.
Thanks to a mistake in configure, if you build outside of srcdir, you'll also
get a valid roms/pcbios.
Alexander Graf [Fri, 30 Oct 2009 21:27:00 +0000 (22:27 +0100)]
Unbreak tap compilation on OS X
Currently compiling the tap sources breaks on Mac OS X. This is because of:
1) tap-linux.h requiring Linux includes
2) typos
3) missing #includes
This patch adds what's necessary to compile tap happily on Mac OS X.
I haven't tested if using tap actually works, but I don't think that's a
major issue as that code was probably seriously untested before already.
I didn't split the patch, because it's only a few lines of code and
splitting is probably not worth the effort here.
Anthony Liguori [Fri, 30 Oct 2009 14:06:09 +0000 (09:06 -0500)]
Switch pc bios from pc-bios to seabios
SeaBIOS is a port of pc-bios to GCC. Besides using a more modern tool chain,
SeaBIOS introduces a number of new features including PMM support, better
BEV and BCV support, and better PnP support.
Mark McLoughlin [Wed, 28 Oct 2009 14:07:23 +0000 (14:07 +0000)]
virtio-net: fix macaddr config regression
This commit:
commit 97b15621
virtio: use qdev properties for configuration.
Signed-off-by: Gerd Hoffmann <[email protected]> Signed-off-by: Anthony Liguori <[email protected]>
makes a guest using virtio-net see an empty macaddr because we never
copy the macaddr into the location that virtio_net_get_config() uses.
Fixes a couple of issues with msix table access:
- With misbehaving guests, misaligned 4 byte access could overflow
msix table and cause qemu to segfault. Since PCI spec requires
host to only issue dword-aligned accesses, as a fix,
it's enough to mask the address low bits.
- Tables use pci format, not native format, and so
we must use pci_[sg]et_long on read/write.
wmb must be at least a compiler barrier, even without SMP.
Further, we likely need some rmb()/mb() as well:
I have not audited the code but lguest has mb(),
add a comment for now.
Kevin Wolf [Mon, 26 Oct 2009 12:03:08 +0000 (13:03 +0100)]
Remove aio_ctx from paio_* interface
The context parameter in paio_submit isn't used anyway, so there is no reason
why block drivers should need to remember it. This also avoids passing a Linux
AIO context to paio_submit (which doesn't do any harm as long as the parameter
is unused, but it is highly confusing).
Gerd Hoffmann [Mon, 26 Oct 2009 14:56:50 +0000 (15:56 +0100)]
usb-host: use qdev for -usbdevice + rework.
Changes:
* We don't create/delete devices, we attach/detach them instead.
* The separate autofilter list is gone, we simply walk the list
of devices directly instead.
* Autofiltering is done unconditionally now. Non-auto device scan
code got dropped.
* Autofiltering turns off the timer if there is nothing to do, it
runs only in case there are unattached host devices.
Gerd Hoffmann [Mon, 26 Oct 2009 14:56:48 +0000 (15:56 +0100)]
usb: make attach optional.
Add a auto_attach field to USBDevice, which is enabled by default.
USB drivers can clear this field in case they do *not* want the device
being attached (i.e. plugged into a usb port) automatically after
successfull init().
Use cases (see next patches):
* attaching encrypted mass storage devices.
* -usbdevice host:...
Gerd Hoffmann [Mon, 26 Oct 2009 14:56:45 +0000 (15:56 +0100)]
usb core: use qdev for -usbdevice
This patchs adds infrastructure to handle -usbdevice via qdev callbacks.
USBDeviceInfo gets a name field (for the -usbdevice driver name) and a
callback for -usbdevice parameter parsing.
The new usbdevice_create() function walks the qdev driver list and looks
for a usb driver with a matching name. When a parameter parsing
callback is present it is called, otherwise the device is created via
usb_create_simple().
Gerd Hoffmann [Mon, 26 Oct 2009 11:18:27 +0000 (12:18 +0100)]
use rom loader for pc bios.
The pc bios shows up in 'info roms' now.
Note that the BIOS is mapped to two places: The complete rom at the top
of the memory, and the first 128k at 0xe0000. Only the first place is
listed in 'info roms'.
Gerd Hoffmann [Mon, 26 Oct 2009 11:18:25 +0000 (12:18 +0100)]
rom loader: make vga+rom loading configurable.
The rom_add_vga() and rom_add_option() macros are transformed into
functions. They look at the new rom_enable_driver_roms variable
and only do something if it is set to non-zero, making vga+option rom
loading runtime option. pc_init() sets rom_enable_driver_roms to 1.
With this in place we can move the rom loading calls from pc.c to the
individual drivers.
Mark McLoughlin [Thu, 22 Oct 2009 16:49:10 +0000 (17:49 +0100)]
net: move AIX code into net/tap-aix.c
Okay, this makes the tap options available on AIX even though there's
no support, but if we want to do it right we should have not compile
the tap code at all on AIX using e.g. CONFIG_TAP.
Anthony Liguori [Thu, 22 Oct 2009 16:43:48 +0000 (17:43 +0100)]
Work around dhclient brokenness
With the latest GSO/csum offload patches, any guest using an unpatched version
of dhclient (any Ubuntu guest, for instance), will no longer be able to get
a DHCP address.
dhclient is actually at fault here. It uses AF_PACKET to receive DHCP responses
but does not check auxdata to see if the packet has a valid csum. This causes
it to throw out the DHCP responses it gets from the virtio interface as there
is not a valid checksum.
Fedora has carried a patch to fix their dhclient (it's needed for Xen too) but
this patch has not made it into a release of dhclient. AFAIK, the patch is in
the dhclient CVS but I cannot confirm since their CVS is not public.
This patch, suggested by Rusty, looks for UDP packets (of a normal MTU) and
explicitly adds a checksum to them if they are missing one.
This allows unpatched dhclients to continue to work without needing to update
the guest kernels.
Mark McLoughlin [Thu, 22 Oct 2009 16:43:46 +0000 (17:43 +0100)]
net: add tap_set_offload()
This API allows virtio-net to enable various offload features on a
tap interface - e.g. to tell the host kernel it can pass up partial
checksums to userspace.
Mark McLoughlin [Thu, 22 Oct 2009 16:43:45 +0000 (17:43 +0100)]
virtio-net: add vnet_hdr support
With '-netdev tap,id=foo -nic model=virtio,netdev=foo' virtio-net can
detect that its peer (i.e. the tap backend) supports vnet headers
and advertise to the guest that it can send packets with partial
checksums and/or TSO packets.
One complication is that if we're migrating and the source host
supports IFF_VNET_HDR but the destination host doesn't, we can't then
stop the guest from using those features. In this scenario, we just
fail the migration.
[v2:
- add has_vnet_hdr uint32_t field for ease of vmstate conversion
- use qemu_error()
]
Gleb Natapov [Thu, 22 Oct 2009 16:43:43 +0000 (17:43 +0100)]
net: use qemu_send_packet_raw() in qemu_announce_self()
Use qemu_send_packet_raw to send gratuitous arp. This will ensure that
vnet header is handled properly.
Also, avoid sending the gratuitous packet to the guest. There doesn't
appear to be any reason for doing that and the code will currently just
crash if the NIC is not associated with a vlan.
projects / qemu.git / log