Like all block drivers virtio-blk should not allow small than block size
granularity access. But given that the protocol specifies a
byte unit length field we currently accept such requests, which cause
qemu to abort() in lower layers. Add checks to the main read and
write handlers to catch them early.
Reported-by: Conor Murphy <conor_murphy_virt@hotmail.com>
Tested-by: Conor Murphy <conor_murphy_virt@hotmail.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
vhost was passing a physical address to cpu_physical_memory_set_dirty,
which is wrong: we need to translate to ram address first.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Note: this lead to crashes during migration, so the patch
is needed on the stable branch too.
On Mon, 4 Apr 2011 20:15:30 +0200, Aurelien Jarno <aurelien@aurel32.net> wrote:
> Is it really safe ignoring write to this register? If yes, it's probably
> a good idea to explain why in a comment. In any case, if supporting this
> register is easy to do, it would be the best option.
I think it is safe. Please see an updated comment below.
And though implementing this register might be possible, I suppose it
is not worth to supporting FrameTooLong detection, for now at least.
Thank you for comments.
>8---------------------------------------------------------------------
From: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Date: Tue, 5 Apr 2011 23:12:07 +0900
Subject: [PATCH] lan9118: Ignore write to MAC_VLAN1 register
Since linux 2.6.38, smsc911x driver writes to VLAN1 registger.
Since this register only affects FrameTooLong detection, ignoring
write to this register should be safe.
Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
(cherry picked from commit a0313c00fc)
If these messages are not handled correctly the guest driver may hang.
Always mandatory:
- ABORT
- BUS DEVICE RESET
Mandatory if tagged queuing is implemented (which disks usually do):
- ABORT TAG
- CLEAR QUEUE
Signed-off-by: Bernhard Kohl <bernhard.kohl@nsn.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
(cherry picked from commit 508240c0da)
Commit c81131db15
detects old guests by comparing virtio and
PCI status. It attempts to do this on load,
as well, but load_config callback in a binding
is invoked too early and so the virtio status
isn't set yet.
We could add yet another callback to the
binding, to invoke after load, but it
seems easier to reuse the existing vmstate
callback.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Cc: Alexander Graf <agraf@suse.de>
(cherry picked from commit 89c473fd82)
Update not only dbc but also dnad when skipping bytes during the MSGOUT
phase. Previously only dbc was updated which is probably wrong and
could lead to bogus message codes being read.
Tested on Linux and Windows Server 2003.
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
(cherry picked from commit 444dd39b5f)
Nothing prevented IRQ sharing on the ISA bus in principle. Not all
boards supported this, neither each and every card nor driver and OS.
Still, there existed valid IRQ sharing scenarios, (at least) two of them
can also be found in QEMU: >2 PC UARTs and the PREP IDE buses.
So remove this artificial restriction from our ISA model.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
(cherry picked from commit ee951a37d8)
A pointer to a size_t variable was passed as the void * pointer to
lduw_p() in virtio_net_receive(). Instead of acting on the 16-bit value
this caused failure on big-endian hosts.
Avoid this issue in the future by using stw_p() instead. In general we
should use ld*_p() for loading from target memory and st*_p() for
storing to target memory anyway, not the other way around.
Also tighten up a correct use of lduw_p() when stw_p() should be used
instead in virtio_net_get_config().
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
(cherry picked from commit b46d97f2d2)
The SD_STATUS and SEND_NUM_WR_BLOCKS commands are supposed to cause
the card to send data back to the host. However sd.c was missing the
state change to sd_sendingdata_state for these commands, with the effect
that the Linux driver would either hang indefinitely waiting for
nonexistent data (pl181) or read zeroes and provoke a qemu warning
message (omap).
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
TPR blocks all interrupts in a priority class, so simple "less or
equal" check is not enough.
Signed-off-by: Gleb Natapov <gleb@redhat.com>
Reviewed-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
The host part of a block device can be deleted with in progress
block migration.
To fix this, add a reference count to DriveInfo, freeing resources
on last reference.
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
CC: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
(cherry picked from commit 84fb392526)
Different AHCI controllers have a different number of ports, so the core
shouldn't care about the amount of ports available.
This patch makes the number of ports available to the AHCI core runtime
configurable, allowing us to have multiple different AHCI implementations
with different amounts of ports.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
(cherry picked from commit 2c4b9d0ea4)
The ahci code was missing its soft reset functionality. This wasn't really an
issue for Linux guests, but Windows gets confused when the controller doesn't
reset when it tells it so.
Using this patch I can now successfully boot Windows 7 from AHCI using AHCI
enabled SeaBIOS.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
(cherry picked from commit 760c3e44d3)
The drive sends a d2h init fis on initialization. Usually, the guest doesn't
receive fises yet at that point though, so the delivery is deferred.
Let's reflect that by sending the init fis on fis receive enablement.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
(cherry picked from commit 87e62065bb)
Sebastian's patch already did a pretty good job at splitting up ICH-9
AHCI code and the AHCI core. We need some more though. Copyright was missing,
the lspci dump belongs to ICH-9, we don't need the AHCI core to have its
own qdev device duplicate.
So let's split them a bit more in this patch, making things easier to
read an understand.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
(cherry picked from commit 7fb6577b13)
Due to popular request, this patch adds a license header to ahci.h
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
(cherry picked from commit f83a40dcd7)
There are multiple ahci devices out there. The currently implemented ich-9
is only one of the many. So let's split that one out into a separate file
to stress the difference.
Signed-off-by: Sebastian Herbszt <herbszt@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
(cherry picked from commit 03c7a6a8e7)
Fix a few style issues and convert magic numbers into prober symbolic
constants, also fixing the wrong but unused IOAPIC_DM_SIPI value.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
(cherry picked from commit 1f5e71a8e6)
qemu-kvm carries the IOAPIC base address in its v2 vmstate. We only
support the default base address so far, and saving even that in the
device state was rejected.
Add a padding field to be able to read qemu-kvm's old state, but
increase our version to 3, indicating that we are not saving a valid
address. This also gives downstream the chance to change to stop
evaluating the base_address and move to v3 as well.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
(cherry picked from commit 5dce499948)
This is a guest modifiable state that must be saved/restored properly.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
(cherry picked from commit 35a74c5c59)
Add the missing EOI broadcast from local APIC to the IOAPICs on
completion of level-triggered IRQs. This ensures that a still asserted
IRQ source properly re-triggers an APIC IRQ.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
(cherry picked from commit 0280b571c1)
This can happen if a port gets unplugged before guest has chance to
initialise vqs.
Reported-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
(cherry picked from commit 7185f9315b)
When MSI is off, each interrupt needs to be bounced through the io
thread when it's set/cleared, so vhost-net causes more context switches and
higher CPU utilization than userspace virtio which handles networking in
the same thread.
We'll need to fix this by adding level irq support in kvm irqfd,
for now disable vhost-net in these configurations.
Added a vhostforce flag to force vhost-net back on.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
The Ubuntu 10.10 gcc for ARM complains that we might be overrunning
the cpu_irqs[][] array: silence this by correcting the bounds on the
loop. (In fact we would not have overrun the array because bit
MAX_PILS in pil_pending and irl_out will always be 0.)
Also add a comment about why the loop's lower bound is OK.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
The "leon3_cache_control_int" (op_helper.c) function is called within leon3.c
which leads to segfault error with the global "env".
Now cache control is a CPU feature and everything is handled in op_helper.c.
Signed-off-by: Fabien Chouteau <chouteau@adacore.com>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
Watch this:
(qemu) drive_add 0 if=none
(qemu) info block
none0: type=hd removable=0 [not inserted]
(qemu) drive_del none0
Segmentation fault (core dumped)
add_init_drive() is confused about drive_init()'s failure modes, and
cleans up when it shouldn't. This leaves the DriveInfo with member
opts dangling. drive_del attempts to free it, and dies.
drive_init() behaves as follows:
* If it created a drive with media, it returns its DriveInfo.
* If it created a drive without media, it clears *fatal_error and
returns NULL.
* If it couldn't create a drive, it sets *fatal_error and returns
NULL.
Of its three callers:
* drive_init_func() is correct.
* usb_msd_init() assumes drive_init() failed when it returns NULL.
This is correct only because it always passes option "file", and
"drive without media" can't happen then.
* add_init_drive() assumes drive_init() failed when it returns NULL.
This is incorrect.
Clean up drive_init() to return NULL on failure and only on failure.
Drop its parameter fatal_error.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Before, type & index were hidden in printf-like fmt, ... parameters,
which get expanded into an option string. Rather inconvenient for
uses later in this series.
New IF_DEFAULT to ask for the machine's default interface. Before,
that was done by having no option "if" in the option string.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Before commit 622b520f, index=12 meant bus=1,unit=5.
Since the commit, it means bus=0,unit=12. The drive is created, but
not the guest device. That's because the controllers we use with
if=scsi drives (lsi53c895a and esp) support only 7 units, and
scsi_bus_legacy_handle_cmdline() ignores drives with unit numbers
exceeding that limit.
Changing the mapping of index to bus, unit is a regression. Breaking
-drive invocations that used to work just makes it worse.
Revert the part of commit 622b520f that causes this, and clean up
some.
Note that the fix only affects if=scsi. You can still put more than 7
units on a SCSI bus with -device & friends.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
qdev_init_bdrv() doesn't belong into qdev.c; it's about drives, not
qdevs. Rename to drive_get_next, move to blockdev.c, drop the bogus
DeviceState argument, and return DriveInfo instead of
BlockDriverState.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
drive_init() picks the first free bus and unit number, unless the user
specifies them.
This isn't a good fit for the drive_add monitor command, because there
we specify the controller by PCI address instead of using bus number
set by drive_init().
scsi_hot_add() takes care to replace the unit number set by
drive_init() by the real one, but it neglects to replace the bus
number. Thus, bus/unit in DriveInfo may be bogus. Affects
drive_get() and drive_get_max_bus(). I'm not aware of anything bad
happening because of that; looks like by the time we're hot-plugging,
the two functions aren't used anymore. Fix it anyway.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The len and is_write arguments to cpu_physical_memory_unmap() were
swapped. This patch changes calls to use the correct argument ordering.
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Raise a config change interrupt when the size changed. This allows
virtio-blk guest drivers to read-read the information from the
config space once it got the config chaged interrupt.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Extend the change_cb callback with a reason argument, and use it
to tell drivers about size changes.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
r3480 added this check to account for the entry vector 0xfff00100 to be
available for CPUs that need it. Today however, the NIP is not yet
initialized at this point (zero), so the check always triggers.
Moreover, BIOS size check is already done previously, so this part can
be removed too.
Cc: Alexander Graf <agraf@suse.de>
Signed-off-by: Andreas Färber <andreas.faerber@web.de>
Signed-off-by: Hervé Poussineau <hpoussin@reactos.org>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
For some unknown reason, the MIPS kernel briefly changes the RTC to
binary mode during boot, switch back to BCD mode and read the time. As
the registers are updated only every second, they may still be in the
old format when they are read.
This patch forces a register update immediately after a format change
(BCD/binary or 12/24H). This avoid long fsck during boot due to time
wrap.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
virtio-blk doesn't work on cross-endian configuration, as endianness is
not handled correctly.
This patch adds missing endianness conversions to make virtio-blk
working. Tested on the following configurations:
- i386 guest on x86_64 host
- ppc guest on x86_64 host
- i386 guest on mips host
- ppc guest on mips host
Reviewed-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
virtio-net used to work on cross-endianness configurations, but doesn't
anymore with recent guest kernels, as the new features don't handle
endianness correctly.
This patch fixes wrong conversion, and add missing ones to make
virtio-net working. Tested on the following configurations:
- i386 guest on x86_64 host
- ppc guest on x86_64 host
- i386 guest on mips host
- ppc guest on mips host
Cc: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Recent PowerPC kernel end up in kernel panic during boot in -nographic
mode. In this mode the second serial port is used as the udbg console,
and thus a few characters are sent on this port. This activates the
tx interrupt flag, and later choke the Linux kernel, as it was not
expecting such a flag to be set.
The problem here comes from the fact that contrary to most devices the
interrupt flags are only set if the interrupt is enabled. Quoting the
datasheet: "If the corresponding IE bit is not set, the IP for that
source of interrupt will never be set."
This patch fixes that by enabling the interrupt flag only when the
corresponding interrupt is enabled.
Cc: Blue Swirl <blauwirbel@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
As noted by Markus Armbruster pxa2xx_gpio vmstate version bumped
because of a change in the or .ilevel / .olevel arrays are saved,
for convenience.
Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
Signed-off-by: Andrzej Zaborowski <andrew.zaborowski@intel.com>
Switch sl-nand emulation to use qdev and vmstate. Also drop ecc_get/_put
functions as sl-nand was the only user of that code.
Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
Signed-off-by: Andrzej Zaborowski <andrew.zaborowski@intel.com>
Christoph Hellwig