The VFIO common code doesn't provide the possibility to modify a
previous mapping entry in another way than unmapping and mapping again
with new properties.
To avoid -EEXIST DMA mapping error, we introduce a GHashTable to store
S390IOTLBEntry instances in order to cache the mapped entries. When
intercepting rpcit instruction, ignore the identical mapped entries to
avoid doing map operations multiple times and do unmap and re-map
operations for the case of updating the valid entries.
Acked-by: Pierre Morel <pmorel@linux.vnet.ibm.com>
Signed-off-by: Yi Min Zhao <zyimin@linux.vnet.ibm.com>
Message-Id: <20180205072258.5968-3-zyimin@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Current s390x PCI IOMMU code is lack of flags' checking, including:
1) protection bit
2) table length
3) table offset
4) intermediate tables' invalid bit
5) format control bit
This patch introduces a new struct named S390IOTLBEntry, and makes up
these missed checkings. At the same time, inform the guest with the
corresponding error number when the check fails. Finally, in order to
get the error number, we export s390_guest_io_table_walk().
Reviewed-by: Pierre Morel <pmorel@linux.vnet.ibm.com>
Signed-off-by: Yi Min Zhao <zyimin@linux.vnet.ibm.com>
Message-Id: <20180205072258.5968-2-zyimin@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
For now, the kernel does not properly indicate configured CPU subfunctions
to the guest, but simply uses the host values (as support in KVM is still
missing). That's why we missed to model the PTFF subfunctions that come
with Multiple-epoch facility.
Let's properly add these, along with a new feature group.
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180205102935.14736-1-david@redhat.com>
Tested-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
AEN and AIS can be provided unconditionally, ZPCI should be turned on
manually.
With -cpu qemu,zpci=on, the guest kernel can now successfully detect
virtio-pci devices under tcg.
Also fixup the order of the MSA_EXT_{3,4} flags while at it.
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
On s390x, pci support is implemented via a set of instructions
(no mmio). Unfortunately, none of them are documented in the
PoP; the code is based upon the existing implementation for KVM
and the Linux zpci driver.
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
This avoids tons of conversions when handling interrupts.
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180129125623.21729-19-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
This avoids tons of conversions when handling interrupts.
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180129125623.21729-18-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
This avoids tons of conversions when handling interrupts.
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180129125623.21729-17-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
s390x is ready. Most likely we are missing some pieces, but it should
already be in pretty good shape now.
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180129125623.21729-16-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
We should be pretty good in shape now. Floating interrupts are working
and atomic instructions should be atomic.
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180129125623.21729-15-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Current STSI implementation is a mess, so let's rewrite it.
Problems fixed by this patch:
1) The order of exceptions/when recognized is wrong.
2) We have to store to virtual address space, not absolute.
3) Alignment check of the block is missing.
3) The SMP information is not indicated.
While at it:
a) Make the code look nicer
- get rid of nesting levels
- use struct initialization instead of initializing to zero
- rename a misspelled field and rename function code defines
- use a union and have only one write statement
- use cpu_to_beX()
b) Indicate the VM name/extended name + UUID just like KVM does
c) Indicate that all LPAR CPUs we fake are dedicated
d) Add a comment why we fake being a KVM guest
e) Give our guest as default the name "TCGguest"
f) Fake the same CPU information we have in our Guest for all layers
While at it, get rid of "potential_page_fault()" by forwarding the
retaddr properly.
The result is best verified by looking at "/proc/sysinfo" in the guest
when specifying on the qemu command line
-uuid "74738ff5-5367-5958-9aee-98fffdcd1876" \
-name "extra long guest name"
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180129125623.21729-14-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
All blocks are 4k in size, which is only true for two of them right now.
Also some reserved fields were wrong, fix it and convert all reserved
fields to u8.
This also fixes the LPAR part output in /proc/sysinfo under TCG. (for
now, everything was indicated as 0)
While at it, introduce typedefs for these structs and use them in TCG/KVM
code.
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180129125623.21729-13-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Kicking all CPUs on every floating interrupt is far from efficient.
Let's optimize it at least a little bit.
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180129125623.21729-12-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Now that we have access to the io interrupts, we can implement
clear_io_irq() for TCG.
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180129125623.21729-11-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Use s390_cpu_virt_mem_write() so we can actually revert what we did
(re-inject the dequeued IO interrupt).
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180129125623.21729-10-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Move floating interrupt handling into the flic. Floating interrupts
will now be considered by all CPUs, not just CPU #0. While at it, convert
I/O interrupts to use a list and make sure we properly consider I/O
sub-classes in s390_cpu_has_io_int().
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180129125623.21729-9-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
This is a preparation for floating interrupt support and only applies to
MTTCG, single threaded TCG works just fine. If a floating interrupt wakes
up a VCPU and the CPU thinks it can run (clearing cs->halted), at
the point where the interrupt would be delivered, already another VCPU
might have picked up the interrupt, resulting in a wakeup without an
interrupt (executing wrong code).
It is wrong to let the VCPU continue to execute (the WAIT PSW). Instead,
we have to put the VCPU back to sleep.
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180129125623.21729-8-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
We can directly call the right function.
Suggested-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180129125623.21729-7-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Let the flic device handle it internally. This will allow us to later
on store floating interrupts in the flic for the TCG case.
This now also simplifies kvm.c. All that's left is the fallback
interface for floating interrupts, which is now triggered directly via
the flic in case anything goes wrong.
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180129125623.21729-6-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
We currently only support CRW machine checks. This is a preparation for
real floating interrupt support.
Get rid of the queue and handle it via the bit INTERRUPT_MCHK. We don't
rename it for now, as it will be soon gone (when moving crw machine checks
into the flic).
Please note that this is the same way also KVM handles it: only one
instance of a machine check can be pending at a time. So no need for a
queue.
While at it, make sure we try to deliver only if env->cregs[14]
actually indicates that CRWs are accepted.
Drop two unused defines on the way (we already have PSW_MASK_...).
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180129125623.21729-5-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
We can simply search for an object of our common type.
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180129125623.21729-4-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
This makes it clearer, which device is used for which accelerator.
Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180129125623.21729-3-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
We have to consider all deliverable interrupts.
We now have to take care of the special scenario, where we first
inject an interrupt with a WAIT PSW, followed by a !WAIT PSW. (very
unlikely but possible)
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180129125623.21729-2-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
In alpine docker image the qemu-system-s390x build is broken and
it throws this error:
qemu-system-s390x: Initialization of device s390-ipl failed: could not
load bootloader 's390-ccw.img'
The grep command of busybox uses regex. This fails on binary data
(e.g. stops on every \0), so it does not identify the string
BiGeNdIaN in the test case big/little. Therefore, it assumes
that the architecture is little endian.
This fix solves the grep problem by printing the content of
TMPO with strings
Signed-off-by: Alice Frosi <alice@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
[some changes to patch description, add -a option to strings]
Message-Id: <20180130133828.77336-2-borntraeger@de.ibm.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20180201111846.21846-4-armbru@redhat.com>
Clean up includes so that osdep.h is included first and headers
which it implies are not included manually.
This commit was created with scripts/clean-includes, with the change
to target/s390x/gen-features.c manually reverted, and blank lines
around deletions collapsed.
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20180201111846.21846-3-armbru@redhat.com>
System headers should be included with <...>, our own headers with
"...". Offenders tracked down with an ugly, brittle and probably
buggy Perl script. Previous iteration was commit a9c94277f0.
Delete inclusions of "string.h" and "strings.h" instead of fixing them
to <string.h> and <strings.h>, because we always include these via
osdep.h.
Put the cleaned up system header includes first.
While there, separate #include from file comment with exactly one
blank line.
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20180201111846.21846-2-armbru@redhat.com>
according to Eduardo Habkost's commit fd3b02c889 all PCIEs now implement
INTERFACE_PCIE_DEVICE so we don't need is_express field anymore.
Devices that implements only INTERFACE_PCIE_DEVICE (is_express == 1)
or
devices that implements only INTERFACE_CONVENTIONAL_PCI_DEVICE (is_express == 0)
where not affected by the change.
The only devices that were affected are those that are hybrid and also
had (is_express == 1) - therefor only:
- hw/vfio/pci.c
- hw/usb/hcd-xhci.c
- hw/xen/xen_pt.c
For those 3 I made sure that QEMU_PCI_CAP_EXPRESS is on in instance_init()
Reviewed-by: Marcel Apfelbaum <marcel@redhat.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Yoni Bettan <ybettan@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Currently virtio-pci driver hardcoded 2 vectors for virtio-blk device,
for multiple I/O queues scenario, all the I/O queues will share one
interrupt vector, while here, enable multiple vectors according to
the number of I/O queues.
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
In the past, we prioritized IOMMU migration so that we have such a
priority order:
IOMMU > PCI Devices
When migrating a guest with both vIOMMU and a pcie-root-port, we'll
always migrate vIOMMU first, since pci buses will be seen to have the
same priority of general PCI devices.
That's problematic.
The thing is that PCI bus number information is stored in the root port,
and that is needed by vIOMMU during post_load(), e.g., to figure out
context entry for a device. If we don't have correct bus numbers for
devices, we won't be able to recover device state of the DMAR memory
regions, and things will be messed up.
So let's boost the PCIe root ports to be even with higher priority:
PCIe Root Port > IOMMU > PCI Devices
A smoke test shows that this patch fixes bug 1538953.
Also, apply this rule to all the PCI bus/bridge devices: ioh3420,
xio3130_downstream, xio3130_upstream, pcie_pci_bridge, pci-pci bridge,
i82801b11.
I noted that we set pcie_pci_bridge_dev_vmstate twice. Clean that up
together.
CC: Alex Williamson <alex.williamson@redhat.com>
CC: Marcel Apfelbaum <marcel@redhat.com>
CC: Michael S. Tsirkin <mst@redhat.com>
CC: Dr. David Alan Gilbert <dgilbert@redhat.com>
CC: Juan Quintela <quintela@redhat.com>
CC: Laurent Vivier <lvivier@redhat.com>
Bug: https://bugzilla.redhat.com/show_bug.cgi?id=1538953
Reported-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Marcel Apfelbaum <marcel@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
The "i82801b11-bridge" device model is a descendant of "base-pci-bridge"
(TYPE_PCI_BRIDGE). However, unlike other similar devices, such as
- pci-bridge,
- pcie-pci-bridge,
- PCIE Root Port,
- xio3130 switch upstream and downstream ports,
- dec-21154-p2p-bridge,
- pbm-bridge,
- xilinx-pcie-root,
"i82801b11-bridge" does not clear the bridge specific registers at
platform reset.
This is a problem because devices on "i82801b11-bridge" continue to
respond to config space cycles after platform reset, when addressed with
the bus number that was previously programmed into the secondary bus
number register of "i82801b11-bridge". This error breaks OVMF's search for
extra (PXB) root buses, for example.
The device class reset method for "i82801b11-bridge" is currently NULL;
set it directly to pci_bridge_reset(), like the last three bridge models
in the above listing do.
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Marcel Apfelbaum <marcel@redhat.com>
Cc: qemu-stable@nongnu.org
Ref: https://bugzilla.redhat.com/show_bug.cgi?id=1541839
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Marcel Apfelbaum <marcel@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Move the log_dirty check into vhost_section.
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Now that the olf vhost_set_memory code is gone, the _nop and _add
callbacks are identical and can be merged. The _del callback is
no longer needed.
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Remove the old update mechanism, vhost_set_memory, and the functions
and flags it used.
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Compare the sections list that's just been generated, and if it's
different from the old one regenerate the region list.
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
As sections are reported by the listener to the _nop and _add
methods, add them to the temporary section list but now merge them
with the previous section if the new one abuts and the backend allows.
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
vhost_verify_ring_mappings() were used to verify that
rings are still accessible and related memory hasn't
been moved after flatview is updated.
It was doing checks by mapping ring's GPA+len and
checking that HVA hadn't changed with new memory map.
To avoid maybe expensive mapping call, we were
identifying address range that changed and were doing
mapping only if ring was in changed range.
However it's not neccessary to perform ring's GPA
mapping as we already have its current HVA and all
we need is to verify that ring's GPA translates to
the same HVA in updated flatview.
This will allow the following patches to simplify the range
comparison that was previously needed to avoid expensive
verify_ring_mapping calls.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
with modifications by:
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor spotted that there's a race, where a region that's unref'd
in a _del callback might be free'd before the set_mem_table call in
the _commit callback, and thus the vhost might end up using free memory.
Fix this by building a complete temporary sections list, ref'ing every
section (during add and nop) and then unref'ing the whole list right
at the end of commit.
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
The loading time of a VM is quite significant when its virtio
devices use a large amount of virt-queues (e.g. a virtio-serial
device with max_ports=511). Most of the time is spend in the
creation of all the required event notifiers (ioeventfd and memory
regions).
This patch pack all the changes to the memory regions in a
single memory transaction.
Reported-by: Sitong Liu
Reported-by: Xiaoling Gao
Signed-off-by: Gal Hammer <ghammer@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Tested-by: Greg Kurz <groug@kaod.org>
The virtio_bus_set_host_notifier function no longer calls
event_notifier_cleanup when a event notifier is removed.
The commit updates the code to match the new behavior and calls
virtio_bus_cleanup_host_notifier after the notifier was de-assign
and no longer in use.
This change is a preparation to allow executing the
virtio_bus_set_host_notifier function in a memory region
transaction.
Signed-off-by: Gal Hammer <ghammer@redhat.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Tested-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
This reverts commit 0750b06021.
Follow up patches are reworking the memory listeners, the new mechanism
will add its own set of traces.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
The x86 vector instruction set is extremely irregular. With newer
editions, Intel has filled in some of the blanks. However, we don't
get many 64-bit operations until SSE4.2, introduced in 2009.
The subsequent edition was for AVX1, introduced in 2011, which added
three-operand addressing, and adjusts how all instructions should be
encoded.
Given the relatively narrow 2 year window between possible to support
and desirable to support, and to vastly simplify code maintainence,
I am only planning to support AVX1 and later cpus.
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Yi Min Zhao