In order to guarantee compatibility on migration, QEMU should have
complete control over the features it announces to the guest via CPUID.
However, the availability of Hyper-V frequency MSRs
(HV_X64_MSR_TSC_FREQUENCY and HV_X64_MSR_APIC_FREQUENCY) depends solely
on the support for them in the underlying KVM.
Introduce "hv-frequencies" cpu property (off by default) which gives
QEMU full control over whether these MSRs are announced.
While at this, drop the redundant check of the cpu tsc frequency, and
decouple this feature from hv-time.
Signed-off-by: Roman Kagan <rkagan@virtuozzo.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20180330170209.20627-2-rkagan@virtuozzo.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Implements the CPUID trap for CPUID 1 to include the
CPUID_EXT_HYPERVISOR flag in the ECX results. This was preventing some
older linux kernels from booting when trying to access MSR's that dont
make sense when virtualized.
Signed-off-by: Justin Terry (VM) <juterry@microsoft.com>
Message-Id: <20180326170658.606-1-juterry@microsoft.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Check device having the feature of VIRTIO_CONSOLE_F_EMERG_WRITE before
get config->emerg_wr. It is neccessary because sizeof(virtio_console_config)
is 8 byte if VirtIOSerial doesn't have the feature of
VIRTIO_CONSOLE_F_EMERG_WRITE(see virtio_serial_device_realize),
read/write emerg_wr will lead to heap-over-flow.
Signed-off-by: linzhecheng <linzhecheng@huawei.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
My rework of section adding combines overlapping or adjoining regions,
but checks they're actually the same underlying RAM block.
Fix the case where two blocks adjoin but don't overlap; that new region
should get added (but not combined), but my previous patch was disallowing it.
Fixes: c1ece84e7c
Reported-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Tested-by: Alex Williamson <alex.williamson@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Without a dedicated protocol feature, QEMU cannot know whether
the backend can handle VHOST_USER_SET_CONFIG and
VHOST_USER_GET_CONFIG messages.
This patch adds a protocol feature that is only advertised by
QEMU if the device implements the config ops. Vhost user init
fails if the device support the feature but the backend doesn't.
The backend should only send VHOST_USER_SLAVE_CONFIG_CHANGE_MSG
requests if the protocol feature has been negotiated.
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Changpeng Liu <changpeng.liu@intel.com>
As soon as vhost-user init is done, the backend may send
VHOST_USER_SLAVE_CONFIG_CHANGE_MSG, so let's set the
notification callback before it.
Also, it will be used to know whether the device supports
the config feature to advertize it or not.
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Changpeng Liu <changpeng.liu@intel.com>
memtohex() adds an extra trailing NUL character.
Reported-by: AddressSanitizer
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20180408145933.1149-1-f4bug@amsat.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Let's simplify it a bit. On some weird circumstances we would have
tried to recompute watchpoints when running under KVM. load_psw() is
called from do_restart_interrupt() during a SIGP RESTART if the target
CPU is STOPPED. Let's touch watchpoints only in the TCG case - where
they are used for PER emulation.
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180409113019.14568-3-david@redhat.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
If we already triggered another exception, don't overwrite it with a
protection exception.
Only applies to old KVM instances without the virtual memory access
IOCTL in KVM.
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180409113019.14568-2-david@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
If the subchannel is already attached or if vfio_get_device() fails, the
code jumps to the 'out_device_err' label and doesn't free the string it
has just allocated.
The code should be reworked so that vcdev->vdev.name only gets set when
the device has been attached, and freed when it is about to be detached.
This could be achieved with the addition of a vfio_ccw_get_device()
function that would be the counterpart of vfio_put_device(). But this is
a more elaborate cleanup that should be done in a follow-up. For now,
let's just add calls to g_free() on the buggy error paths.
Signed-off-by: Greg Kurz <groug@kaod.org>
Message-Id: <152311222681.203086.8874800175539040298.stgit@bahia>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Contains the following commits:
- s390: Do not pass inofficial IPL type to the guest
For s390-netboot.img, this also contains the following commits (update
was forgotten last time):
- pc-bios/s390-ccw: Move string arrays from bootmap header to .c file
- pc-bios/s390-ccw: Increase virtio timeout to 30 seconds
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
IPL over a virtio-scsi device requires special handling not
available in the real architecture. For this purpose the IPL
type 0xFF has been chosen as means of communication between
QEMU and the pc-bios. However, a guest OS could be confused
by seeing an unknown IPL type.
This change sets the IPL parameter type to 0x02 (CCW) to prevent
this. Pre-existing Linux has looked up the IPL parameters only in
the case of FCP IPL. This means that the behavior should stay
the same even if Linux checks for the IPL type unconditionally.
Signed-off-by: Viktor Mihajlovski <mihajlov@linux.vnet.ibm.com>
Message-Id: <1522940844-12336-4-git-send-email-mihajlov@linux.vnet.ibm.com>
Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Operating systems may request an IPL from a virtio-scsi device
by specifying an IPL parameter type of CCW. In this case QEMU
won't set up the IPLB correctly. The BIOS will still detect
it's a SCSI device to boot from, but it will now have to search
for the first LUN and attempt to boot from there.
However this may not be the original boot LUN if there's more than
one SCSI disk attached to the HBA.
With this change QEMU will detect that the request is for a
SCSI device and will rebuild the initial IPL parameter info
if it's the SCSI device used for the first boot. In consequence
the BIOS can use the boot LUN from the IPL information block.
In case a different SCSI device has been set, the BIOS will find
and use the first available LUN.
Signed-off-by: Viktor Mihajlovski <mihajlov@linux.vnet.ibm.com>
Message-Id: <1522940844-12336-3-git-send-email-mihajlov@linux.vnet.ibm.com>
Reviewed-by: Farhan Ali <alifm@linux.vnet.ibm.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Splitting out the the CCW device extraction allows reuse.
Signed-off-by: Viktor Mihajlovski <mihajlov@linux.vnet.ibm.com>
Message-Id: <1522940844-12336-2-git-send-email-mihajlov@linux.vnet.ibm.com>
Reviewed-by: Farhan Ali <alifm@linux.vnet.ibm.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Manually having to use cpu_synchronize_state() is error prone. And as
Christian Borntraeger discovered, e.g. handle_diag() is currently
missing a cpu_synchronize_state(), as decode_basedisp_s() uses a
general purpose register value internally.
So let's do an overall cpu_synchronize_state(), which fixes at least the
one mentioned BUG. We will clean up the superfluous cpu_synchronize_state()
calls later.
We now also call it (although maybe not neded) for
- KVM_EXIT_S390_RESET -> s390_reipl_request()
- KVM_EXIT_DEBUG -> kvm_arch_handle_debug_exit()
- unmanagable/unimplemented intercepts
- ICPT_CPU_STOP -> do_stop_interrupt() -> cpu gets halted
- Scenarios where we inject an operation exception
- handle_stsi()
I don't think any of these are performance critical. Especially as we
have all information directly contained in kvm_run, there are no
additional IOCTLs to issue on modern kernels.
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20180406093552.13016-1-david@redhat.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Not needed, kbd_put_qcode_console() will handle that for us.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Message-id: 20180321135041.15768-5-kraxel@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
For both grapical and text consoles.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Message-id: 20180321135041.15768-3-kraxel@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Install optional dependencies of QEMU to get better coverage.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <1520942752-19449-1-git-send-email-pbonzini@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
We don't source common.rc where prep_fail is defined, so spell out the
commands and do what was intended.
Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <20180326090350.30014-1-famz@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
The apt-get commands we run through ssh expect certain features of the
tty, and refuses to work if /dev/null is used. It is ugly, but easy to
satisfy.
Actually, there is no reason to hide the output. It just makes things
harder to diagnose. We can always redirect in the Makefile, so don't do
it conditionally here.
Reported-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <20180322034753.6301-1-famz@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Tested-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Suggested-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20180315142713.30960-1-f4bug@amsat.org>
Signed-off-by: Fam Zheng <famz@redhat.com>
A mistake in the type passed to sizeof, that happens to work
when the out-of-line fallback itself is using host vectors,
but fails when using only the base types.
Tested-by: Emilio G. Cota <cota@braap.org>
Reported-by: Emilio G. Cota <cota@braap.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
The TLS test cert generation relies on a fixed set of algorithms that are
only usable under GNUTLS' default priority setting. When building QEMU
with a custom distro specific priority setting, this can cause the TLS
tests to fail. By forcing the tests to always use "NORMAL" priority we
can make them more robust.
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
This was added in 13f1243, but is missing from qemu-options.hx
Signed-off-by: Christophe Fergeau <cfergeau@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
The current docs for TLS assume only VNC is using TLS. Some of the information
is also outdated (ie lacking subject alt name info for certs). Rewrite it to
more accurately reflect the current situation.
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kashyap Chamarthy <kchamart@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Commit 567b5b309a ("vfio/pci: Relax DMA map errors for MMIO regions")
added an error message if a passed memory section address or size
is not aligned to the page size and thus cannot be DMA mapped.
This patch fixes the trace by printing the region name and the
memory region section offset within the address space (instead of
offset_within_region).
We also turn the error_report into a trace event. Indeed, In some
cases, the traces can be confusing to non expert end-users and
let think the use case does not work (whereas it works as before).
This is the case where a BAR is successively mapped at different
GPAs and its sections are not compatible with dma map. The listener
is called several times and traces are issued for each intermediate
mapping. The end-user cannot easily match those GPAs against the
final GPA output by lscpi. So let's keep those information to
informed users. In mid term, the plan is to advise the user about
BAR relocation relevance.
Fixes: 567b5b309a ("vfio/pci: Relax DMA map errors for MMIO regions")
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Commit 2b9aef6fcd introduced a regression:
checkpatch.pl started complaining about the following valid pattern:
do {
/* something */
} while (condition);
Fix the script to once again permit this pattern.
Signed-off-by: Su Hang <suhang16@mails.ucas.ac.cn>
Message-Id: <1522029982-4650-1-git-send-email-suhang16@mails.ucas.ac.cn>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
In commit 7073fbada7, the `andn` instruction
was implemented via `tcg_gen_andc` but passes the operands in the wrong
order:
- X86 defines `andn dest,src1,src2` as: dest = ~src1 & src2
- TCG defines `andc dest,src1,src2` as: dest = src1 & ~src2
The following simple test shows the issue:
#include <stdio.h>
#include <stdint.h>
int main(void) {
uint32_t ret = 0;
__asm (
"mov $0xFF00, %%ecx\n"
"mov $0x0F0F, %%eax\n"
"andn %%ecx, %%eax, %%ecx\n"
"mov %%ecx, %0\n"
: "=r" (ret));
printf("%08X\n", ret);
return 0;
}
This patch fixes the problem by simply swapping the order of the two last
arguments in `tcg_gen_andc_tl`.
Reported-by: Alexandro Sanchez Bach <alexandro@phi.nz>
Signed-off-by: Alexandro Sanchez Bach <alexandro@phi.nz>
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Our rule right now is to use <> for external headers only.
util/sys_membarrier.c violates that. Fix it up.
Signed-off-by: Bruce Rogers <brogers@suse.com>
Message-Id: <20180329151018.15319-1-brogers@suse.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Commit 4bfb274 added some QAPIfication of option parsing in
qemu_rbd_open(). We need to remove all the options we processed,
otherwise in bdrv_open_inherit() we will think the remaining options are
invalid.
(This needs to go in 2.12 to avoid a regression that prevents rbd
from being opened.)
Suggested-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Jeff Cody <jcody@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
configure tries to detect if the compiler
supports 16-byte vector operations.
As stated in the comment of the detection
program, there is a problem with the system
compiler on GCC on Centos 7.
This program doesn't actually detect the problem
with GCC on RHEL7 on PPC64LE (Red Hat 4.8.5-28).
This patch updates the test to look more like
it is in QEMU helpers, and now detects the problem.
The error reported is:
CC ppc64-softmmu/accel/tcg/tcg-runtime-gvec.o
..//accel/tcg/tcg-runtime-gvec.c: In function ‘helper_gvec_shl8i’:
../accel/tcg/tcg-runtime-gvec.c:558:26: internal compiler error: in emit_move_insn, at expr.c:3495
*(vec8 *)(d + i) = *(vec8 *)(a + i) << shift;
^
Fixes: db43267 "tcg: Add generic vector expanders"
Signed-off-by: Laurent Vivier <lvivier@redhat.com>
Reviewed-by: Miroslav Rezanina <mrezanin@redhat.com>
Message-id: 20180328133152.24623-1-lvivier@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
