sdk/emulator/qemu.git
10 years agopc-bios/s390-ccw Really big EAV ECKD DASD handling
Eugene (jno) Dvurechenski [Fri, 29 Aug 2014 09:01:39 +0000 (11:01 +0200)]
pc-bios/s390-ccw Really big EAV ECKD DASD handling

For EAV ECKD DASD, the cylinder count will have the magic value
0xfffeU. Therefore, use the block number to test for valid eckd
addresses instead.

Signed-off-by: Eugene (jno) Dvurechenski <jno@linux.vnet.ibm.com>
Acked-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Jens Freimann <jfrei@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
10 years agopc-bios/s390-ccw Improve ECKD informational message
Eugene (jno) Dvurechenski [Fri, 29 Aug 2014 09:01:38 +0000 (11:01 +0200)]
pc-bios/s390-ccw Improve ECKD informational message

Add block size display to ECKD scheme report.

Signed-off-by: Eugene (jno) Dvurechenski <jno@linux.vnet.ibm.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Acked-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Jens Freimann <jfrei@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
10 years agopc-bios/s390-ccw: handle more ECKD DASD block sizes
Eugene (jno) Dvurechenski [Fri, 29 Aug 2014 09:01:37 +0000 (11:01 +0200)]
pc-bios/s390-ccw: handle more ECKD DASD block sizes

Using dasdfmt(8) to format a DASD allows to choose a block size.
There are four supported values: 512, 1024, 2048, and 4096 bytes
per block. Each block size leads to selection of new count of
sectors per track. The head count remains always the same: 15.

This empiric knowledge is used to detect ECKD DASD to IPL from.

Signed-off-by: Eugene (jno) Dvurechenski <jno@linux.vnet.ibm.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Jens Freimann <jfrei@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
10 years agopc-bios/s390-ccw: support all virtio block size
Eugene (jno) Dvurechenski [Fri, 29 Aug 2014 09:01:36 +0000 (11:01 +0200)]
pc-bios/s390-ccw: support all virtio block size

The block size value may be given "as is" OR as a base value and
a shift count (exponent). So, we have to use calculation to get
the proper number in the code.

The main expression reads as
        (blk_cfg.blk_size << blk_cfg.physical_block_exp)

E.g., various combinations between blk_size=1/physical_block_exp=12
and blk_size=4096/physical_block_exp=0 are valid for 4K blocks.

Signed-off-by: Eugene (jno) Dvurechenski <jno@linux.vnet.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Jens Freimann <jfrei@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
10 years agos390x/kvm: execute the first cpu reset on the vcpu thread
David Hildenbrand [Thu, 28 Aug 2014 11:58:52 +0000 (13:58 +0200)]
s390x/kvm: execute the first cpu reset on the vcpu thread

As all full cpu resets currently call into the kernel to do initial cpu reset,
let's run this reset (triggered by cpu_s390x_init()) on the proper vcpu thread.

Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Jens Freimann <jfrei@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
10 years agos390x/kvm: execute "system reset" cpu resets on the vcpu thread
David Hildenbrand [Thu, 28 Aug 2014 11:58:51 +0000 (13:58 +0200)]
s390x/kvm: execute "system reset" cpu resets on the vcpu thread

Let's execute resets triggered by qemu system resets on the target vcpu thread.
This will avoid synchronize_rcu's in the kernel.

Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Jens Freimann <jfrei@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
10 years agos390x/kvm: execute sigp orders on the target vcpu thread
David Hildenbrand [Thu, 28 Aug 2014 11:58:50 +0000 (13:58 +0200)]
s390x/kvm: execute sigp orders on the target vcpu thread

All sigp orders that can result in ioctls on the target vcpu should be executed
on the associated vcpu thread.

Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Jens Freimann <jfrei@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
10 years agos390x/kvm: run guest triggered resets on the target vcpu thread
David Hildenbrand [Thu, 28 Aug 2014 11:58:49 +0000 (13:58 +0200)]
s390x/kvm: run guest triggered resets on the target vcpu thread

Currently, load_normal_reset() and modified_clear_reset() as triggered
by a guest vcpu will initiate cpu resets on the current vcpu thread for
all cpus. The reset should happen on the individual vcpu thread
instead, so let's use run_on_cpu() for this.

This avoids calls to synchronize_rcu() in the kernel.

Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Jens Freimann <jfrei@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
10 years agoMerge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-20140829' into...
Peter Maydell [Fri, 29 Aug 2014 14:48:15 +0000 (15:48 +0100)]
Merge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-20140829' into staging

target-arm queue:
 * support PMCCNTR in ARMv8
 * various GIC fixes and cleanups
 * Correct Cortex-A57 ISAR5 and AA64ISAR0 ID register values
 * Fix regression that disabled VFP for ARMv5 CPUs
 * Update to upstream VIXL 1.5

# gpg: Signature made Fri 29 Aug 2014 15:34:47 BST using RSA key ID 14360CDE
# gpg: Good signature from "Peter Maydell <peter.maydell@linaro.org>"

* remotes/pmaydell/tags/pull-target-arm-20140829:
  target-arm: Implement pmccfiltr_write function
  target-arm: Remove old code and replace with new functions
  target-arm: Implement pmccntr_sync function
  target-arm: Add arm_ccnt_enabled function
  target-arm: Implement PMCCNTR_EL0 and related registers
  arm: Implement PMCCNTR 32b read-modify-write
  target-arm: Make the ARM PMCCNTR register 64-bit
  hw/intc/arm_gic: honor target mask in gic_update()
  aarch64: raise max_cpus to 8
  arm_gic: Use GIC_NR_SGIS constant
  arm_gic: Do not force PPIs to edge-triggered mode
  arm_gic: GICD_ICFGR: Write model only for pre v1 GICs
  arm_gic: Fix read of GICD_ICFGR
  target-arm: Correct Cortex-A57 ISAR5 and AA64ISAR0 ID register values
  target-arm: Fix regression that disabled VFP for ARMv5 CPUs
  disas/libvixl: Update to upstream VIXL 1.5

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agotarget-arm: Implement pmccfiltr_write function
Alistair Francis [Fri, 29 Aug 2014 14:00:30 +0000 (15:00 +0100)]
target-arm: Implement pmccfiltr_write function

This is the function that is called when writing to the
PMCCFILTR_EL0 register

Signed-off-by: Alistair Francis <alistair.francis@xilinx.com>
Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: 73da3da6404855b17d5ae82975a32ff3a4dcae3d.1409025949.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agotarget-arm: Remove old code and replace with new functions
Alistair Francis [Fri, 29 Aug 2014 14:00:30 +0000 (15:00 +0100)]
target-arm: Remove old code and replace with new functions

Remove the old PMCCNTR code and replace it with calls to the new
pmccntr_sync() and arm_ccnt_enabled() functions.

Signed-off-by: Alistair Francis <alistair.francis@xilinx.com>
Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: 693a6e437d915c2195fd3dc7303f384ca538b7bf.1409025949.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agotarget-arm: Implement pmccntr_sync function
Alistair Francis [Fri, 29 Aug 2014 14:00:29 +0000 (15:00 +0100)]
target-arm: Implement pmccntr_sync function

This is used to synchronise the PMCCNTR counter and swap its
state between enabled and disabled if required. It must always
be called twice, both before and after any logic that could
change the state of the PMCCNTR counter.

Signed-off-by: Alistair Francis <alistair.francis@xilinx.com>
Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: 62811d4c0f7b1384f7aab62ea2fcfda3dcb0db50.1409025949.git.peter.crosthwaite@xilinx.com
[PMM: fixed minor typos in pmccntr_sync doc comment]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agotarget-arm: Add arm_ccnt_enabled function
Alistair Francis [Fri, 29 Aug 2014 14:00:29 +0000 (15:00 +0100)]
target-arm: Add arm_ccnt_enabled function

Include a helper function to determine if the CCNT counter
is enabled.

Signed-off-by: Alistair Francis <alistair.francis@xilinx.com>
Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: e1a64f17a756e06c8bda8238ad4826d705049f7a.1409025949.git.peter.crosthwaite@xilinx.com
[ PC changes
  * Remove EL based checks
]
Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agotarget-arm: Implement PMCCNTR_EL0 and related registers
Alistair Francis [Fri, 29 Aug 2014 14:00:29 +0000 (15:00 +0100)]
target-arm: Implement PMCCNTR_EL0 and related registers

This patch adds support for the ARMv8 version of the PMCCNTR and
related registers. It also starts to implement the PMCCFILTR_EL0
register.

Signed-off-by: Alistair Francis <alistair.francis@xilinx.com>
Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: b5d1094764a5416363ee95216799b394ecd011e8.1409025949.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agoarm: Implement PMCCNTR 32b read-modify-write
Peter Crosthwaite [Fri, 29 Aug 2014 14:00:29 +0000 (15:00 +0100)]
arm: Implement PMCCNTR 32b read-modify-write

The register is now 64bit, however a 32 bit write to the register
should leave the higher bits unchanged. The open coded write handler
does not implement this, so we need to read-modify-write accordingly.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Alistair Francis <alistair23@gmail.com>
Message-id: ec350573424bb2adc1701c3b9278d26598e2f2d1.1409025949.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agotarget-arm: Make the ARM PMCCNTR register 64-bit
Alistair Francis [Fri, 29 Aug 2014 14:00:29 +0000 (15:00 +0100)]
target-arm: Make the ARM PMCCNTR register 64-bit

This makes the PMCCNTR register 64-bit to allow for the
64-bit ARMv8 version.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Signed-off-by: Alistair Francis <alistair.francis@xilinx.com>
Message-id: 6c5bac5fd0ea54963b1fc0e7f9464909f2e19a73.1409025949.git.peter.crosthwaite@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agohw/intc/arm_gic: honor target mask in gic_update()
Sergey Fedorov [Fri, 29 Aug 2014 14:00:29 +0000 (15:00 +0100)]
hw/intc/arm_gic: honor target mask in gic_update()

Take IRQ target mask into account when determining the highest priority
pending interrupt.

Signed-off-by: Sergey Fedorov <serge.fdrv@gmail.com>
Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Message-id: 1407947471-26981-1-git-send-email-serge.fdrv@gmail.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agoaarch64: raise max_cpus to 8
Joel Schopp [Fri, 29 Aug 2014 14:00:29 +0000 (15:00 +0100)]
aarch64: raise max_cpus to 8

I'm running on a system with 8 cpus and it would be nice to have qemu
support all of them.  The attached patch does that and has been tested.

That said, I'm not sure if 8 is enough or if we want to bump this even higher
now before systems with many more cpus come along. 255 anyone?

Cc: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Joel Schopp <joel.schopp@amd.com>
Message-id: 20140819213304.19537.2834.stgit@joelaarch64.amd.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agoarm_gic: Use GIC_NR_SGIS constant
Adam Lackorzynski [Fri, 29 Aug 2014 14:00:29 +0000 (15:00 +0100)]
arm_gic: Use GIC_NR_SGIS constant

Use constant rather than a plain number.

Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Signed-off-by: Adam Lackorzynski <adam@os.inf.tu-dresden.de>
Message-id: 1408372255-12358-5-git-send-email-adam@os.inf.tu-dresden.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agoarm_gic: Do not force PPIs to edge-triggered mode
Adam Lackorzynski [Fri, 29 Aug 2014 14:00:28 +0000 (15:00 +0100)]
arm_gic: Do not force PPIs to edge-triggered mode

Only SGIs must be WI, done by forcing them to their default
(edge-triggered).

Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Signed-off-by: Adam Lackorzynski <adam@os.inf.tu-dresden.de>
Message-id: 1408372255-12358-4-git-send-email-adam@os.inf.tu-dresden.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agoarm_gic: GICD_ICFGR: Write model only for pre v1 GICs
Adam Lackorzynski [Fri, 29 Aug 2014 14:00:28 +0000 (15:00 +0100)]
arm_gic: GICD_ICFGR: Write model only for pre v1 GICs

Setting the model is only available in pre-v1 GIC models.

Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Signed-off-by: Adam Lackorzynski <adam@os.inf.tu-dresden.de>
Message-id: 1408372255-12358-3-git-send-email-adam@os.inf.tu-dresden.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agoarm_gic: Fix read of GICD_ICFGR
Adam Lackorzynski [Fri, 29 Aug 2014 14:00:28 +0000 (15:00 +0100)]
arm_gic: Fix read of GICD_ICFGR

The GICD_ICFGR register covers 4 interrupts per byte.

Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Signed-off-by: Adam Lackorzynski <adam@os.inf.tu-dresden.de>
Message-id: 1408372255-12358-2-git-send-email-adam@os.inf.tu-dresden.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agotarget-arm: Correct Cortex-A57 ISAR5 and AA64ISAR0 ID register values
Peter Maydell [Fri, 29 Aug 2014 14:00:28 +0000 (15:00 +0100)]
target-arm: Correct Cortex-A57 ISAR5 and AA64ISAR0 ID register values

We implement the crypto extensions but were incorrectly reporting
ID register values for the Cortex-A57 which did not advertise
crypto. Use the correct values as described in the TRM.
With this fix Linux correctly detects presence of the crypto
features and advertises them in /proc/cpuinfo.

Reported-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1408718660-7295-1-git-send-email-peter.maydell@linaro.org
Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agotarget-arm: Fix regression that disabled VFP for ARMv5 CPUs
Peter Maydell [Fri, 29 Aug 2014 14:00:28 +0000 (15:00 +0100)]
target-arm: Fix regression that disabled VFP for ARMv5 CPUs

Commit 2c7ffc414 added support for honouring the CPACR coprocessor
access control register bits which may disable access to VFP
and Neon instructions. However it failed to account for the
fact that the CPACR is only present starting from the ARMv6
architecture version, so it accidentally disabled VFP completely
for ARMv5 CPUs like the ARM926. Linux would detect this as
"no VFP present" and probably fall back to its own emulation,
but other guest OSes might crash or misbehave.

This fixes bug LP:1359930.

Reported-by: Jakub Jermar <jakub@jermar.eu>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1408714940-7192-1-git-send-email-peter.maydell@linaro.org
Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agodisas/libvixl: Update to upstream VIXL 1.5
Peter Maydell [Fri, 29 Aug 2014 14:00:27 +0000 (15:00 +0100)]
disas/libvixl: Update to upstream VIXL 1.5

Update our copy of libvixl to upstream's 1.5 release.
This includes the upstream versions of the fixes we
were carrying locally (commit ffebe899).

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1407162987-4659-1-git-send-email-peter.maydell@linaro.org

10 years agoMerge remote-tracking branch 'remotes/kraxel/tags/pull-usb-20140829-1' into staging
Peter Maydell [Fri, 29 Aug 2014 12:08:04 +0000 (13:08 +0100)]
Merge remote-tracking branch 'remotes/kraxel/tags/pull-usb-20140829-1' into staging

usb: bugfix collection.
usb: add cleanup functions for host adapters,
     in preparation for hotplug support.
usb: add simple qtests for uhci,ohci,xhci.

# gpg: Signature made Fri 29 Aug 2014 12:56:20 BST using RSA key ID D3E87138
# gpg: Good signature from "Gerd Hoffmann (work) <kraxel@redhat.com>"
# gpg:                 aka "Gerd Hoffmann <gerd@kraxel.org>"
# gpg:                 aka "Gerd Hoffmann (private) <kraxel@gmail.com>"

* remotes/kraxel/tags/pull-usb-20140829-1:
  tests: add xHCI qtest
  tests: add UHCI qtest
  tests: add OHCI qtest
  usb: add usb host adapters exit trace
  usb-xhci: add exit function
  usb-ehci: add ehci-pci device exit function
  usb-ehci: add ehci unrealize funciton
  usb-ehci: add vmstate properity for EHCIState
  usb-uhci: clean up uhci resource when pci-uhci exit
  usb-ohci: add exit function
  usb-ohci: Fix memory leak for ohci timer
  usb: add usb_bus_release function
  Revert "xhci: Fix number of streams allocated when using streams"
  xhci: use (1u << i)
  Fix OHCI ISO TD state never being written back.
  xhci: fix debug print compiling error
  usb: Fix bootindex for portnr > 9

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agotests: add xHCI qtest
Gonglei [Mon, 23 Jun 2014 11:53:53 +0000 (19:53 +0800)]
tests: add xHCI qtest

Signed-off-by: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
10 years agotests: add UHCI qtest
Gonglei [Mon, 23 Jun 2014 11:53:52 +0000 (19:53 +0800)]
tests: add UHCI qtest

Signed-off-by: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
10 years agotests: add OHCI qtest
Gonglei [Mon, 23 Jun 2014 11:53:51 +0000 (19:53 +0800)]
tests: add OHCI qtest

Signed-off-by: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
10 years agousb: add usb host adapters exit trace
Gonglei [Wed, 4 Jun 2014 08:31:55 +0000 (16:31 +0800)]
usb: add usb host adapters exit trace

Signed-off-by: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
10 years agousb-xhci: add exit function
Gonglei [Wed, 4 Jun 2014 08:31:53 +0000 (16:31 +0800)]
usb-xhci: add exit function

clean up xhci resource when xhci pci device exit.

Signed-off-by: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
10 years agousb-ehci: add ehci-pci device exit function
Gonglei [Wed, 4 Jun 2014 08:31:52 +0000 (16:31 +0800)]
usb-ehci: add ehci-pci device exit function

clean up ehci resource when ehci pci device exit.

Signed-off-by: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
10 years agousb-ehci: add ehci unrealize funciton
Gonglei [Wed, 4 Jun 2014 08:31:51 +0000 (16:31 +0800)]
usb-ehci: add ehci unrealize funciton

cleanup ehci controller resource, both pci and sysbus
if they're necessary.

Signed-off-by: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
10 years agousb-ehci: add vmstate properity for EHCIState
Gonglei [Wed, 4 Jun 2014 08:31:50 +0000 (16:31 +0800)]
usb-ehci: add vmstate properity for EHCIState

since hotunplug the ehci host adapter, we should
delete vm_change_state_handler also, so the
VMChangeStateEntry should be saved in EHCIState.

Signed-off-by: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
10 years agousb-uhci: clean up uhci resource when pci-uhci exit
Gonglei [Wed, 4 Jun 2014 08:31:49 +0000 (16:31 +0800)]
usb-uhci: clean up uhci resource when pci-uhci exit

clean up uhci resource when uhci pci device exit.

Signed-off-by: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
10 years agousb-ohci: add exit function
Gonglei [Wed, 4 Jun 2014 08:31:48 +0000 (16:31 +0800)]
usb-ohci: add exit function

clean up ohci resource when ohci pci device exit.

Signed-off-by: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
10 years agousb-ohci: Fix memory leak for ohci timer
Gonglei [Wed, 4 Jun 2014 08:31:47 +0000 (16:31 +0800)]
usb-ohci: Fix memory leak for ohci timer

Signed-off-by: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
10 years agousb: add usb_bus_release function
Gonglei [Wed, 4 Jun 2014 08:31:46 +0000 (16:31 +0800)]
usb: add usb_bus_release function

add global variables releasing logic when the usb buses
were removed or hot-unpluged.

Signed-off-by: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
10 years agoRevert "xhci: Fix number of streams allocated when using streams"
Gerd Hoffmann [Fri, 29 Aug 2014 10:40:55 +0000 (12:40 +0200)]
Revert "xhci: Fix number of streams allocated when using streams"

This reverts commit d063c3112c4cd23a479ee18720c2bd119da2d315.

"2 << x" is the same as "2 ^ (x + 1)", so the old code is correct.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
10 years agoxhci: use (1u << i)
Gerd Hoffmann [Thu, 28 Aug 2014 08:51:35 +0000 (10:51 +0200)]
xhci: use (1u << i)

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
10 years agoFix OHCI ISO TD state never being written back.
Jack Un [Sat, 9 Aug 2014 20:34:36 +0000 (23:34 +0300)]
Fix OHCI ISO TD state never being written back.

There appears to be typo in OHCI with isochronous transfers
resulting in isoch. transfer descriptor state never being written back.
The'put_words' function is in a OR statement hence it is never called.

Signed-off-by: Jack Un <jack.un@gmail.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
10 years agoxhci: fix debug print compiling error
Gonglei [Thu, 21 Aug 2014 12:48:58 +0000 (20:48 +0800)]
xhci: fix debug print compiling error

after commit 003e15a180373048f0c1f4df0bfe303746eb2676
the DPRINTF will broke compiling, adjust its location.

Signed-off-by: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
10 years agousb: Fix bootindex for portnr > 9
Markus Armbruster [Fri, 15 Aug 2014 11:32:36 +0000 (13:32 +0200)]
usb: Fix bootindex for portnr > 9

We identify devices by their Open Firmware device paths.  The encoding
of the host controller and hub port numbers is incorrect:
usb_get_fw_dev_path() formats them in decimal, while SeaBIOS uses
hexadecimal.  When some port number > 9, SeaBIOS will miss the
bootindex (lucky case), or apply it to another device (unlucky case).

The relevant spec[*] agrees with SeaBIOS (and OVMF, for that matter).
Change %d to %x.

Bug can bite only with host controllers or hubs sporting more than ten
ports.  I'm not aware of any.

[*] Open Firmware Recommended Practice: Universal Serial Bus,
Version 1, Section 3.2.1 Device Node Address Representation
http://www.openfirmware.org/1275/bindings/usb/usb-1_0.ps

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Note: xhci can be configured with up to 15 ports (default is 4 ports).

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
10 years agoMerge remote-tracking branch 'remotes/bonzini/tags/for-upstream' into staging
Peter Maydell [Thu, 28 Aug 2014 16:08:13 +0000 (17:08 +0100)]
Merge remote-tracking branch 'remotes/bonzini/tags/for-upstream' into staging

SCSI patches include bug fixes from Fam and Peter, improved error
reporting from Fam and a fix for DPRINTF bitrot.  Memory patches try
again to initialize name from the QOM name.

# gpg: Signature made Thu 28 Aug 2014 15:10:31 BST using RSA key ID 9B4D86F2
# gpg: Good signature from "Paolo Bonzini <pbonzini@redhat.com>"
# gpg:                 aka "Paolo Bonzini <bonzini@gnu.org>"

* remotes/bonzini/tags/for-upstream:
  memory: Lazy init name from QOM name as needed
  xen: hvm: Abstract away memory region name ref
  xen-hvm: Constify string
  virtio-scsi: Report error if num_queues is 0 or too large
  scsi-generic: remove superfluous DPRINTF avoid to break compiling
  block/iscsi: fix memory corruption on iscsi resize
  scsi-bus: Convert DeviceClass init to realize
  block: Pass errp in blkconf_geometry

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agoMerge remote-tracking branch 'remotes/kvm/tags/for-upstream' into staging
Peter Maydell [Thu, 28 Aug 2014 15:07:23 +0000 (16:07 +0100)]
Merge remote-tracking branch 'remotes/kvm/tags/for-upstream' into staging

Mostly bugfixes + Alexey's interface-based implementation
of the NMI monitor command.

# gpg: Signature made Thu 28 Aug 2014 15:07:22 BST using RSA key ID 9B4D86F2
# gpg: Good signature from "Paolo Bonzini <pbonzini@redhat.com>"
# gpg:                 aka "Paolo Bonzini <bonzini@gnu.org>"

* remotes/kvm/tags/for-upstream:
  mc146818rtc: reinitialize irq_reinject_on_ack_count on reset
  target-i386: Add "tsc_adjust" CPU feature name
  target-i386: Add "mpx" CPU feature name
  vl: process -object after other backend options
  checkpatch.pl: adjust typedef definition to QEMU coding style
  x86: Clear MTRRs on vCPU reset
  x86: kvm: Add MTRR support for kvm_get|put_msrs()
  x86: Use common variable range MTRR counts
  target-i386: Don't forbid NX bit on PAE PDEs and PTEs
  spapr: Add support for new NMI interface
  s390x: Migrate to new NMI interface
  s390x: Convert QEMUMachine to MachineClass
  cpus: Define callback for QEMU "nmi" command
  kvm: run cpu state synchronization on target vcpu thread

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agomemory: Lazy init name from QOM name as needed
Peter Crosthwaite [Tue, 26 Aug 2014 03:10:24 +0000 (20:10 -0700)]
memory: Lazy init name from QOM name as needed

To support name retrieval of MemoryRegions that were created
dynamically (that is, not via memory_region_init and friends). We
cache the name in MemoryRegion's state as
object_get_canonical_path_component mallocs the returned value
so it's not suitable for direct return to callers. Memory already
frees the name field, so this will be garbage collected along with
the MR object.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 years agoxen: hvm: Abstract away memory region name ref
Peter Crosthwaite [Tue, 26 Aug 2014 03:09:48 +0000 (20:09 -0700)]
xen: hvm: Abstract away memory region name ref

The mr->name field is removed. This slipped through compile testing.
Fix.

Reviewed-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 years agoxen-hvm: Constify string
Peter Crosthwaite [Tue, 26 Aug 2014 03:09:13 +0000 (20:09 -0700)]
xen-hvm: Constify string

It's constant, and sourced from existing const strings. Avoid dodgy
casts by converting to const.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 years agoMerge remote-tracking branch 'remotes/stefanha/tags/fix-buildbot-12082014-pull-reques...
Peter Maydell [Thu, 28 Aug 2014 13:51:12 +0000 (14:51 +0100)]
Merge remote-tracking branch 'remotes/stefanha/tags/fix-buildbot-12082014-pull-request' into staging

Pull request

# gpg: Signature made Thu 28 Aug 2014 13:43:00 BST using RSA key ID 81AB73C8
# gpg: Good signature from "Stefan Hajnoczi <stefanha@redhat.com>"
# gpg:                 aka "Stefan Hajnoczi <stefanha@gmail.com>"

* remotes/stefanha/tags/fix-buildbot-12082014-pull-request:
  Revert "qemu-img: sort block formats in help message"
  block: sort formats alphabetically in bdrv_iterate_format()
  mirror: fix uninitialized variable delay_ns warnings
  trace: avoid Python 2.5 all() in tracetool
  libqtest: launch QEMU with QEMU_AUDIO_DRV=none
  qapi.py: avoid Python 2.5+ any() function

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agoRevert "qemu-img: sort block formats in help message"
Stefan Hajnoczi [Wed, 27 Aug 2014 11:08:56 +0000 (12:08 +0100)]
Revert "qemu-img: sort block formats in help message"

This reverts commit 1a443c1b8b4314d365e82bddeb1de5b4b1c15fb3 and the
later commit 395071a76328189f50c778f4dee6dabb90503dd9.

GSequence was introduced in glib 2.14.  RHEL 5 fails to compile since it
uses glib 2.12.3.

Now that bdrv_iterate_format() invokes the iteration callback in sorted
order these commits are unnecessary.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Benoît Canet <benoit.canet@nodalink.com>
10 years agoblock: sort formats alphabetically in bdrv_iterate_format()
Stefan Hajnoczi [Wed, 27 Aug 2014 11:08:55 +0000 (12:08 +0100)]
block: sort formats alphabetically in bdrv_iterate_format()

Format names are best consumed in alphabetical order.  This makes
human-readable output easy to produce.

bdrv_iterate_format() already has an array of format strings.  Sort them
before invoking the iteration callback.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Benoît Canet <benoit.canet@nodalink.com>
10 years agomirror: fix uninitialized variable delay_ns warnings
Stefan Hajnoczi [Wed, 27 Aug 2014 11:08:54 +0000 (12:08 +0100)]
mirror: fix uninitialized variable delay_ns warnings

The gcc 4.1.2 compiler warns that delay_ns may be uninitialized in
mirror_iteration().

There are two break statements in the do ... while loop that skip over
the delay_ns assignment.  These are probably the cause of the warning.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Benoît Canet <benoit.canet@nodalink.com>
10 years agotrace: avoid Python 2.5 all() in tracetool
Stefan Hajnoczi [Wed, 27 Aug 2014 11:08:53 +0000 (12:08 +0100)]
trace: avoid Python 2.5 all() in tracetool

Red Hat Enterprise Linux 5 ships Python 2.4.3.  The all() function was
added in Python 2.5 so we cannot use it.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Benoît Canet <benoit.canet@nodalink.com>
10 years agolibqtest: launch QEMU with QEMU_AUDIO_DRV=none
Stefan Hajnoczi [Wed, 27 Aug 2014 11:08:52 +0000 (12:08 +0100)]
libqtest: launch QEMU with QEMU_AUDIO_DRV=none

No test case actually uses the audio backend.  Disable audio to prevent
warnings on hosts with no sound hardware present:

  GTESTER check-qtest-aarch64
  sdl: SDL_OpenAudio failed
  sdl: Reason: No available audio device
  sdl: SDL_OpenAudio failed
  sdl: Reason: No available audio device
  audio: Failed to create voice `lm4549.out'

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
10 years agoqapi.py: avoid Python 2.5+ any() function
Stefan Hajnoczi [Wed, 27 Aug 2014 11:08:51 +0000 (12:08 +0100)]
qapi.py: avoid Python 2.5+ any() function

There is one instance of any() in qapi.py that breaks builds on older
distros that ship Python 2.4 (like RHEL5):

  GEN   qmp-commands.h
Traceback (most recent call last):
  File "build/scripts/qapi-commands.py", line 445, in ?
    exprs = parse_schema(input_file)
  File "build/scripts/qapi.py", line 329, in parse_schema
    schema = QAPISchema(open(input_file, "r"))
  File "build/scripts/qapi.py", line 110, in __init__
    if any(include_path == elem[1]
NameError: global name 'any' is not defined

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Benoît Canet <benoit.canet@nodalink.com>
10 years agomc146818rtc: reinitialize irq_reinject_on_ack_count on reset
Paolo Bonzini [Wed, 27 Aug 2014 15:54:52 +0000 (17:54 +0200)]
mc146818rtc: reinitialize irq_reinject_on_ack_count on reset

This field was forgotten, and it makes the state after reset
non-deterministic.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 years agoMerge remote-tracking branch 'remotes/mcayland/qemu-openbios' into staging
Peter Maydell [Tue, 26 Aug 2014 13:18:40 +0000 (14:18 +0100)]
Merge remote-tracking branch 'remotes/mcayland/qemu-openbios' into staging

* remotes/mcayland/qemu-openbios:
  Update OpenBIOS images

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agotarget-i386: Add "tsc_adjust" CPU feature name
Eduardo Habkost [Mon, 25 Aug 2014 20:02:13 +0000 (17:02 -0300)]
target-i386: Add "tsc_adjust" CPU feature name

tsc_adjust migration support is already implemented (commit
f28558d3d37ad3bc4e35e8ac93f7bf81a0d5622c), so we can add it to the list
of known feature names.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 years agotarget-i386: Add "mpx" CPU feature name
Eduardo Habkost [Mon, 25 Aug 2014 20:02:12 +0000 (17:02 -0300)]
target-i386: Add "mpx" CPU feature name

Migration support for MPX is already implemented (commit
79e9ebebbf2a00c46fcedb6dc7dd5e12bbd30216), so we can add it to the list
of known feature names.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 years agoUpdate OpenBIOS images
Mark Cave-Ayland [Tue, 26 Aug 2014 12:52:15 +0000 (13:52 +0100)]
Update OpenBIOS images

Update OpenBIOS images to SVN r1316 built from submodule.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
10 years agovl: process -object after other backend options
Paolo Bonzini [Mon, 25 Aug 2014 11:47:00 +0000 (13:47 +0200)]
vl: process -object after other backend options

QOM backends can refer to chardevs, but not vice versa.  So
process -chardev and -fsdev options before -object

This fixes the rng-egd backend to virtio-rng.

Reported-by: Amos Kong <akong@redhat.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 years agocheckpatch.pl: adjust typedef definition to QEMU coding style
Paolo Bonzini [Tue, 10 Jun 2014 08:52:02 +0000 (10:52 +0200)]
checkpatch.pl: adjust typedef definition to QEMU coding style

Most QEMU typedefs are camelcase, starting with one uppercase letter
and containing at least one lowercase letter.  There are a few
all-uppercase types, add the most common too.

This fixes recognition of types in lines such as

    static __attribute__((unused)) inline void tcg_out8(TCGContext *s, uint8_t v)

(Example provided by Peter Maydell).

Reported-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Cc: Peter Maydell <peter.maydell@linaro.org>
Cc: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 years agovirtio-scsi: Report error if num_queues is 0 or too large
Fam Zheng [Tue, 26 Aug 2014 06:30:30 +0000 (14:30 +0800)]
virtio-scsi: Report error if num_queues is 0 or too large

No cmd vq surprises guest (Linux panics in virtscsi_probe), too many
queues abort qemu (in the following virtio_add_queue).

Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 years agoscsi-generic: remove superfluous DPRINTF avoid to break compiling
Gonglei [Fri, 22 Aug 2014 02:01:50 +0000 (10:01 +0800)]
scsi-generic: remove superfluous DPRINTF avoid to break compiling

variables lun and tag had been eliminated, break compiling
when enable debug switch. Meanwhile traces provide the same
information with this DPRINTF, so remove it.

Signed-off-by: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 years agoblock/iscsi: fix memory corruption on iscsi resize
Peter Lieven [Fri, 22 Aug 2014 08:08:49 +0000 (10:08 +0200)]
block/iscsi: fix memory corruption on iscsi resize

bs->total_sectors is not yet updated at this point. resulting
in memory corruption if the volume has grown and data is written
to the newly availble areas.

CC: qemu-stable@nongnu.org
Signed-off-by: Peter Lieven <pl@kamp.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 years agoscsi-bus: Convert DeviceClass init to realize
Fam Zheng [Tue, 12 Aug 2014 02:12:55 +0000 (10:12 +0800)]
scsi-bus: Convert DeviceClass init to realize

Replace "init/destroy" with "realize/unrealize" in SCSIDeviceClass,
which has errp as a parameter. So all the implementations now use
error_setg instead of error_report for reporting error.

Also in scsi_bus_legacy_handle_cmdline, report the error when
initializing the if=scsi devices, before returning it, because in the
callee, error_report is changed to error_setg. And the callers don't
have the right locations (e.g. "-drive if=scsi").

Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 years agoblock: Pass errp in blkconf_geometry
Fam Zheng [Tue, 12 Aug 2014 02:12:54 +0000 (10:12 +0800)]
block: Pass errp in blkconf_geometry

This allows us to pass error information to caller.

Reviewed-by: Andreas Färber <afaerber@suse.de>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 years agoMerge remote-tracking branch 'remotes/awilliam/tags/vfio-pci-for-qemu-20140825.0...
Peter Maydell [Tue, 26 Aug 2014 09:42:06 +0000 (10:42 +0100)]
Merge remote-tracking branch 'remotes/awilliam/tags/vfio-pci-for-qemu-20140825.0' into staging

VFIO: Enable primary NVIDIA quirk regardless of VGA support

# gpg: Signature made Mon 25 Aug 2014 20:29:37 BST using RSA key ID 3BB08B22
# gpg: Can't check signature: public key not found

* remotes/awilliam/tags/vfio-pci-for-qemu-20140825.0:
  vfio: Enable NVIDIA 88000 region quirk regardless of VGA

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agovfio: Enable NVIDIA 88000 region quirk regardless of VGA
Alex Williamson [Mon, 25 Aug 2014 18:10:15 +0000 (12:10 -0600)]
vfio: Enable NVIDIA 88000 region quirk regardless of VGA

If we make use of OVMF for the BIOS then we can use GPUs without VGA
space access, but we still need this quirk.  Disassociate it from the
x-vga option and enable it on all NVIDIA VGA display class devices.

Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
10 years agoMerge remote-tracking branch 'remotes/mst/tags/for_upstream' into staging
Peter Maydell [Mon, 25 Aug 2014 17:49:25 +0000 (18:49 +0100)]
Merge remote-tracking branch 'remotes/mst/tags/for_upstream' into staging

pci, pc fixes, features

A bunch of bugfixes - these will make sense for 2.1.1

ACPI support for TPM and partial ARI support for PCIE.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
# gpg: Signature made Sun 24 Aug 2014 23:16:35 BST using RSA key ID D28D5469
# gpg: Good signature from "Michael S. Tsirkin <mst@kernel.org>"
# gpg:                 aka "Michael S. Tsirkin <mst@redhat.com>"

* remotes/mst/tags/for_upstream:
  pcie: fix trailing whitespace
  ioh3420: Enable ARI forwarding
  ioh3420: Remove obsoleted, unused ioh3420_init function
  pcie: Rename the pcie_cap_ari_* functions to pcie_cap_arifwd_*
  pcie: Fix incorrect write to the ari capability next function field
  ssdt-tpm: add generated hex file to git
  Add ACPI tables for TPM
  pc: reserve more memory for ACPI for new machine types
  pcihp: fix possible array out of bounds
  pci_bridge: manually destroy memory regions within PCIBridgeWindows
  hostmem: set MPOL_MF_MOVE

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agox86: Clear MTRRs on vCPU reset
Alex Williamson [Thu, 14 Aug 2014 21:39:39 +0000 (15:39 -0600)]
x86: Clear MTRRs on vCPU reset

The SDM specifies (June 2014 Vol3 11.11.5):

    On a hardware reset, the P6 and more recent processors clear the
    valid flags in variable-range MTRRs and clear the E flag in the
    IA32_MTRR_DEF_TYPE MSR to disable all MTRRs. All other bits in the
    MTRRs are undefined.

We currently do none of that, so whatever MTRR settings you had prior
to reset is what you have after reset.  Usually this doesn't matter
because KVM often ignores the guest mappings and uses write-back
anyway.  However, if you have an assigned device and an IOMMU that
allows NoSnoop for that device, KVM defers to the guest memory
mappings which are now stale after reset.  The result is that OVMF
rebooting on such a configuration takes a full minute to LZMA
decompress the firmware volume, a process that is nearly instant on
the initial boot.

Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 years agox86: kvm: Add MTRR support for kvm_get|put_msrs()
Alex Williamson [Thu, 14 Aug 2014 21:39:33 +0000 (15:39 -0600)]
x86: kvm: Add MTRR support for kvm_get|put_msrs()

The MTRR state in KVM currently runs completely independent of the
QEMU state in CPUX86State.mtrr_*.  This means that on migration, the
target loses MTRR state from the source.  Generally that's ok though
because KVM ignores it and maps everything as write-back anyway.  The
exception to this rule is when we have an assigned device and an IOMMU
that doesn't promote NoSnoop transactions from that device to be cache
coherent.  In that case KVM trusts the guest mapping of memory as
configured in the MTRR.

This patch updates kvm_get|put_msrs() so that we retrieve the actual
vCPU MTRR settings and therefore keep CPUX86State synchronized for
migration.  kvm_put_msrs() is also used on vCPU reset and therefore
allows future modificaitons of MTRR state at reset to be realized.

Note that the entries array used by both functions was already
slightly undersized for holding every possible MSR, so this patch
increases it beyond the 28 new entries necessary for MTRR state.

Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 years agox86: Use common variable range MTRR counts
Alex Williamson [Thu, 14 Aug 2014 21:39:27 +0000 (15:39 -0600)]
x86: Use common variable range MTRR counts

We currently define the number of variable range MTRR registers as 8
in the CPUX86State structure and vmstate, but use MSR_MTRRcap_VCNT
(also 8) to report to guests the number available.  Change this to
use MSR_MTRRcap_VCNT consistently.

Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 years agotarget-i386: Don't forbid NX bit on PAE PDEs and PTEs
William Grant [Sun, 24 Aug 2014 05:13:48 +0000 (15:13 +1000)]
target-i386: Don't forbid NX bit on PAE PDEs and PTEs

Commit e8f6d00c30ed88910d0d985f4b2bf41654172ceb ("target-i386: raise
page fault for reserved physical address bits") added a check that the
NX bit is not set on PAE PDPEs, but it also added it to rsvd_mask for
the rest of the function. This caused any PDEs or PTEs with NX set to be
erroneously rejected, making PAE guests with NX support unusable.

Signed-off-by: William Grant <wgrant@ubuntu.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 years agoMerge remote-tracking branch 'remotes/mjt/tags/trivial-patches-2014-08-24' into staging
Peter Maydell [Mon, 25 Aug 2014 16:34:30 +0000 (17:34 +0100)]
Merge remote-tracking branch 'remotes/mjt/tags/trivial-patches-2014-08-24' into staging

trivial patches for 2014-08-24

# gpg: Signature made Sun 24 Aug 2014 14:28:49 BST using RSA key ID A4C3D7DB
# gpg: Good signature from "Michael Tokarev <mjt@tls.msk.ru>"
# gpg:                 aka "Michael Tokarev <mjt@corpit.ru>"
# gpg:                 aka "Michael Tokarev <mjt@debian.org>"
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 6EE1 95D1 886E 8FFB 810D  4324 457C E0A0 8044 65C5
#      Subkey fingerprint: 6F67 E18E 7C91 C5B1 5514  66A7 BEE5 9D74 A4C3 D7DB

* remotes/mjt/tags/trivial-patches-2014-08-24:
  vmxnet3: Pad short frames to minimum size (60 bytes)
  libdecnumber: Fix warnings from smatch (missing static, boolean operations)
  linux-user: fix file descriptor leaks
  po: Fix Makefile rules for in-tree builds without configuration
  slirp/misc: Use the GLib memory allocation APIs
  configure: no need to mkdir QMP
  dma: axidma: Variablise repeated s->streams[i] sub-expr
  microblaze: ml605: Get rid of ddr_base variable
  tests/bios-tables-test: check the value returned by fopen()
  tcg: dump op count into qemu log
  util/path: Use the GLib memory allocation routines

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agospapr: Add support for new NMI interface
Alexey Kardashevskiy [Wed, 20 Aug 2014 12:16:36 +0000 (22:16 +1000)]
spapr: Add support for new NMI interface

This implements an NMI interface POWERPC SPAPR machine.
This enables an "nmi" HMP/QMP command supported on SPAPR.

This calls POWERPC_EXCP_RESET (vector 0x100) in the guest to deliver NMI
to every CPU. The expected result is XMON (in-kernel debugger) invocation.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 years agos390x: Migrate to new NMI interface
Alexey Kardashevskiy [Wed, 20 Aug 2014 12:16:35 +0000 (22:16 +1000)]
s390x: Migrate to new NMI interface

This implements an NMI interface for s390 and s390-ccw machines.

This removes #ifdef s390 branch in qmp_inject_nmi so new s390's
nmi_monitor_handler() callback is going to be used for NMI.

Since nmi_monitor_handler()-calling code is platform independent,
CPUState::cpu_index is used instead of S390CPU::env.cpu_num.
There should not be any change in behaviour as both @cpu_index and
@cpu_num are global CPU numbers.

Note that s390_cpu_restart() already takes care of the specified cpu,
so we don't need to schedule via async_run_on_cpu().

Since the only error s390_cpu_restart() can return is ENOSYS, convert
it to QERR_UNSUPPORTED.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: Alexander Graf <agraf@suse.de>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 years agos390x: Convert QEMUMachine to MachineClass
Alexey Kardashevskiy [Wed, 20 Aug 2014 12:16:34 +0000 (22:16 +1000)]
s390x: Convert QEMUMachine to MachineClass

This converts s390-virtio and s390-ccw-virtio machines to QOM MachineClass.
This brings ability to add interfaces to the machine classes. The first
interface for addition will be NMI.

The patch is mechanical so no change in behavior is expected.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 years agocpus: Define callback for QEMU "nmi" command
Alexey Kardashevskiy [Wed, 20 Aug 2014 12:16:33 +0000 (22:16 +1000)]
cpus: Define callback for QEMU "nmi" command

This introduces an NMI (Non Maskable Interrupt) interface with
a single nmi_monitor_handler() method. A machine or a device can
implement it. This searches for an QOM object with this interface
and if it is implemented, calls it. The callback implements an action
required to cause debug crash dump on in-kernel debugger invocation.
The callback returns Error**.

This adds a nmi_monitor_handle() helper which walks through
all objects to find the interface. The interface method is called
for all found instances.

This adds support for it in qmp_inject_nmi(). Since no architecture
supports it at the moment, there is no change in behaviour.

This changes inject-nmi command description for HMP and QMP.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 years agopcie: fix trailing whitespace
Michael S. Tsirkin [Sun, 24 Aug 2014 20:45:29 +0000 (22:45 +0200)]
pcie: fix trailing whitespace

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoioh3420: Enable ARI forwarding
Knut Omang [Sun, 24 Aug 2014 13:32:20 +0000 (15:32 +0200)]
ioh3420: Enable ARI forwarding

Signed-off-by: Knut Omang <knut.omang@oracle.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoioh3420: Remove obsoleted, unused ioh3420_init function
Knut Omang [Sun, 24 Aug 2014 13:32:19 +0000 (15:32 +0200)]
ioh3420: Remove obsoleted, unused ioh3420_init function

Signed-off-by: Knut Omang <knut.omang@oracle.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agopcie: Rename the pcie_cap_ari_* functions to pcie_cap_arifwd_*
Knut Omang [Sun, 24 Aug 2014 13:32:18 +0000 (15:32 +0200)]
pcie: Rename the pcie_cap_ari_* functions to pcie_cap_arifwd_*

Rename helper functions to make a clearer distinction between
the PCIe capability/control register feature ARI forwarding and a
device that supports the ARI feature via an ARI extended PCIe capability.

Signed-off-by: Knut Omang <knut.omang@oracle.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agopcie: Fix incorrect write to the ari capability next function field
Knut Omang [Sun, 24 Aug 2014 13:32:17 +0000 (15:32 +0200)]
pcie: Fix incorrect write to the ari capability next function field

PCI_ARI_CAP_NFN, a macro for reading next function was used instead of
the intended write.

Signed-off-by: Knut Omang <knut.omang@oracle.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agossdt-tpm: add generated hex file to git
Michael S. Tsirkin [Wed, 20 Aug 2014 21:20:13 +0000 (23:20 +0200)]
ssdt-tpm: add generated hex file to git

Needed for systems without IASL.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoAdd ACPI tables for TPM
Stefan Berger [Mon, 11 Aug 2014 20:33:36 +0000 (16:33 -0400)]
Add ACPI tables for TPM

Add an SSDT ACPI table for the TPM device.
Add a TCPA table for BIOS logging area when a TPM is being used.

The latter follows this spec here:

http://www.trustedcomputinggroup.org/files/static_page_files/DCD4188E-1A4B-B294-D050A155FB6F7385/TCG_ACPIGeneralSpecification_PublicReview.pdf

This patch has Michael Tsirkin's patches folded in.

Signed-off-by: Stefan Berger <stefanb@linux.vnet.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
10 years agopc: reserve more memory for ACPI for new machine types
Michael S. Tsirkin [Wed, 20 Aug 2014 19:58:12 +0000 (21:58 +0200)]
pc: reserve more memory for ACPI for new machine types

commit 868270f23d8db2cce83e4f082fe75e8625a5fbf9
    acpi-build: tweak acpi migration limits
broke kernel loading with -kernel/-initrd: it doubled
the size of ACPI tables but did not reserve
enough memory.

As a result, issues on boot and halt are observed.

Fix this up by doubling reserved memory for new machine types.

Cc: qemu-stable@nongnu.org
Reported-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agopcihp: fix possible array out of bounds
Gonglei [Wed, 20 Aug 2014 05:52:30 +0000 (13:52 +0800)]
pcihp: fix possible array out of bounds

Prevent out-of-bounds array access on
acpi_pcihp_pci_status.

Signed-off-by: Gonglei <arei.gonglei@huawei.com>
Reviewed-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Cc: qemu-stable@nongnu.org
Reviewed-by: Marcel Apfelbaum <marcel@redhat.com>
10 years agopci_bridge: manually destroy memory regions within PCIBridgeWindows
Paolo Bonzini [Wed, 20 Aug 2014 15:50:05 +0000 (17:50 +0200)]
pci_bridge: manually destroy memory regions within PCIBridgeWindows

The regions are destroyed and recreated on configuration space accesses.
We need to destroy them before the containing PCIBridgeWindows object
is freed.

Reported-by: Gonglei <arei.gonglei@huawei.com>
Reported-by: Knut Omang <knut.omang@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agovmxnet3: Pad short frames to minimum size (60 bytes)
Ben Draper [Wed, 20 Aug 2014 12:27:14 +0000 (13:27 +0100)]
vmxnet3: Pad short frames to minimum size (60 bytes)

When running VMware ESXi under qemu-kvm the guest discards frames
that are too short. Short ARP Requests will be dropped, this prevents
guests on the same bridge as VMware ESXi from communicating. This patch
simply adds the padding on the network device itself.

Signed-off-by: Ben Draper <ben@xrsa.net>
Reviewed-by: Dmitry Fleytman <dmitry@daynix.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
10 years agolibdecnumber: Fix warnings from smatch (missing static, boolean operations)
Stefan Weil [Wed, 20 Aug 2014 09:02:10 +0000 (11:02 +0200)]
libdecnumber: Fix warnings from smatch (missing static, boolean operations)

Signed-off-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
10 years agolinux-user: fix file descriptor leaks
zhanghailiang [Fri, 22 Aug 2014 08:23:51 +0000 (16:23 +0800)]
linux-user: fix file descriptor leaks

Handle variable "fd_orig" going out of scope leaks the handle.

Signed-off-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
Reviewed-by: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
10 years agopo: Fix Makefile rules for in-tree builds without configuration
Stefan Weil [Fri, 18 Jul 2014 14:52:29 +0000 (16:52 +0200)]
po: Fix Makefile rules for in-tree builds without configuration

Adding 'update' to the phony targets fixes this error:

$ LANG=C make -C po update
make: Entering directory `/qemu/po'
  LINK  update
/qemu/po/de_DE.po: file not recognized: File format not recognized
collect2: error: ld returned 1 exit status
make: *** [update] Error 1
make: Leaving directory `/qemu/po'

Some other phony targets (build, install) were also added, and the
existing .PHONY statement was moved to a more prominent position at
the beginning of the Makefile.

The patch also fixes a 2nd bug. The default target should be 'all',
but instead 'modules' (from rules.mak) was the default. Fix this by
adding 'all' as a target before any include statement.

Signed-off-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
10 years agoslirp/misc: Use the GLib memory allocation APIs
zhanghailiang [Tue, 19 Aug 2014 08:30:17 +0000 (16:30 +0800)]
slirp/misc: Use the GLib memory allocation APIs

Here we don't check the return value of malloc() which may fail.
Use the g_new() instead, which will abort the program when
there is not enough memory.

Also, use g_strdup instead of strdup and remove the unnecessary
strdup function.

Signed-off-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Benoît Canet <benoit.canet@nodalink.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
10 years agoconfigure: no need to mkdir QMP
Liming Wang [Tue, 19 Aug 2014 02:24:05 +0000 (10:24 +0800)]
configure: no need to mkdir QMP

commit 7537fe04 QMP: QMP/ -> docs/qmp/

Above commit has moved last QMP files to docs/qmp and it's not necessary
to create QMP directory. So remove it from configure.

Signed-off-by: Liming Wang <liming.wang@canonical.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
10 years agodma: axidma: Variablise repeated s->streams[i] sub-expr
Peter Crosthwaite [Mon, 18 Aug 2014 00:53:12 +0000 (17:53 -0700)]
dma: axidma: Variablise repeated s->streams[i] sub-expr

This have 6 inline usages. Make it a bit more readable by using a local
variable.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
10 years agomicroblaze: ml605: Get rid of ddr_base variable
Peter Crosthwaite [Mon, 18 Aug 2014 00:52:38 +0000 (17:52 -0700)]
microblaze: ml605: Get rid of ddr_base variable

It's a constant based on a macro. Just use the macro in place.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
10 years agotests/bios-tables-test: check the value returned by fopen()
zhanghailiang [Mon, 18 Aug 2014 07:54:33 +0000 (15:54 +0800)]
tests/bios-tables-test: check the value returned by fopen()

The function fopen() may fail, so check its return value.

Signed-off-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
Signed-off-by: Li Liu <john.liuli@huawei.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
10 years agotcg: dump op count into qemu log
zhanghailiang [Mon, 18 Aug 2014 07:58:08 +0000 (15:58 +0800)]
tcg: dump op count into qemu log

fopen() may fail and it does not check its return vaule here,
it is better to dump op count to the normal log file.

Signed-off-by: Li Liu <john.liuli@huawei.com>
Signed-off-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
10 years agoutil/path: Use the GLib memory allocation routines
zhanghailiang [Mon, 18 Aug 2014 07:49:22 +0000 (15:49 +0800)]
util/path: Use the GLib memory allocation routines

In this file, we don't check the return value of malloc/strdup/realloc which may fail.
Instead of using these routines, we use the GLib memory APIs g_malloc/g_strdup/g_realloc.
They will exit on allocation failure, so there is no need to test for failure,
which would be fine for setup.

Signed-off-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>