qemu.git
3 years agoMerge remote-tracking branch 'remotes/dg-gitlab/tags/ppc-for-6.1-20210709' into staging
Peter Maydell [Sat, 10 Jul 2021 15:06:24 +0000 (16:06 +0100)]
Merge remote-tracking branch 'remotes/dg-gitlab/tags/ppc-for-6.1-20210709' into staging

ppc patch queue 2021-07-09

Here's a (probably) final pull request before the qemu-6.1 soft
freeze.  Includes:
  * Implementation of the new H_RPT_INVALIDATE hypercall
  * Virtual Open Firmware for pSeries and pegasos2 machine types.
    This is an experimental minimal Open Firmware implementation which
    works by delegating nearly everything to qemu itself via a special
    hypercall.
  * A number of cleanups to the ppc soft MMU code
  * Fix to handling of two-level radix mode translations for the
    powernv machine type
  * Update the H_GET_CPU_CHARACTERISTICS call with newly defined bits.
    This will allow more flexible handling of possible future CPU
    Spectre-like flaws
  * Correctly treat mtmsrd as an illegal instruction on BookE cpus
  * Firmware update for the ppce500 machine type

# gpg: Signature made Fri 09 Jul 2021 06:16:42 BST
# gpg:                using RSA key 75F46586AE61A66CC44E87DC6C38CACA20D9B392
# gpg: Good signature from "David Gibson <david@gibson.dropbear.id.au>" [full]
# gpg:                 aka "David Gibson (Red Hat) <dgibson@redhat.com>" [full]
# gpg:                 aka "David Gibson (ozlabs.org) <dgibson@ozlabs.org>" [full]
# gpg:                 aka "David Gibson (kernel.org) <dwg@kernel.org>" [unknown]
# Primary key fingerprint: 75F4 6586 AE61 A66C C44E  87DC 6C38 CACA 20D9 B392

* remotes/dg-gitlab/tags/ppc-for-6.1-20210709: (33 commits)
  target/ppc: Support for H_RPT_INVALIDATE hcall
  linux-headers: Update
  spapr: Fix implementation of Open Firmware client interface
  target/ppc: Don't compile ppc_tlb_invalid_all without TCG
  ppc/pegasos2: Implement some RTAS functions with VOF
  ppc/pegasos2: Fix use of && instead of &
  ppc/pegasos2: Use Virtual Open Firmware as firmware replacement
  target/ppc/spapr: Update H_GET_CPU_CHARACTERISTICS L1D cache flush bits
  target/ppc: Allow virtual hypervisor on CPU without HV
  ppc/pegasos2: Introduce Pegasos2MachineState structure
  target/ppc: mtmsrd is an illegal instruction on BookE
  spapr: Implement Open Firmware client interface
  docs/system: ppc: Update ppce500 documentation with eTSEC support
  roms/u-boot: Bump ppce500 u-boot to v2021.07 to add eTSEC support
  target/ppc: change ppc_hash32_xlate to use mmu_idx
  target/ppc: introduce mmu-books.h
  target/ppc: changed ppc_hash64_xlate to use mmu_idx
  target/ppc: fix address translation bug for radix mmus
  target/ppc: Fix compilation with DEBUG_BATS debug option
  target/ppc: Fix compilation with FLUSH_ALL_TLBS debug option
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
3 years agoMerge remote-tracking branch 'remotes/ehabkost-gl/tags/machine-next-pull-request...
Peter Maydell [Fri, 9 Jul 2021 16:58:38 +0000 (17:58 +0100)]
Merge remote-tracking branch 'remotes/ehabkost-gl/tags/machine-next-pull-request' into staging

Machine queue, 2021-07-07

Deprecation:
* Deprecate pmem=on with non-DAX capable backend file
  (Igor Mammedov)

Feature:
* virtio-mem: vfio support (David Hildenbrand)

Cleanup:
* vmbus: Don't make QOM property registration conditional
  (Eduardo Habkost)

# gpg: Signature made Thu 08 Jul 2021 20:55:04 BST
# gpg:                using RSA key 5A322FD5ABC4D3DBACCFD1AA2807936F984DC5A6
# gpg:                issuer "ehabkost@redhat.com"
# gpg: Good signature from "Eduardo Habkost <ehabkost@redhat.com>" [full]
# Primary key fingerprint: 5A32 2FD5 ABC4 D3DB ACCF  D1AA 2807 936F 984D C5A6

* remotes/ehabkost-gl/tags/machine-next-pull-request:
  vfio: Disable only uncoordinated discards for VFIO_TYPE1 iommus
  virtio-mem: Require only coordinated discards
  softmmu/physmem: Extend ram_block_discard_(require|disable) by two discard types
  softmmu/physmem: Don't use atomic operations in ram_block_discard_(disable|require)
  vfio: Support for RamDiscardManager in the vIOMMU case
  vfio: Sanity check maximum number of DMA mappings with RamDiscardManager
  vfio: Query and store the maximum number of possible DMA mappings
  vfio: Support for RamDiscardManager in the !vIOMMU case
  virtio-mem: Implement RamDiscardManager interface
  virtio-mem: Don't report errors when ram_block_discard_range() fails
  virtio-mem: Factor out traversing unplugged ranges
  memory: Helpers to copy/free a MemoryRegionSection
  memory: Introduce RamDiscardManager for RAM memory regions
  Deprecate pmem=on with non-DAX capable backend file
  vmbus: Don't make QOM property registration conditional

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
3 years agoMerge remote-tracking branch 'remotes/mst/tags/for_upstream' into staging
Peter Maydell [Fri, 9 Jul 2021 13:30:01 +0000 (14:30 +0100)]
Merge remote-tracking branch 'remotes/mst/tags/for_upstream' into staging

pc,pci,virtio: bugfixes, improvements

vhost-user-rng support.
Fixes all over the place.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
# gpg: Signature made Wed 07 Jul 2021 14:29:30 BST
# gpg:                using RSA key 5D09FD0871C8F85B94CA8A0D281F0DB8D28D5469
# gpg:                issuer "mst@redhat.com"
# gpg: Good signature from "Michael S. Tsirkin <mst@kernel.org>" [full]
# gpg:                 aka "Michael S. Tsirkin <mst@redhat.com>" [full]
# Primary key fingerprint: 0270 606B 6F3C DF3D 0B17  0970 C350 3912 AFBE 8E67
#      Subkey fingerprint: 5D09 FD08 71C8 F85B 94CA  8A0D 281F 0DB8 D28D 5469

* remotes/mst/tags/for_upstream:
  MAINTAINERS: Add maintainer for vhost-user RNG implementation
  docs: add slot when adding new PCIe root port
  acpi/ged: fix reset cause
  tests: acpi: pc: update expected DSDT blobs
  acpi: pc: revert back to v5.2 PCI slot enumeration
  tests: acpi: prepare for changing DSDT tables
  migration: failover: reset partially_hotplugged
  virtio-pci: Changed return values for "notify", "device" and "isr" read.
  virtio-pci: Added check for virtio device in PCI config cbs.
  virtio-pci: Added check for virtio device presence in mm callbacks.
  hw/pci-host/q35: Ignore write of reserved PCIEXBAR LENGTH field
  virtio: Clarify MR transaction optimization
  virtio: disable ioeventfd for record/replay

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
3 years agotarget/ppc: Support for H_RPT_INVALIDATE hcall
Bharata B Rao [Tue, 6 Jul 2021 11:24:40 +0000 (16:54 +0530)]
target/ppc: Support for H_RPT_INVALIDATE hcall

If KVM_CAP_RPT_INVALIDATE KVM capability is enabled, then

- indicate the availability of H_RPT_INVALIDATE hcall to the guest via
  ibm,hypertas-functions property.
- Enable the hcall

Both the above are done only if the new sPAPR machine capability
cap-rpt-invalidate is set.

Signed-off-by: Bharata B Rao <bharata@linux.ibm.com>
Message-Id: <20210706112440.1449562-3-bharata@linux.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agolinux-headers: Update
Bharata B Rao [Tue, 6 Jul 2021 11:24:39 +0000 (16:54 +0530)]
linux-headers: Update

Update to mainline commit: 79160a603bdb ("Merge tag 'usb-5.14-rc1' of
git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb"

Signed-off-by: Bharata B Rao <bharata@linux.ibm.com>
Message-Id: <20210706112440.1449562-2-bharata@linux.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agospapr: Fix implementation of Open Firmware client interface
Alexey Kardashevskiy [Thu, 8 Jul 2021 06:56:25 +0000 (16:56 +1000)]
spapr: Fix implementation of Open Firmware client interface

This addresses the comments from v22.

The functional changes are (the VOF ones need retesting with Pegasos2):

(VOF) setprop will start failing if the machine class callback
did not handle it;
(VOF) unit addresses are lowered in path_offset();
(SPAPR) /chosen/bootargs is initialized from kernel_cmdline if
the client did not change it.

Fixes: 5c991e5d4378 ("spapr: Implement Open Firmware client interface")
Cc: BALATON Zoltan <balaton@eik.bme.hu>
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Message-Id: <20210708065625.548396-1-aik@ozlabs.ru>
Tested-by: BALATON Zoltan <balaton@eik.bme.hu>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc: Don't compile ppc_tlb_invalid_all without TCG
Lucas Mateus Castro (alqotel) [Thu, 8 Jul 2021 16:49:54 +0000 (13:49 -0300)]
target/ppc: Don't compile ppc_tlb_invalid_all without TCG

The function ppc_tlb_invalid_all is not compiled anymore in a TCG-less
environment, and the call to that function has been disabled in this
situation

Signed-off-by: Lucas Mateus Castro (alqotel) <lucas.araujo@eldorado.org.br>
Message-Id: <20210708164957.28096-2-lucas.araujo@eldorado.org.br>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agoppc/pegasos2: Implement some RTAS functions with VOF
BALATON Zoltan [Thu, 8 Jul 2021 21:46:14 +0000 (23:46 +0200)]
ppc/pegasos2: Implement some RTAS functions with VOF

Linux uses RTAS functions to access PCI devices so we need to provide
these with VOF. Implement some of the most important functions to
allow booting Linux with VOF. With this the board is now usable
without a binary ROM image and we can enable it by default as other
boards.

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Message-Id: <20210708215113.B3F747456E3@zero.eik.bme.hu>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agoppc/pegasos2: Fix use of && instead of &
David Gibson [Thu, 8 Jul 2021 05:40:21 +0000 (15:40 +1000)]
ppc/pegasos2: Fix use of && instead of &

This is obviously intended to be a mask, not a logical operation.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agoppc/pegasos2: Use Virtual Open Firmware as firmware replacement
BALATON Zoltan [Sun, 27 Jun 2021 16:27:13 +0000 (18:27 +0200)]
ppc/pegasos2: Use Virtual Open Firmware as firmware replacement

The pegasos2 board comes with an Open Firmware compliant ROM based on
SmartFirmware but it has some changes that are not open source
therefore the ROM binary cannot be included in QEMU. Guests running on
the board however depend on services provided by the firmware. The
Virtual Open Firmware recently added to QEMU implements a minimal set
of these services to allow some guests to boot without the original
firmware. This patch adds VOF as the default firmware for pegasos2
which allows booting Linux and MorphOS via -kernel option while a ROM
image can still be used with -bios for guests that don't run with VOF.

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Message-Id: <1d6ed6f290c5c1f0b5a1e1c51cf1151452d70d9a.1624811233.git.balaton@eik.bme.hu>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc/spapr: Update H_GET_CPU_CHARACTERISTICS L1D cache flush bits
Nicholas Piggin [Tue, 15 Jun 2021 04:41:07 +0000 (14:41 +1000)]
target/ppc/spapr: Update H_GET_CPU_CHARACTERISTICS L1D cache flush bits

There are several new L1D cache flush bits added to the hcall which reflect
hardware security features for speculative cache access issues.

These behaviours are now being specified as negative in order to simplify
patched kernel compatibility with older firmware (a new problem found in
existing systems would automatically be vulnerable).

[dwg: Technically this changes behaviour for existing machine types.
 After discussion with Nick, we've determined this is safe, because
 the worst that will happen if a guest gets the wrong information due
 to a migration is that it will perform some unnecessary workarounds,
 but will remain correct and secure (well, as secure as it was going
 to be anyway).  In addition the change only affects cap-cfpc=safe
 which is not enabled by default, and in fact is not possible to set
 on any current hardware (though it's expected it will be possible on
 POWER10)]

Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Message-Id: <20210615044107.1481608-1-npiggin@gmail.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc: Allow virtual hypervisor on CPU without HV
BALATON Zoltan [Sun, 27 Jun 2021 16:27:13 +0000 (18:27 +0200)]
target/ppc: Allow virtual hypervisor on CPU without HV

Change the assert in ppc_store_sdr1() to allow vhyp to be set on CPUs
without HV bit. This allows using the vhyp interface for firmware
emulation on pegasos2.

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Message-Id: <21c7745aabbb68fcc50bb2ffaf16b939ba21261c.1624811233.git.balaton@eik.bme.hu>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agoppc/pegasos2: Introduce Pegasos2MachineState structure
BALATON Zoltan [Sun, 27 Jun 2021 16:27:13 +0000 (18:27 +0200)]
ppc/pegasos2: Introduce Pegasos2MachineState structure

Add own machine state structure which will be used to store state
needed for firmware emulation.

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <7f6d5fbf4f70c64dba001483174a2921dd616ecd.1624811233.git.balaton@eik.bme.hu>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc: mtmsrd is an illegal instruction on BookE
Nicholas Piggin [Tue, 6 Jul 2021 05:13:21 +0000 (15:13 +1000)]
target/ppc: mtmsrd is an illegal instruction on BookE

MSR is a 32-bit register in BookE and there is no mtmsrd instruction.

Cc: Christian Zigotzky <chzigotzky@xenosoft.de>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Message-Id: <20210706051321.609046-1-npiggin@gmail.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agospapr: Implement Open Firmware client interface
Alexey Kardashevskiy [Fri, 25 Jun 2021 05:51:55 +0000 (15:51 +1000)]
spapr: Implement Open Firmware client interface

The PAPR platform describes an OS environment that's presented by
a combination of a hypervisor and firmware. The features it specifies
require collaboration between the firmware and the hypervisor.

Since the beginning, the runtime component of the firmware (RTAS) has
been implemented as a 20 byte shim which simply forwards it to
a hypercall implemented in qemu. The boot time firmware component is
SLOF - but a build that's specific to qemu, and has always needed to be
updated in sync with it. Even though we've managed to limit the amount
of runtime communication we need between qemu and SLOF, there's some,
and it has become increasingly awkward to handle as we've implemented
new features.

This implements a boot time OF client interface (CI) which is
enabled by a new "x-vof" pseries machine option (stands for "Virtual Open
Firmware). When enabled, QEMU implements the custom H_OF_CLIENT hcall
which implements Open Firmware Client Interface (OF CI). This allows
using a smaller stateless firmware which does not have to manage
the device tree.

The new "vof.bin" firmware image is included with source code under
pc-bios/. It also includes RTAS blob.

This implements a handful of CI methods just to get -kernel/-initrd
working. In particular, this implements the device tree fetching and
simple memory allocator - "claim" (an OF CI memory allocator) and updates
"/memory@0/available" to report the client about available memory.

This implements changing some device tree properties which we know how
to deal with, the rest is ignored. To allow changes, this skips
fdt_pack() when x-vof=on as not packing the blob leaves some room for
appending.

In absence of SLOF, this assigns phandles to device tree nodes to make
device tree traversing work.

When x-vof=on, this adds "/chosen" every time QEMU (re)builds a tree.

This adds basic instances support which are managed by a hash map
ihandle -> [phandle].

Before the guest started, the used memory is:
0..e60 - the initial firmware
8000..10000 - stack
400000.. - kernel
3ea0000.. - initramdisk

This OF CI does not implement "interpret".

Unlike SLOF, this does not format uninitialized nvram. Instead, this
includes a disk image with pre-formatted nvram.

With this basic support, this can only boot into kernel directly.
However this is just enough for the petitboot kernel and initradmdisk to
boot from any possible source. Note this requires reasonably recent guest
kernel with:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=df5be5be8735

The immediate benefit is much faster booting time which especially
crucial with fully emulated early CPU bring up environments. Also this
may come handy when/if GRUB-in-the-userspace sees light of the day.

This separates VOF and sPAPR in a hope that VOF bits may be reused by
other POWERPC boards which do not support pSeries.

This assumes potential support for booting from QEMU backends
such as blockdev or netdev without devices/drivers used.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Message-Id: <20210625055155.2252896-1-aik@ozlabs.ru>
Reviewed-by: BALATON Zoltan <balaton@eik.bme.hu>
[dwg: Adjusted some includes which broke compile in some more obscure
 compilation setups]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agodocs/system: ppc: Update ppce500 documentation with eTSEC support
Bin Meng [Tue, 6 Jul 2021 03:19:01 +0000 (11:19 +0800)]
docs/system: ppc: Update ppce500 documentation with eTSEC support

This adds eTSEC support to the PowerPC `ppce500` machine documentation.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agoroms/u-boot: Bump ppce500 u-boot to v2021.07 to add eTSEC support
Bin Meng [Tue, 6 Jul 2021 02:46:41 +0000 (10:46 +0800)]
roms/u-boot: Bump ppce500 u-boot to v2021.07 to add eTSEC support

Update the QEMU shipped u-boot.e500 image built from U-Boot mainline
v2021.07 release, which added eTSEC support to the QEMU ppce500 target,
via the following U-Boot series:

  http://patchwork.ozlabs.org/project/uboot/list/?series=233875&state=*

The cross-compilation toolchain used to build the U-Boot image is:
https://mirrors.edge.kernel.org/pub/tools/crosstool/files/bin/x86_64/10.1.0/x86_64-gcc-10.1.0-nolibc-powerpc-linux.tar.xz

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc: change ppc_hash32_xlate to use mmu_idx
Bruno Larsen (billionai) [Tue, 6 Jul 2021 15:03:16 +0000 (12:03 -0300)]
target/ppc: change ppc_hash32_xlate to use mmu_idx

Changed hash32 address translation to use the supplied mmu_idx, instead
of using what was stored in the msr, for parity purposes (radix64
already uses that) and for conceptual correctness, all the relevant
functions should always use the supplied mmu_idx, as there are no
guarantees that the mmu_idx stored in the CPU variable will not desync.

Signed-off-by: Bruno Larsen (billionai) <bruno.larsen@eldorado.org.br>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Message-Id: <20210706150316.21005-3-bruno.larsen@eldorado.org.br>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc: introduce mmu-books.h
Bruno Larsen (billionai) [Tue, 6 Jul 2021 15:03:15 +0000 (12:03 -0300)]
target/ppc: introduce mmu-books.h

Intrudoce a header common to all BookS MMUs, that can hold code that is
common to hash32 and book3s-v3 MMUs.

Suggested-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Bruno Larsen (billionai) <bruno.larsen@eldorado.org.br>
Message-Id: <20210706150316.21005-2-bruno.larsen@eldorado.org.br>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc: changed ppc_hash64_xlate to use mmu_idx
Bruno Larsen (billionai) [Mon, 28 Jun 2021 13:36:10 +0000 (10:36 -0300)]
target/ppc: changed ppc_hash64_xlate to use mmu_idx

Changed hash64 address translation to use the supplied mmu_idx instead
of using the one stored in the msr, for parity purposes (other book3s
MMUs already use it).

Signed-off-by: Bruno Larsen (billionai) <bruno.larsen@eldorado.org.br>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20210628133610.1143-4-bruno.larsen@eldorado.org.br>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc: fix address translation bug for radix mmus
Bruno Larsen (billionai) [Mon, 28 Jun 2021 13:36:08 +0000 (10:36 -0300)]
target/ppc: fix address translation bug for radix mmus

This commit attempts to fix a technical hiccup first mentioned by Richard
Henderson in
https://lists.nongnu.org/archive/html/qemu-devel/2021-05/msg06247.html

To sumarize the hiccup here, when radix-style mmus are translating an
address, they might need to call a second level of translation, with
hypervisor privileges. However, the way it was being done up until
this point meant that the second level translation had the same
privileges as the first level. It could lead to a bug in address
translation when running KVM inside a TCG guest, but this bug was never
experienced by users, so this isn't as much a bug fix as it is a
correctness cleanup.

This patch attempts that cleanup by making radix64_*_xlate functions
receive the mmu_idx, and passing one with the correct permission for the
second level translation.

The mmuidx macros added by this patch are only correct for non-bookE
mmus, because BookE style set the IS and DS bits inverted and there
might be other subtle differences. However, there doesn't seem to be
BookE cpus that have radix-style mmus, so we left a comment there to
document the issue, in case a machine does have that and was missed.

As part of this cleanup, we now need to send the correct mmmu_idx
when calling get_phys_page_debug, otherwise we might not be able to see the
memory that the CPU could

Suggested-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Bruno Larsen (billionai) <bruno.larsen@eldorado.org.br>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Tested-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20210628133610.1143-2-bruno.larsen@eldorado.org.br>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc: Fix compilation with DEBUG_BATS debug option
Fabiano Rosas [Fri, 2 Jul 2021 21:52:35 +0000 (18:52 -0300)]
target/ppc: Fix compilation with DEBUG_BATS debug option

../target/ppc/mmu-hash32.c: In function 'ppc_hash32_bat_lookup':
../target/ppc/mmu-hash32.c:204:13: error: 'BATu' undeclared (first use in this function);
  204 |             BATu = &BATut[i];
      |             ^~~~
      |             BATut
../target/ppc/mmu-hash32.c:205:13: error: 'BATl' undeclared (first use in this function);
  205 |             BATl = &BATlt[i];
      |             ^~~~
      |             BATlt
../target/ppc/mmu-hash32.c:206:13: error: 'BEPIu' undeclared (first use in this function)
  206 |             BEPIu = *BATu & BATU32_BEPIU;
      |             ^~~~~
../target/ppc/mmu-hash32.c:206:29: error: 'BATU32_BEPIU' undeclared (first use in this function);
  206 |             BEPIu = *BATu & BATU32_BEPIU;
      |                             ^~~~~~~~~~~~
      |                             BATU32_BEPI
../target/ppc/mmu-hash32.c:207:13: error: 'BEPIl' undeclared (first use in this function)
  207 |             BEPIl = *BATu & BATU32_BEPIL;
      |             ^~~~~
../target/ppc/mmu-hash32.c:207:29: error: 'BATU32_BEPIL' undeclared (first use in this function);
  207 |             BEPIl = *BATu & BATU32_BEPIL;
      |                             ^~~~~~~~~~~~
      |                             BATU32_BEPI
../target/ppc/mmu-hash32.c:208:13: error: 'bl' undeclared (first use in this function)
  208 |             bl = (*BATu & 0x00001FFC) << 15;
      |             ^~

Fixes: 9813279664 ("target-ppc: Disentangle BAT code for 32-bit hash MMUs")
Signed-off-by: Fabiano Rosas <farosas@linux.ibm.com>
Message-Id: <20210702215235.1941771-4-farosas@linux.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc: Fix compilation with FLUSH_ALL_TLBS debug option
Fabiano Rosas [Fri, 2 Jul 2021 21:52:34 +0000 (18:52 -0300)]
target/ppc: Fix compilation with FLUSH_ALL_TLBS debug option

../target/ppc/mmu_helper.c: In function 'helper_store_ibatu':
../target/ppc/mmu_helper.c:1802:17: error: unused variable 'cpu' [-Werror=unused-variable]
 1802 |     PowerPCCPU *cpu = env_archcpu(env);
      |                 ^~~
../target/ppc/mmu_helper.c: In function 'helper_store_dbatu':
../target/ppc/mmu_helper.c:1838:17: error: unused variable 'cpu' [-Werror=unused-variable]
 1838 |     PowerPCCPU *cpu = env_archcpu(env);
      |                 ^~~
../target/ppc/mmu_helper.c: In function 'helper_store_601_batu':
../target/ppc/mmu_helper.c:1874:17: error: unused variable 'cpu' [-Werror=unused-variable]
 1874 |     PowerPCCPU *cpu = env_archcpu(env);
      |                 ^~~
../target/ppc/mmu_helper.c: In function 'helper_store_601_batl':
../target/ppc/mmu_helper.c:1919:17: error: unused variable 'cpu' [-Werror=unused-variable]
 1919 |     PowerPCCPU *cpu = env_archcpu(env);

Fixes: db70b31144 ("target/ppc: Use env_cpu, env_archcpu")
Signed-off-by: Fabiano Rosas <farosas@linux.ibm.com>
Message-Id: <20210702215235.1941771-3-farosas@linux.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc: Fix compilation with DUMP_PAGE_TABLES debug option
Fabiano Rosas [Fri, 2 Jul 2021 21:52:33 +0000 (18:52 -0300)]
target/ppc: Fix compilation with DUMP_PAGE_TABLES debug option

../target/ppc/mmu_helper.c: In function 'get_segment_6xx_tlb':
../target/ppc/mmu_helper.c:514:46: error: passing argument 1 of
'ppc_hash32_hpt_mask' from incompatible pointer type [-Werror=incompatible-pointer-types]

  514 |                          ppc_hash32_hpt_mask(env) + 0x80);
      |                                              ^~~
      |                                              |
      |                                              CPUPPCState *

Fixes: 36778660d7 ("target/ppc: Eliminate htab_base and htab_mask variables")
Signed-off-by: Fabiano Rosas <farosas@linux.ibm.com>
Message-Id: <20210702215235.1941771-2-farosas@linux.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc: Restrict ppc_cpu_tlb_fill to TCG
Richard Henderson [Mon, 21 Jun 2021 12:51:14 +0000 (09:51 -0300)]
target/ppc: Restrict ppc_cpu_tlb_fill to TCG

This function is used by TCGCPUOps, and is thus TCG specific.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20210621125115.67717-10-bruno.larsen@eldorado.org.br>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc: Introduce ppc_xlate
Richard Henderson [Mon, 21 Jun 2021 12:51:13 +0000 (09:51 -0300)]
target/ppc: Introduce ppc_xlate

Create one common dispatch for all of the ppc_*_xlate functions.
Use ppc64_v3_radix to directly dispatch between ppc_radix64_xlate
and ppc_hash64_xlate.

Remove the separate *_handle_mmu_fault and *_get_phys_page_debug
functions, using common code for ppc_cpu_tlb_fill and
ppc_cpu_get_phys_page_debug.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20210621125115.67717-9-bruno.larsen@eldorado.org.br>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc: Split out ppc_jumbo_xlate
Richard Henderson [Mon, 21 Jun 2021 12:51:12 +0000 (09:51 -0300)]
target/ppc: Split out ppc_jumbo_xlate

Mirror the interface of ppc_radix64_xlate (mostly), putting all
of the logic for older mmu translation into a single entry point.
For booke, we need to add mmu_idx to the xlate-style interface.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20210621125115.67717-8-bruno.larsen@eldorado.org.br>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc: Split out ppc_hash32_xlate
Richard Henderson [Mon, 21 Jun 2021 12:51:11 +0000 (09:51 -0300)]
target/ppc: Split out ppc_hash32_xlate

Mirror the interface of ppc_radix64_xlate, putting all of
the logic for hash32 translation into a single entry point.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20210621125115.67717-7-bruno.larsen@eldorado.org.br>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc: Split out ppc_hash64_xlate
Richard Henderson [Mon, 21 Jun 2021 12:51:10 +0000 (09:51 -0300)]
target/ppc: Split out ppc_hash64_xlate

Mirror the interface of ppc_radix64_xlate, putting all of
the logic for hash64 translation into a single function.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20210621125115.67717-6-bruno.larsen@eldorado.org.br>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc: Use bool success for ppc_radix64_xlate
Richard Henderson [Mon, 21 Jun 2021 12:51:09 +0000 (09:51 -0300)]
target/ppc: Use bool success for ppc_radix64_xlate

Instead of returning non-zero for failure, return true for success.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20210621125115.67717-5-bruno.larsen@eldorado.org.br>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc: Push real-mode handling into ppc_radix64_xlate
Richard Henderson [Mon, 21 Jun 2021 12:51:08 +0000 (09:51 -0300)]
target/ppc: Push real-mode handling into ppc_radix64_xlate

This removes some incomplete duplication between
ppc_radix64_handle_mmu_fault and ppc_radix64_get_phys_page_debug.
The former was correct wrt SPR_HRMOR and the latter was not.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20210621125115.67717-4-bruno.larsen@eldorado.org.br>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc: Use MMUAccessType with *_handle_mmu_fault
Richard Henderson [Mon, 21 Jun 2021 12:51:07 +0000 (09:51 -0300)]
target/ppc: Use MMUAccessType with *_handle_mmu_fault

These changes were waiting until we didn't need to match
the function type of PowerPCCPUClass.handle_mmu_fault.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20210621125115.67717-3-bruno.larsen@eldorado.org.br>
Reviewed-by: Greg Kurz <groug@kaod.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc: Remove PowerPCCPUClass.handle_mmu_fault
Richard Henderson [Mon, 21 Jun 2021 12:51:06 +0000 (09:51 -0300)]
target/ppc: Remove PowerPCCPUClass.handle_mmu_fault

Instead, use a switch on env->mmu_model.  This avoids some
replicated information in cpu setup.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20210621125115.67717-2-bruno.larsen@eldorado.org.br>
Reviewed-by: Greg Kurz <groug@kaod.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agospapr: tune rtas-size
Alexey Kardashevskiy [Tue, 22 Jun 2021 07:03:36 +0000 (17:03 +1000)]
spapr: tune rtas-size

QEMU reserves space for RTAS via /rtas/rtas-size which tells the client
how much space the RTAS requires to work which includes the RTAS binary
blob implementing RTAS runtime. Because pseries supports FWNMI which
requires plenty of space, QEMU reserves more than 2KB which is
enough for the RTAS blob as it is just 20 bytes (under QEMU).

Since FWNMI reset delivery was added, RTAS_SIZE macro is not used anymore.
This replaces RTAS_SIZE with RTAS_MIN_SIZE and uses it in
the /rtas/rtas-size calculation to account for the RTAS blob.

Fixes: 0e236d347790 ("ppc/spapr: Implement FWNMI System Reset delivery")
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Message-Id: <20210622070336.1463250-1-aik@ozlabs.ru>
Reviewed-by: Greg Kurz <groug@kaod.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc: Drop PowerPCCPUClass::interrupts_big_endian()
Greg Kurz [Tue, 22 Jun 2021 14:09:26 +0000 (16:09 +0200)]
target/ppc: Drop PowerPCCPUClass::interrupts_big_endian()

This isn't used anymore.

Signed-off-by: Greg Kurz <groug@kaod.org>
Message-Id: <20210622140926.677618-3-groug@kaod.org>
Reviewed-by: Fabiano Rosas <farosas@linux.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agotarget/ppc: Introduce ppc_interrupts_little_endian()
Greg Kurz [Tue, 22 Jun 2021 14:09:25 +0000 (16:09 +0200)]
target/ppc: Introduce ppc_interrupts_little_endian()

PowerPC CPUs use big endian by default but starting with POWER7,
server grade CPUs use the ILE bit of the LPCR special purpose
register to decide on the endianness to use when handling
interrupts. This gives a clue to QEMU on the endianness the
guest kernel is running, which is needed when generating an
ELF dump of the guest or when delivering an FWNMI machine
check interrupt.

Commit 382d2db62bcb ("target-ppc: Introduce callback for interrupt
endianness") added a class method to PowerPCCPUClass to modelize
this : default implementation returns a fixed "big endian" value,
while POWER7 and newer do the LPCR_ILE check. This is suboptimal
as it forces to implement the method for every new CPU family, and
it is very unlikely that this will ever be different than what we
have today.

We basically only have three cases to consider:
a) CPU doesn't have an LPCR => big endian
b) CPU has an LPCR but doesn't support the ILE bit => big endian
c) CPU has an LPCR and supports the ILE bit => little or big endian

Instead of class methods, introduce an inline helper that checks the
ILE bit in the LPCR_MASK to decide on the outcome. The new helper
words little endian instead of big endian. This allows to drop a !
operator in ppc_cpu_do_fwnmi_machine_check().

Signed-off-by: Greg Kurz <groug@kaod.org>
Message-Id: <20210622140926.677618-2-groug@kaod.org>
Reviewed-by: Fabiano Rosas <farosas@linux.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 years agoMerge remote-tracking branch 'remotes/stefanha-gitlab/tags/block-pull-request' into...
Peter Maydell [Thu, 8 Jul 2021 21:17:28 +0000 (22:17 +0100)]
Merge remote-tracking branch 'remotes/stefanha-gitlab/tags/block-pull-request' into staging

Pull request

# gpg: Signature made Thu 08 Jul 2021 14:11:37 BST
# gpg:                using RSA key 8695A8BFD3F97CDAAC35775A9CA4ABB381AB73C8
# gpg: Good signature from "Stefan Hajnoczi <stefanha@redhat.com>" [full]
# gpg:                 aka "Stefan Hajnoczi <stefanha@gmail.com>" [full]
# Primary key fingerprint: 8695 A8BF D3F9 7CDA AC35  775A 9CA4 ABB3 81AB 73C8

* remotes/stefanha-gitlab/tags/block-pull-request:
  block/io: Merge discard request alignments
  block: Add backend_defaults property
  block/file-posix: Optimize for macOS
  util/async: print leaked BH name when AioContext finalizes
  util/async: add a human-readable name to BHs for debugging

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
3 years agovfio: Disable only uncoordinated discards for VFIO_TYPE1 iommus
David Hildenbrand [Tue, 13 Apr 2021 09:55:31 +0000 (11:55 +0200)]
vfio: Disable only uncoordinated discards for VFIO_TYPE1 iommus

We support coordinated discarding of RAM using the RamDiscardManager for
the VFIO_TYPE1 iommus. Let's unlock support for coordinated discards,
keeping uncoordinated discards (e.g., via virtio-balloon) disabled if
possible.

This unlocks virtio-mem + vfio on x86-64. Note that vfio used via "nvme://"
by the block layer has to be implemented/unlocked separately. For now,
virtio-mem only supports x86-64; we don't restrict RamDiscardManager to
x86-64, though: arm64 and s390x are supposed to work as well, and we'll
test once unlocking virtio-mem support. The spapr IOMMUs will need special
care, to be tackled later, e.g.., once supporting virtio-mem.

Note: The block size of a virtio-mem device has to be set to sane sizes,
depending on the maximum hotplug size - to not run out of vfio mappings.
The default virtio-mem block size is usually in the range of a couple of
MBs. The maximum number of mapping is 64k, shared with other users.
Assume you want to hotplug 256GB using virtio-mem - the block size would
have to be set to at least 8 MiB (resulting in 32768 separate mappings).

Acked-by: Alex Williamson <alex.williamson@redhat.com>
Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Alex Williamson <alex.williamson@redhat.com>
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Auger Eric <eric.auger@redhat.com>
Cc: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: teawater <teawaterz@linux.alibaba.com>
Cc: Marek Kedzierski <mkedzier@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210413095531.25603-14-david@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
3 years agovirtio-mem: Require only coordinated discards
David Hildenbrand [Tue, 13 Apr 2021 09:55:30 +0000 (11:55 +0200)]
virtio-mem: Require only coordinated discards

We implement the RamDiscardManager interface and only require coordinated
discarding of RAM to work.

Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Pankaj Gupta <pankaj.gupta@cloud.ionos.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Alex Williamson <alex.williamson@redhat.com>
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Auger Eric <eric.auger@redhat.com>
Cc: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: teawater <teawaterz@linux.alibaba.com>
Cc: Marek Kedzierski <mkedzier@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210413095531.25603-13-david@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
3 years agosoftmmu/physmem: Extend ram_block_discard_(require|disable) by two discard types
David Hildenbrand [Tue, 13 Apr 2021 09:55:29 +0000 (11:55 +0200)]
softmmu/physmem: Extend ram_block_discard_(require|disable) by two discard types

We want to separate the two cases whereby we discard ram
- uncoordinated: e.g., virito-balloon
- coordinated: e.g., virtio-mem coordinated via the RamDiscardManager

Reviewed-by: Pankaj Gupta <pankaj.gupta@cloud.ionos.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Alex Williamson <alex.williamson@redhat.com>
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Auger Eric <eric.auger@redhat.com>
Cc: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: teawater <teawaterz@linux.alibaba.com>
Cc: Marek Kedzierski <mkedzier@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210413095531.25603-12-david@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
3 years agosoftmmu/physmem: Don't use atomic operations in ram_block_discard_(disable|require)
David Hildenbrand [Tue, 13 Apr 2021 09:55:28 +0000 (11:55 +0200)]
softmmu/physmem: Don't use atomic operations in ram_block_discard_(disable|require)

We have users in migration context that don't hold the BQL (when
finishing migration). To prepare for further changes, use a dedicated mutex
instead of atomic operations. Keep using qatomic_read ("READ_ONCE") for the
functions that only extract the current state (e.g., used by
virtio-balloon), locking isn't necessary.

While at it, split up the counter into two variables to make it easier
to understand.

Suggested-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Pankaj Gupta <pankaj.gupta@cloud.ionos.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Alex Williamson <alex.williamson@redhat.com>
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Auger Eric <eric.auger@redhat.com>
Cc: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: teawater <teawaterz@linux.alibaba.com>
Cc: Marek Kedzierski <mkedzier@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210413095531.25603-11-david@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
3 years agovfio: Support for RamDiscardManager in the vIOMMU case
David Hildenbrand [Tue, 13 Apr 2021 09:55:27 +0000 (11:55 +0200)]
vfio: Support for RamDiscardManager in the vIOMMU case

vIOMMU support works already with RamDiscardManager as long as guests only
map populated memory. Both, populated and discarded memory is mapped
into &address_space_memory, where vfio_get_xlat_addr() will find that
memory, to create the vfio mapping.

Sane guests will never map discarded memory (e.g., unplugged memory
blocks in virtio-mem) into an IOMMU - or keep it mapped into an IOMMU while
memory is getting discarded. However, there are two cases where a malicious
guests could trigger pinning of more memory than intended.

One case is easy to handle: the guest trying to map discarded memory
into an IOMMU.

The other case is harder to handle: the guest keeping memory mapped in
the IOMMU while it is getting discarded. We would have to walk over all
mappings when discarding memory and identify if any mapping would be a
violation. Let's keep it simple for now and print a warning, indicating
that setting RLIMIT_MEMLOCK can mitigate such attacks.

We have to take care of incoming migration: at the point the
IOMMUs get restored and start creating mappings in vfio, RamDiscardManager
implementations might not be back up and running yet: let's add runstate
priorities to enforce the order when restoring.

Acked-by: Alex Williamson <alex.williamson@redhat.com>
Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Alex Williamson <alex.williamson@redhat.com>
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Auger Eric <eric.auger@redhat.com>
Cc: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: teawater <teawaterz@linux.alibaba.com>
Cc: Marek Kedzierski <mkedzier@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210413095531.25603-10-david@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
3 years agovfio: Sanity check maximum number of DMA mappings with RamDiscardManager
David Hildenbrand [Tue, 13 Apr 2021 09:55:26 +0000 (11:55 +0200)]
vfio: Sanity check maximum number of DMA mappings with RamDiscardManager

Although RamDiscardManager can handle running into the maximum number of
DMA mappings by propagating errors when creating a DMA mapping, we want
to sanity check and warn the user early that there is a theoretical setup
issue and that virtio-mem might not be able to provide as much memory
towards a VM as desired.

As suggested by Alex, let's use the number of KVM memory slots to guess
how many other mappings we might see over time.

Acked-by: Alex Williamson <alex.williamson@redhat.com>
Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Alex Williamson <alex.williamson@redhat.com>
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Auger Eric <eric.auger@redhat.com>
Cc: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: teawater <teawaterz@linux.alibaba.com>
Cc: Marek Kedzierski <mkedzier@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210413095531.25603-9-david@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
3 years agovfio: Query and store the maximum number of possible DMA mappings
David Hildenbrand [Tue, 13 Apr 2021 09:55:25 +0000 (11:55 +0200)]
vfio: Query and store the maximum number of possible DMA mappings

Let's query the maximum number of possible DMA mappings by querying the
available mappings when creating the container (before any mappings are
created). We'll use this informaton soon to perform some sanity checks
and warn the user.

Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
Acked-by: Alex Williamson <alex.williamson@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Alex Williamson <alex.williamson@redhat.com>
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Auger Eric <eric.auger@redhat.com>
Cc: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: teawater <teawaterz@linux.alibaba.com>
Cc: Marek Kedzierski <mkedzier@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210413095531.25603-8-david@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
3 years agovfio: Support for RamDiscardManager in the !vIOMMU case
David Hildenbrand [Tue, 13 Apr 2021 09:55:24 +0000 (11:55 +0200)]
vfio: Support for RamDiscardManager in the !vIOMMU case

Implement support for RamDiscardManager, to prepare for virtio-mem
support. Instead of mapping the whole memory section, we only map
"populated" parts and update the mapping when notified about
discarding/population of memory via the RamDiscardListener. Similarly, when
syncing the dirty bitmaps, sync only the actually mapped (populated) parts
by replaying via the notifier.

Using virtio-mem with vfio is still blocked via
ram_block_discard_disable()/ram_block_discard_require() after this patch.

Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
Acked-by: Alex Williamson <alex.williamson@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Alex Williamson <alex.williamson@redhat.com>
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Auger Eric <eric.auger@redhat.com>
Cc: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: teawater <teawaterz@linux.alibaba.com>
Cc: Marek Kedzierski <mkedzier@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210413095531.25603-7-david@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
3 years agovirtio-mem: Implement RamDiscardManager interface
David Hildenbrand [Tue, 13 Apr 2021 09:55:23 +0000 (11:55 +0200)]
virtio-mem: Implement RamDiscardManager interface

Let's properly notify when (un)plugging blocks, after discarding memory
and before allowing the guest to consume memory. Handle errors from
notifiers gracefully (e.g., no remaining VFIO mappings) when plugging,
rolling back the change and telling the guest that the VM is busy.

One special case to take care of is replaying all notifications after
restoring the vmstate. The device starts out with all memory discarded,
so after loading the vmstate, we have to notify about all plugged
blocks.

Acked-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Alex Williamson <alex.williamson@redhat.com>
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Auger Eric <eric.auger@redhat.com>
Cc: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: teawater <teawaterz@linux.alibaba.com>
Cc: Marek Kedzierski <mkedzier@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210413095531.25603-6-david@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
3 years agovirtio-mem: Don't report errors when ram_block_discard_range() fails
David Hildenbrand [Tue, 13 Apr 2021 09:55:22 +0000 (11:55 +0200)]
virtio-mem: Don't report errors when ram_block_discard_range() fails

Any errors are unexpected and ram_block_discard_range() already properly
prints errors. Let's stop manually reporting errors.

Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Alex Williamson <alex.williamson@redhat.com>
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Auger Eric <eric.auger@redhat.com>
Cc: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: teawater <teawaterz@linux.alibaba.com>
Cc: Marek Kedzierski <mkedzier@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210413095531.25603-5-david@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
3 years agovirtio-mem: Factor out traversing unplugged ranges
David Hildenbrand [Tue, 13 Apr 2021 09:55:21 +0000 (11:55 +0200)]
virtio-mem: Factor out traversing unplugged ranges

Let's factor out the core logic, no need to replicate.

Reviewed-by: Pankaj Gupta <pankaj.gupta@cloud.ionos.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Alex Williamson <alex.williamson@redhat.com>
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Auger Eric <eric.auger@redhat.com>
Cc: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: teawater <teawaterz@linux.alibaba.com>
Cc: Marek Kedzierski <mkedzier@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210413095531.25603-4-david@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
3 years agomemory: Helpers to copy/free a MemoryRegionSection
David Hildenbrand [Tue, 13 Apr 2021 09:55:20 +0000 (11:55 +0200)]
memory: Helpers to copy/free a MemoryRegionSection

In case one wants to create a permanent copy of a MemoryRegionSections,
one needs access to flatview_ref()/flatview_unref(). Instead of exposing
these, let's just add helpers to copy/free a MemoryRegionSection and
properly adjust references.

Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Alex Williamson <alex.williamson@redhat.com>
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Auger Eric <eric.auger@redhat.com>
Cc: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: teawater <teawaterz@linux.alibaba.com>
Cc: Marek Kedzierski <mkedzier@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210413095531.25603-3-david@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
3 years agomemory: Introduce RamDiscardManager for RAM memory regions
David Hildenbrand [Tue, 13 Apr 2021 09:55:19 +0000 (11:55 +0200)]
memory: Introduce RamDiscardManager for RAM memory regions

We have some special RAM memory regions (managed by virtio-mem), whereby
the guest agreed to only use selected memory ranges. "unused" parts are
discarded so they won't consume memory - to logically unplug these memory
ranges. Before the VM is allowed to use such logically unplugged memory
again, coordination with the hypervisor is required.

This results in "sparse" mmaps/RAMBlocks/memory regions, whereby only
coordinated parts are valid to be used/accessed by the VM.

In most cases, we don't care about that - e.g., in KVM, we simply have a
single KVM memory slot. However, in case of vfio, registering the
whole region with the kernel results in all pages getting pinned, and
therefore an unexpected high memory consumption - discarding of RAM in
that context is broken.

Let's introduce a way to coordinate discarding/populating memory within a
RAM memory region with such special consumers of RAM memory regions: they
can register as listeners and get updates on memory getting discarded and
populated. Using this machinery, vfio will be able to map only the
currently populated parts, resulting in discarded parts not getting pinned
and not consuming memory.

A RamDiscardManager has to be set for a memory region before it is getting
mapped, and cannot change while the memory region is mapped.

Note: At some point, we might want to let RAMBlock users (esp. vfio used
for nvme://) consume this interface as well. We'll need RAMBlock notifier
calls when a RAMBlock is getting mapped/unmapped (via the corresponding
memory region), so we can properly register a listener there as well.

Reviewed-by: Pankaj Gupta <pankaj.gupta@cloud.ionos.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Alex Williamson <alex.williamson@redhat.com>
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Auger Eric <eric.auger@redhat.com>
Cc: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: teawater <teawaterz@linux.alibaba.com>
Cc: Marek Kedzierski <mkedzier@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210413095531.25603-2-david@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
3 years agoMerge remote-tracking branch 'remotes/vivier2/tags/linux-user-for-6.1-pull-request...
Peter Maydell [Thu, 8 Jul 2021 15:30:18 +0000 (16:30 +0100)]
Merge remote-tracking branch 'remotes/vivier2/tags/linux-user-for-6.1-pull-request' into staging

linux-user pull request 20210708

Code and style cleanup
Add ppid in self/stat
ELF_HWCAP for RISC-V

# gpg: Signature made Thu 08 Jul 2021 11:07:31 BST
# gpg:                using RSA key CD2F75DDC8E3A4DC2E4F5173F30C38BD3F2FBE3C
# gpg:                issuer "laurent@vivier.eu"
# gpg: Good signature from "Laurent Vivier <lvivier@redhat.com>" [full]
# gpg:                 aka "Laurent Vivier <laurent@vivier.eu>" [full]
# gpg:                 aka "Laurent Vivier (Red Hat) <lvivier@redhat.com>" [full]
# Primary key fingerprint: CD2F 75DD C8E3 A4DC 2E4F  5173 F30C 38BD 3F2F BE3C

* remotes/vivier2/tags/linux-user-for-6.1-pull-request:
  linux-user/syscall: Remove hardcoded tabs (code style)
  linux-user/alpha: Remove hardcoded tabs (code style)
  linux-user/mips: Handle TARGET_EWOULDBLOCK as TARGET_EAGAIN
  linux-user/hppa: Handle TARGET_EWOULDBLOCK as TARGET_EAGAIN
  linux-user/alpha: Handle TARGET_EWOULDBLOCK as TARGET_EAGAIN
  linux-user: Fix style problems in linuxload.c
  linux-user: fill ppid field in /proc/self/stat
  linux-user/elfload: Implement ELF_HWCAP for RISC-V

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
3 years agotests/tcg: generalise the disabling of the signals test
Alex Bennée [Tue, 6 Jul 2021 14:57:43 +0000 (15:57 +0100)]
tests/tcg: generalise the disabling of the signals test

It turns out you push down in one place and failures pop-up elsewhere.
Especially on CI. Disable for now for all targets.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
3 years agolinux-user/syscall: Remove hardcoded tabs (code style)
Philippe Mathieu-Daudé [Sun, 4 Jul 2021 18:37:54 +0000 (20:37 +0200)]
linux-user/syscall: Remove hardcoded tabs (code style)

We are going to move this code, fix its style first.

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20210704183755.655002-12-f4bug@amsat.org>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
3 years agolinux-user/alpha: Remove hardcoded tabs (code style)
Philippe Mathieu-Daudé [Sun, 4 Jul 2021 18:37:49 +0000 (20:37 +0200)]
linux-user/alpha: Remove hardcoded tabs (code style)

We are going to move this code, fix its style first.

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20210704183755.655002-7-f4bug@amsat.org>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
3 years agolinux-user/mips: Handle TARGET_EWOULDBLOCK as TARGET_EAGAIN
Philippe Mathieu-Daudé [Sun, 4 Jul 2021 18:37:46 +0000 (20:37 +0200)]
linux-user/mips: Handle TARGET_EWOULDBLOCK as TARGET_EAGAIN

Linux kernel defines EWOULDBLOCK as EAGAIN (since before v2.6.12-rc2).

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20210704183755.655002-4-f4bug@amsat.org>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
3 years agolinux-user/hppa: Handle TARGET_EWOULDBLOCK as TARGET_EAGAIN
Philippe Mathieu-Daudé [Sun, 4 Jul 2021 18:37:45 +0000 (20:37 +0200)]
linux-user/hppa: Handle TARGET_EWOULDBLOCK as TARGET_EAGAIN

Linux kernel defines EWOULDBLOCK as EAGAIN (since before v2.6.12-rc2).

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20210704183755.655002-3-f4bug@amsat.org>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
3 years agolinux-user/alpha: Handle TARGET_EWOULDBLOCK as TARGET_EAGAIN
Philippe Mathieu-Daudé [Sun, 4 Jul 2021 18:37:44 +0000 (20:37 +0200)]
linux-user/alpha: Handle TARGET_EWOULDBLOCK as TARGET_EAGAIN

Linux kernel defines EWOULDBLOCK as EAGAIN (since before v2.6.12-rc2).

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20210704183755.655002-2-f4bug@amsat.org>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
3 years agolinux-user: Fix style problems in linuxload.c
Richard Henderson [Tue, 6 Jul 2021 23:48:58 +0000 (16:48 -0700)]
linux-user: Fix style problems in linuxload.c

Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20210706234932.356913-3-richard.henderson@linaro.org>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
3 years agolinux-user: fill ppid field in /proc/self/stat
Andreas Schwab [Mon, 21 Jun 2021 09:32:59 +0000 (11:32 +0200)]
linux-user: fill ppid field in /proc/self/stat

Signed-off-by: Andreas Schwab <schwab@suse.de>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <mvmwnqnef5g.fsf@suse.de>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
3 years agolinux-user/elfload: Implement ELF_HWCAP for RISC-V
Kito Cheng [Tue, 6 Jul 2021 03:50:15 +0000 (11:50 +0800)]
linux-user/elfload: Implement ELF_HWCAP for RISC-V

Set I, M, A, F, D and C bit for hwcap if misa is set.

Signed-off-by: Kito Cheng <kito.cheng@sifive.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20210706035015.122899-1-kito.cheng@sifive.com>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
3 years agoMAINTAINERS: Add maintainer for vhost-user RNG implementation
Mathieu Poirier [Mon, 14 Jun 2021 20:28:42 +0000 (14:28 -0600)]
MAINTAINERS: Add maintainer for vhost-user RNG implementation

This patch adds entry for the vhost-user-rng related files.

Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Message-Id: <20210614202842.581640-6-mathieu.poirier@linaro.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
3 years agoDeprecate pmem=on with non-DAX capable backend file
Igor Mammedov [Mon, 11 Jan 2021 20:33:32 +0000 (15:33 -0500)]
Deprecate pmem=on with non-DAX capable backend file

It is not safe to pretend that emulated NVDIMM supports
persistence while backend actually failed to enable it
and used non-persistent mapping as fall back.
Instead of falling-back, QEMU should be more strict and
error out with clear message that it's not supported.
So if user asks for persistence (pmem=on), they should
store backing file on NVDIMM.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20210111203332.740815-1-imammedo@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
3 years agovmbus: Don't make QOM property registration conditional
Eduardo Habkost [Fri, 9 Oct 2020 20:07:01 +0000 (16:07 -0400)]
vmbus: Don't make QOM property registration conditional

Having properties registered conditionally makes QOM type
introspection difficult.  Instead of skipping registration of the
"instanceid" property, always register the property but validate
its value against the instance id required by the class.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Acked-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com>
Message-Id: <20201009200701.1830060-1-ehabkost@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
3 years agoblock/io: Merge discard request alignments
Akihiko Odaki [Mon, 5 Jul 2021 13:04:58 +0000 (22:04 +0900)]
block/io: Merge discard request alignments

Signed-off-by: Akihiko Odaki <akihiko.odaki@gmail.com>
Message-id: 20210705130458.97642-3-akihiko.odaki@gmail.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
3 years agoblock: Add backend_defaults property
Akihiko Odaki [Mon, 5 Jul 2021 13:04:57 +0000 (22:04 +0900)]
block: Add backend_defaults property

backend_defaults property allow users to control if default block
properties should be decided with backend information.

If it is off, any backend information will be discarded, which is
suitable if you plan to perform live migration to a different disk backend.

If it is on, a block device may utilize backend information more
aggressively.

By default, it is auto, which uses backend information for block
sizes and ignores the others, which is consistent with the older
versions.

Signed-off-by: Akihiko Odaki <akihiko.odaki@gmail.com>
Message-id: 20210705130458.97642-2-akihiko.odaki@gmail.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
3 years agoblock/file-posix: Optimize for macOS
Akihiko Odaki [Mon, 5 Jul 2021 13:04:56 +0000 (22:04 +0900)]
block/file-posix: Optimize for macOS

This commit introduces "punch hole" operation and optimizes transfer
block size for macOS.

Thanks to Konstantin Nazarov for detailed analysis of a flaw in an
old version of this change:
https://gist.github.com/akihikodaki/87df4149e7ca87f18dc56807ec5a1bc5#gistcomment-3654667

Signed-off-by: Akihiko Odaki <akihiko.odaki@gmail.com>
Message-id: 20210705130458.97642-1-akihiko.odaki@gmail.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
3 years agoMerge remote-tracking branch 'remotes/bonzini-gitlab/tags/for-upstream' into staging
Peter Maydell [Tue, 6 Jul 2021 10:24:58 +0000 (11:24 +0100)]
Merge remote-tracking branch 'remotes/bonzini-gitlab/tags/for-upstream' into staging

* More Meson test conversions and configure cleanups
* Generalize XSAVE area offset so that it matches AMD processors on KVM
* Improvements for -display and deprecation of -no-quit
* Enable SMP configuration as a compound machine property ("-M smp.cpus=...")
* Haiku compilation fix
* Add icon on Darwin

# gpg: Signature made Tue 06 Jul 2021 08:35:23 BST
# gpg:                using RSA key F13338574B662389866C7682BFFBD25F78C7AE83
# gpg:                issuer "pbonzini@redhat.com"
# gpg: Good signature from "Paolo Bonzini <bonzini@gnu.org>" [full]
# gpg:                 aka "Paolo Bonzini <pbonzini@redhat.com>" [full]
# Primary key fingerprint: 46F5 9FBD 57D6 12E7 BFD4  E2F7 7E15 100C CD36 69B1
#      Subkey fingerprint: F133 3857 4B66 2389 866C  7682 BFFB D25F 78C7 AE83

* remotes/bonzini-gitlab/tags/for-upstream: (40 commits)
  config-host.mak: remove unused compiler-related lines
  Set icon for QEMU binary on Mac OS
  qemu-option: remove now-dead code
  machine: add smp compound property
  vl: switch -M parsing to keyval
  keyval: introduce keyval_parse_into
  keyval: introduce keyval_merge
  qom: export more functions for use with non-UserCreatable objects
  configure: convert compiler tests to meson, part 6
  configure: convert compiler tests to meson, part 5
  configure: convert compiler tests to meson, part 4
  configure: convert compiler tests to meson, part 3
  configure: convert compiler tests to meson, part 2
  configure: convert compiler tests to meson, part 1
  configure: convert HAVE_BROKEN_SIZE_MAX to meson
  configure, meson: move CONFIG_IVSHMEM to meson
  meson: store dependency('threads') in a variable
  meson: sort existing compiler tests
  configure, meson: convert libxml2 detection to meson
  configure, meson: convert liburing detection to meson
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
3 years agoconfig-host.mak: remove unused compiler-related lines
Paolo Bonzini [Tue, 6 Jul 2021 06:59:01 +0000 (08:59 +0200)]
config-host.mak: remove unused compiler-related lines

Most of the build is not done via Makefiles, therefore the toolchain
variables are mostly unused.  They are still used by tests/tcg
and pc-bios/roms, but most of them are not needed there.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agoSet icon for QEMU binary on Mac OS
John Arbuckle [Mon, 5 Jul 2021 19:53:28 +0000 (15:53 -0400)]
Set icon for QEMU binary on Mac OS

Before switching the build system over to Meson, an icon was
added to the QEMU binary on Mac OS. This patch adds back that
feature; it piggybacks on the existing scripts/entitlement.sh,
which already does in-place changes to the executable on Darwin.

Signed-off-by: John Arbuckle <programmingkidx@gmail.com>
Message-Id: <20210705195328.36442-1-programmingkidx@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agoqemu-option: remove now-dead code
Paolo Bonzini [Wed, 11 Nov 2020 14:03:37 +0000 (09:03 -0500)]
qemu-option: remove now-dead code

-M was the sole user of qemu_opts_set and qemu_opts_set_defaults,
remove them and the arguments that they used.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agomachine: add smp compound property
Paolo Bonzini [Thu, 13 May 2021 13:03:48 +0000 (09:03 -0400)]
machine: add smp compound property

Make -smp syntactic sugar for a compound property "-machine
smp.{cores,threads,cpu,...}".  machine_smp_parse is replaced by the
setter for the property.

numa-test will now cover the new syntax, while other tests
still use -smp.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agovl: switch -M parsing to keyval
Paolo Bonzini [Thu, 13 May 2021 15:28:34 +0000 (11:28 -0400)]
vl: switch -M parsing to keyval

Switch from QemuOpts to keyval.  This enables the introduction
of non-scalar machine properties, and JSON syntax in the future.

For JSON syntax to be supported right now, we would have to
consider what would happen if string-based dictionaries (produced by
-M key=val) were to be merged with strongly-typed dictionaries
(produced by -M {'key': 123}).

The simplest way out is to never enter the situation, and only allow one
-M option when JSON syntax is in use.  However, we want options such as
-smp to become syntactic sugar for -M, and this is a problem; as soon
as -smp becomes a shortcut for -M, QEMU would forbid using -M '{....}'
together with -smp.  Therefore, allowing JSON syntax right now for -M
would be a forward-compatibility nightmare and it would be impossible
anyway to introduce -M incrementally in tools.

Instead, support for JSON syntax is delayed until after the main
options are converted to QOM compound properties.  These include -boot,
-acpitable, -smbios, -m, -semihosting-config, -rtc and -fw_cfg.  Once JSON
syntax is introduced, these options will _also_ be forbidden together
with -M '{...}'.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agokeyval: introduce keyval_parse_into
Paolo Bonzini [Mon, 2 Nov 2020 12:36:48 +0000 (07:36 -0500)]
keyval: introduce keyval_parse_into

Allow parsing multiple keyval sequences into the same dictionary.
This will be used to simplify the parsing of the -M command line
option, which is currently a .merge_lists = true QemuOpts group.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agokeyval: introduce keyval_merge
Paolo Bonzini [Thu, 12 Nov 2020 13:40:11 +0000 (08:40 -0500)]
keyval: introduce keyval_merge

This patch introduces a function that merges two keyval-produced
(or keyval-like) QDicts.  It can be used to emulate the behavior of
.merge_lists = true QemuOpts groups, merging -readconfig sections and
command-line options in a single QDict, and also to implement -set.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agoqom: export more functions for use with non-UserCreatable objects
Paolo Bonzini [Mon, 2 Nov 2020 16:08:07 +0000 (11:08 -0500)]
qom: export more functions for use with non-UserCreatable objects

Machines and accelerators are not user-creatable but they are going
to share similar command-line parsing machinery.  Export functions
that will be used with -machine and -accel in softmmu/vl.c.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agoconfigure: convert compiler tests to meson, part 6
Paolo Bonzini [Thu, 3 Jun 2021 10:02:00 +0000 (12:02 +0200)]
configure: convert compiler tests to meson, part 6

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agoconfigure: convert compiler tests to meson, part 5
Paolo Bonzini [Thu, 3 Jun 2021 11:04:47 +0000 (13:04 +0200)]
configure: convert compiler tests to meson, part 5

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agoconfigure: convert compiler tests to meson, part 4
Paolo Bonzini [Thu, 3 Jun 2021 10:02:00 +0000 (12:02 +0200)]
configure: convert compiler tests to meson, part 4

And remove them from the summary, since now their outcome is verbosely
included in the meson output.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agoconfigure: convert compiler tests to meson, part 3
Paolo Bonzini [Thu, 3 Jun 2021 10:02:00 +0000 (12:02 +0200)]
configure: convert compiler tests to meson, part 3

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agoconfigure: convert compiler tests to meson, part 2
Paolo Bonzini [Thu, 3 Jun 2021 10:14:48 +0000 (12:14 +0200)]
configure: convert compiler tests to meson, part 2

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agoconfigure: convert compiler tests to meson, part 1
Paolo Bonzini [Thu, 3 Jun 2021 10:10:05 +0000 (12:10 +0200)]
configure: convert compiler tests to meson, part 1

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agoconfigure: convert HAVE_BROKEN_SIZE_MAX to meson
Paolo Bonzini [Thu, 3 Jun 2021 09:57:04 +0000 (11:57 +0200)]
configure: convert HAVE_BROKEN_SIZE_MAX to meson

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agoconfigure, meson: move CONFIG_IVSHMEM to meson
Paolo Bonzini [Thu, 3 Jun 2021 10:50:17 +0000 (12:50 +0200)]
configure, meson: move CONFIG_IVSHMEM to meson

This is a duplicate of CONFIG_EVENTFD, handle it directly in meson.build.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agomeson: store dependency('threads') in a variable
Paolo Bonzini [Thu, 3 Jun 2021 13:01:35 +0000 (15:01 +0200)]
meson: store dependency('threads') in a variable

It can be useful for has_function checks.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agomeson: sort existing compiler tests
Paolo Bonzini [Thu, 3 Jun 2021 09:56:11 +0000 (11:56 +0200)]
meson: sort existing compiler tests

The next patches will add more compiler tests.  Sort and group the
existing tests, keeping similar cc.has_* tests together and sorting them
alphabetically by macro name.  This should make it easier to look for
examples when adding new tests to meson.build.

Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agoconfigure, meson: convert libxml2 detection to meson
Paolo Bonzini [Thu, 3 Jun 2021 09:31:35 +0000 (11:31 +0200)]
configure, meson: convert libxml2 detection to meson

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agoconfigure, meson: convert liburing detection to meson
Paolo Bonzini [Thu, 3 Jun 2021 09:31:35 +0000 (11:31 +0200)]
configure, meson: convert liburing detection to meson

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agoconfigure, meson: convert libpmem detection to meson
Paolo Bonzini [Thu, 3 Jun 2021 09:31:35 +0000 (11:31 +0200)]
configure, meson: convert libpmem detection to meson

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agoconfigure, meson: convert libdaxctl detection to meson
Paolo Bonzini [Thu, 3 Jun 2021 09:31:35 +0000 (11:31 +0200)]
configure, meson: convert libdaxctl detection to meson

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agoconfigure, meson: convert virgl detection to meson
Paolo Bonzini [Thu, 3 Jun 2021 09:31:35 +0000 (11:31 +0200)]
configure, meson: convert virgl detection to meson

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agoconfigure, meson: convert vte detection to meson
Paolo Bonzini [Thu, 3 Jun 2021 09:31:35 +0000 (11:31 +0200)]
configure, meson: convert vte detection to meson

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agoconfigure: drop vte-2.90 check
Paolo Bonzini [Tue, 15 Jun 2021 15:18:42 +0000 (17:18 +0200)]
configure: drop vte-2.90 check

All currently supported distros have vte 0.37 or newer, which is where the
ABI changed from 2.90 to 2.91.  So drop support for the older ABI.

Suggested-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agotarget/i386: Move X86XSaveArea into TCG
David Edmondson [Mon, 5 Jul 2021 10:46:32 +0000 (11:46 +0100)]
target/i386: Move X86XSaveArea into TCG

Given that TCG is now the only consumer of X86XSaveArea, move the
structure definition and associated offset declarations and checks to a
TCG specific header.

Signed-off-by: David Edmondson <david.edmondson@oracle.com>
Message-Id: <20210705104632.2902400-9-david.edmondson@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agotarget/i386: Populate x86_ext_save_areas offsets using cpuid where possible
David Edmondson [Mon, 5 Jul 2021 10:46:31 +0000 (11:46 +0100)]
target/i386: Populate x86_ext_save_areas offsets using cpuid where possible

Rather than relying on the X86XSaveArea structure definition,
determine the offset of XSAVE state areas using CPUID leaf 0xd where
possible (KVM and HVF).

Signed-off-by: David Edmondson <david.edmondson@oracle.com>
Message-Id: <20210705104632.2902400-8-david.edmondson@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agotarget/i386: Observe XSAVE state area offsets
David Edmondson [Mon, 5 Jul 2021 10:46:30 +0000 (11:46 +0100)]
target/i386: Observe XSAVE state area offsets

Rather than relying on the X86XSaveArea structure definition directly,
the routines that manipulate the XSAVE state area should observe the
offsets declared in the x86_ext_save_areas array.

Currently the offsets declared in the array are derived from the
structure definition, resulting in no functional change.

Signed-off-by: David Edmondson <david.edmondson@oracle.com>
Message-Id: <20210705104632.2902400-7-david.edmondson@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agotarget/i386: Make x86_ext_save_areas visible outside cpu.c
David Edmondson [Mon, 5 Jul 2021 10:46:29 +0000 (11:46 +0100)]
target/i386: Make x86_ext_save_areas visible outside cpu.c

Provide visibility of the x86_ext_save_areas array and associated type
outside of cpu.c.

Signed-off-by: David Edmondson <david.edmondson@oracle.com>
Message-Id: <20210705104632.2902400-6-david.edmondson@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agotarget/i386: Pass buffer and length to XSAVE helper
David Edmondson [Mon, 5 Jul 2021 10:46:28 +0000 (11:46 +0100)]
target/i386: Pass buffer and length to XSAVE helper

In preparation for removing assumptions about XSAVE area offsets, pass
a buffer pointer and buffer length to the XSAVE helper functions.

Signed-off-by: David Edmondson <david.edmondson@oracle.com>
Message-Id: <20210705104632.2902400-5-david.edmondson@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agotarget/i386: Clarify the padding requirements of X86XSaveArea
David Edmondson [Mon, 5 Jul 2021 10:46:27 +0000 (11:46 +0100)]
target/i386: Clarify the padding requirements of X86XSaveArea

Replace the hard-coded size of offsets or structure elements with
defined constants or sizeof().

Signed-off-by: David Edmondson <david.edmondson@oracle.com>
Message-Id: <20210705104632.2902400-4-david.edmondson@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agotarget/i386: Consolidate the X86XSaveArea offset checks
David Edmondson [Mon, 5 Jul 2021 10:46:26 +0000 (11:46 +0100)]
target/i386: Consolidate the X86XSaveArea offset checks

Rather than having similar but different checks in cpu.h and kvm.c,
move them all to cpu.h.
Message-Id: <20210705104632.2902400-3-david.edmondson@oracle.com>

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
3 years agotarget/i386: Declare constants for XSAVE offsets
David Edmondson [Mon, 5 Jul 2021 10:46:25 +0000 (11:46 +0100)]
target/i386: Declare constants for XSAVE offsets

Declare and use manifest constants for the XSAVE state component
offsets.

Signed-off-by: David Edmondson <david.edmondson@oracle.com>
Message-Id: <20210705104632.2902400-2-david.edmondson@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>