OSDN Git Service

qmiga/qemu.git
2 years agodocs/system: move x86 CPU configuration to a separate document
Paolo Bonzini [Tue, 7 Sep 2021 14:08:59 +0000 (16:08 +0200)]
docs/system: move x86 CPU configuration to a separate document

Currently, cpu-models-x86.rst.inc is included in target-i386.rst directly.
To make the toctree more homogeneous when adding more documentation,
include it through a first-class .rst file.

Together with the previous changes to the man page skeletons, this also
frees "===" for the headings, so that cpu-models-x86.rst.inc need not
assume anything about the headings used by target-i386.rst.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agodocs/system: standardize man page sections to --- with overline
Paolo Bonzini [Tue, 7 Sep 2021 15:33:27 +0000 (17:33 +0200)]
docs/system: standardize man page sections to --- with overline

Man pages in docs/system use file inclusion heavily.  Use headings with
overlines in the main files, so that the same included file work well
from both manuals and man pages.

This style of heading is a bit more heavy-weight, so it is not used by
the other man pages in interop/ and tools/.  If in the future they
are changed to use include files, for example to avoid having sections
named "synopsis" or "description", they can switch to --- with overline
as well.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agodocs: standardize directory index to --- with overline
Paolo Bonzini [Tue, 7 Sep 2021 14:01:28 +0000 (16:01 +0200)]
docs: standardize directory index to --- with overline

Use a standard heading format for the index.rst file in a directory.
Using overlines makes it clear that individual documents can use e.g.
=== for chapter titles and --- for section titles, as suggested in the
Linux kernel guidelines[1].  They could do it anyway, because documents
included in a toctree are parsed separately and therefore are not tied
to the same conventions for headings.  However, keeping some consistency is
useful since sometimes files are included from multiple places.

[1] https://www.kernel.org/doc/html/latest/doc-guide/sphinx.html

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agodocs: standardize book titles to === with overline
Paolo Bonzini [Tue, 7 Sep 2021 13:58:12 +0000 (15:58 +0200)]
docs: standardize book titles to === with overline

Documents within a Sphinx manual are separate files and therefore can use
different conventions for headings.  However, keeping some consistency is
useful so that included files are easy to get right.

This patch uses a standard heading format for book titles, so that it is
obvious when a file sits at the top level toctree of a book or man page.
The heading is irrelevant for man pages, but keep it consistent as well.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: Added vVMLOAD and vVMSAVE feature
Lara Lazier [Mon, 9 Aug 2021 14:26:28 +0000 (16:26 +0200)]
target/i386: Added vVMLOAD and vVMSAVE feature

The feature allows the VMSAVE and VMLOAD instructions to execute in guest mode without
causing a VMEXIT. (APM2 15.33.1)

Signed-off-by: Lara Lazier <laramglazier@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: Added changed priority check for VIRQ
Lara Lazier [Fri, 13 Aug 2021 09:45:52 +0000 (11:45 +0200)]
target/i386: Added changed priority check for VIRQ

Writes to cr8 affect v_tpr. This could set or unset an interrupt
request as the priority might have changed.

Signed-off-by: Lara Lazier <laramglazier@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: Added ignore TPR check in ctl_has_irq
Lara Lazier [Tue, 17 Aug 2021 17:30:45 +0000 (19:30 +0200)]
target/i386: Added ignore TPR check in ctl_has_irq

The APM2 states that if V_IGN_TPR is nonzero, the current
virtual interrupt ignores the (virtual) TPR.

Signed-off-by: Lara Lazier <laramglazier@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: Added VGIF V_IRQ masking capability
Lara Lazier [Thu, 5 Aug 2021 11:08:23 +0000 (13:08 +0200)]
target/i386: Added VGIF V_IRQ masking capability

VGIF provides masking capability for when virtual interrupts
are taken. (APM2)

Signed-off-by: Lara Lazier <laramglazier@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: Moved int_ctl into CPUX86State structure
Lara Lazier [Sat, 14 Aug 2021 07:51:00 +0000 (09:51 +0200)]
target/i386: Moved int_ctl into CPUX86State structure

Moved int_ctl into the CPUX86State structure.  It removes some
unnecessary stores and loads, and prepares for tracking the vIRQ
state even when it is masked due to vGIF.

Signed-off-by: Lara Lazier <laramglazier@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: Added VGIF feature
Lara Lazier [Fri, 30 Jul 2021 07:07:42 +0000 (09:07 +0200)]
target/i386: Added VGIF feature

VGIF allows STGI and CLGI to execute in guest mode and control virtual
interrupts in guest mode.
When the VGIF feature is enabled then:
 * executing STGI in the guest sets bit 9 of the VMCB offset 60h.
 * executing CLGI in the guest clears bit 9 of the VMCB offset 60h.

Signed-off-by: Lara Lazier <laramglazier@gmail.com>
Message-Id: <20210730070742.9674-1-laramglazier@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: VMRUN and VMLOAD canonicalizations
Lara Lazier [Wed, 4 Aug 2021 11:30:58 +0000 (13:30 +0200)]
target/i386: VMRUN and VMLOAD canonicalizations

APM2 requires that VMRUN and VMLOAD canonicalize (sign extend to 63
from 48/57) all base addresses in the segment registers that have been
respectively loaded.

Signed-off-by: Lara Lazier <laramglazier@gmail.com>
Message-Id: <20210804113058.45186-1-laramglazier@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: add missing bits to CR4_RESERVED_MASK
Daniel P. Berrangé [Tue, 31 Aug 2021 17:50:33 +0000 (18:50 +0100)]
target/i386: add missing bits to CR4_RESERVED_MASK

Booting Fedora kernels with -cpu max hangs very early in boot. Disabling
the la57 CPUID bit fixes the problem. git bisect traced the regression to

  commit 213ff024a2f92020290296cb9dc29c2af3d4a221 (HEAD, refs/bisect/bad)
  Author: Lara Lazier <laramglazier@gmail.com>
  Date:   Wed Jul 21 17:26:50 2021 +0200

    target/i386: Added consistency checks for CR4

    All MBZ bits in CR4 must be zero. (APM2 15.5)
    Added reserved bitmask and added checks in both
    helper_vmrun and helper_write_crN.

Signed-off-by: Lara Lazier <laramglazier@gmail.com>
    Message-Id: <20210721152651.14683-2-laramglazier@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
In this commit CR4_RESERVED_MASK is missing CR4_LA57_MASK and
two others. Adding this lets Fedora kernels boot once again.

Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Tested-by: Richard W.M. Jones <rjones@redhat.com>
Message-Id: <20210831175033.175584-1-berrange@redhat.com>
[Removed VMXE/SMXE, matching the commit message. - Paolo]
Fixes: 213ff024a2 ("target/i386: Added consistency checks for CR4", 2021-07-22)
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoMerge remote-tracking branch 'remotes/mcayland/tags/qemu-openbios-20210908' into...
Peter Maydell [Wed, 8 Sep 2021 10:06:17 +0000 (11:06 +0100)]
Merge remote-tracking branch 'remotes/mcayland/tags/qemu-openbios-20210908' into staging

qemu-openbios queue

# gpg: Signature made Wed 08 Sep 2021 10:34:07 BST
# gpg:                using RSA key CC621AB98E82200D915CC9C45BC2C56FAE0F321F
# gpg:                issuer "mark.cave-ayland@ilande.co.uk"
# gpg: Good signature from "Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>" [full]
# Primary key fingerprint: CC62 1AB9 8E82 200D 915C  C9C4 5BC2 C56F AE0F 321F

* remotes/mcayland/tags/qemu-openbios-20210908:
  Update OpenBIOS images to d657b653 built from submodule.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
2 years agoUpdate OpenBIOS images to d657b653 built from submodule.
Mark Cave-Ayland [Wed, 8 Sep 2021 09:29:29 +0000 (10:29 +0100)]
Update OpenBIOS images to d657b653 built from submodule.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
2 years agoMerge remote-tracking branch 'remotes/thuth-gitlab/tags/s390x-pull-request-2021-09...
Peter Maydell [Tue, 7 Sep 2021 16:46:13 +0000 (17:46 +0100)]
Merge remote-tracking branch 'remotes/thuth-gitlab/tags/s390x-pull-request-2021-09-07' into staging

* Some CSS related fixes
* Storage key related fixes
* Test SIGILL and SIGSEGV handling in usermode emulation
* Fix SETPREFIX instruction
* Replace PAGE_SIZE, PAGE_SHIFT and PAGE_MASK to fix Alpine compilation
* Add more feature to gen16 default model

# gpg: Signature made Tue 07 Sep 2021 14:07:38 BST
# gpg:                using RSA key 27B88847EEE0250118F3EAB92ED9D774FE702DB5
# gpg:                issuer "thuth@redhat.com"
# gpg: Good signature from "Thomas Huth <th.huth@gmx.de>" [full]
# gpg:                 aka "Thomas Huth <thuth@redhat.com>" [full]
# gpg:                 aka "Thomas Huth <huth@tuxfamily.org>" [full]
# gpg:                 aka "Thomas Huth <th.huth@posteo.de>" [unknown]
# Primary key fingerprint: 27B8 8847 EEE0 2501 18F3  EAB9 2ED9 D774 FE70 2DB5

* remotes/thuth-gitlab/tags/s390x-pull-request-2021-09-07:
  s390x/cpumodel: Add more feature to gen16 default model
  s390x: Replace PAGE_SIZE, PAGE_SHIFT and PAGE_MASK
  hw/s390x/s390-skeys: lazy storage key enablement under TCG
  hw/s390x/s390-skeys: rename skeys_enabled to skeys_are_enabled
  hw/s390x/s390-skeys: check if an address is valid before dumping the key
  hw/s390x/s390-skeys: use memory mapping to detect which storage keys to dump
  hw/s390x/s390-skeys: use memory mapping to detect which storage keys to migrate
  s390x/mmu_helper: avoid setting the storage key if nothing changed
  s390x/mmu_helper: move address validation into mmu_translate*()
  s390x/mmu_helper: fixup mmu_translate() documentation
  s390x/mmu_helper: no need to pass access type to mmu_translate_asce()
  s390x/tcg: check for addressing exceptions for RRBE, SSKE and ISKE
  s390x/tcg: convert real to absolute address for RRBE, SSKE and ISKE
  s390x/tcg: fix ignoring bit 63 when setting the storage key in SSKE
  s390x/tcg: wrap address for RRBE
  s390x/ioinst: Fix wrong MSCH alignment check on little endian
  s390x/tcg: fix and optimize SPX (SET PREFIX)
  tests/tcg/s390x: Test SIGILL and SIGSEGV handling
  css: fix actl handling for unit exceptions
  vfio-ccw: forward halt/clear errors

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
2 years agoMerge remote-tracking branch 'remotes/stefanha-gitlab/tags/block-pull-request' into...
Peter Maydell [Tue, 7 Sep 2021 12:24:43 +0000 (13:24 +0100)]
Merge remote-tracking branch 'remotes/stefanha-gitlab/tags/block-pull-request' into staging

Pull request

Userspace NVMe driver patches.

# gpg: Signature made Tue 07 Sep 2021 09:13:57 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/nvme: Only report VFIO error on failed retry
  util/vfio-helpers: Let qemu_vfio_do_mapping() propagate Error
  util/vfio-helpers: Simplify qemu_vfio_dma_map() returning directly
  util/vfio-helpers: Use error_setg in qemu_vfio_find_[fixed/temp]_iova
  util/vfio-helpers: Extract qemu_vfio_water_mark_reached()
  util/vfio-helpers: Pass Error handle to qemu_vfio_dma_map()
  block/nvme: Have nvme_create_queue_pair() report errors consistently
  util/vfio-helpers: Remove unreachable code in qemu_vfio_dma_map()
  util/vfio-helpers: Replace qemu_mutex_lock() calls with QEMU_LOCK_GUARD
  util/vfio-helpers: Let qemu_vfio_verify_mappings() use error_report()
  block/nvme: Use safer trace format string

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
2 years agos390x/cpumodel: Add more feature to gen16 default model
Christian Borntraeger [Tue, 7 Sep 2021 10:10:17 +0000 (10:10 +0000)]
s390x/cpumodel: Add more feature to gen16 default model

Add the new gen16 features to the default model and fence them for
machine version 6.1 and earlier.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210907101017.27126-1-borntraeger@de.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agoMerge remote-tracking branch 'remotes/mjt/tags/patch-fetch' into staging
Peter Maydell [Tue, 7 Sep 2021 09:15:48 +0000 (10:15 +0100)]
Merge remote-tracking branch 'remotes/mjt/tags/patch-fetch' into staging

qemu-socket unix socket bugfix 2021-09-06

# gpg: Signature made Mon 06 Sep 2021 16:19:32 BST
# gpg:                using RSA key 7B73BAD68BE7A2C289314B22701B4F6B1A693E59
# gpg:                issuer "mjt@tls.msk.ru"
# gpg: Good signature from "Michael Tokarev <mjt@tls.msk.ru>" [full]
# gpg:                 aka "Michael Tokarev <mjt@corpit.ru>" [full]
# gpg:                 aka "Michael Tokarev <mjt@debian.org>" [full]
# Primary key fingerprint: 6EE1 95D1 886E 8FFB 810D  4324 457C E0A0 8044 65C5
#      Subkey fingerprint: 7B73 BAD6 8BE7 A2C2 8931  4B22 701B 4F6B 1A69 3E59

* remotes/mjt/tags/patch-fetch:
  qemu-sockets: fix unix socket path copy (again)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
2 years agoblock/nvme: Only report VFIO error on failed retry
Philippe Mathieu-Daudé [Thu, 2 Sep 2021 07:00:25 +0000 (09:00 +0200)]
block/nvme: Only report VFIO error on failed retry

We expect the first qemu_vfio_dma_map() to fail (indicating
DMA mappings exhaustion, see commit 15a730e7a3a). Do not
report the first failure as error, since we are going to
flush the mappings and retry.

This removes spurious error message displayed on the monitor:

  (qemu) c
  (qemu) qemu-kvm: VFIO_MAP_DMA failed: No space left on device
  (qemu) info status
  VM status: running

Reported-by: Tingting Mao <timao@redhat.com>
Reviewed-by: Klaus Jensen <k.jensen@samsung.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20210902070025.197072-12-philmd@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2 years agoutil/vfio-helpers: Let qemu_vfio_do_mapping() propagate Error
Philippe Mathieu-Daudé [Thu, 2 Sep 2021 07:00:24 +0000 (09:00 +0200)]
util/vfio-helpers: Let qemu_vfio_do_mapping() propagate Error

Pass qemu_vfio_do_mapping() an Error* argument so it can propagate
any error to callers. Replace error_report() which only report
to the monitor by the more generic error_setg_errno().

Reviewed-by: Fam Zheng <fam@euphon.net>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Klaus Jensen <k.jensen@samsung.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20210902070025.197072-11-philmd@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2 years agoutil/vfio-helpers: Simplify qemu_vfio_dma_map() returning directly
Philippe Mathieu-Daudé [Thu, 2 Sep 2021 07:00:23 +0000 (09:00 +0200)]
util/vfio-helpers: Simplify qemu_vfio_dma_map() returning directly

To simplify qemu_vfio_dma_map():
- reduce 'ret' (returned value) scope by returning errno directly,
- remove the goto 'out' label.

Reviewed-by: Klaus Jensen <k.jensen@samsung.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20210902070025.197072-10-philmd@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2 years agoutil/vfio-helpers: Use error_setg in qemu_vfio_find_[fixed/temp]_iova
Philippe Mathieu-Daudé [Thu, 2 Sep 2021 07:00:22 +0000 (09:00 +0200)]
util/vfio-helpers: Use error_setg in qemu_vfio_find_[fixed/temp]_iova

Both qemu_vfio_find_fixed_iova() and qemu_vfio_find_temp_iova()
return an errno which is unused (or overwritten). Have them propagate
eventual errors to callers, returning a boolean (which is what the
Error API recommends, see commit e3fe3988d78 "error: Document Error
API usage rules" for rationale).

Suggested-by: Klaus Jensen <k.jensen@samsung.com>
Reviewed-by: Klaus Jensen <k.jensen@samsung.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20210902070025.197072-9-philmd@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2 years agoutil/vfio-helpers: Extract qemu_vfio_water_mark_reached()
Philippe Mathieu-Daudé [Thu, 2 Sep 2021 07:00:21 +0000 (09:00 +0200)]
util/vfio-helpers: Extract qemu_vfio_water_mark_reached()

Extract qemu_vfio_water_mark_reached() for readability,
and have it provide an error hint it its Error* handle.

Suggested-by: Klaus Jensen <k.jensen@samsung.com>
Reviewed-by: Klaus Jensen <k.jensen@samsung.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20210902070025.197072-8-philmd@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2 years agoutil/vfio-helpers: Pass Error handle to qemu_vfio_dma_map()
Philippe Mathieu-Daudé [Thu, 2 Sep 2021 07:00:20 +0000 (09:00 +0200)]
util/vfio-helpers: Pass Error handle to qemu_vfio_dma_map()

Currently qemu_vfio_dma_map() displays errors on stderr.
When using management interface, this information is simply
lost. Pass qemu_vfio_dma_map() an Error** handle so it can
propagate the error to callers.

Reviewed-by: Fam Zheng <fam@euphon.net>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Klaus Jensen <k.jensen@samsung.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20210902070025.197072-7-philmd@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2 years agoblock/nvme: Have nvme_create_queue_pair() report errors consistently
Philippe Mathieu-Daudé [Thu, 2 Sep 2021 07:00:19 +0000 (09:00 +0200)]
block/nvme: Have nvme_create_queue_pair() report errors consistently

nvme_create_queue_pair() does not return a boolean value (indicating
eventual error) but a pointer, and is inconsistent in how it fills the
error handler. To fulfill callers expectations, always set an error
message on failure.

Reported-by: Auger Eric <eric.auger@redhat.com>
Reviewed-by: Klaus Jensen <k.jensen@samsung.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20210902070025.197072-6-philmd@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2 years agoutil/vfio-helpers: Remove unreachable code in qemu_vfio_dma_map()
Philippe Mathieu-Daudé [Thu, 2 Sep 2021 07:00:18 +0000 (09:00 +0200)]
util/vfio-helpers: Remove unreachable code in qemu_vfio_dma_map()

qemu_vfio_add_mapping() returns a pointer to an indexed entry
in pre-allocated QEMUVFIOState::mappings[], thus can not be NULL.
Remove the pointless check.

Reviewed-by: Klaus Jensen <k.jensen@samsung.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20210902070025.197072-5-philmd@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2 years agoutil/vfio-helpers: Replace qemu_mutex_lock() calls with QEMU_LOCK_GUARD
Philippe Mathieu-Daudé [Thu, 2 Sep 2021 07:00:17 +0000 (09:00 +0200)]
util/vfio-helpers: Replace qemu_mutex_lock() calls with QEMU_LOCK_GUARD

Simplify qemu_vfio_dma_[un]map() handlers by replacing a pair of
qemu_mutex_lock/qemu_mutex_unlock calls by the WITH_QEMU_LOCK_GUARD
macro.

Reviewed-by: Klaus Jensen <k.jensen@samsung.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20210902070025.197072-4-philmd@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2 years agoutil/vfio-helpers: Let qemu_vfio_verify_mappings() use error_report()
Philippe Mathieu-Daudé [Thu, 2 Sep 2021 07:00:16 +0000 (09:00 +0200)]
util/vfio-helpers: Let qemu_vfio_verify_mappings() use error_report()

Instead of displaying the error on stderr, use error_report()
which also report to the monitor.

Reviewed-by: Fam Zheng <fam@euphon.net>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Klaus Jensen <k.jensen@samsung.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20210902070025.197072-3-philmd@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2 years agoblock/nvme: Use safer trace format string
Philippe Mathieu-Daudé [Thu, 2 Sep 2021 07:00:15 +0000 (09:00 +0200)]
block/nvme: Use safer trace format string

Fix when building with -Wshorten-64-to-32:

  warning: implicit conversion loses integer precision: 'unsigned long' to 'int' [-Wshorten-64-to-32]

Reviewed-by: Klaus Jensen <k.jensen@samsung.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20210902070025.197072-2-philmd@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2 years agos390x: Replace PAGE_SIZE, PAGE_SHIFT and PAGE_MASK
Thomas Huth [Wed, 1 Sep 2021 12:58:00 +0000 (14:58 +0200)]
s390x: Replace PAGE_SIZE, PAGE_SHIFT and PAGE_MASK

The PAGE_SIZE macro is causing trouble on Alpine Linux since it
clashes with a macro from a system header there. We already have
the TARGET_PAGE_SIZE, TARGET_PAGE_MASK and TARGET_PAGE_BITS macros
in QEMU anyway, so let's simply replace the PAGE_SIZE, PAGE_MASK
and PAGE_SHIFT macro with their TARGET_* counterparts.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/572
Message-Id: <20210901125800.611183-1-thuth@redhat.com>
Reviewed-by: Halil Pasic <pasic@linux.ibm.com>
Reviewed-by: Matthew Rosato <mjrosato@linux.ibm.com>
Reviewed-by: Eric Farman <farman@linux.ibm.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agohw/s390x/s390-skeys: lazy storage key enablement under TCG
David Hildenbrand [Fri, 3 Sep 2021 15:55:14 +0000 (17:55 +0200)]
hw/s390x/s390-skeys: lazy storage key enablement under TCG

Let's enable storage keys lazily under TCG, just as we do under KVM.
Only fairly old Linux versions actually make use of storage keys, so it
can be kind of wasteful to allocate quite some memory and track
changes and references if nobody cares.

We have to make sure to flush the TLB when enabling storage keys after
the VM was already running: otherwise it might happen that we don't
catch references or modifications afterwards.

Add proper documentation to all callbacks.

The kvm-unit-tests skey tests keeps on working with this change.

Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20210903155514.44772-14-david@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agohw/s390x/s390-skeys: rename skeys_enabled to skeys_are_enabled
David Hildenbrand [Fri, 3 Sep 2021 15:55:13 +0000 (17:55 +0200)]
hw/s390x/s390-skeys: rename skeys_enabled to skeys_are_enabled

... and make it return a bool instead.

Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20210903155514.44772-13-david@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agohw/s390x/s390-skeys: check if an address is valid before dumping the key
David Hildenbrand [Fri, 3 Sep 2021 15:55:12 +0000 (17:55 +0200)]
hw/s390x/s390-skeys: check if an address is valid before dumping the key

Let's validate the given address and report a proper error in case it's
not. All call paths now properly check the validity of the given GFN.
Remove the TODO.

The errors inside the getter and setter should only trigger if something
really goes wrong now, for example, with a broken migration stream. Or
when we forget to update the storage key allocation with memory hotplug.

Signed-off-by: David Hildenbrand <david@redhat.com>
Acked-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20210903155514.44772-12-david@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agohw/s390x/s390-skeys: use memory mapping to detect which storage keys to dump
David Hildenbrand [Fri, 3 Sep 2021 15:55:11 +0000 (17:55 +0200)]
hw/s390x/s390-skeys: use memory mapping to detect which storage keys to dump

Handle it similar to migration. Assert that we're holding the BQL, to
make sure we don't see concurrent modifications.

Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20210903155514.44772-11-david@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agohw/s390x/s390-skeys: use memory mapping to detect which storage keys to migrate
David Hildenbrand [Fri, 3 Sep 2021 15:55:10 +0000 (17:55 +0200)]
hw/s390x/s390-skeys: use memory mapping to detect which storage keys to migrate

Let's use the guest_phys_blocks API to get physical memory regions
that are well defined inside our physical address space and migrate the
storage keys of these.

This is a preparation for having memory besides initial ram defined in
the guest physical address space, for example, via memory devices. We
get rid of the ms->ram_size dependency.

Please note that we will usually have very little (--> 1) physical
ranges. With virtio-mem might have significantly more ranges in the
future. If that turns out to be a problem (e.g., total memory
footprint of the list), we could look into a memory mapping
API that avoids creation of a list and instead triggers a callback for
each range.

Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20210903155514.44772-10-david@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agos390x/mmu_helper: avoid setting the storage key if nothing changed
David Hildenbrand [Fri, 3 Sep 2021 15:55:09 +0000 (17:55 +0200)]
s390x/mmu_helper: avoid setting the storage key if nothing changed

Avoid setting the key if nothing changed.

Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20210903155514.44772-9-david@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agos390x/mmu_helper: move address validation into mmu_translate*()
David Hildenbrand [Fri, 3 Sep 2021 15:55:08 +0000 (17:55 +0200)]
s390x/mmu_helper: move address validation into mmu_translate*()

Let's move address validation into mmu_translate() and
mmu_translate_real(). This allows for checking whether an absolute
address is valid before looking up the storage key. We can now get rid of
the ram_size check.

Interestingly, we're already handling LOAD REAL ADDRESS wrong, because
a) We're not supposed to touch storage keys
b) We're not supposed to convert to an absolute address

Let's use a fake, negative MMUAccessType to teach mmu_translate() to
fix that handling and to not perform address validation.

Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20210903155514.44772-8-david@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agos390x/mmu_helper: fixup mmu_translate() documentation
David Hildenbrand [Fri, 3 Sep 2021 15:55:07 +0000 (17:55 +0200)]
s390x/mmu_helper: fixup mmu_translate() documentation

Looks like we forgot to adjust documentation of one parameter.

Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20210903155514.44772-7-david@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agos390x/mmu_helper: no need to pass access type to mmu_translate_asce()
David Hildenbrand [Fri, 3 Sep 2021 15:55:06 +0000 (17:55 +0200)]
s390x/mmu_helper: no need to pass access type to mmu_translate_asce()

The access type is unused since commit 81d7e3bc45 ("s390x/mmu: Inject
DAT exceptions from a single place").

Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20210903155514.44772-6-david@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agos390x/tcg: check for addressing exceptions for RRBE, SSKE and ISKE
David Hildenbrand [Fri, 3 Sep 2021 15:55:05 +0000 (17:55 +0200)]
s390x/tcg: check for addressing exceptions for RRBE, SSKE and ISKE

Let's replace the ram_size check by a proper physical address space
check (for example, to prepare for memory hotplug), trigger addressing
exceptions and trace the return value of the storage key getter/setter.

Provide an helper mmu_absolute_addr_valid() to be used in other context
soon. Always test for "read" instead of "write" as we are not actually
modifying the page itself.

Signed-off-by: David Hildenbrand <david@redhat.com>
Acked-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20210903155514.44772-5-david@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agos390x/tcg: convert real to absolute address for RRBE, SSKE and ISKE
David Hildenbrand [Fri, 3 Sep 2021 15:55:04 +0000 (17:55 +0200)]
s390x/tcg: convert real to absolute address for RRBE, SSKE and ISKE

For RRBE, SSKE, and ISKE, we're dealing with real addresses, so we have to
convert to an absolute address first.

In the future, when adding EDAT1 support, we'll have to pay attention to
SSKE handling, as we'll be dealing with absolute addresses when the
multiple-block control is one.

Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20210903155514.44772-4-david@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agos390x/tcg: fix ignoring bit 63 when setting the storage key in SSKE
David Hildenbrand [Fri, 3 Sep 2021 15:55:03 +0000 (17:55 +0200)]
s390x/tcg: fix ignoring bit 63 when setting the storage key in SSKE

Right now we could set an 8-bit storage key via SSKE and retrieve it
again via ISKE, which is against the architecture description:

SSKE:
"
The new seven-bit storage-key value, or selected bits
thereof, is obtained from bit positions 56-62 of gen-
eral register R 1 . The contents of bit positions 0-55
and 63 of the register are ignored.
"

ISKE:
"
The seven-bit storage key is inserted in bit positions
56-62 of general register R 1 , and bit 63 is set to zero.
"

Let's properly ignore bit 63 to create the correct seven-bit storage key.

Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20210903155514.44772-3-david@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agos390x/tcg: wrap address for RRBE
David Hildenbrand [Fri, 3 Sep 2021 15:55:02 +0000 (17:55 +0200)]
s390x/tcg: wrap address for RRBE

Let's wrap the address just like for SSKE and ISKE.

Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20210903155514.44772-2-david@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agos390x/ioinst: Fix wrong MSCH alignment check on little endian
David Hildenbrand [Thu, 5 Aug 2021 14:37:53 +0000 (16:37 +0200)]
s390x/ioinst: Fix wrong MSCH alignment check on little endian

schib->pmcw.chars is 32bit, not 16bit. This fixes the kvm-unit-tests
"css" test, which fails with:

  FAIL: Channel Subsystem: measurement block format1: Unaligned MB origin:
  Program interrupt: expected(21) == received(0)

Because we end up not injecting an operand program exception.

Fixes: a54b8ac340c2 ("css: SCHIB measurement block origin must be aligned")
Cc: Halil Pasic <pasic@linux.ibm.com>
Cc: Cornelia Huck <cohuck@redhat.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Richard Henderson <richard.henderson@linaro.org>
Cc: Thomas Huth <thuth@redhat.com>
Cc: Pierre Morel <pmorel@linux.ibm.com>
Cc: qemu-s390x@nongnu.org
Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Halil Pasic <pasic@linux.ibm.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Pierre Morel <pmorel@linux.ibm.com>
Message-Id: <20210805143753.86520-1-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agos390x/tcg: fix and optimize SPX (SET PREFIX)
David Hildenbrand [Thu, 5 Aug 2021 12:59:38 +0000 (14:59 +0200)]
s390x/tcg: fix and optimize SPX (SET PREFIX)

We not only invalidate the translation of the range 0x0-0x2000, we also
invalidate the translation of the new prefix range and the translation
of the old prefix range -- because real2abs would return different
results for all of these ranges when changing the prefix location.

This fixes the kvm-unit-tests "edat" test that just hangs before this
patch because we end up clearing the new prefix area instead of the old
prefix area.

While at it, let's not do anything in case the prefix doesn't change.

Cc: Richard Henderson <richard.henderson@linaro.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: Cornelia Huck <cohuck@redhat.com>
Cc: Thomas Huth <thuth@redhat.com>
Cc: Claudio Imbrenda <imbrenda@linux.ibm.com>
Cc: qemu-s390x@nongnu.org
Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Claudio Imbrenda <imbrenda@linux.ibm.com>
Message-Id: <20210805125938.74034-1-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agotests/tcg/s390x: Test SIGILL and SIGSEGV handling
Ilya Leoshkevich [Wed, 4 Aug 2021 22:51:46 +0000 (00:51 +0200)]
tests/tcg/s390x: Test SIGILL and SIGSEGV handling

Verify that s390x-specific uc_mcontext.psw.addr is reported correctly
and that signal handling interacts properly with debugging.

Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Acked-by: Alex Bennée <alex.bennee@linaro.org>
Acked-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210804225146.154513-1-iii@linux.ibm.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agocss: fix actl handling for unit exceptions
Cornelia Huck [Mon, 5 Jul 2021 16:39:52 +0000 (18:39 +0200)]
css: fix actl handling for unit exceptions

When a subchannel becomes pending with unit exception, start
pending (and for that matter, halt or clear pending) are not
removed in the actl. Device active and subchannel active,
however, are (due to the subchannel becoming status pending
with primary respectively secondary status).

The other conditions in the actl are only cleared when the
guest executes tsch on the subchannel.

Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Matthew Rosato <mjrosato@linux.ibm.com>
Tested-by: Jared Rossi <jrossi@linux.ibm.com>
Message-Id: <20210705163952.736020-3-cohuck@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agovfio-ccw: forward halt/clear errors
Cornelia Huck [Mon, 5 Jul 2021 16:39:51 +0000 (18:39 +0200)]
vfio-ccw: forward halt/clear errors

hsch and csch basically have two parts: execute the command,
and perform the halt/clear function. For fully emulated
subchannels, it is pretty clear how it will work: check the
subchannel state, and actually 'perform the halt/clear function'
and set cc 0 if everything looks good.

For passthrough subchannels, some of the checking is done
within QEMU, but some has to be done within the kernel. QEMU's
subchannel state may be such that we can perform the async
function, but the kernel may still get a cc != 0 when it is
actually executing the instruction. In that case, we need to
set the condition actually encountered by the kernel; if we
set cc 0 on error, we would actually need to inject an interrupt
as well.

Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Matthew Rosato <mjrosato@linux.ibm.com>
Tested-by: Jared Rossi <jrossi@linux.ibm.com>
Message-Id: <20210705163952.736020-2-cohuck@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agoqemu-sockets: fix unix socket path copy (again)
Michael Tokarev [Wed, 1 Sep 2021 13:16:24 +0000 (16:16 +0300)]
qemu-sockets: fix unix socket path copy (again)

Commit 4cfd970ec188558daa6214f26203fe553fb1e01f added an
assert which ensures the path within an address of a unix
socket returned from the kernel is at least one byte and
does not exceed sun_path buffer. Both of this constraints
are wrong:

A unix socket can be unnamed, in this case the path is
completely empty (not even \0)

And some implementations (notable linux) can add extra
trailing byte (\0) _after_ the sun_path buffer if we
passed buffer larger than it (and we do).

So remove the assertion (since it causes real-life breakage)
but at the same time fix the usage of sun_path. Namely,
we should not access sun_path[0] if kernel did not return
it at all (this is the case for unnamed sockets),
and use the returned salen when copyig actual path as an
upper constraint for the amount of bytes to copy - this
will ensure we wont exceed the information provided by
the kernel, regardless whenever there is a trailing \0
or not. This also helps with unnamed sockets.

Note the case of abstract socket, the sun_path is actually
a blob and can contain \0 characters, - it should not be
passed to g_strndup and the like, it should be accessed by
memcpy-like functions.

Fixes: 4cfd970ec188558daa6214f26203fe553fb1e01f
Fixes: http://bugs.debian.org/993145
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
CC: qemu-stable@nongnu.org
2 years agoMerge remote-tracking branch 'remotes/thuth-gitlab/tags/pull-request-2021-09-06'...
Peter Maydell [Mon, 6 Sep 2021 11:38:07 +0000 (12:38 +0100)]
Merge remote-tracking branch 'remotes/thuth-gitlab/tags/pull-request-2021-09-06' into staging

* Add definitions of terms for CI/testing
* Fix g_setenv problem discovered by Coverity
* Gitlab CI improvements
* Build system improvements (configure script + meson.build)
* Removal of the show-fixed-bugs.sh script
* Clean up of the sdl and curses options

# gpg: Signature made Mon 06 Sep 2021 10:51:49 BST
# gpg:                using RSA key 27B88847EEE0250118F3EAB92ED9D774FE702DB5
# gpg:                issuer "thuth@redhat.com"
# gpg: Good signature from "Thomas Huth <th.huth@gmx.de>" [full]
# gpg:                 aka "Thomas Huth <thuth@redhat.com>" [full]
# gpg:                 aka "Thomas Huth <huth@tuxfamily.org>" [full]
# gpg:                 aka "Thomas Huth <th.huth@posteo.de>" [unknown]
# Primary key fingerprint: 27B8 8847 EEE0 2501 18F3  EAB9 2ED9 D774 FE70 2DB5

* remotes/thuth-gitlab/tags/pull-request-2021-09-06:
  softmmu/vl: Deprecate the -sdl and -curses option
  softmmu/vl: Deprecate the old grab options
  softmmu/vl: Add a "grab-mod" parameter to the -display sdl option
  scripts: Remove the "show-fixed-bugs.sh" file
  configure / meson: Move the GBM handling to meson.build
  meson.build: Don't use internal libfdt if the user requested the system libfdt
  meson.build: Fix the check for a usable libfdt
  gitlab-ci: Don't try to use the system libfdt in the debian job
  libqtest: check for g_setenv() failure
  docs: add definitions of terms for CI/testing

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
2 years agosoftmmu/vl: Deprecate the -sdl and -curses option
Thomas Huth [Wed, 25 Aug 2021 09:20:23 +0000 (11:20 +0200)]
softmmu/vl: Deprecate the -sdl and -curses option

It's not that much complicated to type "-display sdl" or "-display curses",
so we should not clutter our main option name space with such simple
wrapper options and rather present the users with a concise interface
instead. Thus let's deprecate the "-sdl" and "-curses" wrapper options now.

Message-Id: <20210825092023.81396-4-thuth@redhat.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Peter Krempa <pkrempa@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agosoftmmu/vl: Deprecate the old grab options
Thomas Huth [Wed, 25 Aug 2021 09:20:22 +0000 (11:20 +0200)]
softmmu/vl: Deprecate the old grab options

The alt_grab and ctrl_grab parameter of the -display sdl option prevent
the QAPIfication of the "sdl" part of the -display option, so we should
eventually remove them. And since this feature is also rather niche anyway,
we should not clutter the top-level option list with these, so let's
also deprecate the "-alt-grab" and the "-ctrl-grab" options while we're
at it.

Once the deprecation period of "alt_grab" and "ctrl_grab" is over, we
then can finally switch the -display sdl option to use QAPI internally,
too.

Message-Id: <20210825092023.81396-3-thuth@redhat.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Peter Krempa <pkrempa@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agosoftmmu/vl: Add a "grab-mod" parameter to the -display sdl option
Thomas Huth [Wed, 25 Aug 2021 09:20:21 +0000 (11:20 +0200)]
softmmu/vl: Add a "grab-mod" parameter to the -display sdl option

The -display sdl option is not using QAPI internally yet, and uses hand-
crafted parsing instead (see parse_display() in vl.c), which is quite
ugly, since most of the other code is using the QAPIfied DisplayOption
already. Unfortunately, the "alt_grab" and "ctrl_grab" use underscores in
their names which has recently been forbidden in new QAPI code, so
a straight conversion is not possible. While we could add some exceptions
to the QAPI schema parser for this, the way these parameters have been
designed was maybe a bad idea anyway: First, it's not possible to enable
both parameters at the same time, thus instead of two boolean parameters
it would be better to have only one multi-choice parameter instead.
Second, the naming is also somewhat unfortunate since the "alt_grab"
parameter is not about the ALT key, but rather about the left SHIFT key
that has to be used additionally when the parameter is enabled.

So instead of trying to QAPIfy "alt_grab" and "ctrl_grab", let's rather
introduce an alternative to these parameters instead, a new parameter
called "grab-mod" which can either be set to "lshift-lctrl-lalt" or to
"rctrl". In case we ever want to support additional modes later, we can
then also simply extend the list of supported strings here.

Message-Id: <20210825092023.81396-2-thuth@redhat.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agoscripts: Remove the "show-fixed-bugs.sh" file
Thomas Huth [Wed, 25 Aug 2021 14:21:43 +0000 (16:21 +0200)]
scripts: Remove the "show-fixed-bugs.sh" file

Since we are not using Launchpad anymore, there is no more need for
this script.

Message-Id: <20210825142143.142037-1-thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agoconfigure / meson: Move the GBM handling to meson.build
Thomas Huth [Tue, 13 Jul 2021 11:09:02 +0000 (13:09 +0200)]
configure / meson: Move the GBM handling to meson.build

The GBM library detection does not need to be in the configure script,
since it does not have any user-facing options (there are no
--enable-gbm or --disable-gbm switches). Let's move it to meson.build
instead, so we don't have to clutter config-host.mak with the related
switches.

Additionally, only check for GBM if it is really required, i.e. if we
either compile with OpenGL or with virglrenderer support.

Message-Id: <20210714085045.797168-1-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agomeson.build: Don't use internal libfdt if the user requested the system libfdt
Thomas Huth [Fri, 27 Aug 2021 12:09:00 +0000 (14:09 +0200)]
meson.build: Don't use internal libfdt if the user requested the system libfdt

If the users ran configure with --enable-libfdt=system, they likely did
that on purpose. We should not silently fall back to the internal libfdt
if the system libfdt is not usable, but report the problem with a proper
message instead.

Message-Id: <20210827120901.150276-3-thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agomeson.build: Fix the check for a usable libfdt
Thomas Huth [Fri, 27 Aug 2021 12:08:59 +0000 (14:08 +0200)]
meson.build: Fix the check for a usable libfdt

The check for libfdt currently has a flaw: If there is a system libfdt, the
meson.build code initialized the fdt variable with fdt = cc.find_library(...).
However, if this libfdt is too old and there is no internal dtc module
available, it continues with "fdt" pointing to the old and unusable version.
The check later in the file that tries to detect whether libfdt is necessary
then fails to trigger:

 if not fdt.found() and fdt_required.length() > 0
  error('fdt not available but required by targets ' + ', '.join(fdt_required))
 endif

The build fails then during compilation instead, which is of course bad
since this is quite confusing and already wasted quite some time of the user.
Thus if libfdt is not usable, we should unset the "fdt" variable immediately
again, so that the build already fails during the configuration phase.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/255
Message-Id: <20210827120901.150276-2-thuth@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agogitlab-ci: Don't try to use the system libfdt in the debian job
Thomas Huth [Fri, 27 Aug 2021 15:17:18 +0000 (17:17 +0200)]
gitlab-ci: Don't try to use the system libfdt in the debian job

libfdt in Debian is too old to be usable for QEMU. So far we were
silently falling back to the internal dtc submodule, but since
this is wrong, let's remove the --enable-fdt=system switch here now.

Message-Id: <20210827151718.178988-1-thuth@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Acked-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agolibqtest: check for g_setenv() failure
Peter Maydell [Fri, 20 Aug 2021 16:37:50 +0000 (17:37 +0100)]
libqtest: check for g_setenv() failure

g_setenv() can fail; check for it when starting a QEMU process
when we set the QEMU_AUDIO_DRV environment variable.

Because this happens after fork() reporting an exact message
via printf() is a bad idea; just exit(1), as we already do
for the case of execlp() failure.

Fixes: Coverity CID 1460117
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20210820163750.9106-1-peter.maydell@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agodocs: add definitions of terms for CI/testing
Willian Rampazzo [Tue, 31 Aug 2021 15:29:39 +0000 (12:29 -0300)]
docs: add definitions of terms for CI/testing

To understand the current state of QEMU CI/testing and have a base to
discuss the plans for the future, it is important to define some usual
terms. This patch defines the terms for "Automated tests", "Unit
testing", "Functional testing", "System testing", "Flaky tests",
"Gating", and "Continuous Integration".

Signed-off-by: Willian Rampazzo <willianr@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20210831152939.97570-2-willianr@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
2 years agoMerge remote-tracking branch 'remotes/mst/tags/for_upstream' into staging
Peter Maydell [Sun, 5 Sep 2021 14:48:42 +0000 (15:48 +0100)]
Merge remote-tracking branch 'remotes/mst/tags/for_upstream' into staging

pc,pci,virtio: fixes, cleanups

Fixes, cleanups all over the place.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
# gpg: Signature made Sat 04 Sep 2021 22:34:10 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: (35 commits)
  vhost-vdpa: remove the unncessary queue_index assignment
  vhost-vdpa: fix the wrong assertion in vhost_vdpa_init()
  vhost-vdpa: tweak the error label in vhost_vdpa_add()
  vhost-vdpa: fix leaking of vhost_net in vhost_vdpa_add()
  vhost-vdpa: don't cleanup twice in vhost_vdpa_add()
  vhost-vdpa: remove the unnecessary check in vhost_vdpa_add()
  vhost_net: do not assume nvqs is always 2
  vhost: use unsigned int for nvqs
  vhost_net: remove the meaningless assignment in vhost_net_start_one()
  vhost-vdpa: correctly return err in vhost_vdpa_set_backend_cap()
  vhost-vdpa: remove unused variable "acked_features"
  tests/vhost-user-bridge.c: Fix typo in help message
  tests/vhost-user-bridge.c: Sanity check socket path length
  hw/virtio: Add flatview update in vhost_user_cleanup()
  hw/virtio: Remove NULL check in virtio_free_region_cache()
  hw/virtio: Document virtio_queue_packed_empty_rcu is called within RCU
  MAINTAINERS: Added myself as a reviewer for acpi/smbios subsystem
  hw/acpi: use existing references to pci device struct within functions
  hw/pci: remove all references to find_i440fx function
  hw/i386/acpi-build: Get NUMA information from struct NumaState
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
2 years agovhost-vdpa: remove the unncessary queue_index assignment
Jason Wang [Fri, 3 Sep 2021 09:10:21 +0000 (17:10 +0800)]
vhost-vdpa: remove the unncessary queue_index assignment

The queue_index of NetClientState should be assigned in set_netdev()
afterwards, so trying to net_vhost_vdpa_init() is meaningless. This
patch removes this.

Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20210903091031.47303-12-jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovhost-vdpa: fix the wrong assertion in vhost_vdpa_init()
Jason Wang [Fri, 3 Sep 2021 09:10:20 +0000 (17:10 +0800)]
vhost-vdpa: fix the wrong assertion in vhost_vdpa_init()

Vhost_vdpa_add() can fail for various reasons, so the assertion of the
succeed is wrong. Instead, we should free the NetClientState and
propagate the error to the caller

Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20210903091031.47303-11-jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovhost-vdpa: tweak the error label in vhost_vdpa_add()
Jason Wang [Fri, 3 Sep 2021 09:10:19 +0000 (17:10 +0800)]
vhost-vdpa: tweak the error label in vhost_vdpa_add()

Introduce new error label to avoid the unnecessary checking of net
pointer.

Fixes: 1e0a84ea49b68 ("vhost-vdpa: introduce vhost-vdpa net client")
Signed-off-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20210903091031.47303-10-jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovhost-vdpa: fix leaking of vhost_net in vhost_vdpa_add()
Jason Wang [Fri, 3 Sep 2021 09:10:18 +0000 (17:10 +0800)]
vhost-vdpa: fix leaking of vhost_net in vhost_vdpa_add()

Fixes: 1e0a84ea49b68 ("vhost-vdpa: introduce vhost-vdpa net client")
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20210903091031.47303-9-jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovhost-vdpa: don't cleanup twice in vhost_vdpa_add()
Jason Wang [Fri, 3 Sep 2021 09:10:17 +0000 (17:10 +0800)]
vhost-vdpa: don't cleanup twice in vhost_vdpa_add()

The previous vhost_net_cleanup is sufficient for freeing, calling
vhost_vdpa_del() in this case will lead an extra round of free. Note
that this kind of "double free" is safe since vhost_dev_cleanup() zero
the whole structure.

Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20210903091031.47303-8-jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovhost-vdpa: remove the unnecessary check in vhost_vdpa_add()
Jason Wang [Fri, 3 Sep 2021 09:10:16 +0000 (17:10 +0800)]
vhost-vdpa: remove the unnecessary check in vhost_vdpa_add()

The VhostVDPAState is just allocated by qemu_new_net_client() via
g_malloc0() in net_vhost_vdpa_init(). So s->vhost_net is NULL for
sure, let's remove this unnecessary check in vhost_vdpa_add().

Signed-off-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20210903091031.47303-7-jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovhost_net: do not assume nvqs is always 2
Jason Wang [Fri, 3 Sep 2021 09:10:15 +0000 (17:10 +0800)]
vhost_net: do not assume nvqs is always 2

This patch switches to initialize dev.nvqs from the VhostNetOptions
instead of assuming it was 2. This is useful for implementing control
virtqueue support which will be a single vhost_net structure with a
single cvq.

Note that nvqs is still set to 2 for all users and this patch does not
change functionality.

Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Reviewed-by: Eli Cohen <elic@nvidia.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20210903091031.47303-6-jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovhost: use unsigned int for nvqs
Jason Wang [Fri, 3 Sep 2021 09:10:14 +0000 (17:10 +0800)]
vhost: use unsigned int for nvqs

Switch to use unsigned int for nvqs since it's not expected to be
negative.

Reviewed-by: Eli Cohen <elic@nvidia.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20210903091031.47303-5-jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovhost_net: remove the meaningless assignment in vhost_net_start_one()
Jason Wang [Fri, 3 Sep 2021 09:10:13 +0000 (17:10 +0800)]
vhost_net: remove the meaningless assignment in vhost_net_start_one()

The nvqs and vqs have been initialized during vhost_net_init() and are
not expected to change during the life cycle of vhost_net
structure. So this patch removes the meaningless assignment.

Reviewed-by: Eli Cohen <elic@nvidia.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20210903091031.47303-4-jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovhost-vdpa: correctly return err in vhost_vdpa_set_backend_cap()
Jason Wang [Fri, 3 Sep 2021 09:10:12 +0000 (17:10 +0800)]
vhost-vdpa: correctly return err in vhost_vdpa_set_backend_cap()

We should return error code instead of zero, otherwise there's no way
for the caller to detect the failure.

Signed-off-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20210903091031.47303-3-jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovhost-vdpa: remove unused variable "acked_features"
Jason Wang [Fri, 3 Sep 2021 09:10:11 +0000 (17:10 +0800)]
vhost-vdpa: remove unused variable "acked_features"

"acked_features" is unused, let's remove that.

Signed-off-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20210903091031.47303-2-jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agotests/vhost-user-bridge.c: Fix typo in help message
Peter Maydell [Wed, 1 Sep 2021 15:27:13 +0000 (16:27 +0100)]
tests/vhost-user-bridge.c: Fix typo in help message

Fix a typo in the help message printed by vhost-user-bridge.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20210901152713.25701-1-peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agotests/vhost-user-bridge.c: Sanity check socket path length
Peter Maydell [Wed, 1 Sep 2021 15:26:32 +0000 (16:26 +0100)]
tests/vhost-user-bridge.c: Sanity check socket path length

The vhost-user-bridge binary accepts a UNIX socket path on
the command line. Sanity check that this is short enough to
fit into a sockaddr_un before copying it in.

Fixes: Coverity CID 1432866
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20210901152632.25511-1-peter.maydell@linaro.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
2 years agohw/virtio: Add flatview update in vhost_user_cleanup()
Yuwei Zhang [Mon, 30 Aug 2021 12:34:33 +0000 (20:34 +0800)]
hw/virtio: Add flatview update in vhost_user_cleanup()

Qemu will crash on vhost backend unexpected exit and re-connect                                                                          │
in some case due to access released memory.

Signed-off-by: Yuwei Zhang <zhangyuwei.9149@bytedance.com>
Message-Id: <20210830123433.45727-1-zhangyuwei.9149@bytedance.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agohw/virtio: Remove NULL check in virtio_free_region_cache()
Philippe Mathieu-Daudé [Thu, 26 Aug 2021 17:26:57 +0000 (19:26 +0200)]
hw/virtio: Remove NULL check in virtio_free_region_cache()

virtio_free_region_cache() is called within call_rcu(),
always with a non-NULL argument. Ensure new code keep it
that way by replacing the NULL check by an assertion.
Add a comment this function is called within call_rcu().

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20210826172658.2116840-3-philmd@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
2 years agohw/virtio: Document virtio_queue_packed_empty_rcu is called within RCU
Philippe Mathieu-Daudé [Thu, 26 Aug 2021 17:26:56 +0000 (19:26 +0200)]
hw/virtio: Document virtio_queue_packed_empty_rcu is called within RCU

While virtio_queue_packed_empty_rcu() uses the '_rcu' suffix,
it is not obvious it is called within rcu_read_lock(). All other
functions from this file called with the RCU locked have a comment
describing it. Document this one similarly for consistency.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20210826172658.2116840-2-philmd@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
2 years agoMAINTAINERS: Added myself as a reviewer for acpi/smbios subsystem
Ani Sinha [Wed, 25 Aug 2021 03:19:47 +0000 (08:49 +0530)]
MAINTAINERS: Added myself as a reviewer for acpi/smbios subsystem

I have developed an interest in this space and hopefully can lend some
helping hand to Igor and Michael in reviewing simpler patches.

Signed-off-by: Ani Sinha <ani@anisinha.ca>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Acked-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20210825031949.919376-4-ani@anisinha.ca>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agohw/acpi: use existing references to pci device struct within functions
Ani Sinha [Wed, 25 Aug 2021 03:19:46 +0000 (08:49 +0530)]
hw/acpi: use existing references to pci device struct within functions

There is no need to use fresh typecasts to get references to pci device structs
when there is an existing reference to pci device struct. Use existing reference.
Minor cleanup.

Signed-off-by: Ani Sinha <ani@anisinha.ca>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20210825031949.919376-3-ani@anisinha.ca>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agohw/pci: remove all references to find_i440fx function
Ani Sinha [Wed, 25 Aug 2021 03:19:45 +0000 (08:49 +0530)]
hw/pci: remove all references to find_i440fx function

commit c0e427d6eb5fefc538 ("hw/acpi/ich9: Enable ACPI PCI hot-plug") removed all
uses of find_i440fx() function. This has been replaced by the more generic call
acpi_get_i386_pci_host() which maybe able to find the root bus both for i440fx
machine type as well as for the q35 machine type. There seems to be no more any
need to maintain a i440fx specific version of the api call. Remove it.

Tested by building from a clean tree successfully.

Signed-off-by: Ani Sinha <ani@anisinha.ca>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20210825031949.919376-2-ani@anisinha.ca>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agohw/i386/acpi-build: Get NUMA information from struct NumaState
Jingqi Liu [Mon, 23 Aug 2021 01:12:54 +0000 (09:12 +0800)]
hw/i386/acpi-build: Get NUMA information from struct NumaState

Since commits aa57020774b ("numa: move numa global variable
nb_numa_nodes into MachineState") and 7e721e7b10e ("numa: move
numa global variable numa_info into MachineState"), we can get
NUMA information completely from MachineState::numa_state.

Remove PCMachineState::numa_nodes and PCMachineState::node_mem,
since they are just copied from MachineState::numa_state.

Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Jingqi Liu <jingqi.liu@intel.com>
Message-Id: <20210823011254.28506-1-jingqi.liu@intel.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovhost: correctly detect the enabling IOMMU
Jason Wang [Wed, 4 Aug 2021 03:48:03 +0000 (11:48 +0800)]
vhost: correctly detect the enabling IOMMU

Vhost used to compare the dma_as against the address_space_memory to
detect whether the IOMMU is enabled or not. This might not work well
since the virito-bus may call get_dma_as if VIRTIO_F_IOMMU_PLATFORM is
set without an actual IOMMU enabled when device is plugged. In the
case of PCI where pci_get_address_space() is used, the bus master as
is returned. So vhost actually tries to enable device IOTLB even if
the IOMMU is not enabled. This will lead a lots of unnecessary
transactions between vhost and Qemu and will introduce a huge drop of
the performance.

For PCI, an ideal approach is to use pci_device_iommu_address_space()
just for get_dma_as. But Qemu may choose to initialize the IOMMU after
the virtio-pci which lead a wrong address space is returned during
device plugged. So this patch switch to use transport specific way via
iommu_enabled() to detect the IOMMU during vhost start. In this case,
we are fine since we know the IOMMU is initialized correctly.

Signed-off-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20210804034803.1644-4-jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovirtio-pci: implement iommu_enabled()
Jason Wang [Wed, 4 Aug 2021 03:48:02 +0000 (11:48 +0800)]
virtio-pci: implement iommu_enabled()

This patch implements the PCI transport version of iommu_enabled. This
is done by comparing the address space returned by
pci_device_iommu_address_space() against address_space_memory.

Note that an ideal approach is to use pci_device_iommu_address_space()
in get_dma_as(), but it might not work well since the IOMMU could be
initialized after the virtio-pci device is initialized.

Signed-off-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20210804034803.1644-3-jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovirtio-bus: introduce iommu_enabled()
Jason Wang [Wed, 4 Aug 2021 03:48:01 +0000 (11:48 +0800)]
virtio-bus: introduce iommu_enabled()

This patch introduce a new method for the virtio-bus for the transport
to report whether or not the IOMMU is enabled for the device.

Signed-off-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20210804034803.1644-2-jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovirtio-balloon: free page hinting cleanups
David Hildenbrand [Thu, 8 Jul 2021 09:53:39 +0000 (11:53 +0200)]
virtio-balloon: free page hinting cleanups

Let's compress the code a bit to improve readability. We can drop the
vm_running check in virtio_balloon_free_page_start() as it's already
properly checked in the single caller.

Cc: Wei Wang <wei.w.wang@intel.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Philippe Mathieu-Daudé <philmd@redhat.com>
Cc: Alexander Duyck <alexander.duyck@gmail.com>
Cc: Juan Quintela <quintela@redhat.com>
Cc: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
Cc: Peter Xu <peterx@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210708095339.20274-3-david@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovirtio-balloon: don't start free page hinting if postcopy is possible
David Hildenbrand [Thu, 8 Jul 2021 09:53:38 +0000 (11:53 +0200)]
virtio-balloon: don't start free page hinting if postcopy is possible

Postcopy never worked properly with 'free-page-hint=on', as there are
at least two issues:

1) With postcopy, the guest will never receive a VIRTIO_BALLOON_CMD_ID_DONE
   and consequently won't release free pages back to the OS once
   migration finishes.

   The issue is that for postcopy, we won't do a final bitmap sync while
   the guest is stopped on the source and
   virtio_balloon_free_page_hint_notify() will only call
   virtio_balloon_free_page_done() on the source during
   PRECOPY_NOTIFY_CLEANUP, after the VM state was already migrated to
   the destination.

2) Once the VM touches a page on the destination that has been excluded
   from migration on the source via qemu_guest_free_page_hint() while
   postcopy is active, that thread will stall until postcopy finishes
   and all threads are woken up. (with older Linux kernels that won't
   retry faults when woken up via userfaultfd, we might actually get a
   SEGFAULT)

   The issue is that the source will refuse to migrate any pages that
   are not marked as dirty in the dirty bmap -- for example, because the
   page might just have been sent. Consequently, the faulting thread will
   stall, waiting for the page to be migrated -- which could take quite
   a while and result in guest OS issues.

While we could fix 1) comparatively easily, 2) is harder to get right and
might require more involved RAM migration changes on source and destination
[1].

As it never worked properly, let's not start free page hinting in the
precopy notifier if the postcopy migration capability was enabled to fix
it easily. Capabilities cannot be enabled once migration is already
running.

Note 1: in the future we might either adjust migration code on the source
        to track pages that have actually been sent or adjust
        migration code on source and destination  to eventually send
        pages multiple times from the source and and deal with pages
        that are sent multiple times on the destination.

Note 2: virtio-mem has similar issues, however, access to "unplugged"
        memory by the guest is very rare and we would have to be very
        lucky for it to happen during migration. The spec states
        "The driver SHOULD NOT read from unplugged memory blocks ..."
        and "The driver MUST NOT write to unplugged memory blocks".
        virtio-mem will move away from virtio_balloon_free_page_done()
        soon and handle this case explicitly on the destination.

[1] https://lkml.kernel.org/r/e79fd18c-aa62-c1d8-c7f3-ba3fc2c25fc8@redhat.com

Fixes: c13c4153f76d ("virtio-balloon: VIRTIO_BALLOON_F_FREE_PAGE_HINT")
Cc: qemu-stable@nongnu.org
Cc: Wei Wang <wei.w.wang@intel.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Philippe Mathieu-Daudé <philmd@redhat.com>
Cc: Alexander Duyck <alexander.duyck@gmail.com>
Cc: Juan Quintela <quintela@redhat.com>
Cc: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
Cc: Peter Xu <peterx@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210708095339.20274-2-david@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
2 years agoMerge remote-tracking branch 'remotes/armbru/tags/pull-qapi-2021-09-03' into staging
Peter Maydell [Sat, 4 Sep 2021 18:21:19 +0000 (19:21 +0100)]
Merge remote-tracking branch 'remotes/armbru/tags/pull-qapi-2021-09-03' into staging

QAPI patches patches for 2021-09-03

# gpg: Signature made Fri 03 Sep 2021 16:20:49 BST
# gpg:                using RSA key 354BC8B3D7EB2A6B68674E5F3870B400EB918653
# gpg:                issuer "armbru@redhat.com"
# gpg: Good signature from "Markus Armbruster <armbru@redhat.com>" [full]
# gpg:                 aka "Markus Armbruster <armbru@pond.sub.org>" [full]
# Primary key fingerprint: 354B C8B3 D7EB 2A6B 6867  4E5F 3870 B400 EB91 8653

* remotes/armbru/tags/pull-qapi-2021-09-03:
  qapi: Tweak error messages for unknown / conflicting 'if' keys
  qapi: Tweak error messages for missing / conflicting meta-type
  tests/qapi-schema: Hide OrderedDict in test output
  qapi: Use re.fullmatch() where appropriate
  qapi: Use "not COND" instead of "!COND" for generated documentation
  qapi: Avoid redundant parens in code generated for conditionals
  qapi: Factor common recursion out of cgen_ifcond(), docgen_ifcond()
  qapi: Fix C code generation for 'if'
  tests/qapi-schema: Demonstrate broken C code for 'if'
  tests/qapi-schema: Correct two 'if' conditionals
  qapi: Simplify how QAPISchemaIfCond represents "no condition"
  qapi: Simplify QAPISchemaIfCond's interface for generating C
  qapi: Set boolean value correctly in examples

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
2 years agoUse PCI_HOST_BRIDGE macro
Eduardo Habkost [Thu, 5 Aug 2021 19:34:31 +0000 (15:34 -0400)]
Use PCI_HOST_BRIDGE macro

OBJECT_CHECK(PciHostState, ..., TYPE_PCI_HOST_BRIDGE) is exactly
what the PCI_HOST_BRIDGE macro does.  We can just use the macro
instead of using OBJECT_CHECK manually.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20210805193431.307761-7-ehabkost@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agoacpi: Delete broken ACPI_GED_X86 macro
Eduardo Habkost [Thu, 5 Aug 2021 19:34:26 +0000 (15:34 -0400)]
acpi: Delete broken ACPI_GED_X86 macro

The macro never worked and never will, because the
AcpiGedX86State type never existed.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20210805193431.307761-2-ehabkost@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovhost-user: add missing space in error message
Alyssa Ross [Fri, 6 Aug 2021 14:39:26 +0000 (14:39 +0000)]
vhost-user: add missing space in error message

This would previously give error messages like

> Received unexpected msg type.Expected 0 received 1

Signed-off-by: Alyssa Ross <hi@alyssa.is>
Message-Id: <20210806143926.315725-1-hi@alyssa.is>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agohw/virtio: move vhost_set_backend_type() to vhost.c
Tiberiu Georgescu [Mon, 9 Aug 2021 13:40:15 +0000 (13:40 +0000)]
hw/virtio: move vhost_set_backend_type() to vhost.c

Just a small refactor patch.

vhost_set_backend_type() gets called only in vhost.c, so we can move the
function there and make it static. We can then extern the visibility of
kernel_ops, to match the other VhostOps in vhost-backend.h.
The VhostOps constants now make more sense in vhost.h

Suggested-by: Raphael Norwitz <raphael.norwitz@nutanix.com>
Signed-off-by: Tiberiu Georgescu <tiberiu.georgescu@nutanix.com>
Message-Id: <20210809134015.67941-1-tiberiu.georgescu@nutanix.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agohw/acpi: refactor acpi hp modules so that targets can just use what they need
Ani Sinha [Thu, 12 Aug 2021 07:14:09 +0000 (12:44 +0530)]
hw/acpi: refactor acpi hp modules so that targets can just use what they need

Currently various acpi hotplug modules like cpu hotplug, memory hotplug, pci
hotplug, nvdimm hotplug are all pulled in when CONFIG_ACPI_X86 is turned on.
This brings in support for whole lot of subsystems that some targets like
mips does not need. They are added just to satisfy symbol dependencies. This
is ugly and should be avoided. Targets should be able to pull in just what they
need and no more. For example, mips only needs support for PIIX4 and does not
need acpi pci hotplug support or cpu hotplug support or memory hotplug support
etc. This change is an effort to clean this up.
In this change, new config variables are added for various acpi hotplug
subsystems. Targets like mips can only enable PIIX4 support and not the rest
of all the other modules which were being previously pulled in as a part of
CONFIG_ACPI_X86. Function stubs make sure that symbols which piix4 needs but
are not required by mips (for example, symbols specific to pci hotplug etc)
are available to satisfy the dependencies.

Currently, this change only addresses issues with mips malta targets. In future
we might be able to clean up other targets which are similarly pulling in lot
of unnecessary hotplug modules by enabling ACPI_X86.

This change should also address issues such as the following:
https://gitlab.com/qemu-project/qemu/-/issues/221
https://gitlab.com/qemu-project/qemu/-/issues/193

Signed-off-by: Ani Sinha <ani@anisinha.ca>
Message-Id: <20210812071409.492299-1-ani@anisinha.ca>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agoq35: catch invalid cpu hotplug configuration
Gerd Hoffmann [Thu, 12 Aug 2021 10:23:41 +0000 (12:23 +0200)]
q35: catch invalid cpu hotplug configuration

Related: https://bugzilla.redhat.com//show_bug.cgi?id=1985924
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Message-Id: <20210812102341.3316254-1-kraxel@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agohw/acpi: define PIIX4 acpi pci hotplug property strings at a single place
Ani Sinha [Mon, 16 Aug 2021 08:32:14 +0000 (14:02 +0530)]
hw/acpi: define PIIX4 acpi pci hotplug property strings at a single place

Now that we have "acpi-pci-hotplug-with-bridge-support" PIIX4 PM property being
used for both q35 and i440fx machine types, it is better that we defined this
property string at a single place within a header file like other PIIX4
properties. We can then use this single definition at all the places that needs
it instead of duplicating the string everywhere. While at it, this change also
adds a definition for "acpi-root-pci-hotplug" PIIX4 PM property and uses
this definition at all places that were formally using the string value.

Signed-off-by: Ani Sinha <ani@anisinha.ca>
Message-Id: <20210816083214.105740-1-ani@anisinha.ca>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovhost: make SET_VRING_ADDR, SET_FEATURES send replies
Denis Plotnikov [Mon, 9 Aug 2021 10:48:24 +0000 (13:48 +0300)]
vhost: make SET_VRING_ADDR, SET_FEATURES send replies

On vhost-user-blk migration, qemu normally sends a number of commands
to enable logging if VHOST_USER_PROTOCOL_F_LOG_SHMFD is negotiated.
Qemu sends VHOST_USER_SET_FEATURES to enable buffers logging and
VHOST_USER_SET_VRING_ADDR per each started ring to enable "used ring"
data logging.
The issue is that qemu doesn't wait for reply from the vhost daemon
for these commands which may result in races between qemu expectation
of logging starting and actual login starting in vhost daemon.

The race can appear as follows: on migration setup, qemu enables dirty page
logging by sending VHOST_USER_SET_FEATURES. The command doesn't arrive to a
vhost-user-blk daemon immediately and the daemon needs some time to turn the
logging on internally. If qemu doesn't wait for reply, after sending the
command, qemu may start migrateing memory pages to a destination. At this time,
the logging may not be actually turned on in the daemon but some guest pages,
which the daemon is about to write to, may have already been transferred
without logging to the destination. Since the logging wasn't turned on,
those pages won't be transferred again as dirty. So we may end up with
corrupted data on the destination.
The same scenario is applicable for "used ring" data logging, which is
turned on with VHOST_USER_SET_VRING_ADDR command.

To resolve this issue, this patch makes qemu wait for the command result
explicitly if VHOST_USER_PROTOCOL_F_REPLY_ACK is negotiated and logging enabled.

Signed-off-by: Denis Plotnikov <den-plotnikov@yandex-team.ru>
Message-Id: <20210809104824.78830-1-den-plotnikov@yandex-team.ru>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agohw/virtio: Fix leak of host-notifier memory-region
Yajun Wu [Mon, 16 Aug 2021 01:32:35 +0000 (04:32 +0300)]
hw/virtio: Fix leak of host-notifier memory-region

If call virtio_queue_set_host_notifier_mr fails, should free
host-notifier memory-region.

Fixes: 44866521bd ("vhost-user: support registering external host notifiers")
Signed-off-by: Yajun Wu <yajunw@nvidia.com>
Message-Id: <1629077555-19907-1-git-send-email-yajunw@nvidia.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovhost-vdpa: Do not send empty IOTLB update batches
Eugenio Pérez [Thu, 12 Aug 2021 14:09:33 +0000 (16:09 +0200)]
vhost-vdpa: Do not send empty IOTLB update batches

With the introduction of the batch hinting, meaningless batches can be
created with no IOTLB updates if the memory region was skipped by
vhost_vdpa_listener_skipped_section. This is the case of host notifiers
memory regions, device un/realize, and others. This causes the vdpa
device to receive dma mapping settings with no changes, a possibly
expensive operation for nothing.

To avoid that, VHOST_IOTLB_BATCH_BEGIN hint is delayed until we have a
meaningful (not skipped section) mapping or unmapping operation, and
VHOST_IOTLB_BATCH_END is not written unless at least one of _UPDATE /
_INVALIDATE has been issued.

v3:
  * Use a bool instead of a counter avoiding potential number wrapping
  * Fix bad check on _commit
  * Move VHOST_BACKEND_F_IOTLB_BATCH check to
    vhost_vdpa_iotlb_batch_begin_once

v2 (from RFC):
  * Rename misleading name
  * Abstract start batching function for listener_add/del

Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Message-Id: <20210812140933.226288-1-eperezma@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agoqapi: Tweak error messages for unknown / conflicting 'if' keys
Markus Armbruster [Tue, 31 Aug 2021 12:38:09 +0000 (14:38 +0200)]
qapi: Tweak error messages for unknown / conflicting 'if' keys

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20210831123809.1107782-13-armbru@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
2 years agoqapi: Tweak error messages for missing / conflicting meta-type
Markus Armbruster [Tue, 31 Aug 2021 12:38:08 +0000 (14:38 +0200)]
qapi: Tweak error messages for missing / conflicting meta-type

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20210831123809.1107782-12-armbru@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
2 years agotests/qapi-schema: Hide OrderedDict in test output
Markus Armbruster [Tue, 31 Aug 2021 12:38:07 +0000 (14:38 +0200)]
tests/qapi-schema: Hide OrderedDict in test output

Since commit 5d83b9a130 "qapi: replace if condition list with dict
{'all': [...]}", we represent if conditionals as trees consisting of
OrderedDict, list and str.  This results in less than legible test
output.  For instance:

    if OrderedDict([('not', OrderedDict([('any', [OrderedDict([('not', 'TEST_IF_EVT')]), OrderedDict([('not', 'TEST_IF_STRUCT')])])]))])

We intend to replace OrderedDict by dict when we get Python 3.7, which
will result in more legible output:

    if {'not': {'any': [{'not': 'TEST_IF_EVT'}, {'not': 'TEST_IF_STRUCT'}]}}

Can't wait: put in a hack to get that now, with a comment to revert it
when we replace OrderedDict.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20210831123809.1107782-11-armbru@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>