]> git.maquefel.me Git - qemu.git/log
qemu.git
4 years agoppc/pnv: move PCI registers to PnvPHB4
Daniel Henrique Barboza [Tue, 18 Jan 2022 11:56:30 +0000 (12:56 +0100)]
ppc/pnv: move PCI registers to PnvPHB4

Previous patch changed pnv_pec_stk_pci_xscom_read() and
pnv_pec_stk_pci_xscom_write() to use a PnvPHB4 opaque, making it easier
to move both pci_regs[] and the pci_regs_mr MemoryRegion to the PnvHB4
object.

Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220113192952.911188-3-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
4 years agoppc/pnv: use PHB4 obj in pnv_pec_stk_pci_xscom_ops
Daniel Henrique Barboza [Tue, 18 Jan 2022 11:56:30 +0000 (12:56 +0100)]
ppc/pnv: use PHB4 obj in pnv_pec_stk_pci_xscom_ops

The current relationship between PnvPhb4PecStack and PnvPHB4 objects is
overly complex. Recent work done in pnv_phb4.c and pnv_phb4_pec.c shows
that the stack obj role in the overall design is more of a placeholder for
its 'phb' object, having no atributes that stand on its own. This became
clearer after pnv-phb4 user creatable devices were implemented.

What remains now are a lot of stack->phb and phb->stack pointers
throughout .read and .write callbacks of MemoryRegionOps that are being
initialized in phb4_realize() time. stk_realize() is a no-op if the
machine is being run with -nodefaults.

The first step of trying to decouple the stack and phb relationship is
to move the MemoryRegionOps that belongs to PnvPhb4PecStack to PhbPHB4.
Unfortunately this can't be done  without some preliminary steps to
change the usage of 'stack' and replace it with 'phb' in these
read/write callbacks.

This patch starts this process by using a PnvPHB4 opaque in
pnv_pec_stk_pci_xscom_ops instead of PnvPhb4PecStack.

Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220113192952.911188-2-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
4 years agotests/avocado: ppc: Add smoke tests for MPC7400 and MPC7450 families
Fabiano Rosas [Tue, 18 Jan 2022 11:56:30 +0000 (12:56 +0100)]
tests/avocado: ppc: Add smoke tests for MPC7400 and MPC7450 families

These tests ensure that our emulation for these cpus is not completely
broken and we can at least run OpenBIOS on them.

$ make check-avocado AVOCADO_TESTS=../tests/avocado/ppc_74xx.py

Signed-off-by: Fabiano Rosas <farosas@linux.ibm.com>
Reviewed-by: Willian Rampazzo <willianr@redhat.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220117144757.782441-1-farosas@linux.ibm.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
4 years agotarget/ppc: Fix 7448 support
Cédric Le Goater [Tue, 18 Jan 2022 11:56:30 +0000 (12:56 +0100)]
target/ppc: Fix 7448 support

The 7448 CPU is an evolution of the PowerPC 7447A and the last of the
G4 family. Change its family to reflect correctly its features. This
fixes Linux boot.

Cc: Fabiano Rosas <farosas@linux.ibm.com>
Reviewed-by: Fabiano Rosas <farosas@linux.ibm.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220117092555.1616512-1-clg@kaod.org>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
4 years agotarget/ppc: Finish removal of 401/403 CPUs
Cédric Le Goater [Tue, 18 Jan 2022 11:56:30 +0000 (12:56 +0100)]
target/ppc: Finish removal of 401/403 CPUs

Commit c8f49e6b938e ("target/ppc: remove 401/403 CPUs") left a few
things behind.

Reviewed-by: Fabiano Rosas <farosas@linux.ibm.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Message-Id: <20220117091541.1615807-1-clg@kaod.org>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220118104150.1899661-3-clg@kaod.org>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
4 years agotarget/ppc: Remove last user of .load_state_old
Cédric Le Goater [Tue, 18 Jan 2022 11:56:30 +0000 (12:56 +0100)]
target/ppc: Remove last user of .load_state_old

This breaks migration compatibility from (very) old versions of
QEMU. This should not be a problem for the pseries machine for which
migration is only supported on recent QEMUs ( > 2.x). There is no
clear status on what is supported or not for the other machines. Let's
move forward and remove the .load_state_old handler.

Signed-off-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Message-Id: <20220118104150.1899661-2-clg@kaod.org>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
4 years agoLink new ppc-spapr-uv-hcalls.rst to pseries.rst.
Leonardo Garcia [Tue, 18 Jan 2022 11:56:30 +0000 (12:56 +0100)]
Link new ppc-spapr-uv-hcalls.rst to pseries.rst.

Signed-off-by: Leonardo Garcia <lagarcia@br.ibm.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <a24d2c260442e44aa99d4c18b1717f2f4a73529a.1642446876.git.lagarcia@br.ibm.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
4 years agoRename ppc-spapr-uv-hcalls.txt to ppc-spapr-uv-hcalls.rst.
Leonardo Garcia [Tue, 18 Jan 2022 11:56:30 +0000 (12:56 +0100)]
Rename ppc-spapr-uv-hcalls.txt to ppc-spapr-uv-hcalls.rst.

Signed-off-by: Leonardo Garcia <lagarcia@br.ibm.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <ea1554e3a02e093eaa64d1260faf01d525c9ecdc.1642446876.git.lagarcia@br.ibm.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
4 years agorSTify ppc-spapr-uv-hcalls.txt.
Leonardo Garcia [Tue, 18 Jan 2022 11:56:30 +0000 (12:56 +0100)]
rSTify ppc-spapr-uv-hcalls.txt.

Signed-off-by: Leonardo Garcia <lagarcia@br.ibm.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <243a714d3861f7539d29b02a899ffc376757d668.1642446876.git.lagarcia@br.ibm.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
4 years agoLink new ppc-spapr-hotplug.rst file to pseries.rst.
Leonardo Garcia [Tue, 18 Jan 2022 11:56:30 +0000 (12:56 +0100)]
Link new ppc-spapr-hotplug.rst file to pseries.rst.

Signed-off-by: Leonardo Garcia <lagarcia@br.ibm.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <e54aad58e3cfbd96647430d33cdb59fd87d199b0.1641995058.git.lagarcia@br.ibm.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
4 years agodocs: Rename ppc-spapr-hotplug.txt to ppc-spapr-hotplug.rst.
Leonardo Garcia [Tue, 18 Jan 2022 11:56:30 +0000 (12:56 +0100)]
docs: Rename ppc-spapr-hotplug.txt to ppc-spapr-hotplug.rst.

Signed-off-by: Leonardo Garcia <lagarcia@br.ibm.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <1f5860217273f272fddadc68b5d205b4090f6b04.1641995058.git.lagarcia@br.ibm.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
4 years agodocs: rSTify ppc-spapr-hotplug.txt.
Leonardo Garcia [Tue, 18 Jan 2022 11:56:30 +0000 (12:56 +0100)]
docs: rSTify ppc-spapr-hotplug.txt.

While working on this file, also removed and unused reference in the end of the file. The reference in the text was removed by commit 9f992cca93d (spapr: update spapr hotplug documentation), but the link in the end of the document was not removed then.

Signed-off-by: Leonardo Garcia <lagarcia@br.ibm.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <50ed30232e0e6eafb580c17adec3fba17b873014.1641995058.git.lagarcia@br.ibm.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
4 years agodocs/devel: add some clarifying text for aliases
Alex Bennée [Mon, 10 Jan 2022 17:51:04 +0000 (17:51 +0000)]
docs/devel: add some clarifying text for aliases

We do mention the limitation of single parenthood for
memory_region_add_subregion but lets also make it clear how aliases
help solve that conundrum.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220110175104.2908956-7-alex.bennee@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
4 years agohw/dma: Let dma_buf_read() / dma_buf_write() propagate MemTxResult
Philippe Mathieu-Daudé [Thu, 16 Dec 2021 08:36:38 +0000 (09:36 +0100)]
hw/dma: Let dma_buf_read() / dma_buf_write() propagate MemTxResult

Since commit 292e13142d2, dma_buf_rw() returns a MemTxResult type.
Do not discard it, return it to the caller. Pass the previously
returned value (the QEMUSGList residual size, which was rarely used)
as an optional argument.

With this new API, SCSIRequest::residual might now be accessed via
a pointer. Since the size_t type does not have the same size on
32 and 64-bit host architectures, convert it to a uint64_t, which
is big enough to hold the residual size, and the type is constant
on both 32/64-bit hosts.

Update the few dma_buf_read() / dma_buf_write() callers to the new
API.

Reviewed-by: Klaus Jensen <k.jensen@samsung.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Acked-by: Peter Xu <peterx@redhat.com>
Message-Id: <20220117125130.131828-1-f4bug@amsat.org>

4 years agohw/dma: Use dma_addr_t type definition when relevant
Philippe Mathieu-Daudé [Fri, 31 Dec 2021 10:33:29 +0000 (11:33 +0100)]
hw/dma: Use dma_addr_t type definition when relevant

Update the obvious places where dma_addr_t should be used
(instead of uint64_t, hwaddr, size_t, int32_t types).

This allows to have &dma_addr_t type portable on 32/64-bit
hosts.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220111184309.28637-11-f4bug@amsat.org>

4 years agohw/dma: Move ScatterGatherEntry / QEMUSGList declarations around
Philippe Mathieu-Daudé [Tue, 11 Jan 2022 10:32:22 +0000 (11:32 +0100)]
hw/dma: Move ScatterGatherEntry / QEMUSGList declarations around

In the next commit we will use the dma_addr_t type in the QEMUSGList
structure. Since currently dma_addr_t is defined after QEMUSGList,
move the declarations to have dma_addr_t defined first. This is a
pure code-movement patch.

Suggested-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Message-Id: <20220111184309.28637-10-f4bug@amsat.org>

4 years agohw/dma: Fix format string issues using dma_addr_t
Philippe Mathieu-Daudé [Tue, 4 Jan 2022 08:42:21 +0000 (09:42 +0100)]
hw/dma: Fix format string issues using dma_addr_t

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220111184309.28637-10-f4bug@amsat.org>

4 years agohw/scsi: Rename SCSIRequest::resid as 'residual'
Philippe Mathieu-Daudé [Fri, 31 Dec 2021 10:13:34 +0000 (11:13 +0100)]
hw/scsi: Rename SCSIRequest::resid as 'residual'

The 'resid' field is slightly confusing and could be
interpreted as some ID. Rename it as 'residual' which
is clearer to review. No logical change.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Message-Id: <20220111184309.28637-8-f4bug@amsat.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
4 years agohw/rdma/rdma_utils: Rename rdma_pci_dma_map 'len' argument
Philippe Mathieu-Daudé [Fri, 31 Dec 2021 10:19:08 +0000 (11:19 +0100)]
hw/rdma/rdma_utils: Rename rdma_pci_dma_map 'len' argument

Various APIs use 'pval' naming for 'pointer to val'.
rdma_pci_dma_map() uses 'plen' for 'PCI length', but since
'PCI' is already explicit in the function name, simplify
and rename the argument 'len'. No logical change.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Yuval Shaia <yuval.shaia.ml@gmail.com>
Tested-by: Yuval Shaia <yuval.shaia.ml@gmail.com>
Message-Id: <20220111184309.28637-7-f4bug@amsat.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
4 years agolinux-user: Remove MAX_SIGQUEUE_SIZE
Warner Losh [Sun, 16 Jan 2022 20:44:22 +0000 (13:44 -0700)]
linux-user: Remove MAX_SIGQUEUE_SIZE

It's been unused for 7 years since 907f5fddaa67 when linux-user stopped
queueing any signals.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220116204423.16133-2-imp@bsdimp.com>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
4 years agolinux-user: Return void from queue_signal()
Peter Maydell [Fri, 14 Jan 2022 15:37:32 +0000 (15:37 +0000)]
linux-user: Return void from queue_signal()

The linux-user queue_signal() function always returns 1, and none of
its callers check the return value.  Give it a void return type
instead.

The return value is a leftover from the old pre-2016 linux-user
signal handling code, which really did have a queue of signals and so
might return a failure indication if too many signals were queued at
once.  The current design avoids having to ever have more than one
signal queued via queue_signal() at once, so it can never fail.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220114153732.3767229-4-peter.maydell@linaro.org>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
4 years agolinux-user: Rename user_force_sig tracepoint to match function name
Peter Maydell [Fri, 14 Jan 2022 15:37:31 +0000 (15:37 +0000)]
linux-user: Rename user_force_sig tracepoint to match function name

In commit c599d4d6d6e9bfdb64 in 2016 we renamed the old force_sig()
function to dump_core_and_abort(), but we forgot to rename the
associated tracepoint.  Rename the tracepoint to to match the
function it's called from.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220114153732.3767229-3-peter.maydell@linaro.org>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
4 years agolinux-user: Fix comment typo in arm cpu_loop code
Peter Maydell [Fri, 14 Jan 2022 18:25:35 +0000 (18:25 +0000)]
linux-user: Fix comment typo in arm cpu_loop code

Fix a typo in a comment in the arm cpu_loop code.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Warner Losh <imp@bsdimp.com>
Message-Id: <20220114182535.3804783-1-peter.maydell@linaro.org>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
4 years agosoftmmu: Provide a clue as to why device tree loading failed
Bernhard Beschow [Sun, 16 Jan 2022 11:46:49 +0000 (12:46 +0100)]
softmmu: Provide a clue as to why device tree loading failed

fdt_open_into() obligingly returns an error code in case the operation
failed. So be obliging as well and use it in the error message.

Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220116114649.40859-1-shentey@gmail.com>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
4 years agotests: Fix typo in check-help output
Philippe Mathieu-Daudé [Tue, 11 Jan 2022 17:55:28 +0000 (18:55 +0100)]
tests: Fix typo in check-help output

Fix typo in 'make check-help' output.

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220111175528.22294-1-f4bug@amsat.org>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
4 years agoqdev-core.h: Fix wrongly named reference to TYPE_SPLIT_IRQ
Peter Maydell [Tue, 11 Jan 2022 17:26:55 +0000 (17:26 +0000)]
qdev-core.h: Fix wrongly named reference to TYPE_SPLIT_IRQ

Fix a comment in qdev-core.h where we incorrectly referred
to TYPE_IRQ_SPLIT when we meant TYPE_SPLIT_IRQ.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20220111172655.3546766-1-peter.maydell@linaro.org>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
4 years agohw/scsi/megasas: Simplify using the ldst API
Philippe Mathieu-Daudé [Sat, 18 Dec 2021 11:19:12 +0000 (12:19 +0100)]
hw/scsi/megasas: Simplify using the ldst API

This code is easier to review using the load/store API.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20211218111912.1499377-1-philmd@redhat.com>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
4 years agohw/timer/etraxfs_timer: Add vmstate for ETRAX timers
Philippe Mathieu-Daudé [Sat, 6 Nov 2021 10:56:23 +0000 (11:56 +0100)]
hw/timer/etraxfs_timer: Add vmstate for ETRAX timers

Add the vmstate for the ETRAX timers.
This is in theory a migration compatibility break
for the 'AXIS devboard 88' CRIS machine.

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20211106105623.510868-1-f4bug@amsat.org>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
4 years agoseabios: update binaries to snapshot
Gerd Hoffmann [Tue, 18 Jan 2022 11:15:49 +0000 (12:15 +0100)]
seabios: update binaries to snapshot

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agoseabios: update submodule to snapshot
Gerd Hoffmann [Tue, 18 Jan 2022 10:56:16 +0000 (11:56 +0100)]
seabios: update submodule to snapshot

Most important update is smbios3 support being added, this update should
help getting the qemu side of things (flip to smbios3 by default for new
machine types) merged and tested.

Not fully clear yet when the next 1.16 seabios version will be released,
but it should be no later than end of February so we can pick it up for
qemu 7.0.

full shortlog
=============

Andy Pei (3):
      virtio-blk: add feature VIRTIO_BLK_F_SIZE_MAX and VIRTIO_BLK_F_SEG_MAX
      virtio-blk: abstract a function named virtio_blk_op_one_segment to handle r/w request
      virtio-blk: split large IO according to size_max

Eduardo Habkost (19):
      biostables: copy_fseg_table() function
      util.h: Delete unused get_smbios_entry_point() prototype
      smbios: Rename code specific for SMBIOS 2.1 entry points
      smbios: Generic smbios_next() function
      smbios: smbios_get_tables() function
      smbios: Use smbios_get_tables()/smbios_next() at display_uuid()
      smbios: smbios_major_version()/smbios_minor_version() helpers
      tpm: Use smbios_get_tables()
      csm: Don't check SMBios21Addr before calling copy_smbios_21()
      smbios: Make SMBios21Addr variable static
      smbios: Use smbios_next() at smbios_romfile_setup()
      smbios: Extract SMBIOS table building code to separate function
      smbios: Make smbios_build_tables() more generic
      smbios: smbios_21_setup_entry_point() function
      smbios: Make some smbios_build_tables() arguments optional
      smbios: Make smbios_build_tables() ready for 64-bit tables
      smbios: copy_smbios_30() function
      smbios: Support SMBIOS 3.0 entry point at copy_table()
      smbios: Support SMBIOS 3.0 entry point at smbios_romfile_setup()

Gerd Hoffmann (1):
      svgamodes: add standard 4k modes

Igor Mammedov (2):
      pci: reserve resources for pcie-pci-bridge to fix regressed hotplug on q35
      pci: let firmware reserve IO for pcie-pci-bridge

Kevin O'Connor (4):
      vgasrc: Don't use VAR16 in header files to fix gcc warning
      memmap: Fix gcc out-of-bounds warning
      readserial: Improve Python3 compatibility
      scripts: Remove python23compat.py

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agohw/dma: Remove CONFIG_USER_ONLY check
Philippe Mathieu-Daudé [Fri, 31 Dec 2021 10:18:15 +0000 (11:18 +0100)]
hw/dma: Remove CONFIG_USER_ONLY check

DMA API should not be included in user-mode emulation.
If so, build should fail. Remove the CONFIG_USER_ONLY check.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Message-Id: <20220111184309.28637-6-f4bug@amsat.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
4 years agohw/pci: Document pci_dma_map()
Philippe Mathieu-Daudé [Fri, 31 Dec 2021 10:51:52 +0000 (11:51 +0100)]
hw/pci: Document pci_dma_map()

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220111184309.28637-5-f4bug@amsat.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
4 years agohw/pci: Restrict pci-bus stub to sysemu
Philippe Mathieu-Daudé [Fri, 31 Dec 2021 10:10:27 +0000 (11:10 +0100)]
hw/pci: Restrict pci-bus stub to sysemu

Neither tools nor user-mode emulation require the PCI bus stub.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Message-Id: <20220111184309.28637-4-f4bug@amsat.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
4 years agohw/nvram: Restrict fw_cfg QOM interface to sysemu and tools
Philippe Mathieu-Daudé [Fri, 31 Dec 2021 11:13:57 +0000 (12:13 +0100)]
hw/nvram: Restrict fw_cfg QOM interface to sysemu and tools

fw_cfg QOM interface is required by system emulation and
qemu-storage-daemon. User-mode emulation doesn't need it.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Message-Id: <20220111184309.28637-3-f4bug@amsat.org>

4 years agostubs: Restrict fw_cfg to system emulation
Philippe Mathieu-Daudé [Tue, 11 Jan 2022 10:44:18 +0000 (11:44 +0100)]
stubs: Restrict fw_cfg to system emulation

fw_cfg_arch_key_name() stub is only required for sysemu.

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Message-Id: <20220111184309.28637-2-f4bug@amsat.org>

4 years agomemory: Fix incorrect calls of log_global_start/stop
Peter Xu [Tue, 30 Nov 2021 08:00:28 +0000 (16:00 +0800)]
memory: Fix incorrect calls of log_global_start/stop

We should only call the log_global_start/stop when the global dirty track
bitmask changes from zero<->non-zero.

No real issue reported for this yet probably because no immediate user to
enable both dirty rate measurement and migration at the same time.  However
it'll be good to be prepared for it.

Fixes: 63b41db4bc ("memory: make global_dirty_tracking a bitmask")
Cc: qemu-stable@nongnu.org
Cc: Hyman Huang <huangy81@chinatelecom.cn>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Juan Quintela <quintela@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Message-Id: <20211130080028.6474-1-peterx@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
4 years agomemory: Update description of memory_region_is_mapped()
David Hildenbrand [Tue, 2 Nov 2021 16:43:17 +0000 (17:43 +0100)]
memory: Update description of memory_region_is_mapped()

Let's update the documentation, making it clearer what the semantics
of memory_region_is_mapped() actually are.

Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20211102164317.45658-4-david@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
4 years agomemory: Make memory_region_is_mapped() succeed when mapped via an alias
David Hildenbrand [Tue, 2 Nov 2021 16:43:16 +0000 (17:43 +0100)]
memory: Make memory_region_is_mapped() succeed when mapped via an alias

memory_region_is_mapped() currently does not return "true" when a memory
region is mapped via an alias.

Assuming we have:
    alias (A0) -> alias (A1) -> region (R0)
Mapping A0 would currently only make memory_region_is_mapped() succeed
on A0, but not on A1 and R0.

Let's fix that by adding a "mapped_via_alias" counter to memory regions and
updating it accordingly when an alias gets (un)mapped.

I am not aware of actual issues, this is rather a cleanup to make it
consistent.

Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20211102164317.45658-3-david@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
4 years agomachine: Use host_memory_backend_is_mapped() in machine_consume_memdev()
David Hildenbrand [Tue, 2 Nov 2021 16:43:15 +0000 (17:43 +0100)]
machine: Use host_memory_backend_is_mapped() in machine_consume_memdev()

memory_region_is_mapped() is the wrong check, we actually want to check
whether the backend is already marked mapped.

For example, memory regions mapped via an alias, such as NVDIMMs,
currently don't make memory_region_is_mapped() return "true". As the
machine is initialized before any memory devices (and thereby before
NVDIMMs are initialized), this isn't a fix but merely a cleanup.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20211102164317.45658-2-david@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
4 years agomemory: Have 'info mtree' remove duplicated Address Space information
Philippe Mathieu-Daudé [Fri, 20 Aug 2021 10:34:14 +0000 (12:34 +0200)]
memory: Have 'info mtree' remove duplicated Address Space information

Per Peter Maydell [*]:

  'info mtree' monitor command was designed on the assumption that
  there's really only one or two interesting address spaces, and
  with more recent developments that's just not the case any more.

Similarly about how the FlatView are sorted using a GHashTable,
sort the AddressSpace objects to remove the duplications (AS
using the same root MemoryRegion).

This drastically reduces the output of 'info mtree' on some boards.

Before:

  $ (echo info mtree; echo q) \
    | qemu-system-aarch64 -S -monitor stdio -M raspi3b \
    | wc -l
  423

After:

  $ (echo info mtree; echo q) \
    | qemu-system-aarch64 -S -monitor stdio -M raspi3b \
    | wc -l
  106

  (qemu) info mtree
  address-space: I/O
    0000000000000000-000000000000ffff (prio 0, i/o): io

  address-space: cpu-memory-0
  address-space: cpu-memory-1
  address-space: cpu-memory-2
  address-space: cpu-memory-3
  address-space: cpu-secure-memory-0
  address-space: cpu-secure-memory-1
  address-space: cpu-secure-memory-2
  address-space: cpu-secure-memory-3
  address-space: memory
    0000000000000000-ffffffffffffffff (prio 0, i/o): system
      0000000000000000-000000003fffffff (prio 0, ram): ram
      000000003f000000-000000003fffffff (prio 1, i/o): bcm2835-peripherals
        000000003f003000-000000003f00301f (prio 0, i/o): bcm2835-sys-timer
        000000003f004000-000000003f004fff (prio -1000, i/o): bcm2835-txp
        000000003f006000-000000003f006fff (prio 0, i/o): mphi
        000000003f007000-000000003f007fff (prio 0, i/o): bcm2835-dma
        000000003f00b200-000000003f00b3ff (prio 0, i/o): bcm2835-ic
        000000003f00b400-000000003f00b43f (prio -1000, i/o): bcm2835-sp804
        000000003f00b800-000000003f00bbff (prio 0, i/o): bcm2835-mbox
        000000003f100000-000000003f1001ff (prio 0, i/o): bcm2835-powermgt
        000000003f101000-000000003f102fff (prio 0, i/o): bcm2835-cprman
        000000003f104000-000000003f10400f (prio 0, i/o): bcm2835-rng
        000000003f200000-000000003f200fff (prio 0, i/o): bcm2835_gpio
        000000003f201000-000000003f201fff (prio 0, i/o): pl011
        000000003f202000-000000003f202fff (prio 0, i/o): bcm2835-sdhost
        000000003f203000-000000003f2030ff (prio -1000, i/o): bcm2835-i2s
        000000003f204000-000000003f20401f (prio -1000, i/o): bcm2835-spi0
        000000003f205000-000000003f20501f (prio -1000, i/o): bcm2835-i2c0
        000000003f20f000-000000003f20f07f (prio -1000, i/o): bcm2835-otp
        000000003f212000-000000003f212007 (prio 0, i/o): bcm2835-thermal
        000000003f214000-000000003f2140ff (prio -1000, i/o): bcm2835-spis
        000000003f215000-000000003f2150ff (prio 0, i/o): bcm2835-aux
        000000003f300000-000000003f3000ff (prio 0, i/o): sdhci
        000000003f600000-000000003f6000ff (prio -1000, i/o): bcm2835-smi
        000000003f804000-000000003f80401f (prio -1000, i/o): bcm2835-i2c1
        000000003f805000-000000003f80501f (prio -1000, i/o): bcm2835-i2c2
        000000003f900000-000000003f907fff (prio -1000, i/o): bcm2835-dbus
        000000003f910000-000000003f917fff (prio -1000, i/o): bcm2835-ave0
        000000003f980000-000000003f990fff (prio 0, i/o): dwc2
          000000003f980000-000000003f980fff (prio 0, i/o): dwc2-io
          000000003f981000-000000003f990fff (prio 0, i/o): dwc2-fifo
        000000003fc00000-000000003fc00fff (prio -1000, i/o): bcm2835-v3d
        000000003fe00000-000000003fe000ff (prio -1000, i/o): bcm2835-sdramc
        000000003fe05000-000000003fe050ff (prio 0, i/o): bcm2835-dma-chan15
      0000000040000000-00000000400000ff (prio 0, i/o): bcm2836-control

  address-space: bcm2835-dma-memory
  address-space: bcm2835-fb-memory
  address-space: bcm2835-property-memory
  address-space: dwc2
    0000000000000000-00000000ffffffff (prio 0, i/o): bcm2835-gpu
      0000000000000000-000000003fffffff (prio 0, ram): alias bcm2835-gpu-ram-alias[*] @ram 0000000000000000-000000003fffffff
      0000000040000000-000000007fffffff (prio 0, ram): alias bcm2835-gpu-ram-alias[*] @ram 0000000000000000-000000003fffffff
      000000007e000000-000000007effffff (prio 1, i/o): alias bcm2835-peripherals @bcm2835-peripherals 0000000000000000-0000000000ffffff
      0000000080000000-00000000bfffffff (prio 0, ram): alias bcm2835-gpu-ram-alias[*] @ram 0000000000000000-000000003fffffff
      00000000c0000000-00000000ffffffff (prio 0, ram): alias bcm2835-gpu-ram-alias[*] @ram 0000000000000000-000000003fffffff

  address-space: bcm2835-mbox-memory
    0000000000000000-000000000000008f (prio 0, i/o): bcm2835-mbox
      0000000000000010-000000000000001f (prio 0, i/o): bcm2835-fb
      0000000000000080-000000000000008f (prio 0, i/o): bcm2835-property

  memory-region: ram
    0000000000000000-000000003fffffff (prio 0, ram): ram

  memory-region: bcm2835-peripherals
    000000003f000000-000000003fffffff (prio 1, i/o): bcm2835-peripherals
      000000003f003000-000000003f00301f (prio 0, i/o): bcm2835-sys-timer
      000000003f004000-000000003f004fff (prio -1000, i/o): bcm2835-txp
      000000003f006000-000000003f006fff (prio 0, i/o): mphi
      000000003f007000-000000003f007fff (prio 0, i/o): bcm2835-dma
      000000003f00b200-000000003f00b3ff (prio 0, i/o): bcm2835-ic
      000000003f00b400-000000003f00b43f (prio -1000, i/o): bcm2835-sp804
      000000003f00b800-000000003f00bbff (prio 0, i/o): bcm2835-mbox
      000000003f100000-000000003f1001ff (prio 0, i/o): bcm2835-powermgt
      000000003f101000-000000003f102fff (prio 0, i/o): bcm2835-cprman
      000000003f104000-000000003f10400f (prio 0, i/o): bcm2835-rng
      000000003f200000-000000003f200fff (prio 0, i/o): bcm2835_gpio
      000000003f201000-000000003f201fff (prio 0, i/o): pl011
      000000003f202000-000000003f202fff (prio 0, i/o): bcm2835-sdhost
      000000003f203000-000000003f2030ff (prio -1000, i/o): bcm2835-i2s
      000000003f204000-000000003f20401f (prio -1000, i/o): bcm2835-spi0
      000000003f205000-000000003f20501f (prio -1000, i/o): bcm2835-i2c0
      000000003f20f000-000000003f20f07f (prio -1000, i/o): bcm2835-otp
      000000003f212000-000000003f212007 (prio 0, i/o): bcm2835-thermal
      000000003f214000-000000003f2140ff (prio -1000, i/o): bcm2835-spis
      000000003f215000-000000003f2150ff (prio 0, i/o): bcm2835-aux
      000000003f300000-000000003f3000ff (prio 0, i/o): sdhci
      000000003f600000-000000003f6000ff (prio -1000, i/o): bcm2835-smi
      000000003f804000-000000003f80401f (prio -1000, i/o): bcm2835-i2c1
      000000003f805000-000000003f80501f (prio -1000, i/o): bcm2835-i2c2
      000000003f900000-000000003f907fff (prio -1000, i/o): bcm2835-dbus
      000000003f910000-000000003f917fff (prio -1000, i/o): bcm2835-ave0
      000000003f980000-000000003f990fff (prio 0, i/o): dwc2
        000000003f980000-000000003f980fff (prio 0, i/o): dwc2-io
        000000003f981000-000000003f990fff (prio 0, i/o): dwc2-fifo
      000000003fc00000-000000003fc00fff (prio -1000, i/o): bcm2835-v3d
      000000003fe00000-000000003fe000ff (prio -1000, i/o): bcm2835-sdramc
      000000003fe05000-000000003fe050ff (prio 0, i/o): bcm2835-dma-chan15

  (qemu) q

[*] https://www.mail-archive.com/qemu-devel@nongnu.org/msg829821.html

Suggested-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20210904231101.1071929-2-philmd@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
4 years agomemory: Split mtree_info() as mtree_info_flatview() + mtree_info_as()
Philippe Mathieu-Daudé [Wed, 1 Sep 2021 15:45:48 +0000 (17:45 +0200)]
memory: Split mtree_info() as mtree_info_flatview() + mtree_info_as()

While mtree_info() handles both ASes and flatviews cases,
the two cases share basically no code. Split mtree_info()
as mtree_info_flatview() + mtree_info_as() to simplify.

Suggested-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20210904231101.1071929-2-philmd@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
4 years agomemory: Directly dispatch alias accesses on origin memory region
Philippe Mathieu-Daudé [Sun, 16 Aug 2020 17:07:11 +0000 (19:07 +0200)]
memory: Directly dispatch alias accesses on origin memory region

Since commit 2cdfcf272d ("memory: assign MemoryRegionOps to all
regions"), all newly created regions are assigned with
unassigned_mem_ops (which might be then overwritten).

When using aliased container regions, and there is no region mapped
at address 0 in the container, the memory_region_dispatch_read()
and memory_region_dispatch_write() calls incorrectly return the
container unassigned_mem_ops, because the alias offset is not used.

Consider the following setup:

    +--------------------+ < - - - - - - - - - - - +
    |     Container      |  mr
    |  (unassigned_mem)  |                         |
    |                    |
    |                    |                         |
    |                    |  alias_offset
    +                    + <- - - - - - +----------+---------+
    | +----------------+ |              |                    |
    | |  MemoryRegion0 | |              |                    |
    | +----------------+ |              |       Alias        |  addr1
    | |  MemoryRegion1 | | <~ ~  ~  ~ ~ |                    | <~~~~~~
    | +----------------+ |              |                    |
    |                    |              +--------------------+
    |                    |
    |                    |
    |                    |
    |                    |
    | +----------------+ |
    | |  MemoryRegionX | |
    | +----------------+ |
    | |  MemoryRegionY | |
    | +----------------+ |
    | |  MemoryRegionZ | |
    | +----------------+ |
    +--------------------+

The memory_region_init_alias() flow is:

  memory_region_init_alias()
  -> memory_region_init()
     -> object_initialize(TYPE_MEMORY_REGION)
        -> memory_region_initfn()
           -> mr->ops = &unassigned_mem_ops;

Later when accessing offset=addr1 via the alias, we expect to hit
MemoryRegion1. The memory_region_dispatch_read() flow is:

  memory_region_dispatch_read(addr1)
  -> memory_region_access_valid(mr)   <- addr1 offset is ignored
     -> mr->ops->valid.accepts()
        -> unassigned_mem_accepts()
        <- false
     <- false
   <- MEMTX_DECODE_ERROR

The caller gets a MEMTX_DECODE_ERROR while the access is OK.

Fix by dispatching aliases recursively, accessing its origin region
after adding the alias offset.

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Peter Xu <peterx@redhat.com>
Message-Id: <20210418055708.820980-1-f4bug@amsat.org>

4 years agoMAINTAINERS: Add myself to s390 I/O areas
Eric Farman [Wed, 12 Jan 2022 16:40:44 +0000 (17:40 +0100)]
MAINTAINERS: Add myself to s390 I/O areas

After the recent restructuring, I'd like to volunteer to help
in some of the s390 I/O areas.

Built on "[PATCH RFC v2] MAINTAINERS: split out s390x sections"

Signed-off-by: Eric Farman <farman@linux.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@linux.ibm.com>
Acked-by: Halil Pasic <pasic@linux.ibm.com>
Message-Id: <20220112164044.2210508-1-farman@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
4 years agoMAINTAINERS: split out s390x sections
Cornelia Huck [Wed, 22 Dec 2021 10:55:48 +0000 (11:55 +0100)]
MAINTAINERS: split out s390x sections

Split out some more specialized devices etc., so that we can build
smarter lists of people to be put on cc: in the future.

Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Acked-by: David Hildenbrand <david@redhat.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Acked-by: Thomas Huth <thuth@redhat.com>
Acked-by: Halil Pasic <pasic@linux.ibm.com>
Acked-by: Eric Farman <farman@linux.ibm.com>
Message-Id: <20211222105548.356852-1-cohuck@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
4 years agotests/tcg/s390x: Test shift instructions
Ilya Leoshkevich [Wed, 12 Jan 2022 16:50:16 +0000 (17:50 +0100)]
tests/tcg/s390x: Test shift instructions

Add a test for each shift instruction in order to to prevent
regressions.

Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Acked-by: David Hildenbrand <david@redhat.com>
Message-Id: <20220112165016.226996-6-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
4 years agotarget/s390x: Fix shifting 32-bit values for more than 31 bits
Ilya Leoshkevich [Wed, 12 Jan 2022 16:50:15 +0000 (17:50 +0100)]
target/s390x: Fix shifting 32-bit values for more than 31 bits

According to PoP, both 32- and 64-bit shifts use lowest 6 address
bits. The current code special-cases 32-bit shifts to use only 5 bits,
which is not correct. For example, shifting by 32 bits currently
preserves the initial value, however, it's supposed zero it out
instead.

Fix by merging sh32 and sh64 and adapting CC calculation to shift
values greater than 31.

Fixes: cbe24bfa91d2 ("target-s390: Convert SHIFT, ROTATE SINGLE")
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Message-Id: <20220112165016.226996-5-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
4 years agotarget/s390x: Fix cc_calc_sla_64() missing overflows
Ilya Leoshkevich [Wed, 12 Jan 2022 16:50:14 +0000 (17:50 +0100)]
target/s390x: Fix cc_calc_sla_64() missing overflows

An overflow occurs for SLAG when at least one shifted bit is not equal
to sign bit. Therefore, we need to check that `shift + 1` bits are
neither all 0s nor all 1s. The current code checks only `shift` bits,
missing some overflows.

Fixes: cbe24bfa91d2 ("target-s390: Convert SHIFT, ROTATE SINGLE")
Co-developed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Message-Id: <20220112165016.226996-4-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
4 years agotarget/s390x: Fix SRDA CC calculation
Ilya Leoshkevich [Wed, 12 Jan 2022 16:50:13 +0000 (17:50 +0100)]
target/s390x: Fix SRDA CC calculation

SRDA uses r1_D32 for binding the first operand and s64 for setting CC.
cout_s64() relies on o->out being the shift result, however,
wout_r1_D32() clobbers it.

Fix by using a temporary.

Fixes: a79ba3398a0a ("target-s390: Convert SHIFT DOUBLE")
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Message-Id: <20220112165016.226996-3-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
4 years agotarget/s390x: Fix SLDA sign bit index
Ilya Leoshkevich [Wed, 12 Jan 2022 16:50:12 +0000 (17:50 +0100)]
target/s390x: Fix SLDA sign bit index

SLDA operates on 64-bit values, so its sign bit index should be 63,
not 31.

Fixes: a79ba3398a0a ("target-s390: Convert SHIFT DOUBLE")
Reported-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Message-Id: <20220112165016.226996-2-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
4 years agos390x/css: fix PMCW invalid mask
Nico Boehr [Thu, 16 Dec 2021 13:16:57 +0000 (14:16 +0100)]
s390x/css: fix PMCW invalid mask

Previously, we required bits 5, 6 and 7 to be zero (0x07 == 0b111). But,
as per the principles of operation, bit 5 is ignored in MSCH and bits 0,
1, 6 and 7 need to be zero.

As both PMCW_FLAGS_MASK_INVALID and ioinst_schib_valid() are only used
by ioinst_handle_msch(), adjust the mask accordingly.

Fixes: db1c8f53bfb1 ("s390: Channel I/O basic definitions.")
Signed-off-by: Nico Boehr <nrb@linux.ibm.com>
Reviewed-by: Pierre Morel <pmorel@linux.ibm.com>
Reviewed-by: Halil Pasic <pasic@linux.ibm.com>
Reviewed-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Message-Id: <20211216131657.1057978-1-nrb@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
4 years agoMerge remote-tracking branch 'remotes/mcayland/tags/qemu-openbios-20220115' into...
Peter Maydell [Sun, 16 Jan 2022 20:12:23 +0000 (20:12 +0000)]
Merge remote-tracking branch 'remotes/mcayland/tags/qemu-openbios-20220115' into staging

qemu-openbios queue

# gpg: Signature made Sat 15 Jan 2022 17:35:58 GMT
# 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-20220115:
  roms/openbios: update OpenBIOS images to 04dfc98 built from submodule

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
4 years agoMerge remote-tracking branch 'remotes/konstantin/tags/qga-win32-pull-2022-01-10'...
Peter Maydell [Sun, 16 Jan 2022 16:32:34 +0000 (16:32 +0000)]
Merge remote-tracking branch 'remotes/konstantin/tags/qga-win32-pull-2022-01-10' into staging

[PULL 0/9] qemu-ga-win patches

# gpg: Signature made Sat 15 Jan 2022 22:04:01 GMT
# gpg:                using RSA key C2C2C109EA43C63C1423EB84EF5D5E8161BA84E7
# gpg: Good signature from "Kostiantyn Kostiuk (Upstream PR sign) <kkostiuk@redhat.com>" [unknown]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: C2C2 C109 EA43 C63C 1423  EB84 EF5D 5E81 61BA 84E7

* remotes/konstantin/tags/qga-win32-pull-2022-01-10:
  qga-win: Detect Windows 11 by build number
  qga-win: Detect OS based on Windows 10 by first build number
  gqa-win: get_pci_info: Replace 'while' with 2 calls of the function
  gqa-win: get_pci_info: Add g_autofree for few variables
  gqa-win: get_pci_info: Split logic to separate functions
  gqa-win: get_pci_info: Free parent_dev_info properly
  gqa-win: get_pci_info: Use common 'end' label
  gqa-win: get_pci_info: Clean dev_info if handle is valid
  MAINTAINERS: Add entry for QEMU Guest Agent Windows components

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
4 years agoroms/openbios: update OpenBIOS images to 04dfc98 built from submodule
Mark Cave-Ayland [Sat, 15 Jan 2022 14:24:36 +0000 (14:24 +0000)]
roms/openbios: update OpenBIOS images to 04dfc98 built from submodule

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
4 years agoMerge remote-tracking branch 'remotes/kevin/tags/for-upstream' into staging
Peter Maydell [Fri, 14 Jan 2022 15:56:30 +0000 (15:56 +0000)]
Merge remote-tracking branch 'remotes/kevin/tags/for-upstream' into staging

Block layer patches

- qemu-storage-daemon: Add vhost-user-blk help
- block-backend: Fix use-after-free for BDS pointers after aio_poll()
- qemu-img: Fix sparseness of output image with unaligned ranges
- vvfat: Fix crashes in read-write mode
- Fix device deletion events with -device JSON syntax
- Code cleanups

# gpg: Signature made Fri 14 Jan 2022 13:50:16 GMT
# gpg:                using RSA key DC3DEB159A9AF95D3D7456FE7F09B272C88F2FD6
# gpg:                issuer "kwolf@redhat.com"
# gpg: Good signature from "Kevin Wolf <kwolf@redhat.com>" [full]
# Primary key fingerprint: DC3D EB15 9A9A F95D 3D74  56FE 7F09 B272 C88F 2FD6

* remotes/kevin/tags/for-upstream:
  iotests/testrunner.py: refactor test_field_width
  block: drop BLK_PERM_GRAPH_MOD
  qemu-img: make is_allocated_sectors() more efficient
  iotests: Test qemu-img convert of zeroed data cluster
  vvfat: Fix vvfat_write() for writes before the root directory
  vvfat: Fix size of temporary qcow file
  iotests/308: Fix for CAP_DAC_OVERRIDE
  iotests/stream-error-on-reset: New test
  block-backend: prevent dangling BDS pointers across aio_poll()
  qapi/block: Restrict vhost-user-blk to CONFIG_VHOST_USER_BLK_SERVER
  qemu-storage-daemon: Add vhost-user-blk help
  docs: Correct 'vhost-user-blk' spelling
  softmmu: fix device deletion events with -device JSON syntax
  include/sysemu/blockdev.h: remove drive_get_max_devs
  include/sysemu/blockdev.h: remove drive_mark_claimed_by_board and inline drive_def
  block_int: make bdrv_backing_overridden static

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
4 years agoMerge remote-tracking branch 'remotes/kraxel/tags/kraxel-20220114-pull-request' into...
Peter Maydell [Fri, 14 Jan 2022 13:21:41 +0000 (13:21 +0000)]
Merge remote-tracking branch 'remotes/kraxel/tags/kraxel-20220114-pull-request' into staging

- bugfixes for ui, usb, audio, display
- change default display resolution
- add horizontal scrolling support

# gpg: Signature made Fri 14 Jan 2022 06:52:53 GMT
# gpg:                using RSA key A0328CFFB93A17A79901FE7D4CB6D8EED3E87138
# gpg: Good signature from "Gerd Hoffmann (work) <kraxel@redhat.com>" [full]
# gpg:                 aka "Gerd Hoffmann <gerd@kraxel.org>" [full]
# gpg:                 aka "Gerd Hoffmann (private) <kraxel@gmail.com>" [full]
# Primary key fingerprint: A032 8CFF B93A 17A7 9901  FE7D 4CB6 D8EE D3E8 7138

* remotes/kraxel/tags/kraxel-20220114-pull-request:
  ui/input-legacy: pass horizontal scroll information
  ui/sdl2: pass horizontal scroll information to the device code
  ui/gtk: pass horizontal scroll information to the device code
  ui/cocoa: pass horizontal scroll information to the device code
  ps2: Initial horizontal scroll support
  edid: Added support for 4k@60 Hz monitor
  edid: set default resolution to 1280x800 (WXGA)
  hw/mips/jazz: Inline vga_mmio_init() and remove it
  hw/display/vga-mmio: QOM'ify vga_mmio_init() as TYPE_VGA_MMIO
  hw/display/vga-mmio: Inline vga_mm_init()
  hw/display: Rename VGA_ISA_MM -> VGA_MMIO
  uas: add missing return
  ui: fix gtk clipboard clear assertion
  ui/dbus: fix buffer-overflow detected by ASAN
  hw/audio/intel-hda: fix stream reset
  dsoundaudio: fix crackling audio recordings
  jackaudio: use ifdefs to hide unavailable functions
  ui/vnc.c: Fixed a deadlock bug.
  usb: allow max 8192 bytes for desc
  hw/usb/dev-wacom: add missing HID descriptor

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
4 years agoiotests/testrunner.py: refactor test_field_width
Vladimir Sementsov-Ogievskiy [Fri, 10 Dec 2021 20:14:50 +0000 (21:14 +0100)]
iotests/testrunner.py: refactor test_field_width

A lot of Optional[] types doesn't make code beautiful.
test_field_width defaults to 8, but that is never used in the code.

More over, if we want some default behavior for single call of
test_run(), it should just print the whole test name, not limiting or
expanding its width, so 8 is bad default.

So, just drop the default as unused for now.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20211210201450.101576-1-vsementsov@virtuozzo.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
4 years agoblock: drop BLK_PERM_GRAPH_MOD
Vladimir Sementsov-Ogievskiy [Thu, 2 Sep 2021 09:37:54 +0000 (12:37 +0300)]
block: drop BLK_PERM_GRAPH_MOD

First, this permission never protected a node from being changed, as
generic child-replacing functions don't check it.

Second, it's a strange thing: it presents a permission of parent node
to change its child. But generally, children are replaced by different
mechanisms, like jobs or qmp commands, not by nodes.

Graph-mod permission is hard to understand. All other permissions
describe operations which done by parent node on its child: read,
write, resize. Graph modification operations are something completely
different.

The only place where BLK_PERM_GRAPH_MOD is used as "perm" (not shared
perm) is mirror_start_job, for s->target. Still modern code should use
bdrv_freeze_backing_chain() to protect from graph modification, if we
don't do it somewhere it may be considered as a bug. So, it's a bit
risky to drop GRAPH_MOD, and analyzing of possible loss of protection
is hard. But one day we should do it, let's do it now.

One more bit of information is that locking the corresponding byte in
file-posix doesn't make sense at all.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20210902093754.2352-1-vsementsov@virtuozzo.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
4 years agoqemu-img: make is_allocated_sectors() more efficient
Vladimir Sementsov-Ogievskiy [Fri, 17 Dec 2021 16:46:54 +0000 (17:46 +0100)]
qemu-img: make is_allocated_sectors() more efficient

Consider the case when the whole buffer is zero and end is unaligned.

If i <= tail, we return 1 and do one unaligned WRITE, RMW happens.

If i > tail, we do on aligned WRITE_ZERO (or skip if target is zeroed)
and again one unaligned WRITE, RMW happens.

Let's do better: don't fragment the whole-zero buffer and report it as
ZERO: in case of zeroed target we just do nothing and avoid RMW. If
target is not zeroes, one unaligned WRITE_ZERO should not be much worse
than one unaligned WRITE.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20211217164654.1184218-3-vsementsov@virtuozzo.com>
Tested-by: Peter Lieven <pl@kamp.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
4 years agoiotests: Test qemu-img convert of zeroed data cluster
Kevin Wolf [Fri, 17 Dec 2021 16:46:53 +0000 (17:46 +0100)]
iotests: Test qemu-img convert of zeroed data cluster

This demonstrates what happens when the block status changes in
sub-min_sparse granularity, but all of the parts are zeroed out. The
alignment logic in is_allocated_sectors() prevents that the target image
remains fully sparse as expected, but turns it into a data cluster of
explicit zeros.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20211217164654.1184218-2-vsementsov@virtuozzo.com>
Tested-by: Peter Lieven <pl@kamp.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
4 years agovvfat: Fix vvfat_write() for writes before the root directory
Kevin Wolf [Thu, 9 Dec 2021 15:22:31 +0000 (16:22 +0100)]
vvfat: Fix vvfat_write() for writes before the root directory

The calculation in sector2cluster() is done relative to the offset of
the root directory. Any writes to blocks before the start of the root
directory (in particular, writes to the FAT) result in negative values,
which are not handled correctly in vvfat_write().

This changes sector2cluster() to return a signed value, and makes sure
that vvfat_write() doesn't try to find mappings for negative cluster
number. It clarifies the code in vvfat_write() to make it more obvious
that the cluster numbers can be negative.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20211209152231.23756-1-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
4 years agovvfat: Fix size of temporary qcow file
Kevin Wolf [Thu, 9 Dec 2021 15:18:15 +0000 (16:18 +0100)]
vvfat: Fix size of temporary qcow file

The size of the qcow size was calculated so that only the FAT partition
would fit on it, but not the whole disk. However, offsets relative to
the whole disk are used to access it, so increase its size to be large
enough for that.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20211209151815.23495-1-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
4 years agoiotests/308: Fix for CAP_DAC_OVERRIDE
Hanna Reitz [Mon, 3 Jan 2022 12:00:14 +0000 (13:00 +0100)]
iotests/308: Fix for CAP_DAC_OVERRIDE

With CAP_DAC_OVERRIDE (which e.g. root generally has), permission checks
will be bypassed when opening files.

308 in one instance tries to open a read-only file (FUSE export) with
qemu-io as read/write, and expects this to fail.  However, when running
it as root, opening will succeed (thanks to CAP_DAC_OVERRIDE) and only
the actual write operation will fail.

Note this as "Case not run", but have the test pass in either case.

Reported-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Fixes: 2c7dd057aa7bd7a875e9b1a53975c220d6380bc4
       ("export/fuse: Pass default_permissions for mount")
Signed-off-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20220103120014.13061-1-hreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
4 years agoiotests/stream-error-on-reset: New test
Hanna Reitz [Tue, 11 Jan 2022 15:36:13 +0000 (15:36 +0000)]
iotests/stream-error-on-reset: New test

Test the following scenario:
- Simple stream block in two-layer backing chain (base and top)
- The job is drained via blk_drain(), then an error occurs while the job
  settles the ongoing request
- And so the job completes while in blk_drain()

This was reported as a segfault, but is fixed by "block-backend: prevent
dangling BDS pointers across aio_poll()".

Buglink: https://bugzilla.redhat.com/show_bug.cgi?id=2036178
Signed-off-by: Hanna Reitz <hreitz@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-Id: <20220111153613.25453-3-stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
4 years agoblock-backend: prevent dangling BDS pointers across aio_poll()
Stefan Hajnoczi [Tue, 11 Jan 2022 15:36:12 +0000 (15:36 +0000)]
block-backend: prevent dangling BDS pointers across aio_poll()

The BlockBackend root child can change when aio_poll() is invoked. This
happens when a temporary filter node is removed upon blockjob
completion, for example.

Functions in block/block-backend.c must be aware of this when using a
blk_bs() pointer across aio_poll() because the BlockDriverState refcnt
may reach 0, resulting in a stale pointer.

One example is scsi_device_purge_requests(), which calls blk_drain() to
wait for in-flight requests to cancel. If the backup blockjob is active,
then the BlockBackend root child is a temporary filter BDS owned by the
blockjob. The blockjob can complete during bdrv_drained_begin() and the
last reference to the BDS is released when the temporary filter node is
removed. This results in a use-after-free when blk_drain() calls
bdrv_drained_end(bs) on the dangling pointer.

Explicitly hold a reference to bs across block APIs that invoke
aio_poll().

Buglink: https://bugzilla.redhat.com/show_bug.cgi?id=2021778
Buglink: https://bugzilla.redhat.com/show_bug.cgi?id=2036178
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-Id: <20220111153613.25453-2-stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
4 years agoqapi/block: Restrict vhost-user-blk to CONFIG_VHOST_USER_BLK_SERVER
Philippe Mathieu-Daudé [Fri, 7 Jan 2022 10:54:20 +0000 (11:54 +0100)]
qapi/block: Restrict vhost-user-blk to CONFIG_VHOST_USER_BLK_SERVER

When building QEMU with --disable-vhost-user and using introspection,
query-qmp-schema lists vhost-user-blk even though it's not actually
available:

  { "execute": "query-qmp-schema" }
  {
      "return": [
          ...
          {
              "name": "312",
              "members": [
                  {
                      "name": "nbd"
                  },
                  {
                      "name": "vhost-user-blk"
                  }
              ],
              "meta-type": "enum",
              "values": [
                  "nbd",
                  "vhost-user-blk"
              ]
          },

Restrict vhost-user-blk in BlockExportType when
CONFIG_VHOST_USER_BLK_SERVER is disabled, so it
doesn't end listed by query-qmp-schema.

Fixes: 90fc91d50b7 ("convert vhost-user-blk server to block export API")
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220107105420.395011-4-f4bug@amsat.org>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
4 years agoqemu-storage-daemon: Add vhost-user-blk help
Philippe Mathieu-Daudé [Fri, 7 Jan 2022 10:54:19 +0000 (11:54 +0100)]
qemu-storage-daemon: Add vhost-user-blk help

Add missing vhost-user-blk help:

  $ qemu-storage-daemon -h
  ...
    --export [type=]vhost-user-blk,id=<id>,node-name=<node-name>,
             addr.type=unix,addr.path=<socket-path>[,writable=on|off]
             [,logical-block-size=<block-size>][,num-queues=<num-queues>]
                           export the specified block node as a
                           vhosts-user-blk device over UNIX domain socket
    --export [type=]vhost-user-blk,id=<id>,node-name=<node-name>,
             fd,addr.str=<fd>[,writable=on|off]
             [,logical-block-size=<block-size>][,num-queues=<num-queues>]
                           export the specified block node as a
                           vhosts-user-blk device over file descriptor
  ...

Fixes: 90fc91d50b7 ("convert vhost-user-blk server to block export API")
Reported-by: Qing Wang <qinwang@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220107105420.395011-3-f4bug@amsat.org>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
4 years agodocs: Correct 'vhost-user-blk' spelling
Philippe Mathieu-Daudé [Fri, 7 Jan 2022 10:54:18 +0000 (11:54 +0100)]
docs: Correct 'vhost-user-blk' spelling

Reported-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220107105420.395011-2-f4bug@amsat.org>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
4 years agosoftmmu: fix device deletion events with -device JSON syntax
Daniel P. Berrangé [Wed, 5 Jan 2022 12:38:47 +0000 (12:38 +0000)]
softmmu: fix device deletion events with -device JSON syntax

The -device JSON syntax impl leaks a reference on the created
DeviceState instance. As a result when you hot-unplug the
device, the device_finalize method won't be called and thus
it will fail to emit the required DEVICE_DELETED event.

A 'json-cli' feature was previously added against the
'device_add' QMP command QAPI schema to indicated to mgmt
apps that -device supported JSON syntax. Given the hotplug
bug that feature flag is not usable for its purpose, so
we add a new 'json-cli-hotplug' feature to indicate the
-device supports JSON without breaking hotplug.

Fixes: 5dacda5167560b3af8eadbce5814f60ba44b467e
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/802
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20220105123847.4047954-2-berrange@redhat.com>
Reviewed-by: Laurent Vivier <lvivier@redhat.com>
Tested-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
4 years agoinclude/sysemu/blockdev.h: remove drive_get_max_devs
Emanuele Giuseppe Esposito [Wed, 15 Dec 2021 12:11:40 +0000 (07:11 -0500)]
include/sysemu/blockdev.h: remove drive_get_max_devs

Remove drive_get_max_devs, as it is not used by anyone.

Last use was removed in commit 8f2d75e81d5
("hw: Drop superfluous special checks for orphaned -drive").

Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-Id: <20211215121140.456939-4-eesposit@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
4 years agoinclude/sysemu/blockdev.h: remove drive_mark_claimed_by_board and inline drive_def
Emanuele Giuseppe Esposito [Wed, 15 Dec 2021 12:11:39 +0000 (07:11 -0500)]
include/sysemu/blockdev.h: remove drive_mark_claimed_by_board and inline drive_def

drive_def is only a particular use case of
qemu_opts_parse_noisily, so it can be inlined.

Also remove drive_mark_claimed_by_board, as it is only defined
but not implemented (nor used) anywhere.

Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Message-Id: <20211215121140.456939-3-eesposit@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
4 years agoblock_int: make bdrv_backing_overridden static
Emanuele Giuseppe Esposito [Wed, 15 Dec 2021 12:11:38 +0000 (07:11 -0500)]
block_int: make bdrv_backing_overridden static

bdrv_backing_overridden is only used in block.c, so there is
no need to leave it in block_int.h

Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-Id: <20211215121140.456939-2-eesposit@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
4 years agoMerge remote-tracking branch 'remotes/stefanha-gitlab/tags/block-pull-request' into...
Peter Maydell [Fri, 14 Jan 2022 10:43:32 +0000 (10:43 +0000)]
Merge remote-tracking branch 'remotes/stefanha-gitlab/tags/block-pull-request' into staging

Pull request

# gpg: Signature made Wed 12 Jan 2022 17:13:54 GMT
# 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:
  virtio: unify dataplane and non-dataplane ->handle_output()
  virtio: use ->handle_output() instead of ->handle_aio_output()
  virtio-scsi: prepare virtio_scsi_handle_cmd for dataplane
  virtio-blk: drop unused virtio_blk_handle_vq() return value
  virtio: get rid of VirtIOHandleAIOOutput
  aio-posix: split poll check from ready handler

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
4 years agoui/input-legacy: pass horizontal scroll information
Dmitry Petrov [Sat, 8 Jan 2022 15:39:47 +0000 (16:39 +0100)]
ui/input-legacy: pass horizontal scroll information

This code seems to be used by vmport hack, passing these values allows
to implement horizontal scroll support even when using vmport.
In case it's not supported horizontal scroll will act as a vertical one.

Signed-off-by: Dmitry Petrov <dpetroff@gmail.com>
Message-Id: <20220108153947.171861-6-dpetroff@gmail.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agoui/sdl2: pass horizontal scroll information to the device code
Dmitry Petrov [Sat, 8 Jan 2022 15:39:46 +0000 (16:39 +0100)]
ui/sdl2: pass horizontal scroll information to the device code

Signed-off-by: Dmitry Petrov <dpetroff@gmail.com>
Message-Id: <20220108153947.171861-5-dpetroff@gmail.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agoui/gtk: pass horizontal scroll information to the device code
Dmitry Petrov [Sat, 8 Jan 2022 15:39:45 +0000 (16:39 +0100)]
ui/gtk: pass horizontal scroll information to the device code

Signed-off-by: Dmitry Petrov <dpetroff@gmail.com>
Message-Id: <20220108153947.171861-4-dpetroff@gmail.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agoui/cocoa: pass horizontal scroll information to the device code
Dmitry Petrov [Sat, 8 Jan 2022 15:39:44 +0000 (16:39 +0100)]
ui/cocoa: pass horizontal scroll information to the device code

Signed-off-by: Dmitry Petrov <dpetroff@gmail.com>
Message-Id: <20220108153947.171861-3-dpetroff@gmail.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agops2: Initial horizontal scroll support
Dmitry Petrov [Sat, 8 Jan 2022 15:39:43 +0000 (16:39 +0100)]
ps2: Initial horizontal scroll support

This change adds support for horizontal scroll to ps/2 mouse device
code. The code is implemented to match the logic of linux kernel
which is used as a reference.

Signed-off-by: Dmitry Petrov <dpetroff@gmail.com>
Message-Id: <20220108153947.171861-2-dpetroff@gmail.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agoMerge remote-tracking branch 'remotes/bonzini-gitlab/tags/for-upstream' into staging
Peter Maydell [Thu, 13 Jan 2022 13:59:56 +0000 (13:59 +0000)]
Merge remote-tracking branch 'remotes/bonzini-gitlab/tags/for-upstream' into staging

* configure and meson cleanups
* KVM_GET/SET_SREGS2 support for x86

# gpg: Signature made Wed 12 Jan 2022 13:09:19 GMT
# 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:
  meson: reenable filemonitor-inotify compilation
  meson: build all modules by default
  configure: do not create roms/seabios/config.mak if SeaBIOS not present
  tests/tcg: Fix target-specific Makefile variables path for user-mode
  KVM: x86: ignore interrupt_bitmap field of KVM_GET/SET_SREGS
  KVM: use KVM_{GET|SET}_SREGS2 when supported.
  meson: add comments in the target-specific flags section
  configure, meson: move config-poison.h to meson
  meson: build contrib/ executables after generated headers
  configure: move non-command-line variables away from command-line parsing section
  configure: parse --enable/--disable-strip automatically, flip default
  configure, makefile: remove traces of really old files
  configure: do not set bsd_user/linux_user early
  configure: simplify creation of plugin symbol list
  block/file-posix: Simplify the XFS_IOC_DIOINFO handling
  meson: cleanup common-user/ build
  user: move common-user includes to a subdirectory of {bsd,linux}-user/
  meson: reuse common_user_inc when building files specific to user-mode emulators

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
4 years agoMerge remote-tracking branch 'remotes/legoater/tags/pull-ppc-20220112' into staging
Peter Maydell [Thu, 13 Jan 2022 11:18:24 +0000 (11:18 +0000)]
Merge remote-tracking branch 'remotes/legoater/tags/pull-ppc-20220112' into staging

ppc 7.0 queue:

* New SLOF for PPC970 and POWER5+ (Alexey)
* Fixes for POWER5+ pseries (Cedric)
* Updates of documentation (Leonardo and Thomas)
* First step of exception model cleanup (Fabiano)
* User created PHB3/PHB4 devices (Daniel and Cedric)

# gpg: Signature made Wed 12 Jan 2022 10:43:21 GMT
# gpg:                using RSA key A0F66548F04895EBFE6B0B6051A343C7CFFBECA1
# gpg: Good signature from "Cédric Le Goater <clg@kaod.org>" [undefined]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: A0F6 6548 F048 95EB FE6B  0B60 51A3 43C7 CFFB ECA1

* remotes/legoater/tags/pull-ppc-20220112: (34 commits)
  ppc/pnv: use stack->pci_regs[] in pnv_pec_stk_pci_xscom_write()
  ppc/pnv: turn pnv_phb4_update_regions() into static
  ppc/pnv: Introduce user creatable pnv-phb4 devices
  ppc/pnv: turn 'phb' into a pointer in struct PnvPhb4PecStack
  ppc/pnv: move PHB4 XSCOM init to phb4_realize()
  ppc/pnv: set phb4 properties in stk_realize()
  pnv_phb4_pec: use pnv_phb4_pec_get_phb_id() in pnv_pec_dt_xscom()
  pnv_phb4_pec.c: move pnv_pec_phb_offset() to pnv_phb4.c
  pnv_phb4.c: change TYPE_PNV_PHB4_ROOT_BUS name
  pnv_phb3.h: change TYPE_PNV_PHB3_ROOT_BUS name
  ppc/pnv: Move num_phbs under Pnv8Chip
  ppc/pnv: Complete user created PHB3 devices
  ppc/pnv: Reparent user created PHB3 devices to the PnvChip
  ppc/pnv: Introduce support for user created PHB3 devices
  pnv_phb4.c: check if root port exists in rc_config functions
  pnv_phb4.c: make pnv-phb4-root-port user creatable
  ppc/pnv: Attach PHB3 root port device when defaults are enabled
  pnv_phb4.c: add unique chassis and slot for pnv_phb4_root_port
  pnv_phb3.c: add unique chassis and slot for pnv_phb3_root_port
  target/ppc: Set the correct endianness for powernv memory dumps
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
4 years agoedid: Added support for 4k@60 Hz monitor
Satyeshwar Singh [Tue, 16 Nov 2021 22:11:03 +0000 (14:11 -0800)]
edid: Added support for 4k@60 Hz monitor

Previously, the large modes (>1080p) that were generated by Qemu in its EDID
were all 50 Hz. If we provide them to a Guest OS and the user selects
one of these modes, then the OS by default only gets 50 FPS. This is
especially true for Windows OS. With this patch, we are now exposing a
3840x2160@60 Hz which will allow the guest OS to get 60 FPS.

Cc: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Satyeshwar Singh <satyeshwar.singh@intel.com>
Message-Id: <20211116221103.27128-1-dongwon.kim@intel.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agoedid: set default resolution to 1280x800 (WXGA)
Daniel P. Berrangé [Mon, 29 Nov 2021 14:05:08 +0000 (14:05 +0000)]
edid: set default resolution to 1280x800 (WXGA)

Currently QEMU defaults to a resolution of 1024x768 when exposing EDID
info to the guest OS. The EDID default info is important as this will
influence what resolution many guest OS will configure the screen with
on boot. It can also potentially influence what resolution the firmware
will configure the screen with, though until very recently EDK2 would
not handle EDID info.

One important thing to bear in mind is that the default graphics card
driver provided by Windows will leave the display set to whatever
resolution was enabled by the firmware on boot. Even if sufficient
VRAM is available, the resolution can't be changed without installing
new drivers. IOW, the default resolution choice is quite important
for usability of Windows.

Modern real world monitor hardware for desktop/laptop has supported
resolutions higher than 1024x768 for a long time now, perhaps as long
as 15+ years. There are quite a wide variety of native resolutions in
use today, however, and in wide screen form factors the height may not
be all that tall.

None the less, it is considered that there is scope for making the
QEMU default resolution slightly larger.

In considering what possible new default could be suitable, choices
considered were 1280x720 (720p), 1280x800 (WXGA) and 1280x1024 (SXGA).

In many ways, vertical space is the most important, and so 720p was
discarded due to loosing vertical space, despite being 25% wider.

The SXGA resolution would be good, but when taking into account
window titlebars/toolbars and window manager desktop UI, this might
be a little too tall for some users to fit the guest on their physical
montior.

This patch thus suggests a modest change to 1280x800 (WXGA). This
only consumes 1 MB per colour channel, allowing double buffered
framebuffer in 8 MB of VRAM. Width wise this is 25% larger than
QEMU's current default, but height wise this only adds 5%, so the
difference isn't massive on the QEMU side.

Overall there doesn't appear to be a compelling reason to stick
with 1024x768 resolution.

Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Message-Id: <20211129140508.1745130-1-berrange@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agohw/mips/jazz: Inline vga_mmio_init() and remove it
Philippe Mathieu-Daudé [Mon, 6 Dec 2021 22:45:28 +0000 (23:45 +0100)]
hw/mips/jazz: Inline vga_mmio_init() and remove it

vga_mmio_init() is used only one time and not very helpful,
inline and remove it.

Reviewed-by: BALATON Zoltan <balaton@eik.bme.hu>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20211206224528.563588-5-f4bug@amsat.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agohw/display/vga-mmio: QOM'ify vga_mmio_init() as TYPE_VGA_MMIO
Philippe Mathieu-Daudé [Mon, 6 Dec 2021 22:45:27 +0000 (23:45 +0100)]
hw/display/vga-mmio: QOM'ify vga_mmio_init() as TYPE_VGA_MMIO

Introduce TYPE_VGA_MMIO, a sysbus device.

While there is no change in the vga_mmio_init()
interface, this is a migration compatibility break
of the MIPS Acer Pica 61 Jazz machine (pica61).

Suggested-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: BALATON Zoltan <balaton@eik.bme.hu>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20211206224528.563588-4-f4bug@amsat.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agohw/display/vga-mmio: Inline vga_mm_init()
Philippe Mathieu-Daudé [Mon, 6 Dec 2021 22:45:26 +0000 (23:45 +0100)]
hw/display/vga-mmio: Inline vga_mm_init()

Inline vga_mm_init() in vga_mmio_init() to simplify the
next patch review. Kind of.

Reviewed-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20211206224528.563588-3-f4bug@amsat.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agohw/display: Rename VGA_ISA_MM -> VGA_MMIO
Philippe Mathieu-Daudé [Mon, 6 Dec 2021 22:45:25 +0000 (23:45 +0100)]
hw/display: Rename VGA_ISA_MM -> VGA_MMIO

There is no ISA bus part in the MMIO VGA device, so rename:

 *  hw/display/vga-isa-mm.c -> hw/display/vga-mmio.c
 *  CONFIG_VGA_ISA_MM -> CONFIG_VGA_MMIO
 *  ISAVGAMMState -> VGAMmioState
 *  isa_vga_mm_init() -> vga_mmio_init()

Reviewed-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20211206224528.563588-2-f4bug@amsat.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agouas: add missing return
Gerd Hoffmann [Fri, 10 Dec 2021 08:06:59 +0000 (09:06 +0100)]
uas: add missing return

Otherwise we run the error handling code even for successful requests.

Fixes: 13b250b12ad3 ("uas: add stream number sanity checks.")
Reported-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20211210080659.2537084-1-kraxel@redhat.com>

4 years agoui: fix gtk clipboard clear assertion
Marc-André Lureau [Thu, 16 Dec 2021 08:32:33 +0000 (12:32 +0400)]
ui: fix gtk clipboard clear assertion

When closing the QEMU Gtk display window, it can occasionaly warn:
qemu-system-x86_64: Gtk: gtk_clipboard_set_with_data: assertion 'targets != NULL' failed

#3  0x00007ffff4f02f22 in gtk_clipboard_set_with_data (clipboard=<optimized out>, targets=<optimized out>, n_targets=<optimized out>, get_func=<optimized out>, clear_func=<optimized out>, user_data=<optimized out>) at /usr/src/debug/gtk3-3.24.30-4.fc35.x86_64/gtk/gtkclipboard.c:672
#4  0x00007ffff552cd75 in gd_clipboard_update_info (gd=0x5555579a9e00, info=0x555557ba4b50) at ../ui/gtk-clipboard.c:98
#5  0x00007ffff552ce00 in gd_clipboard_notify (notifier=0x5555579aaba8, data=0x7fffffffd720) at ../ui/gtk-clipboard.c:128
#6  0x000055555603e0ff in notifier_list_notify (list=0x555556657470 <clipboard_notifiers>, data=0x7fffffffd720) at ../util/notify.c:39
#7  0x000055555594e8e0 in qemu_clipboard_update (info=0x555557ba4b50) at ../ui/clipboard.c:54
#8  0x000055555594e840 in qemu_clipboard_peer_release (peer=0x55555684a5b0, selection=QEMU_CLIPBOARD_SELECTION_PRIMARY) at ../ui/clipboard.c:40
#9  0x000055555594e786 in qemu_clipboard_peer_unregister (peer=0x55555684a5b0) at ../ui/clipboard.c:19
#10 0x000055555595f044 in vdagent_disconnect (vd=0x55555684a400) at ../ui/vdagent.c:852
#11 0x000055555595f262 in vdagent_chr_fini (obj=0x55555684a400) at ../ui/vdagent.c:908

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20211216083233.1166504-1-marcandre.lureau@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agoui/dbus: fix buffer-overflow detected by ASAN
Marc-André Lureau [Wed, 22 Dec 2021 14:40:32 +0000 (18:40 +0400)]
ui/dbus: fix buffer-overflow detected by ASAN

On the last added dbus patch, I left a tiny BO:

==441487==ERROR: AddressSanitizer: heap-buffer-overflow on address 0x611000025a70 at pc 0x7f0817bb764c bp 0x7ffde672ae60 sp 0x7ffde672ae58
WRITE of size 8 at 0x611000025a70 thread T0
    #0 0x7f0817bb764b in dbus_vc_class_init ../ui/dbus.c:401

A cookie for ASAN! not you C :)

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Fixes: 7f767ca35e5 ("ui/dbus: register D-Bus VC handler")
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20211222144032.443424-1-marcandre.lureau@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agohw/audio/intel-hda: fix stream reset
Volker Rümelin [Sun, 26 Dec 2021 15:40:17 +0000 (16:40 +0100)]
hw/audio/intel-hda: fix stream reset

Quote from:
High Definition Audio Specification 1.0a, section 3.3.35

Offset 80: {IOB}SDnCTL Stream Reset (SRST): Writing a 1 causes
the corresponding stream to be reset. The Stream Descriptor
registers (except the SRST bit itself) ... are reset.

Change the code to reset the Stream Descriptor Control and Status
registers except the SRST bit.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/757
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20211226154017.6067-3-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agodsoundaudio: fix crackling audio recordings
Volker Rümelin [Sun, 26 Dec 2021 15:40:16 +0000 (16:40 +0100)]
dsoundaudio: fix crackling audio recordings

Audio recordings with the DirectSound backend don't sound right.
A look a the Microsoft online documentation tells us why.

From the DirectSound Programming Guide, Capture Buffer Information:
'You can safely copy data from the buffer only up to the read
cursor.'

Change the code to read up to the read cursor instead of the
capture cursor.

Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20211226154017.6067-2-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agojackaudio: use ifdefs to hide unavailable functions
Volker Rümelin [Sun, 26 Dec 2021 15:40:15 +0000 (16:40 +0100)]
jackaudio: use ifdefs to hide unavailable functions

On Windows the jack_set_thread_creator() function and on MacOS the
pthread_setname_np() function with a thread pointer paramater is
not available. Use #ifdefs to remove the jack_set_thread_creator()
function call and the qjack_thread_creator() function in both
cases.

The qjack_thread_creator() function just sets the name of the
created thread for debugging purposes and isn't really necessary.

From the jack_set_thread_creator() documentation:
(...)

No normal application/client should consider calling this. (...)

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/785
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Message-Id: <20211226154017.6067-1-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agoui/vnc.c: Fixed a deadlock bug.
Rao Lei [Wed, 5 Jan 2022 02:08:08 +0000 (10:08 +0800)]
ui/vnc.c: Fixed a deadlock bug.

The GDB statck is as follows:
(gdb) bt
0  __lll_lock_wait (futex=futex@entry=0x56211df20360, private=0) at lowlevellock.c:52
1  0x00007f263caf20a3 in __GI___pthread_mutex_lock (mutex=0x56211df20360) at ../nptl/pthread_mutex_lock.c:80
2  0x000056211a757364 in qemu_mutex_lock_impl (mutex=0x56211df20360, file=0x56211a804857 "../ui/vnc-jobs.h", line=60)
    at ../util/qemu-thread-posix.c:80
3  0x000056211a0ef8c7 in vnc_lock_output (vs=0x56211df14200) at ../ui/vnc-jobs.h:60
4  0x000056211a0efcb7 in vnc_clipboard_send (vs=0x56211df14200, count=1, dwords=0x7ffdf1701338) at ../ui/vnc-clipboard.c:138
5  0x000056211a0f0129 in vnc_clipboard_notify (notifier=0x56211df244c8, data=0x56211dd1bbf0) at ../ui/vnc-clipboard.c:209
6  0x000056211a75dde8 in notifier_list_notify (list=0x56211afa17d0 <clipboard_notifiers>, data=0x56211dd1bbf0) at ../util/notify.c:39
7  0x000056211a0bf0e6 in qemu_clipboard_update (info=0x56211dd1bbf0) at ../ui/clipboard.c:50
8  0x000056211a0bf05d in qemu_clipboard_peer_release (peer=0x56211df244c0, selection=QEMU_CLIPBOARD_SELECTION_CLIPBOARD)
    at ../ui/clipboard.c:41
9  0x000056211a0bef9b in qemu_clipboard_peer_unregister (peer=0x56211df244c0) at ../ui/clipboard.c:19
10 0x000056211a0d45f3 in vnc_disconnect_finish (vs=0x56211df14200) at ../ui/vnc.c:1358
11 0x000056211a0d4c9d in vnc_client_read (vs=0x56211df14200) at ../ui/vnc.c:1611
12 0x000056211a0d4df8 in vnc_client_io (ioc=0x56211ce70690, condition=G_IO_IN, opaque=0x56211df14200) at ../ui/vnc.c:1649
13 0x000056211a5b976c in qio_channel_fd_source_dispatch
    (source=0x56211ce50a00, callback=0x56211a0d4d71 <vnc_client_io>, user_data=0x56211df14200) at ../io/channel-watch.c:84
14 0x00007f263ccede8e in g_main_context_dispatch () at /lib/x86_64-linux-gnu/libglib-2.0.so.0
15 0x000056211a77d4a1 in glib_pollfds_poll () at ../util/main-loop.c:232
16 0x000056211a77d51f in os_host_main_loop_wait (timeout=958545) at ../util/main-loop.c:255
17 0x000056211a77d630 in main_loop_wait (nonblocking=0) at ../util/main-loop.c:531
18 0x000056211a45bc8e in qemu_main_loop () at ../softmmu/runstate.c:726
19 0x000056211a0b45fa in main (argc=69, argv=0x7ffdf1701778, envp=0x7ffdf17019a8) at ../softmmu/main.c:50

From the call trace, we can see it is a deadlock bug.
vnc_disconnect_finish will acquire the output_mutex.
But, the output_mutex will be acquired again in vnc_clipboard_send.
Repeated locking will cause deadlock. So, I move
qemu_clipboard_peer_unregister() behind vnc_unlock_output();

Fixes: 0bf41cab93e ("ui/vnc: clipboard support")
Signed-off-by: Lei Rao <lei.rao@intel.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220105020808.597325-1-lei.rao@intel.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agousb: allow max 8192 bytes for desc
zhenwei pi [Wed, 12 Jan 2022 01:58:35 +0000 (09:58 +0800)]
usb: allow max 8192 bytes for desc

A device of USB video class usually uses larger desc structure, so
use larger buffer to avoid failure. (dev-video.c is ready)

This is an unlikely code path:
1, during guest startup, guest tries to probe device.
2, run 'lsusb' command in guest(or other similar commands).

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: zhenwei pi <pizhenwei@bytedance.com>
Message-Id: <20220112015835.900619-1-pizhenwei@bytedance.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agohw/usb/dev-wacom: add missing HID descriptor
Dario Binacchi [Wed, 12 Jan 2022 09:01:25 +0000 (10:01 +0100)]
hw/usb/dev-wacom: add missing HID descriptor

Linux need to fill up the HID descriptor in order to let the driver be
emulated. The descriptor was downloaded from [1]. The patch was tested
with evtest tool on top of qemu 5.2.0 with linux kernel 4.19.208.

[1] https://github.com/linuxwacom/wacom-hid-descriptors/tree/master/Wacom%20PenPartner

Signed-off-by: Michael Trimarchi <michael@amarulasolutions.com>
Co-developed-by: Michael Trimarchi <michael@amarulasolutions.com>
Signed-off-by: Dario Binacchi <dario.binacchi@amarulasolutions.com>
Message-Id: <20220112090125.381364-1-dario.binacchi@amarulasolutions.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agovirtio: unify dataplane and non-dataplane ->handle_output()
Stefan Hajnoczi [Tue, 7 Dec 2021 13:23:36 +0000 (13:23 +0000)]
virtio: unify dataplane and non-dataplane ->handle_output()

Now that virtio-blk and virtio-scsi are ready, get rid of
the handle_aio_output() callback. It's no longer needed.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Message-id: 20211207132336.36627-7-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
4 years agovirtio: use ->handle_output() instead of ->handle_aio_output()
Stefan Hajnoczi [Tue, 7 Dec 2021 13:23:35 +0000 (13:23 +0000)]
virtio: use ->handle_output() instead of ->handle_aio_output()

The difference between ->handle_output() and ->handle_aio_output() was
that ->handle_aio_output() returned a bool return value indicating
progress. This was needed by the old polling API but now that the bool
return value is gone, the two functions can be unified.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Message-id: 20211207132336.36627-6-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
4 years agovirtio-scsi: prepare virtio_scsi_handle_cmd for dataplane
Stefan Hajnoczi [Tue, 7 Dec 2021 13:23:34 +0000 (13:23 +0000)]
virtio-scsi: prepare virtio_scsi_handle_cmd for dataplane

Prepare virtio_scsi_handle_cmd() to be used by both dataplane and
non-dataplane by making the condition for starting ioeventfd more
specific. This way it won't trigger when dataplane has already been
started.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Message-id: 20211207132336.36627-5-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
4 years agovirtio-blk: drop unused virtio_blk_handle_vq() return value
Stefan Hajnoczi [Tue, 7 Dec 2021 13:23:33 +0000 (13:23 +0000)]
virtio-blk: drop unused virtio_blk_handle_vq() return value

The return value of virtio_blk_handle_vq() is no longer used. Get rid of
it. This is a step towards unifying the dataplane and non-dataplane
virtqueue handler functions.

Prepare virtio_blk_handle_output() to be used by both dataplane and
non-dataplane by making the condition for starting ioeventfd more
specific. This way it won't trigger when dataplane has already been
started.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Message-id: 20211207132336.36627-4-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
4 years agovirtio: get rid of VirtIOHandleAIOOutput
Stefan Hajnoczi [Tue, 7 Dec 2021 13:23:32 +0000 (13:23 +0000)]
virtio: get rid of VirtIOHandleAIOOutput

The virtqueue host notifier API
virtio_queue_aio_set_host_notifier_handler() polls the virtqueue for new
buffers. AioContext previously required a bool progress return value
indicating whether an event was handled or not. This is no longer
necessary because the AioContext polling API has been split into a poll
check function and an event handler function. The event handler is only
run when we know there is work to do, so it doesn't return bool.

The VirtIOHandleAIOOutput function signature is now the same as
VirtIOHandleOutput. Get rid of the bool return value.

Further simplifications will be made for virtio-blk and virtio-scsi in
the next patch.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Message-id: 20211207132336.36627-3-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
4 years agoaio-posix: split poll check from ready handler
Stefan Hajnoczi [Tue, 7 Dec 2021 13:23:31 +0000 (13:23 +0000)]
aio-posix: split poll check from ready handler

Adaptive polling measures the execution time of the polling check plus
handlers called when a polled event becomes ready. Handlers can take a
significant amount of time, making it look like polling was running for
a long time when in fact the event handler was running for a long time.

For example, on Linux the io_submit(2) syscall invoked when a virtio-blk
device's virtqueue becomes ready can take 10s of microseconds. This
can exceed the default polling interval (32 microseconds) and cause
adaptive polling to stop polling.

By excluding the handler's execution time from the polling check we make
the adaptive polling calculation more accurate. As a result, the event
loop now stays in polling mode where previously it would have fallen
back to file descriptor monitoring.

The following data was collected with virtio-blk num-queues=2
event_idx=off using an IOThread. Before:

168k IOPS, IOThread syscalls:

  9837.115 ( 0.020 ms): IO iothread1/620155 io_submit(ctx_id: 140512552468480, nr: 16, iocbpp: 0x7fcb9f937db0)    = 16
  9837.158 ( 0.002 ms): IO iothread1/620155 write(fd: 103, buf: 0x556a2ef71b88, count: 8)                         = 8
  9837.161 ( 0.001 ms): IO iothread1/620155 write(fd: 104, buf: 0x556a2ef71b88, count: 8)                         = 8
  9837.163 ( 0.001 ms): IO iothread1/620155 ppoll(ufds: 0x7fcb90002800, nfds: 4, tsp: 0x7fcb9f1342d0, sigsetsize: 8) = 3
  9837.164 ( 0.001 ms): IO iothread1/620155 read(fd: 107, buf: 0x7fcb9f939cc0, count: 512)                        = 8
  9837.174 ( 0.001 ms): IO iothread1/620155 read(fd: 105, buf: 0x7fcb9f939cc0, count: 512)                        = 8
  9837.176 ( 0.001 ms): IO iothread1/620155 read(fd: 106, buf: 0x7fcb9f939cc0, count: 512)                        = 8
  9837.209 ( 0.035 ms): IO iothread1/620155 io_submit(ctx_id: 140512552468480, nr: 32, iocbpp: 0x7fca7d0cebe0)    = 32

174k IOPS (+3.6%), IOThread syscalls:

  9809.566 ( 0.036 ms): IO iothread1/623061 io_submit(ctx_id: 140539805028352, nr: 32, iocbpp: 0x7fd0cdd62be0)    = 32
  9809.625 ( 0.001 ms): IO iothread1/623061 write(fd: 103, buf: 0x5647cfba5f58, count: 8)                         = 8
  9809.627 ( 0.002 ms): IO iothread1/623061 write(fd: 104, buf: 0x5647cfba5f58, count: 8)                         = 8
  9809.663 ( 0.036 ms): IO iothread1/623061 io_submit(ctx_id: 140539805028352, nr: 32, iocbpp: 0x7fd0d0388b50)    = 32

Notice that ppoll(2) and eventfd read(2) syscalls are eliminated because
the IOThread stays in polling mode instead of falling back to file
descriptor monitoring.

As usual, polling is not implemented on Windows so this patch ignores
the new io_poll_read() callback in aio-win32.c.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Message-id: 20211207132336.36627-2-stefanha@redhat.com

[Fixed up aio_set_event_notifier() calls in
tests/unit/test-fdmon-epoll.c added after this series was queued.
--Stefan]

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>