OSDN Git Service

qmiga/qemu.git
18 months agotarget/arm: Mark up sysregs for HFGITR bits 12..17
Peter Maydell [Mon, 30 Jan 2023 18:24:53 +0000 (18:24 +0000)]
target/arm: Mark up sysregs for HFGITR bits 12..17

Mark up the sysreg definitions for the system instructions
trapped by HFGITR bits 12..17. These bits cover AT address
translation instructions.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Fuad Tabba <tabba@google.com>
Message-id: 20230130182459.3309057-18-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-18-peter.maydell@linaro.org

18 months agotarget/arm: Mark up sysregs for HFGITR bits 0..11
Peter Maydell [Mon, 30 Jan 2023 18:24:52 +0000 (18:24 +0000)]
target/arm: Mark up sysregs for HFGITR bits 0..11

Mark up the sysreg definitions for the system instructions
trapped by HFGITR bits 0..11. These bits cover various
cache maintenance operations.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Fuad Tabba <tabba@google.com>
Message-id: 20230130182459.3309057-17-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-17-peter.maydell@linaro.org

18 months agotarget/arm: Mark up sysregs for HDFGRTR bits 12..63
Peter Maydell [Mon, 30 Jan 2023 18:24:51 +0000 (18:24 +0000)]
target/arm: Mark up sysregs for HDFGRTR bits 12..63

Mark up the sysreg definitions for the registers trapped
by HDFGRTR/HDFGWTR bits 12..x.

Bits 12..22 and bit 58 are for PMU registers.

The remaining bits in HDFGRTR/HDFGWTR are for traps on
registers that are part of features we don't implement:

Bits 23..32 and 63 : FEAT_SPE
Bits 33..48 : FEAT_ETE
Bits 50..56 : FEAT_TRBE
Bits 59..61 : FEAT_BRBE
Bit 62 : FEAT_SPEv1p2.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Fuad Tabba <tabba@google.com>
Message-id: 20230130182459.3309057-16-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-16-peter.maydell@linaro.org

18 months agotarget/arm: Mark up sysregs for HDFGRTR bits 0..11
Peter Maydell [Mon, 30 Jan 2023 18:24:50 +0000 (18:24 +0000)]
target/arm: Mark up sysregs for HDFGRTR bits 0..11

Mark up the sysreg definitons for the registers trapped
by HDFGRTR/HDFGWTR bits 0..11. These cover various debug
related registers.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Fuad Tabba <tabba@google.com>
Message-id: 20230130182459.3309057-15-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-15-peter.maydell@linaro.org

18 months agotarget/arm: Mark up sysregs for HFGRTR bits 36..63
Peter Maydell [Mon, 30 Jan 2023 18:24:49 +0000 (18:24 +0000)]
target/arm: Mark up sysregs for HFGRTR bits 36..63

Mark up the sysreg definitions for the registers trapped
by HFGRTR/HFGWTR bits 36..63.

Of these, some correspond to RAS registers which we implement as
always-UNDEF: these don't need any extra handling for FGT because the
UNDEF-to-EL1 always takes priority over any theoretical
FGT-trap-to-EL2.

Bit 50 (NACCDATA_EL1) is for the ACCDATA_EL1 register which is part
of the FEAT_LS64_ACCDATA feature which we don't yet implement.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Fuad Tabba <tabba@google.com>
Message-id: 20230130182459.3309057-14-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-14-peter.maydell@linaro.org

18 months agotarget/arm: Mark up sysregs for HFGRTR bits 24..35
Peter Maydell [Mon, 30 Jan 2023 18:24:48 +0000 (18:24 +0000)]
target/arm: Mark up sysregs for HFGRTR bits 24..35

Mark up the sysreg definitions for the registers trapped
by HFGRTR/HFGWTR bits 24..35.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Fuad Tabba <tabba@google.com>
Message-id: 20230130182459.3309057-13-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-13-peter.maydell@linaro.org

18 months agotarget/arm: Mark up sysregs for HFGRTR bits 12..23
Peter Maydell [Mon, 30 Jan 2023 18:24:47 +0000 (18:24 +0000)]
target/arm: Mark up sysregs for HFGRTR bits 12..23

Mark up the sysreg definitions for the registers trapped
by HFGRTR/HFGWTR bits 12..23.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Fuad Tabba <tabba@google.com>
Message-id: 20230130182459.3309057-12-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-12-peter.maydell@linaro.org

18 months agotarget/arm: Mark up sysregs for HFGRTR bits 0..11
Peter Maydell [Mon, 30 Jan 2023 18:24:46 +0000 (18:24 +0000)]
target/arm: Mark up sysregs for HFGRTR bits 0..11

Mark up the sysreg definitions for the registers trapped
by HFGRTR/HFGWTR bits 0..11.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Fuad Tabba <tabba@google.com>
Message-id: 20230130182459.3309057-11-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-11-peter.maydell@linaro.org

18 months agotarget/arm: Implement FGT trapping infrastructure
Peter Maydell [Mon, 30 Jan 2023 18:24:45 +0000 (18:24 +0000)]
target/arm: Implement FGT trapping infrastructure

Implement the machinery for fine-grained traps on normal sysregs.
Any sysreg with a fine-grained trap will set the new field to
indicate which FGT register bit it should trap on.

FGT traps only happen when an AArch64 EL2 enables them for
an AArch64 EL1. They therefore are only relevant for AArch32
cpregs when the cpreg can be accessed from EL0. The logic
in access_check_cp_reg() will check this, so it is safe to
add a .fgt marking to an ARM_CP_STATE_BOTH ARMCPRegInfo.

The DO_BIT and DO_REV_BIT macros define enum constants FGT_##bitname
which can be used to specify the FGT bit, eg
   .fgt = FGT_AFSR0_EL1
(We assume that there is no bit name duplication across the FGT
registers, for brevity's sake.)

Subsequent commits will add the .fgt fields to the relevant register
definitions and define the FGT_nnn values for them.

Note that some of the FGT traps are for instructions that we don't
handle via the cpregs mechanisms (mostly these are instruction traps).
Those we will have to handle separately.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Fuad Tabba <tabba@google.com>
Message-id: 20230130182459.3309057-10-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-10-peter.maydell@linaro.org

18 months agotarget/arm: Define the FEAT_FGT registers
Peter Maydell [Mon, 30 Jan 2023 18:24:44 +0000 (18:24 +0000)]
target/arm: Define the FEAT_FGT registers

Define the system registers which are provided by the
FEAT_FGT fine-grained trap architectural feature:
 HFGRTR_EL2, HFGWTR_EL2, HDFGRTR_EL2, HDFGWTR_EL2, HFGITR_EL2

All these registers are a set of bit fields, where each bit is set
for a trap and clear to not trap on a particular system register
access.  The R and W register pairs are for system registers,
allowing trapping to be done separately for reads and writes; the I
register is for system instructions where trapping is on instruction
execution.

The data storage in the CPU state struct is arranged as a set of
arrays rather than separate fields so that when we're looking up the
bits for a system register access we can just index into the array
rather than having to use a switch to select a named struct member.
The later FEAT_FGT2 will add extra elements to these arrays.

The field definitions for the new registers are in cpregs.h because
in practice the code that needs them is code that also needs
the cpregs information; cpu.h is included in a lot more files.
We're also going to add some FGT-specific definitions to cpregs.h
in the next commit.

We do not implement HAFGRTR_EL2, because we don't implement
FEAT_AMUv1.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Fuad Tabba <tabba@google.com>
Message-id: 20230130182459.3309057-9-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-9-peter.maydell@linaro.org

18 months agotarget/arm: Disable HSTR_EL2 traps if EL2 is not enabled
Peter Maydell [Mon, 30 Jan 2023 18:24:43 +0000 (18:24 +0000)]
target/arm: Disable HSTR_EL2 traps if EL2 is not enabled

The HSTR_EL2 register is not supposed to have an effect unless EL2 is
enabled in the current security state.  We weren't checking for this,
which meant that if the guest set up the HSTR_EL2 register we would
incorrectly trap even for accesses from Secure EL0 and EL1.

Add the missing checks. (Other places where we look at HSTR_EL2
for the not-in-v8A bits TTEE and TJDBX are already checking that
we are in NS EL0 or EL1, so there we alredy know EL2 is enabled.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Fuad Tabba <tabba@google.com>
Message-id: 20230130182459.3309057-8-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-8-peter.maydell@linaro.org

18 months agotarget/arm: Make HSTR_EL2 traps take priority over UNDEF-at-EL1
Peter Maydell [Mon, 30 Jan 2023 18:24:42 +0000 (18:24 +0000)]
target/arm: Make HSTR_EL2 traps take priority over UNDEF-at-EL1

The semantics of HSTR_EL2 require that it traps cpreg accesses
to EL2 for:
 * EL1 accesses
 * EL0 accesses, if the access is not UNDEFINED when the
   trap bit is 0

(You can see this in the I_ZFGJP priority ordering, where HSTR_EL2
traps from EL1 to EL2 are priority 12, UNDEFs are priority 13, and
HSTR_EL2 traps from EL0 are priority 15.)

However, we don't get this right for EL1 accesses which UNDEF because
the register doesn't exist at all or because its ri->access bits
non-configurably forbid the access.  At EL1, check for the HSTR_EL2
trap early, before either of these UNDEF reasons.

We have to retain the HSTR_EL2 check in access_check_cp_reg(),
because at EL0 any kind of UNDEF-to-EL1 (including "no such
register", "bad ri->access" and "ri->accessfn returns 'trap to EL1'")
takes precedence over the trap to EL2.  But we only need to do that
check for EL0 now.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Tested-by: Fuad Tabba <tabba@google.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230130182459.3309057-7-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-7-peter.maydell@linaro.org

18 months agotarget/arm: All UNDEF-at-EL0 traps take priority over HSTR_EL2 traps
Peter Maydell [Mon, 30 Jan 2023 18:24:41 +0000 (18:24 +0000)]
target/arm: All UNDEF-at-EL0 traps take priority over HSTR_EL2 traps

The HSTR_EL2 register has a collection of trap bits which allow
trapping to EL2 for AArch32 EL0 or EL1 accesses to coprocessor
registers.  The specification of these bits is that when the bit is
set we should trap
 * EL1 accesses
 * EL0 accesses, if the access is not UNDEFINED when the
   trap bit is 0

In other words, all UNDEF traps from EL0 to EL1 take precedence over
the HSTR_EL2 trap to EL2.  (Since this is all AArch32, the only kind
of trap-to-EL1 is the UNDEF.)

Our implementation doesn't quite get this right -- we check for traps
in the order:
 * no such register
 * ARMCPRegInfo::access bits
 * HSTR_EL2 trap bits
 * ARMCPRegInfo::accessfn

So UNDEFs that happen because of the access bits or because the
register doesn't exist at all correctly take priority over the
HSTR_EL2 trap, but where a register can UNDEF at EL0 because of the
accessfn we are incorrectly always taking the HSTR_EL2 trap.  There
aren't many of these, but one example is the PMCR; if you look at the
access pseudocode for this register you can see that UNDEFs taken
because of the value of PMUSERENR.EN are checked before the HSTR_EL2
bit.

Rearrange helper_access_check_cp_reg() so that we always call the
accessfn, and use its return value if it indicates that the access
traps to EL0 rather than continuing to do the HSTR_EL2 check.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Fuad Tabba <tabba@google.com>
Message-id: 20230130182459.3309057-6-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-6-peter.maydell@linaro.org

18 months agotarget/arm: Move do_coproc_insn() syndrome calculation earlier
Peter Maydell [Mon, 30 Jan 2023 18:24:40 +0000 (18:24 +0000)]
target/arm: Move do_coproc_insn() syndrome calculation earlier

Rearrange the code in do_coproc_insn() so that we calculate the
syndrome value for a potential trap early; we're about to add a
second check that wants this value earlier than where it is currently
determined.

(Specifically, a trap to EL2 because of HSTR_EL2 should take
priority over an UNDEF to EL1, even when the UNDEF is because
the register does not exist at all or because its ri->access
bits non-configurably fail the access. So the check we put in
for HSTR_EL2 trapping at EL1 (which needs the syndrome) is
going to have to be done before the check "is the ARMCPRegInfo
pointer NULL".)

This commit is just code motion; the change to HSTR_EL2
handling that will use the 'syndrome' variable is in a
subsequent commit.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Fuad Tabba <tabba@google.com>
Message-id: 20230130182459.3309057-5-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-5-peter.maydell@linaro.org

18 months agotarget/arm: Remove CP_ACCESS_TRAP_UNCATEGORIZED_{EL2, EL3}
Peter Maydell [Mon, 30 Jan 2023 18:24:39 +0000 (18:24 +0000)]
target/arm: Remove CP_ACCESS_TRAP_UNCATEGORIZED_{EL2, EL3}

We added the CPAccessResult values CP_ACCESS_TRAP_UNCATEGORIZED_EL2
and CP_ACCESS_TRAP_UNCATEGORIZED_EL3 purely in order to use them in
the ats_access() function, but doing so was incorrect (a bug fixed in
a previous commit).  There aren't any cases where we want an access
function to be able to request a trap to EL2 or EL3 with a zero
syndrome value, so remove these enum values.

As well as cleaning up dead code, the motivation here is that
we'd like to implement fine-grained-trap handling in
helper_access_check_cp_reg(). Although the fine-grained traps
to EL2 are always lower priority than trap-to-same-EL and
higher priority than trap-to-EL3, they are in the middle of
various other kinds of trap-to-EL2. Knowing that a trap-to-EL2
must always for us have the same syndrome (ie that an access
function will return CP_ACCESS_TRAP_EL2 and there is no other
kind of trap-to-EL2 enum value) means we don't have to try
to choose which of the two syndrome values to report if the
access would trap to EL2 both for the fine-grained-trap and
because the access function requires it.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Fuad Tabba <tabba@google.com>
Message-id: 20230130182459.3309057-4-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-4-peter.maydell@linaro.org

18 months agotarget/arm: Correct syndrome for ATS12NSO* at Secure EL1
Peter Maydell [Mon, 30 Jan 2023 18:24:38 +0000 (18:24 +0000)]
target/arm: Correct syndrome for ATS12NSO* at Secure EL1

The AArch32 ATS12NSO* address translation operations are supposed to
trap to either EL2 or EL3 if they're executed at Secure EL1 (which
can only happen if EL3 is AArch64).  We implement this, but we got
the syndrome value wrong: like other traps to EL2 or EL3 on an
AArch32 cpreg access, they should report the 0x3 syndrome, not the
0x0 'uncategorized' syndrome.  This is clear in the access pseudocode
for these instructions.

Fix the syndrome value for these operations by correcting the
returned value from the ats_access() function.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Fuad Tabba <tabba@google.com>
Message-id: 20230130182459.3309057-3-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-3-peter.maydell@linaro.org

18 months agotarget/arm: Name AT_S1E1RP and AT_S1E1WP cpregs correctly
Peter Maydell [Mon, 30 Jan 2023 18:24:37 +0000 (18:24 +0000)]
target/arm: Name AT_S1E1RP and AT_S1E1WP cpregs correctly

The encodings 0,0,C7,C9,0 and 0,0,C7,C9,1 are AT SP1E1RP and AT
S1E1WP, but our ARMCPRegInfo definitions for them incorrectly name
them AT S1E1R and AT S1E1W (which are entirely different
instructions).  Fix the names.

(This has no guest-visible effect as the names are for debug purposes
only.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Fuad Tabba <tabba@google.com>
Message-id: 20230130182459.3309057-2-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-2-peter.maydell@linaro.org

18 months agosbsa-ref: remove cortex-a76 from list of supported cpus
Marcin Juszkiewicz [Thu, 26 Jan 2023 11:44:16 +0000 (12:44 +0100)]
sbsa-ref: remove cortex-a76 from list of supported cpus

Cortex-A76 supports 40bits of address space. sbsa-ref's memory
starts above this limit.

Signed-off-by: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230126114416.2447685-1-marcin.juszkiewicz@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
18 months agohw/arm/virt: Make accels in GIC finalize logic explicit
Alexander Graf [Fri, 23 Dec 2022 09:01:07 +0000 (10:01 +0100)]
hw/arm/virt: Make accels in GIC finalize logic explicit

Let's explicitly list out all accelerators that we support when trying to
determine the supported set of GIC versions. KVM was already separate, so
the only missing one is HVF which simply reuses all of TCG's emulation
code and thus has the same compatibility matrix.

Signed-off-by: Alexander Graf <agraf@csgraf.de>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Zenghui Yu <yuzenghui@huawei.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20221223090107.98888-3-agraf@csgraf.de
[PMM: Added qtest to the list of accelerators]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
18 months agohw/arm/virt: Consolidate GIC finalize logic
Alexander Graf [Fri, 23 Dec 2022 09:01:06 +0000 (10:01 +0100)]
hw/arm/virt: Consolidate GIC finalize logic

Up to now, the finalize_gic_version() code open coded what is essentially
a support bitmap match between host/emulation environment and desired
target GIC type.

This open coding leads to undesirable side effects. For example, a VM with
KVM and -smp 10 will automatically choose GICv3 while the same command
line with TCG will stay on GICv2 and fail the launch.

This patch combines the TCG and KVM matching code paths by making
everything a 2 pass process. First, we determine which GIC versions the
current environment is able to support, then we go through a single
state machine to determine which target GIC mode that means for us.

After this patch, the only user noticable changes should be consolidated
error messages as well as TCG -M virt supporting -smp > 8 automatically.

Signed-off-by: Alexander Graf <agraf@csgraf.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Zenghui Yu <yuzenghui@huawei.com>
Message-id: 20221223090107.98888-2-agraf@csgraf.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
18 months agohvf: arm: Add support for GICv3
Alexander Graf [Sat, 28 Jan 2023 22:44:59 +0000 (23:44 +0100)]
hvf: arm: Add support for GICv3

We currently only support GICv2 emulation. To also support GICv3, we will
need to pass a few system registers into their respective handler functions.

This patch adds support for HVF to call into the TCG callbacks for GICv3
system register handlers. This is safe because the GICv3 TCG code is generic
as long as we limit ourselves to EL0 and EL1 - which are the only modes
supported by HVF.

To make sure nobody trips over that, we also annotate callbacks that don't
work in HVF mode, such as EL state change hooks.

With GICv3 support in place, we can run with more than 8 vCPUs.

Signed-off-by: Alexander Graf <agraf@csgraf.de>
Message-id: 20230128224459.70676-1-agraf@csgraf.de
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
18 months agohw/char/pl011: better handling of FIFO flags on LCR reset
Evgeny Iakovlev [Mon, 23 Jan 2023 16:23:03 +0000 (17:23 +0100)]
hw/char/pl011: better handling of FIFO flags on LCR reset

Current FIFO handling code does not reset RXFE/RXFF flags when guest
resets FIFO by writing to UARTLCR register, although internal FIFO state
is reset to 0 read count. Actual guest-visible flag update will happen
only on next data read or write attempt. As a result of that any guest
that expects RXFE flag to be set (and RXFF to be cleared) after resetting
FIFO will never see that happen.

Signed-off-by: Evgeny Iakovlev <eiakovlev@linux.microsoft.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20230123162304.26254-5-eiakovlev@linux.microsoft.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
18 months agohw/char/pl011: implement a reset method
Evgeny Iakovlev [Mon, 23 Jan 2023 16:23:02 +0000 (17:23 +0100)]
hw/char/pl011: implement a reset method

PL011 currently lacks a reset method. Implement it.

Signed-off-by: Evgeny Iakovlev <eiakovlev@linux.microsoft.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230123162304.26254-4-eiakovlev@linux.microsoft.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
18 months agohw/char/pl011: add post_load hook for backwards-compatibility
Evgeny Iakovlev [Mon, 23 Jan 2023 16:23:01 +0000 (17:23 +0100)]
hw/char/pl011: add post_load hook for backwards-compatibility

Previous change slightly modified the way we handle data writes when
FIFO is disabled. Previously we kept incrementing read_pos and were
storing data at that position, although we only have a
single-register-deep FIFO now. Then we changed it to always store data
at pos 0.

If guest disables FIFO and the proceeds to read data, it will work out
fine, because we still read from current read_pos before setting it to
0.

However, to make code less fragile, introduce a post_load hook for
PL011State and move fixup read FIFO state when FIFO is disabled. Since
we are introducing a post_load hook, also do some sanity checking on
untrusted incoming input state.

Signed-off-by: Evgeny Iakovlev <eiakovlev@linux.microsoft.com>
Message-id: 20230123162304.26254-3-eiakovlev@linux.microsoft.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
18 months agohw/char/pl011: refactor FIFO depth handling code
Evgeny Iakovlev [Mon, 23 Jan 2023 16:23:00 +0000 (17:23 +0100)]
hw/char/pl011: refactor FIFO depth handling code

PL011 can be in either of 2 modes depending guest config: FIFO and
single register. The last mode could be viewed as a 1-element-deep FIFO.

Current code open-codes a bunch of depth-dependent logic. Refactor FIFO
depth handling code to isolate calculating current FIFO depth.

One functional (albeit guest-invisible) side-effect of this change is
that previously we would always increment s->read_pos in UARTDR read
handler even if FIFO was disabled, now we are limiting read_pos to not
exceed FIFO depth (read_pos itself is reset to 0 if user disables FIFO).

Signed-off-by: Evgeny Iakovlev <eiakovlev@linux.microsoft.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230123162304.26254-2-eiakovlev@linux.microsoft.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
18 months agotarget/arm: Fix physical address resolution for Stage2
Richard Henderson [Thu, 26 Jan 2023 23:31:34 +0000 (13:31 -1000)]
target/arm: Fix physical address resolution for Stage2

Conversion to probe_access_full missed applying the page offset.

Cc: qemu-stable@nongnu.org
Reported-by: Sid Manning <sidneym@quicinc.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230126233134.103193-1-richard.henderson@linaro.org
Fixes: f3639a64f602 ("target/arm: Use softmmu tlbs for page table walking")
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
18 months agohw/arm: Use TYPE_ARM_SMMUV3
Richard Henderson [Tue, 24 Jan 2023 23:20:59 +0000 (13:20 -1000)]
hw/arm: Use TYPE_ARM_SMMUV3

Use the macro instead of two explicit string literals.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Message-id: 20230124232059.4017615-1-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
18 months agoMerge tag 'pull-tcg-20230123' of https://gitlab.com/rth7680/qemu into staging
Peter Maydell [Fri, 3 Feb 2023 09:30:45 +0000 (09:30 +0000)]
Merge tag 'pull-tcg-20230123' of https://gitlab.com/rth7680/qemu into staging

common-user: Re-enable ppc32 host
tcg: Avoid recursion in tcg_gen_mulu2_i32
tcg: Mark tcg helpers noinline to avoid an issue with LTO
tcg/arm: Use register pair allocation for qemu_{ld,st}_i64
disas: Enable loongarch disassembler, and fixes
tcg/loongarch64: Improve move immediate
tcg/loongarch64: Improve add immediate
tcg/loongarch64: Improve setcond
tcg/loongarch64: Implement movcond
tcg/loongarch64: Use tcg_pcrel_diff in tcg_out_ldst
tcg/loongarch64: Reorg goto_tb implementation

# -----BEGIN PGP SIGNATURE-----
#
# iQFRBAABCgA7FiEEekgeeIaLTbaoWgXAZN846K9+IV8FAmPPO+0dHHJpY2hhcmQu
# aGVuZGVyc29uQGxpbmFyby5vcmcACgkQZN846K9+IV93jwgAhG+H5XHtJqF2isCc
# a6pYuUWRbhsOFL23FmWKx2O41tHlJ2Seort8M9eIHOu21L9DUJFd291O/4ckiMQM
# 13+KH/Kl5fumM+uEkO9YMyplOddmvygdTd5dCi5y349Gi3CgJH3n4HUl0qnioM/7
# Dy3n8JIvYsBp+8jUsLXo1gSl5P1kLMLwJmP68qgy8z8Xly4bDco1Nb2UKb7qKevO
# lMr6L+2/ALbKLZ6OU50erdUrlbgNs0eiQyJAfJ47SQ57RGuqF4pZ09+9yRI2FPZt
# UlSn+srsec1ieYyM2e5krVWbNcXaj6FouV7CkbgFXoUZt29xA1HTXsso+8vLgDPu
# g8vvuw==
# =Up0b
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 24 Jan 2023 02:01:17 GMT
# gpg:                using RSA key 7A481E78868B4DB6A85A05C064DF38E8AF7E215F
# gpg:                issuer "richard.henderson@linaro.org"
# gpg: Good signature from "Richard Henderson <richard.henderson@linaro.org>" [full]
# Primary key fingerprint: 7A48 1E78 868B 4DB6 A85A  05C0 64DF 38E8 AF7E 215F

* tag 'pull-tcg-20230123' of https://gitlab.com/rth7680/qemu:
  tcg/loongarch64: Reorg goto_tb implementation
  tcg/loongarch64: Use tcg_pcrel_diff in tcg_out_ldst
  tcg/loongarch64: Implement movcond
  tcg/loongarch64: Improve setcond expansion
  tcg/loongarch64: Introduce tcg_out_addi
  tcg/loongarch64: Update tcg-insn-defs.c.inc
  tcg/loongarch64: Optimize immediate loading
  target/loongarch: Disassemble pcadd* addresses
  target/loongarch: Disassemble jirl properly
  target/loongarch: Enable the disassembler for host tcg
  tcg: Mark tcg helpers noinline to avoid an issue with LTO
  linux-user: Implment host/ppc/host-signal.h
  common-user/host/ppc: Implement safe-syscall.inc.S
  tcg/arm: Use register pair allocation for qemu_{ld,st}_i64
  tcg: Avoid recursion in tcg_gen_mulu2_i32

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
18 months agoMerge tag 'pull-jan-omnibus-020223-1' of https://gitlab.com/stsquad/qemu into staging
Peter Maydell [Thu, 2 Feb 2023 18:00:40 +0000 (18:00 +0000)]
Merge tag 'pull-jan-omnibus-020223-1' of https://gitlab.com/stsquad/qemu into staging

Testing, docs, semihosting and plugin updates

  - update playbooks for custom runners
  - add section timing support to gitlab
  - upgrade fedora images to 37
  - purge perl from the build system and deps
  - disable unstable tests in CI
  - improve intro, emulation and semihosting docs
  - semihosting bug fix and O_BINARY default
  - add memory-sve test
  - fix some races in qht
  - improve plugin handling of memory helpers
  - optimise plugin hooks
  - fix some plugin deadlocks
  - reduce win64-cross build time by dropping some targets

# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCgAdFiEEZoWumedRZ7yvyN81+9DbCVqeKkQFAmPb3fgACgkQ+9DbCVqe
# KkQbXAf9Eoc+PdNvafbqzH/blPjvd9ve8pJ+GcPDukNXwxP8OF/jFEJUQ1E7l9O7
# y0qV4akKCdIqVice4R5bK2CAq44Y3aut8SDf56C8E3Riha2zA2RbQWOv/zCvA3OP
# LFF+OaXZyg4JTR48HUKzh9ei2bd1+ccBSUe+xlRi59XaV5K8+5bmcZj10QKUR0lD
# 0HC5auEWWpayvd5D7Da15C7+oVY3LMCFxSdpHwbuIPPan/TRo5yqMI6ChYDKB8QD
# gdwMCL8znj2ADCTBftyBDYDAtjKVyLQidf7KdQHiSF+nmXYopS6SbsPCOMtJqCMH
# tXcKAIxs/MEntPrWTKTdtdnzotJVKw==
# =AtfN
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 02 Feb 2023 15:59:52 GMT
# gpg:                using RSA key 6685AE99E75167BCAFC8DF35FBD0DB095A9E2A44
# gpg: Good signature from "Alex Bennée (Master Work Key) <alex.bennee@linaro.org>" [full]
# Primary key fingerprint: 6685 AE99 E751 67BC AFC8  DF35 FBD0 DB09 5A9E 2A44

* tag 'pull-jan-omnibus-020223-1' of https://gitlab.com/stsquad/qemu: (36 commits)
  gitlab: cut even more from cross-win64-system build
  plugins: Iterate on cb_lists in qemu_plugin_user_exit
  cpu-exec: assert that plugin_mem_cbs is NULL after execution
  tcg: exclude non-memory effecting helpers from instrumentation
  translator: always pair plugin_gen_insn_{start, end} calls
  plugins: fix optimization in plugin_gen_disable_mem_helpers
  plugins: make qemu_plugin_user_exit's locking order consistent with fork_start's
  util/qht: use striped locks under TSAN
  thread: de-const qemu_spin_destroy
  util/qht: add missing atomic_set(hashes[i])
  cpu: free cpu->tb_jmp_cache with RCU
  tests/tcg: add memory-sve test for aarch64
  semihosting: add O_BINARY flag in host_open for NT compatibility
  semihosting: Write back semihosting data before completion callback
  docs: add an introduction to the system docs
  semihosting: add semihosting section to the docs
  docs: add a new section to outline emulation support
  docs: add hotlinks to about preface text
  MAINTAINERS: Fix the entry for tests/tcg/nios2
  gitlab: wrap up test results for custom runners
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
18 months agoMerge tag 'pull-request-2023-01-31' of https://gitlab.com/thuth/qemu into staging
Peter Maydell [Thu, 2 Feb 2023 13:42:56 +0000 (13:42 +0000)]
Merge tag 'pull-request-2023-01-31' of https://gitlab.com/thuth/qemu into staging

* qtest improvements
* Remove the deprecated OTP config of sifive_u
* Add libfdt to some of our CI jobs that were still missing it
* Use __builtin_bswap() everywhere (all compiler versions support it now)
* Deprecate the HAXM accelerator
* Document PCI devices handling on s390x
* Make Audiodev introspectable
* Improve the runtime of some CI jobs

# -----BEGIN PGP SIGNATURE-----
#
# iQJFBAABCAAvFiEEJ7iIR+7gJQEY8+q5LtnXdP5wLbUFAmPY59YRHHRodXRoQHJl
# ZGhhdC5jb20ACgkQLtnXdP5wLbXzhxAAmoq2j2sbAf2Vr9tz6Ez2p9oKNYnzUEWb
# NGXdvQMcVFKIdjvSYt5ozLC53OFIzuS74X7oHKbdLvGzez3nMCijZIbzN6vNnvd9
# HNGum4blNwHEfQcY9hr9y30Iurc7CQu6VtwGF+XXdzQZDbPz1Z4AWvtPTLcTbkxa
# PskYJfFvow/oaTHDA/7t+90cxCOixKvQMKXL5ATCtMRGnjlbOAEoPbXUB+yM24mk
# 9qp1L/8h8pvXfeXlFj+KETmu+eE5ETEOQtqc2KhQqqze2+VMKYxSX2H+sNkJBPDP
# En8Mpy+fEdefu8Jcu+M2kMLhf1f3LVf9uARhLZY4/xmOYFg+F3xzwpshnH1bs+Kw
# IzWP84uHjE77jSy/wKvYiCx2hdCDwO0G+zym67D1fPzvjzKzUNprV4OIuRzTWah3
# 6Zli5uuaLrBNjR8SJB1HDmLGKDFgToH9dzfLPtDmW8UPJGkAGcBbPKktLTe5y/4E
# del99NqpTx5SAqMmbSMRPZ/vZ7ITdfB0Av3a0GdO8j7eSPb9BOsoZOVD2/iUzab/
# P0dBuNqMM8fwywVKqcK+0CJ/npWIJvOqqlwSDqhY1A78G/uRuapOqUwsB/LWRFv5
# /1VvHfA2rv4l9o66N5jssS5/D1v5p/UBB6JvlTUvuoJMFTXa9de9XFxYxfkyiaAz
# LJl+Dh+aeWk=
# =uq7y
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 31 Jan 2023 10:05:10 GMT
# 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

* tag 'pull-request-2023-01-31' of https://gitlab.com/thuth/qemu: (27 commits)
  gitlab-ci.d/buildtest: Merge the --without-default-* jobs
  tests/qtest/display-vga-test: Add proper checks if a device is available
  gitlab-ci.d/buildtest: Remove ppc-softmmu from the clang-system job
  qapi, audio: Make introspection reflect build configuration more closely
  qapi, audio: add query-audiodev command
  docs/s390x/pcidevices: document pci devices on s390x
  tests/qtest/boot-serial-test: Constify tests[] array
  tests/qtest/vnc-display-test: Disable on Darwin
  tests/qtest/vnc-display-test: Use the 'none' machine
  tests/qtest/vnc-display-test: Suppress build warnings on Windows
  tests/tcg: Do not build/run TCG tests if TCG is disabled
  docs/about/deprecated: Mark HAXM in QEMU as deprecated
  MAINTAINERS: Abort HAXM maintenance
  qemu/bswap: Use compiler __builtin_bswap() on NetBSD
  qemu/bswap: Use compiler __builtin_bswap() on FreeBSD
  qemu/bswap: Use compiler __builtin_bswap() on Haiku
  qemu/bswap: Remove <byteswap.h> dependency
  qemu/bswap: Replace bswapXXs() by compiler __builtin_bswap()
  qemu/bswap: Replace bswapXX() by compiler __builtin_bswap()
  tests/docker/dockerfiles: Add libfdt to the i386 and to the riscv64 container
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
18 months agogitlab: cut even more from cross-win64-system build
Alex Bennée [Thu, 2 Feb 2023 13:25:27 +0000 (13:25 +0000)]
gitlab: cut even more from cross-win64-system build

This keeps hitting the time limits so cut even more from the list.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
18 months agoplugins: Iterate on cb_lists in qemu_plugin_user_exit
Richard Henderson [Tue, 24 Jan 2023 18:01:27 +0000 (18:01 +0000)]
plugins: Iterate on cb_lists in qemu_plugin_user_exit

Rather than iterate over all plugins for all events,
iterate over plugins that have registered a given event.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230117035701.168514-4-richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-36-alex.bennee@linaro.org>

18 months agocpu-exec: assert that plugin_mem_cbs is NULL after execution
Emilio Cota [Tue, 24 Jan 2023 18:01:26 +0000 (18:01 +0000)]
cpu-exec: assert that plugin_mem_cbs is NULL after execution

Fixes: #1381

Signed-off-by: Emilio Cota <cota@braap.org>
Message-Id: <20230108165107.62488-1-cota@braap.org>
[AJB: manually applied follow-up fix]
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230124180127.1881110-35-alex.bennee@linaro.org>

18 months agotcg: exclude non-memory effecting helpers from instrumentation
Emilio Cota [Tue, 24 Jan 2023 18:01:25 +0000 (18:01 +0000)]
tcg: exclude non-memory effecting helpers from instrumentation

There are actually a whole bunch of helpers that don't affect memory
that we shouldn't instrument. They are helpfully identified by the
TCG_CALL_NO_SIDE_EFFECTS flag which marks out lookup_tb_ptr as well as
a lot of the maths helpers. To avoid the string compare we introduce a
new flag for plugin internals so we skip that too.

Related: #1381
Signed-off-by: Emilio Cota <cota@braap.org>
Message-Id: <20230108164731.61469-4-cota@braap.org>
[AJB: updated to skip all no SE plugins, add flag for plugin helper]
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230124180127.1881110-34-alex.bennee@linaro.org>

18 months agotranslator: always pair plugin_gen_insn_{start, end} calls
Emilio Cota [Tue, 24 Jan 2023 18:01:24 +0000 (18:01 +0000)]
translator: always pair plugin_gen_insn_{start, end} calls

Related: #1381

Signed-off-by: Emilio Cota <cota@braap.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230108164731.61469-3-cota@braap.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-33-alex.bennee@linaro.org>

18 months agoplugins: fix optimization in plugin_gen_disable_mem_helpers
Emilio Cota [Tue, 24 Jan 2023 18:01:23 +0000 (18:01 +0000)]
plugins: fix optimization in plugin_gen_disable_mem_helpers

We were mistakenly checking tcg_ctx->plugin_insn as a canary to know
whether the TB had emitted helpers that might have accessed memory.

The problem is that tcg_ctx->plugin_insn gets updated on every
instruction in the TB, which results in us wrongly performing the
optimization (i.e. not clearing cpu->plugin_mem_cbs) way too often,
since it's not rare that the last instruction in the TB doesn't
use helpers.

Fix it by tracking a per-TB canary.

While at it, expand documentation.

Related: #1381

Signed-off-by: Emilio Cota <cota@braap.org>
Message-Id: <20230108164731.61469-2-cota@braap.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-32-alex.bennee@linaro.org>

18 months agoplugins: make qemu_plugin_user_exit's locking order consistent with fork_start's
Emilio Cota [Tue, 24 Jan 2023 18:01:22 +0000 (18:01 +0000)]
plugins: make qemu_plugin_user_exit's locking order consistent with fork_start's

To fix potential deadlocks as reported by tsan.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Emilio Cota <cota@braap.org>
Message-Id: <20230111151628.320011-6-cota@braap.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-31-alex.bennee@linaro.org>

18 months agoutil/qht: use striped locks under TSAN
Emilio Cota [Tue, 24 Jan 2023 18:01:21 +0000 (18:01 +0000)]
util/qht: use striped locks under TSAN

Fixes this tsan crash, easy to reproduce with any large enough program:

$ tests/unit/test-qht
1..2
ThreadSanitizer: CHECK failed: sanitizer_deadlock_detector.h:67 "((n_all_locks_)) < (((sizeof(all_locks_with_contexts_)/sizeof((all_locks_with_contexts_)[0]))))" (0x40, 0x40) (tid=1821568)
    #0 __tsan::CheckUnwind() ../../../../src/libsanitizer/tsan/tsan_rtl.cpp:353 (libtsan.so.2+0x90034)
    #1 __sanitizer::CheckFailed(char const*, int, char const*, unsigned long long, unsigned long long) ../../../../src/libsanitizer/sanitizer_common/sanitizer_termination.cpp:86 (libtsan.so.2+0xca555)
    #2 __sanitizer::DeadlockDetectorTLS<__sanitizer::TwoLevelBitVector<1ul, __sanitizer::BasicBitVector<unsigned long> > >::addLock(unsigned long, unsigned long, unsigned int) ../../../../src/libsanitizer/sanitizer_common/sanitizer_deadlock_detector.h:67 (libtsan.so.2+0xb3616)
    #3 __sanitizer::DeadlockDetectorTLS<__sanitizer::TwoLevelBitVector<1ul, __sanitizer::BasicBitVector<unsigned long> > >::addLock(unsigned long, unsigned long, unsigned int) ../../../../src/libsanitizer/sanitizer_common/sanitizer_deadlock_detector.h:59 (libtsan.so.2+0xb3616)
    #4 __sanitizer::DeadlockDetector<__sanitizer::TwoLevelBitVector<1ul, __sanitizer::BasicBitVector<unsigned long> > >::onLockAfter(__sanitizer::DeadlockDetectorTLS<__sanitizer::TwoLevelBitVector<1ul, __sanitizer::BasicBitVector<unsigned long> > >*, unsigned long, unsigned int) ../../../../src/libsanitizer/sanitizer_common/sanitizer_deadlock_detector.h:216 (libtsan.so.2+0xb3616)
    #5 __sanitizer::DD::MutexAfterLock(__sanitizer::DDCallback*, __sanitizer::DDMutex*, bool, bool) ../../../../src/libsanitizer/sanitizer_common/sanitizer_deadlock_detector1.cpp:169 (libtsan.so.2+0xb3616)
    #6 __tsan::MutexPostLock(__tsan::ThreadState*, unsigned long, unsigned long, unsigned int, int) ../../../../src/libsanitizer/tsan/tsan_rtl_mutex.cpp:200 (libtsan.so.2+0xa3382)
    #7 __tsan_mutex_post_lock ../../../../src/libsanitizer/tsan/tsan_interface_ann.cpp:384 (libtsan.so.2+0x76bc3)
    #8 qemu_spin_lock /home/cota/src/qemu/include/qemu/thread.h:259 (test-qht+0x44a97)
    #9 qht_map_lock_buckets ../util/qht.c:253 (test-qht+0x44a97)
    #10 do_qht_iter ../util/qht.c:809 (test-qht+0x45f33)
    #11 qht_iter ../util/qht.c:821 (test-qht+0x45f33)
    #12 iter_check ../tests/unit/test-qht.c:121 (test-qht+0xe473)
    #13 qht_do_test ../tests/unit/test-qht.c:202 (test-qht+0xe473)
    #14 qht_test ../tests/unit/test-qht.c:240 (test-qht+0xe7c1)
    #15 test_default ../tests/unit/test-qht.c:246 (test-qht+0xe828)
    #16 <null> <null> (libglib-2.0.so.0+0x7daed)
    #17 <null> <null> (libglib-2.0.so.0+0x7d80a)
    #18 <null> <null> (libglib-2.0.so.0+0x7d80a)
    #19 g_test_run_suite <null> (libglib-2.0.so.0+0x7dfe9)
    #20 g_test_run <null> (libglib-2.0.so.0+0x7e055)
    #21 main ../tests/unit/test-qht.c:259 (test-qht+0xd2c6)
    #22 __libc_start_call_main ../sysdeps/nptl/libc_start_call_main.h:58 (libc.so.6+0x29d8f)
    #23 __libc_start_main_impl ../csu/libc-start.c:392 (libc.so.6+0x29e3f)
    #24 _start <null> (test-qht+0xdb44)

Signed-off-by: Emilio Cota <cota@braap.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230111151628.320011-5-cota@braap.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-30-alex.bennee@linaro.org>

18 months agothread: de-const qemu_spin_destroy
Emilio Cota [Tue, 24 Jan 2023 18:01:20 +0000 (18:01 +0000)]
thread: de-const qemu_spin_destroy

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Emilio Cota <cota@braap.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230111151628.320011-4-cota@braap.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-29-alex.bennee@linaro.org>

18 months agoutil/qht: add missing atomic_set(hashes[i])
Emilio Cota [Tue, 24 Jan 2023 18:01:19 +0000 (18:01 +0000)]
util/qht: add missing atomic_set(hashes[i])

We forgot to add this one in "a890643958 util/qht: atomically set b->hashes".

Detected with tsan.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Emilio Cota <cota@braap.org>
Message-Id: <20230111151628.320011-3-cota@braap.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-28-alex.bennee@linaro.org>

18 months agocpu: free cpu->tb_jmp_cache with RCU
Emilio Cota [Tue, 24 Jan 2023 18:01:18 +0000 (18:01 +0000)]
cpu: free cpu->tb_jmp_cache with RCU

Fixes the appended use-after-free. The root cause is that
during tb invalidation we use CPU_FOREACH, and therefore
to safely free a vCPU we must wait for an RCU grace period
to elapse.

$ x86_64-linux-user/qemu-x86_64 tests/tcg/x86_64-linux-user/munmap-pthread
=================================================================
==1800604==ERROR: AddressSanitizer: heap-use-after-free on address 0x62d0005f7418 at pc 0x5593da6704eb bp 0x7f4961a7ac70 sp 0x7f4961a7ac60
READ of size 8 at 0x62d0005f7418 thread T2
    #0 0x5593da6704ea in tb_jmp_cache_inval_tb ../accel/tcg/tb-maint.c:244
    #1 0x5593da6704ea in do_tb_phys_invalidate ../accel/tcg/tb-maint.c:290
    #2 0x5593da670631 in tb_phys_invalidate__locked ../accel/tcg/tb-maint.c:306
    #3 0x5593da670631 in tb_invalidate_phys_page_range__locked ../accel/tcg/tb-maint.c:542
    #4 0x5593da67106d in tb_invalidate_phys_range ../accel/tcg/tb-maint.c:614
    #5 0x5593da6a64d4 in target_munmap ../linux-user/mmap.c:766
    #6 0x5593da6dba05 in do_syscall1 ../linux-user/syscall.c:10105
    #7 0x5593da6f564c in do_syscall ../linux-user/syscall.c:13329
    #8 0x5593da49e80c in cpu_loop ../linux-user/x86_64/../i386/cpu_loop.c:233
    #9 0x5593da6be28c in clone_func ../linux-user/syscall.c:6633
    #10 0x7f496231cb42 in start_thread nptl/pthread_create.c:442
    #11 0x7f49623ae9ff  (/lib/x86_64-linux-gnu/libc.so.6+0x1269ff)

0x62d0005f7418 is located 28696 bytes inside of 32768-byte region [0x62d0005f0400,0x62d0005f8400)
freed by thread T148 here:
    #0 0x7f49627b6460 in __interceptor_free ../../../../src/libsanitizer/asan/asan_malloc_linux.cpp:52
    #1 0x5593da5ac057 in cpu_exec_unrealizefn ../cpu.c:180
    #2 0x5593da81f851  (/home/cota/src/qemu/build/qemu-x86_64+0x484851)

Signed-off-by: Emilio Cota <cota@braap.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230111151628.320011-2-cota@braap.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-27-alex.bennee@linaro.org>

18 months agotests/tcg: add memory-sve test for aarch64
Alex Bennée [Tue, 24 Jan 2023 18:01:17 +0000 (18:01 +0000)]
tests/tcg: add memory-sve test for aarch64

This will be helpful in debugging problems with tracking SVE memory
accesses via the TCG plugins system.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Cc: Robert Henry <robhenry@microsoft.com>
Cc: Aaron Lindsay <aaron@os.amperecomputing.com>
Message-Id: <20230124180127.1881110-26-alex.bennee@linaro.org>

18 months agosemihosting: add O_BINARY flag in host_open for NT compatibility
Evgeny Iakovlev [Tue, 24 Jan 2023 18:01:16 +0000 (18:01 +0000)]
semihosting: add O_BINARY flag in host_open for NT compatibility

Windows open(2) implementation opens files in text mode by default and
needs a Windows-only O_BINARY flag to open files as binary. QEMU already
knows about that flag in osdep and it is defined to 0 on non-Windows,
so we can just add it to the host_flags for better compatibility.

Signed-off-by: Evgeny Iakovlev <eiakovlev@linux.microsoft.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Message-Id: <20230106102018.20520-1-eiakovlev@linux.microsoft.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-25-alex.bennee@linaro.org>

18 months agosemihosting: Write back semihosting data before completion callback
Keith Packard [Tue, 24 Jan 2023 18:01:15 +0000 (18:01 +0000)]
semihosting: Write back semihosting data before completion callback

'lock_user' allocates a host buffer to shadow a target buffer,
'unlock_user' copies that host buffer back to the target and frees the
host memory. If the completion function uses the target buffer, it
must be called after unlock_user to ensure the data are present.

This caused the arm-compatible TARGET_SYS_READC to fail as the
completion function, common_semi_readc_cb, pulled data from the target
buffer which would not have been gotten the console data.

I decided to fix all instances of this pattern instead of just the
console_read function to make things consistent and potentially fix
bugs in other cases.

Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20221012014822.1242170-1-keithp@keithp.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230124180127.1881110-24-alex.bennee@linaro.org>

18 months agodocs: add an introduction to the system docs
Alex Bennée [Tue, 24 Jan 2023 18:01:14 +0000 (18:01 +0000)]
docs: add an introduction to the system docs

Drop the frankly misleading quickstart section for a more rounded
introduction section. This new section gives an overview of the
accelerators as well as a high level introduction to some of the key
features of the emulator. We also expand on a general form for a QEMU
command line with a hopefully not too scary worked example of what
this looks like.

Acked-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Kashyap Chamarthy <kchamart@redhat.com>
Message-Id: <20230124180127.1881110-23-alex.bennee@linaro.org>

18 months agosemihosting: add semihosting section to the docs
Alex Bennée [Tue, 24 Jan 2023 18:01:13 +0000 (18:01 +0000)]
semihosting: add semihosting section to the docs

The main reason to do this is to document our O_BINARY implementation
decision somewhere. However I've also moved some of the implementation
details out of qemu-options and added links between the two. As a
bonus I've highlighted the scary warnings about host access with the
appropriate RST tags.

Acked-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-22-alex.bennee@linaro.org>

18 months agodocs: add a new section to outline emulation support
Alex Bennée [Tue, 24 Jan 2023 18:01:12 +0000 (18:01 +0000)]
docs: add a new section to outline emulation support

This affects both system and user mode emulation so we should probably
list it up front.

Acked-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-21-alex.bennee@linaro.org>

18 months agodocs: add hotlinks to about preface text
Alex Bennée [Tue, 24 Jan 2023 18:01:11 +0000 (18:01 +0000)]
docs: add hotlinks to about preface text

Make it easier to navigate the documentation.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Acked-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230124180127.1881110-20-alex.bennee@linaro.org>

18 months agoMAINTAINERS: Fix the entry for tests/tcg/nios2
Thomas Huth [Tue, 24 Jan 2023 18:01:10 +0000 (18:01 +0000)]
MAINTAINERS: Fix the entry for tests/tcg/nios2

tests/tcg/nios2/Makefile.target has accidentally been added
to the Microblaze section. Move it into the correct nios2
section instead - and while we're at it, it should also cover
the whole folder, and not only the Makefile.

Fixes: 67f80eb4d0 ("tests/tcg: enable debian-nios2-cross for test building")
Signed-off-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230119130326.2030297-1-thuth@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-19-alex.bennee@linaro.org>

18 months agogitlab: wrap up test results for custom runners
Alex Bennée [Tue, 24 Jan 2023 18:01:09 +0000 (18:01 +0000)]
gitlab: wrap up test results for custom runners

Instead of spewing the whole log to stdout lets just define them as
build artefacts so we can examine them later. Where we are running
check-tcg run it first as those tests are yet to be integrated into
meson. To avoid confusion we don't run multiple check-tcg tests at
once.

Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-18-alex.bennee@linaro.org>

18 months agotests/tcg: Use SIGKILL for timeout
Richard Henderson [Tue, 24 Jan 2023 18:01:08 +0000 (18:01 +0000)]
tests/tcg: Use SIGKILL for timeout

linux-user blocks all signals while attempting to handle guest
signals (e.g. ABRT), which means that the default TERM sent by timeout
has no effect -- KILL instead.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230117035701.168514-2-richard.henderson@linaro.org>
[AJB: expanded commit message from cover letter]
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230124180127.1881110-17-alex.bennee@linaro.org>

18 months agotests/tcg: skip the vma-pthread test on CI
Alex Bennée [Tue, 24 Jan 2023 18:01:07 +0000 (18:01 +0000)]
tests/tcg: skip the vma-pthread test on CI

We are getting a lot of failures that are not related to changes so
this could be a flaky test.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-16-alex.bennee@linaro.org>

18 months agotests/docker: drop debian-tricore-cross's partial status
Alex Bennée [Tue, 24 Jan 2023 18:01:06 +0000 (18:01 +0000)]
tests/docker: drop debian-tricore-cross's partial status

This image is perfectly capable of building QEMU, and indeed we do
that on gitlab. Drop the DOCKER_PARTIAL_IMAGES setting so we can also
test the gitlab build locally.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-15-alex.bennee@linaro.org>

18 months agotests/docker: Install flex in debian-tricore-cross
Philippe Mathieu-Daudé [Tue, 24 Jan 2023 18:01:05 +0000 (18:01 +0000)]
tests/docker: Install flex in debian-tricore-cross

When flex is not available, binutils sources default to the
'missing' script, but the current script available is not in
the format expected by the 'configure' script:

  $ ./configure
  ...
  /usr/src/binutils/missing: Unknown `--run' option
  Try `/usr/src/binutils/missing --help' for more information
  configure: WARNING: `missing' script is too old or missing
  ...
  checking for bison... bison -y
  checking for flex... no
  checking for lex... no
  checking for flex... /usr/src/binutils/missing flex

  $ make
  ...
  updating ldgram.h
  gcc -DHAVE_CONFIG_H -I. -I. -I. -D_GNU_SOURCE -I. -I. -I../bfd -I./../bfd -I./../include -I./../intl -I../intl  -w -DLOCALEDIR="\"/usr/local/share/locale\""   -W -Wall -Wstrict-prototypes -Wmissing-prototypes -w -c `test -f 'ldgram.c' || echo './'`ldgram.c
  `test -f ldlex.l || echo './'`ldlex.l
  /bin/sh: 1: ldlex.l: not found
  make[3]: *** [Makefile:662: ldlex.c] Error 127
  make[3]: Leaving directory '/usr/src/binutils/ld'
  make[2]: *** [Makefile:799: all-recursive] Error 1

By pass the 'missing' script use by directly installing 'flex'
in the container.

Reported-by: Peter Maydell <peter.maydell@linaro.org>
Suggested-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230112155643.7408-1-philmd@linaro.org>
Reviewed-by: Bastian-Koppelmann <kbastian@mail.uni-paderborn.de>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-14-alex.bennee@linaro.org>

18 months agolcitool: drop texinfo from QEMU project/dependencies
Marc-André Lureau [Tue, 24 Jan 2023 18:01:04 +0000 (18:01 +0000)]
lcitool: drop texinfo from QEMU project/dependencies

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20230110132700.833690-9-marcandre.lureau@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-13-alex.bennee@linaro.org>

18 months agolcitool: drop perl from QEMU project/dependencies
Marc-André Lureau [Tue, 24 Jan 2023 18:01:03 +0000 (18:01 +0000)]
lcitool: drop perl from QEMU project/dependencies

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230110132700.833690-8-marcandre.lureau@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-12-alex.bennee@linaro.org>

18 months agoUpdate lcitool and fedora to 37
Marc-André Lureau [Tue, 24 Jan 2023 18:01:02 +0000 (18:01 +0000)]
Update lcitool and fedora to 37

Fedora 35 is EOL.

Update to upstream lcitool, that dropped f35 and added f37.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230110132700.833690-7-marcandre.lureau@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-11-alex.bennee@linaro.org>

18 months agogitlab: add lsan suppression file to workaround tcmalloc issues
Alex Bennée [Tue, 24 Jan 2023 18:01:01 +0000 (18:01 +0000)]
gitlab: add lsan suppression file to workaround tcmalloc issues

The up-coming upgrade to Fedora 37 will bring in libtcmalloc as a
dependency of libglusterfs which confuses our fuzz run. Rather than
disable the build lets use LSAN's suppression mechanism to prevent the
job from failing.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Cc: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20230124180127.1881110-10-alex.bennee@linaro.org>

18 months agodocs: drop texinfo options
Marc-André Lureau [Tue, 24 Jan 2023 18:01:00 +0000 (18:01 +0000)]
docs: drop texinfo options

It looks like this is no longer wanted, we only build the html output.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20230110132700.833690-6-marcandre.lureau@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-9-alex.bennee@linaro.org>

18 months agomeson: replace Perl usage with Python
Marc-André Lureau [Tue, 24 Jan 2023 18:00:59 +0000 (18:00 +0000)]
meson: replace Perl usage with Python

Let's try to remove Perl usage during build time.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20230110132700.833690-5-marcandre.lureau@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-8-alex.bennee@linaro.org>

18 months ago.gitlab-ci.d/windows: do not disable opengl
Marc-André Lureau [Tue, 24 Jan 2023 18:00:58 +0000 (18:00 +0000)]
.gitlab-ci.d/windows: do not disable opengl

The previous patch should have fixed shader compilation.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230110132700.833690-3-marcandre.lureau@redhat.com>
[AJB: tweak commit message]
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-7-alex.bennee@linaro.org>

18 months agobuild-sys: fix crlf-ending C code
Marc-André Lureau [Tue, 24 Jan 2023 18:00:57 +0000 (18:00 +0000)]
build-sys: fix crlf-ending C code

On msys2, the shader-to-C script produces bad C:
./ui/shader/texture-blit-vert.h:2:5: error: missing terminating " character [-Werror]

Fix it by changing the line ending from crlf to lf, and convert the
script to Python (qemu build seems perl-free after that).

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Acked-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20230110132700.833690-2-marcandre.lureau@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-6-alex.bennee@linaro.org>

18 months agotests/unit: drop hacky race avoidance in test-io-channel-command
Alex Bennée [Tue, 24 Jan 2023 18:00:56 +0000 (18:00 +0000)]
tests/unit: drop hacky race avoidance in test-io-channel-command

We don't need to play timing games to ensure one socat wins over the
other, just create the fifo they both can use before spawning the
processes. However in the process we need to disable two tests for
Windows platforms as we don't have an abstraction for mkfifo().

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1403
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20230124180127.1881110-5-alex.bennee@linaro.org>

18 months agogitlab: just use plain --cc=clang for custom runner build
Alex Bennée [Tue, 24 Jan 2023 18:00:55 +0000 (18:00 +0000)]
gitlab: just use plain --cc=clang for custom runner build

I think this was because older Ubuntu's didn't alias clang to whatever
the latest version was. They do now so lets use that and not break.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-4-alex.bennee@linaro.org>

18 months agogitlab: add FF_SCRIPT_SECTIONS for timings
Mark Cave-Ayland [Tue, 24 Jan 2023 18:00:54 +0000 (18:00 +0000)]
gitlab: add FF_SCRIPT_SECTIONS for timings

Suggested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20230124180127.1881110-3-alex.bennee@linaro.org>

18 months agoscripts/ci: update gitlab-runner playbook to use latest runner
Alex Bennée [Tue, 24 Jan 2023 18:00:53 +0000 (18:00 +0000)]
scripts/ci: update gitlab-runner playbook to use latest runner

We were using quite and old runner on our machines and running into
issues with stalling jobs. Gitlab in the meantime now reliably provide
the latest packaged versions of the runner under a stable URL. This
update:

  - creates a per-arch subdir for builds
  - switches from binary tarballs to deb packages
  - re-uses the same binary for the secondary runner
  - updates distro check for second to 22.04

Note this script isn't fully idempotent as we end up accumulating
runners especially during testing. However we also want to be able to
run twice with different GitLab keys (e.g. project and personal) so I
think we just have to be mindful of that during testing.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Acked-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230124180127.1881110-2-alex.bennee@linaro.org>

18 months agoMerge tag 'for_upstream' of https://git.kernel.org/pub/scm/virt/kvm/mst/qemu into...
Peter Maydell [Thu, 2 Feb 2023 10:10:07 +0000 (10:10 +0000)]
Merge tag 'for_upstream' of https://git./virt/kvm/mst/qemu into staging

virtio,pc,pci: features, cleanups, fixes

lots of fixes, cleanups

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
# -----BEGIN PGP SIGNATURE-----
#
# iQFDBAABCAAtFiEEXQn9CHHI+FuUyooNKB8NuNKNVGkFAmPYJdcPHG1zdEByZWRo
# YXQuY29tAAoJECgfDbjSjVRp08cIAMYq0y++RtepDpLnPjybR0v1G4cPgZS4DXFz
# 8uc/2nkAHe1Q2lJNmk9p3YjLLloSO8yC1bmuuhUpmry9BJokYzY1r7rfXc8jd/Za
# z2FjC9LuYX+sk26NTGUxPq9mhT0p14HXyoxpnQlCweuVL0DJg1Tip6HI4oOG2LJj
# Au6Rl9keMQNqf9qVtsR1djO+8nO4ywbx6D9d2CYSKkQ3pK3uLvNds9vqU16x8wq7
# mNPqV8BIoDgW4zEOL478h6rJcL7pDQo6kAT1wfg7q1JcMMHJfW36VcBeFfskfJFg
# Pej3TEP2rg1LsGfh5XVw5Rp6FZ4K2TEyTK9cPZ9F7CzKdUrgBHU=
# =S0zd
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 30 Jan 2023 20:17:27 GMT
# 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

* tag 'for_upstream' of https://git.kernel.org/pub/scm/virt/kvm/mst/qemu: (56 commits)
  docs/pcie.txt: Replace ioh3420 with pcie-root-port
  Revert "vhost-user: Introduce nested event loop in vhost_user_read()"
  Revert "vhost-user: Monitor slave channel in vhost_user_read()"
  tests/qtest/bios-tables-test: Make the test less verbose by default
  hw: Use TYPE_PCI_BUS definition where appropriate
  vhost-user: Skip unnecessary duplicated VHOST_USER_ADD/REM_MEM_REG requests
  tests: acpi: update expected blobs
  pcihp: generate populated non-hotpluggble slot descriptions on non-hotplug path
  tests: acpi: whitelist DSDT before moving non-hotpluggble slots description from hotplug path
  tests: acpi: update expected blobs
  pcihp: acpi: ignore coldplugged bridges when composing hotpluggable slots
  tests: acpi: whitelist DSDT blobs before removing dynamic _DSM on coldplugged bridges
  tests: acpi: update expected blobs
  pcihp: acpi: decouple hotplug and generic slots description
  tests: acpi: whitelist DSDT before decoupling PCI hotplug code from basic slots description
  pcihp: isolate rule whether slot should be described in DSDT
  pci: make sure pci_bus_is_express() won't error out with "discards ‘const’ qualifier"
  pcihp: make bridge describe itself using AcpiDevAmlIfClass:build_dev_aml
  pci: acpi: wire up AcpiDevAmlIf interface to generic bridge
  x86: pcihp: acpi: prepare slot ignore rule to work with self describing bridges
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
18 months agoMerge tag 'python-pull-request' of https://gitlab.com/jsnow/qemu into staging
Peter Maydell [Wed, 1 Feb 2023 16:15:55 +0000 (16:15 +0000)]
Merge tag 'python-pull-request' of https://gitlab.com/jsnow/qemu into staging

Python

Bits and pieces, kibbles'n'bits

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEE+ber27ys35W+dsvQfe+BBqr8OQ4FAmPQlMIACgkQfe+BBqr8
# OQ5RIxAAqaG8Dx63CXa8WHMsGWc0CKTOcwTcRDw92GT3qhVkebZiNmNlZwckaU/c
# CkVunJnU5T6T2qkploysUXwdlQ+XsY4fQlACNciZeffmT2E4siNQ/4H1uPB4xca6
# 8Sgmg2VH7OF+EWwuBihY1pbe7g+sOJg9w9isRduBnLGrLbOrewGIJBNbiVzFlz5W
# 30RdvfLoUUak5qTlMT/6yl98r6fkkDmfPX653iYmpA/H/Ah+17ZJXB2XNigkqBdD
# Cp8OxtFceKQdZOqNiADJRzT3Gore4lBkPnULKwct/5U0B/tUiBdZ2YDJW8EObUMY
# zFE7giE5mCnyFSmfBmjKu8yS8zJm9NooYEjunTcodop/FDb96c3sh8376ZLamTii
# /p5WSwfo4a6DXPUTx0aiCkqpeCdPncRgwKc5TvqyKLKxQHbfjt6UZrcL6iYbe6O6
# ltBcdvfdzL41TNjS678QqiGuYkADVa/nhig3ano4msx/Tf5e0O8eMoK9bDbVS9KF
# QuONtOcut1YhnAHJp4oYN2Nimtr0t8j07iOOfc4X3+WwdbMCfR+toDM4wWVJ3u/O
# 8Phy8hinfndMXdP9Q4eeFAiJ1zuD/XkpaKoDe0gHcEvp3zMEXmHiEOdv4hFeWTQB
# ivU3oM/j2uVcHU4CSxra3B54vfLc1gudJ2yLPvhwPKoIRbJ/kbc=
# =36NA
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 25 Jan 2023 02:32:34 GMT
# gpg:                using RSA key F9B7ABDBBCACDF95BE76CBD07DEF8106AAFC390E
# gpg: Good signature from "John Snow (John Huston) <jsnow@redhat.com>" [full]
# Primary key fingerprint: FAEB 9711 A12C F475 812F  18F2 88A9 064D 1835 61EB
#      Subkey fingerprint: F9B7 ABDB BCAC DF95 BE76  CBD0 7DEF 8106 AAFC 390E

* tag 'python-pull-request' of https://gitlab.com/jsnow/qemu:
  python/qemu/machine: use socketpair() for QMP by default
  python/qmp/legacy: make QEMUMonitorProtocol accept a socket
  python/qmp/protocol: add open_with_socket()
  python/qmp: increase read buffer size
  python/machine: Fix AF_UNIX path too long on macOS
  python: QEMUMachine: enable qmp accept timeout by default
  Fix some typos

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
18 months agogitlab-ci.d/buildtest: Merge the --without-default-* jobs
Thomas Huth [Mon, 30 Jan 2023 10:44:44 +0000 (11:44 +0100)]
gitlab-ci.d/buildtest: Merge the --without-default-* jobs

Let's safe some CI minutes by merging these two jobs. We can now
also drop "--disable-capstone" since the capstone submodule has
been removed a while ago. We should rather test --disable-fdt now
to check a compilation without the "dtc" submodule (for this we
have to drop i386-softmmu from the target list unfortunately).
Additionally, the qtests with s390x and sh4 are not read for
"--without-default-devices" yet, so we can only test mips64 and
avr here now.

Message-Id: <20230130104446.1286773-5-thuth@redhat.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>
18 months agotests/qtest/display-vga-test: Add proper checks if a device is available
Thomas Huth [Mon, 30 Jan 2023 10:44:43 +0000 (11:44 +0100)]
tests/qtest/display-vga-test: Add proper checks if a device is available

display-vga-test currently tries to guess the usable VGA devices
according to the target architecture that is used for the test.
This of course does not work if QEMU has been built with the
"--without-default-devices" configure switch. To fix this, use the
qtest_has_device() function for the decision instead. This way
we can also consolidate most of the test functions into one single
function (that takes a parameter with the device name now), except
for the multihead test that tries to instantiate two devices and
thus is a little bit different.

Message-Id: <20230130104446.1286773-4-thuth@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>
18 months agogitlab-ci.d/buildtest: Remove ppc-softmmu from the clang-system job
Thomas Huth [Mon, 30 Jan 2023 10:44:41 +0000 (11:44 +0100)]
gitlab-ci.d/buildtest: Remove ppc-softmmu from the clang-system job

We are also compile-testing ppc64-softmmu with clang in the "tsan-build"
job, and ppc64-softmmu covers pretty much the same code as ppc-softmmu,
so we should not lose much test coverage here by removing ppc-softmmu
from the "clang-system" job.

Message-Id: <20230130104446.1286773-2-thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>
18 months agoqapi, audio: Make introspection reflect build configuration more closely
Daniel P. Berrangé [Mon, 23 Jan 2023 08:39:57 +0000 (09:39 +0100)]
qapi, audio: Make introspection reflect build configuration more closely

Currently the -audiodev accepts any audiodev type regardless of what is
built in to QEMU. An error only occurs later at runtime when a sound
device tries to use the audio backend.

With this change QEMU will immediately reject -audiodev args that are
not compiled into the binary. The QMP schema will also be introspectable
to identify what is compiled in.

This also helps to avoid compiling code that is not required in the
binary. Note: When building the audiodevs as modules, the patch only
compiles out code for modules that we don't build at all.

Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
[thuth: Rebase, take sndio and dbus devices into account]
Message-Id: <20230123083957.20349-3-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
18 months agoqapi, audio: add query-audiodev command
Daniel P. Berrangé [Mon, 23 Jan 2023 08:39:56 +0000 (09:39 +0100)]
qapi, audio: add query-audiodev command

Way back in QEMU 4.0, the -audiodev command line option was introduced
for configuring audio backends. This CLI option does not use QemuOpts
so it is not visible for introspection in 'query-command-line-options',
instead using the QAPI Audiodev type.  Unfortunately there is also no
QMP command that uses the Audiodev type, so it is not introspectable
with 'query-qmp-schema' either.

This introduces a 'query-audiodev' command that simply reflects back
the list of configured -audiodev command line options. This alone is
maybe not very useful by itself, but it makes Audiodev introspectable
via 'query-qmp-schema', so that libvirt (and other upper layer tools)
can discover the available audiodevs.

Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
[thuth: Update for upcoming QEMU v8.0, and use QAPI_LIST_PREPEND]
Message-Id: <20230123083957.20349-2-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
18 months agodocs/s390x/pcidevices: document pci devices on s390x
Sebastian Mitterle [Fri, 27 Jan 2023 12:33:49 +0000 (13:33 +0100)]
docs/s390x/pcidevices: document pci devices on s390x

Add some documentation about the zpci device and how
to use it with pci devices on s390x.

Used source: Cornelia Huck's blog post
https://people.redhat.com/~cohuck/2018/02/19/notes-on-pci-on-s390x.html

Signed-off-by: Sebastian Mitterle <smitterl@redhat.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Message-Id: <20230127123349.55294-1-smitterl@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
18 months agodocs/pcie.txt: Replace ioh3420 with pcie-root-port
Marcel Apfelbaum [Mon, 23 Jan 2023 17:42:05 +0000 (17:42 +0000)]
docs/pcie.txt: Replace ioh3420 with pcie-root-port

Do not mention ioh3420 in the "how to" doc.
The device still works and can be used by already
existing setups, but no need to be mentioned.

Suggested-by: Andrew Jones <drjones@redhat.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Signed-off-by: Marcel Apfelbaum <marcel@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20230123174205.683979-1-berrange@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
18 months agoRevert "vhost-user: Introduce nested event loop in vhost_user_read()"
Greg Kurz [Thu, 19 Jan 2023 17:24:24 +0000 (18:24 +0100)]
Revert "vhost-user: Introduce nested event loop in vhost_user_read()"

This reverts commit a7f523c7d114d445c5d83aecdba3efc038e5a692.

The nested event loop is broken by design. It's only user was removed.
Drop the code as well so that nobody ever tries to use it again.

I had to fix a couple of trivial conflicts around return values because
of 025faa872bcf ("vhost-user: stick to -errno error return convention").

Signed-off-by: Greg Kurz <groug@kaod.org>
Message-Id: <20230119172424.478268-3-groug@kaod.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Maxime Coquelin <maxime.coquelin@redhat.com>
18 months agoRevert "vhost-user: Monitor slave channel in vhost_user_read()"
Greg Kurz [Thu, 19 Jan 2023 17:24:23 +0000 (18:24 +0100)]
Revert "vhost-user: Monitor slave channel in vhost_user_read()"

This reverts commit db8a3772e300c1a656331a92da0785d81667dc81.

Motivation : this is breaking vhost-user with DPDK as reported in [0].

Received unexpected msg type. Expected 22 received 40
Fail to update device iotlb
Received unexpected msg type. Expected 40 received 22
Received unexpected msg type. Expected 22 received 11
Fail to update device iotlb
Received unexpected msg type. Expected 11 received 22
vhost VQ 1 ring restore failed: -71: Protocol error (71)
Received unexpected msg type. Expected 22 received 11
Fail to update device iotlb
Received unexpected msg type. Expected 11 received 22
vhost VQ 0 ring restore failed: -71: Protocol error (71)
unable to start vhost net: 71: falling back on userspace virtio

The failing sequence that leads to the first error is :
- QEMU sends a VHOST_USER_GET_STATUS (40) request to DPDK on the master
  socket
- QEMU starts a nested event loop in order to wait for the
  VHOST_USER_GET_STATUS response and to be able to process messages from
  the slave channel
- DPDK sends a couple of legitimate IOTLB miss messages on the slave
  channel
- QEMU processes each IOTLB request and sends VHOST_USER_IOTLB_MSG (22)
  updates on the master socket
- QEMU assumes to receive a response for the latest VHOST_USER_IOTLB_MSG
  but it gets the response for the VHOST_USER_GET_STATUS instead

The subsequent errors have the same root cause : the nested event loop
breaks the order by design. It lures QEMU to expect responses to the
latest message sent on the master socket to arrive first.

Since this was only needed for DAX enablement which is still not merged
upstream, just drop the code for now. A working solution will have to
be merged later on. Likely protect the master socket with a mutex
and service the slave channel with a separate thread, as discussed with
Maxime in the mail thread below.

[0] https://lore.kernel.org/qemu-devel/43145ede-89dc-280e-b953-6a2b436de395@redhat.com/

Reported-by: Yanghang Liu <yanghliu@redhat.com>
Buglink: https://bugzilla.redhat.com/show_bug.cgi?id=2155173
Signed-off-by: Greg Kurz <groug@kaod.org>
Message-Id: <20230119172424.478268-2-groug@kaod.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Acked-by: Maxime Coquelin <maxime.coquelin@redhat.com>
18 months agotests/qtest/bios-tables-test: Make the test less verbose by default
Thomas Huth [Wed, 18 Jan 2023 12:51:32 +0000 (13:51 +0100)]
tests/qtest/bios-tables-test: Make the test less verbose by default

We are facing the issues that our test logs in the gitlab CI are
too big (and thus cut off). The bios-tables-test is one of the few
qtests that prints many lines of output by default when running with
V=1, so it contributes to this problem. Almost all other qtests are
silent with V=1 and only print debug messages with V=2 and higher.
Thus let's change the bios-tables-test to behave more like the
other tests and only print the debug messages with V=2 (or higher).

Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20230118125132.1694469-1-thuth@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
18 months agohw: Use TYPE_PCI_BUS definition where appropriate
Philippe Mathieu-Daudé [Tue, 17 Jan 2023 19:30:14 +0000 (20:30 +0100)]
hw: Use TYPE_PCI_BUS definition where appropriate

Use the proper QOM type definition instead of magic string.
This also helps during eventual refactor while using git-grep.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230117193014.83502-1-philmd@linaro.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
18 months agovhost-user: Skip unnecessary duplicated VHOST_USER_ADD/REM_MEM_REG requests
Minghao Yuan [Mon, 23 Jan 2023 12:21:19 +0000 (20:21 +0800)]
vhost-user: Skip unnecessary duplicated VHOST_USER_ADD/REM_MEM_REG requests

The VHOST_USER_ADD/REM_MEM_REG requests should be categorized into
non-vring specific messages, and should be sent only once.

Signed-off-by: Minghao Yuan <yuanmh12@chinatelecom.cn>
Message-Id: <20230123122119.194347-1-yuanmh12@chinatelecom.cn>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
18 months agotests: acpi: update expected blobs
Igor Mammedov [Thu, 12 Jan 2023 14:03:12 +0000 (15:03 +0100)]
tests: acpi: update expected blobs

Expected change removal of dynamic _DSM AML for non-hotpluggable
hots-bridge, storage, isa bridge devices from PC machine blobs:

  -            Scope (S00)
  -            {
  -                Name (ASUN, Zero)
  -                Method (_DSM, 4, Serialized)  // _DSM: Device-Specific Method
  -                {
  -                    Local0 = Package (0x02)
  -                        {
  -                            BSEL,
  -                            ASUN
  -                        }
  -                    Return (PDSM (Arg0, Arg1, Arg2, Arg3, Local0))
  -                }
  -            }
  -
  -            Scope (S08)
  -            {
  -                Name (ASUN, One)
  -                Method (_DSM, 4, Serialized)  // _DSM: Device-Specific Method
  -                {
  -                    Local0 = Package (0x02)
  -                        {
  -                            BSEL,
  -                            ASUN
  -                        }
  -                    Return (PDSM (Arg0, Arg1, Arg2, Arg3, Local0))
  -                }
  -            }
  -
  -            Scope (S10)
  -            {
  -                Name (ASUN, 0x02)
  -                Method (_DSM, 4, Serialized)  // _DSM: Device-Specific Method
  -                {
  -                    Local0 = Package (0x02)
  -                        {
  -                            BSEL,
  -                            ASUN
  -                        }
  -                    Return (PDSM (Arg0, Arg1, Arg2, Arg3, Local0))
  -                }
  -            }

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-41-imammedo@redhat.com>

18 months agopcihp: generate populated non-hotpluggble slot descriptions on non-hotplug path
Igor Mammedov [Thu, 12 Jan 2023 14:03:11 +0000 (15:03 +0100)]
pcihp: generate populated non-hotpluggble slot descriptions on non-hotplug path

Generating slots descriptions populated by non-hotpluggable devices
is akward at best and complicates hotplug path (build_append_pcihp_slots)
needlessly, and builds only dynamic _DSM for such slots which is overlkill.
Clean it up and let non-hotplug path (build_append_pci_bus_devices)
to handle that task.

Such clean up effectively drops dynamic _DSM methods on non-hotpluggable
slots (even though bus itself is hotpluggable), but in practice it
affects only built-in devices (ide controllers/various bridges) that don't
use acpi-index anyways so effectively it doesn't matter (NICs are hotpluggble).

Follow up series will add static _DSM for non-hotpluggble devices/buses
that will not depend on ACPI PCI hotplug at all, and potentially would
allows us to reuse non-hotplug path elsewhere (PBX/microvm/arm-virt),
including new support for acpi-index for non-hotpluggable devices.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-40-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
18 months agotests: acpi: whitelist DSDT before moving non-hotpluggble slots description from...
Igor Mammedov [Thu, 12 Jan 2023 14:03:10 +0000 (15:03 +0100)]
tests: acpi: whitelist DSDT before moving non-hotpluggble slots description from hotplug path

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-39-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
18 months agotests: acpi: update expected blobs
Igor Mammedov [Thu, 12 Jan 2023 14:03:09 +0000 (15:03 +0100)]
tests: acpi: update expected blobs

expected change is removal of dynamic _DSM bits from slots populated
by coldplugged bridges (something like):

    -            Scope (S18)
    -            {
    -                Name (ASUN, 0x03)
    -                Method (_DSM, 4, Serialized)  // _DSM: Device-Specific Method
    -                {
    -                    Local0 = Package (0x02)
    -                        {
    -                            BSEL,
    -                            ASUN
    -                        }
    -                    Return (PDSM (Arg0, Arg1, Arg2, Arg3, Local0))
    -                }
    -            }

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-38-imammedo@redhat.com>

18 months agopcihp: acpi: ignore coldplugged bridges when composing hotpluggable slots
Igor Mammedov [Thu, 12 Jan 2023 14:03:08 +0000 (15:03 +0100)]
pcihp: acpi: ignore coldplugged bridges when composing hotpluggable slots

coldplugged bridges are not unpluggable, so there is no need
to describe slots where they are plugged as hotpluggable. To
that effect we have a condition that marks slot as non-hotpluggable
if it's populated by coldplugged bridge and prevents generation
_SUN/_EJ0 objects for it. That leaves dynamic _DSM method on
such slot (which also depends on BSEL and pcihp hardware).
This _DSM method provides only dynamic acpi-index support so far,
which is not actually used/supported by linux kernel for bridges
and it's doubtful there will be need for it at all.

So it's rather pointless to generate acpi-index related AML
for bridges and we can simplify hotplug slots generator a bit
more by completely ignoring coldplugged bridges on hotplug path.

Another point in favor of dropping dynamic _DSM support, is
that we can replace it with static _DSM if necessary since
a slot with bridge can't change during VM runtime and without
any dependency on ACPI PCI hotplug at that.
Later I plan to implement bridge specific static _DSM
   PCI Firmware Specification 3.2
   4.6.5.  _DSM for Ignoring PCI Boot Configurations
part of spec, to fix longstanding issue with fixed IO/MEM
resource assignment that often leads to hotplugged device
being in-operational within the guest due limited IO/MEM
windows programmed on bridge at boot time.

Expected change when coldplugged bridge is ignored by hotplug
code, should look like:
-            Scope (S18)
-            {
-                Name (ASUN, 0x03)
-                Method (_DSM, 4, Serialized)  // _DSM: Device-Specific Method
-                {
-                    Local0 = Package (0x02)
-                        {
-                            BSEL,
-                            ASUN
-                        }
-                    Return (PDSM (Arg0, Arg1, Arg2, Arg3, Local0))
-                }
-            }

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-37-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
18 months agotests: acpi: whitelist DSDT blobs before removing dynamic _DSM on coldplugged bridges
Igor Mammedov [Thu, 12 Jan 2023 14:03:07 +0000 (15:03 +0100)]
tests: acpi: whitelist DSDT blobs before removing dynamic _DSM on coldplugged bridges

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-36-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
18 months agotests: acpi: update expected blobs
Igor Mammedov [Thu, 12 Jan 2023 14:03:06 +0000 (15:03 +0100)]
tests: acpi: update expected blobs

Expected change for non-populated slots is that
thay are moved after non-hotpluggable PCI tree description.

And expected change for hotplug capable populated slots is:
  - ...
  +                Name (BSEL, 0x03)
  +                Scope (S00)
  +                {
  +                    Name (ASUN, Zero)
  +                    Method (_DSM, 4, Serialized)  // _DSM: Device-Specific Method
  +                    {
  +                        Local0 = Package (0x02)
  +                            {
  +                                BSEL,
  +                                ASUN
  +                            }
  +                        Return (PDSM (Arg0, Arg1, Arg2, Arg3, Local0))
  +                    }
  [ ... other hotplug depended bits ]
  +                }

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-35-imammedo@redhat.com>

18 months agopcihp: acpi: decouple hotplug and generic slots description
Igor Mammedov [Thu, 12 Jan 2023 14:03:05 +0000 (15:03 +0100)]
pcihp: acpi: decouple hotplug and generic slots description

Split build_append_pci_bus_devices() onto generic part that builds
AML descriptions only for populated slots which is applicable to
both hotplug disabled and enabled bridges. And a hotplug only
part that complements generic AML with hotplug depended bits
(that depend on BSEL), like _SUN/_EJ0 entries, dynamic _DSM.

Hotplug part, will generate full 'Device' descriptors for
non-populated slots (like it used to be) and complementary
'Scope' descriptors for populated slots that are hotplug capable.
i.e. something like this:
  - ...
  +                Name (BSEL, 0x03)
  +                Scope (S00)
  +                {
  +                    Name (ASUN, Zero)
  +                    Method (_DSM, 4, Serialized)  // _DSM: Device-Specific Method
  +                    {
  +                        Local0 = Package (0x02)
  +                            {
  +                                BSEL,
  +                                ASUN
  +                            }
  +                        Return (PDSM (Arg0, Arg1, Arg2, Arg3, Local0))
  +                    }
  +  [ ... other hotplug depended bits ]
  +                }

While generic build_append_pci_bus_devices() still calls hotplug part at
its end it doesn't really depend on any hotplug bits anymore and later
both could be completely separated when it's necessary.

Main benefit though is that both build_append_pci_bus_devices() and
build_append_pcihp_slots() become more readable and it makes easier
to modify them with less risk of affecting another part. Also it opens
possibility to re-use generic part elsewhere (microvm, arm/virt).

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-34-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
18 months agotests: acpi: whitelist DSDT before decoupling PCI hotplug code from basic slots descr...
Igor Mammedov [Thu, 12 Jan 2023 14:03:04 +0000 (15:03 +0100)]
tests: acpi: whitelist DSDT before decoupling PCI hotplug code from basic slots description

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-33-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
18 months agopcihp: isolate rule whether slot should be described in DSDT
Igor Mammedov [Thu, 12 Jan 2023 14:03:03 +0000 (15:03 +0100)]
pcihp: isolate rule whether slot should be described in DSDT

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-32-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
18 months agopci: make sure pci_bus_is_express() won't error out with "discards ‘const’ qualifier"
Igor Mammedov [Thu, 12 Jan 2023 14:03:02 +0000 (15:03 +0100)]
pci: make sure pci_bus_is_express() won't error out with "discards ‘const’ qualifier"

function doesn't need RW aceess to passed in bus pointer,
make it const.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-31-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
18 months agopcihp: make bridge describe itself using AcpiDevAmlIfClass:build_dev_aml
Igor Mammedov [Thu, 12 Jan 2023 14:03:01 +0000 (15:03 +0100)]
pcihp: make bridge describe itself using AcpiDevAmlIfClass:build_dev_aml

simplify build_append_pci_bus_devices() a bit by handling bridge
specific logic in bridge dedicated AcpiDevAmlIfClass::build_dev_aml
callback.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-30-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
18 months agopci: acpi: wire up AcpiDevAmlIf interface to generic bridge
Igor Mammedov [Thu, 12 Jan 2023 14:03:00 +0000 (15:03 +0100)]
pci: acpi: wire up AcpiDevAmlIf interface to generic bridge

... so that the concrete impl. won't has to duplicate it
every time. By default it doesn't do anything unless leaf class
defines and sets AcpiDevAmlIfClass::build_dev_aml handler.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-29-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
18 months agox86: pcihp: acpi: prepare slot ignore rule to work with self describing bridges
Igor Mammedov [Thu, 12 Jan 2023 14:02:59 +0000 (15:02 +0100)]
x86: pcihp: acpi: prepare slot ignore rule to work with self describing bridges

Before switching pci bridges to AcpiDevAmlIf interface, ensure that
ignored slots are handled correctly.
(existing rule works but only if bridge doesn't have AcpiDevAmlIf interface).
While at it rewrite related comments to be less confusing (hopefully).

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-28-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
18 months agotests: acpi: update expected blobs
Igor Mammedov [Thu, 12 Jan 2023 14:02:58 +0000 (15:02 +0100)]
tests: acpi: update expected blobs

previous commit added endpoint devices to bridge testcases,
which exposes extra non-hotpluggable slot in DSDT on bus where
hotplug is not available.
It should look like this (numbers may vary):

+            Device (S28)
+            {
+                Name (_ADR, 0x00050000)  // _ADR: Address
+            }

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-27-imammedo@redhat.com>

18 months agotests: acpi: add endpoint devices to bridges
Igor Mammedov [Thu, 12 Jan 2023 14:02:57 +0000 (15:02 +0100)]
tests: acpi: add endpoint devices to bridges

to make sure that they are enumerated or ignored as expected

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-26-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
18 months agowhitelist DSDT before adding endpoint devices to bridge testcases
Igor Mammedov [Thu, 12 Jan 2023 14:02:56 +0000 (15:02 +0100)]
whitelist DSDT before adding endpoint devices to bridge testcases

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-25-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
18 months agotests: acpi: update expected blobs
Igor Mammedov [Thu, 12 Jan 2023 14:02:55 +0000 (15:02 +0100)]
tests: acpi: update expected blobs

Expected changes:
 * pc/bridge testcase due to
   ("pcihp: compose PCNT callchain right before its user _GPE._E01")
  ...
  +    Scope (\_SB.PCI0)
  +    {
  +        Scope (S18)
  +        {
  +            Scope (S08)
  +            {
  +                Method (PCNT, 0, NotSerialized)
  +                {
  +                    BNUM = 0x02
  +                    DVNT (PCIU, One)
  +                    DVNT (PCID, 0x03)
  +                }
  +            }

               Method (PCNT, 0, NotSerialized)
               {
  -                BNUM = Zero
  +                BNUM = One
                   DVNT (PCIU, One)
                   DVNT (PCID, 0x03)
  -                ^S18.PCNT ()
  +                ^S08.PCNT ()
               }
           }
  +
  +        Method (PCNT, 0, NotSerialized)
  +        {
  +            BNUM = Zero
  +            DVNT (PCIU, One)
  +            DVNT (PCID, 0x03)
  +            ^S18.PCNT ()
  +        }
       }

     Scope (_GPE)

 * due to ("pcihp: do not put empty PCNT in DSDT") in the most Q35 tests
  ...
               {
                   Name (_ADR, 0x001F0003)  // _ADR: Address
               }
  -
  -            Method (PCNT, 0, NotSerialized)
  -            {
  -            }
           }
       }

  ...
       {
           Method (_E01, 0, NotSerialized)  // _Exx: Edge-Triggered GPE
           {
  -            Acquire (\_SB.PCI0.BLCK, 0xFFFF)
  -            \_SB.PCI0.PCNT ()
  -            Release (\_SB.PCI0.BLCK)
           }
       }

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-24-imammedo@redhat.com>

18 months agopcihp: do not put empty PCNT in DSDT
Igor Mammedov [Thu, 12 Jan 2023 14:02:54 +0000 (15:02 +0100)]
pcihp: do not put empty PCNT in DSDT

count number of PCNT methods that actually call Notify
and if there aren't any, drop PCNT altogether.
It mostly affects 'Q35' tests where there is no root-ports
/bridges attached and 'PC' machine when ACPI PCI hotplug is
completely disabled.

Expected ASL change:

-            Method (PCNT, 0, NotSerialized)
-            {
-            }
...
         Method (_E01, 0, NotSerialized)  // _Exx: Edge-Triggered GPE
         {
-            Acquire (\_SB.PCI0.BLCK, 0xFFFF)
-            \_SB.PCI0.PCNT ()
-            Release (\_SB.PCI0.BLCK)
         }

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-23-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
18 months agopcihp: compose PCNT callchain right before its user _GPE._E01
Igor Mammedov [Thu, 12 Jan 2023 14:02:53 +0000 (15:02 +0100)]
pcihp: compose PCNT callchain right before its user _GPE._E01

it's a stepping stone to making build_append_pci_bus_devices() suitable
for AcpiDevAmlIfClass:build_dev_aml callback and lets further simplify
it by separating PCNT generation from slots descriptions.

It also makes PCNT callchain ASL much more readable since callchain
not longer cluttered by slots descriptors.

Plus, move will let next patch easily drop empty PCNT (pc/q35)
when there is nothing hotpluggable.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-22-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>