tests/tcg: Fix target-specific Makefile variables path for user-mode

Commit 812b31d3f91 refactor missed to update this path.

Fixes: 812b31d3f91 ("configs: rename default-configs to configs and reorganise")
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20211226001541.3807919-1-f4bug@amsat.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

KVM: x86: ignore interrupt_bitmap field of KVM_GET/SET_SREGS

This is unnecessary, because the interrupt would be retrieved and queued
anyway by KVM_GET_VCPU_EVENTS and KVM_SET_VCPU_EVENTS respectively,
and it makes the flow more similar to the one for KVM_GET/SET_SREGS2.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

KVM: use KVM_{GET|SET}_SREGS2 when supported.

This allows to make PDPTRs part of the migration
stream and thus not reload them after migration which
is against X86 spec.

Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Message-Id: <20211101132300.192584-2-mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

meson: add comments in the target-specific flags section

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

configure, meson: move config-poison.h to meson

This ensures that the file is regenerated properly whenever config-target.h
or config-devices.h files change.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

meson: build contrib/ executables after generated headers

This will be needed as soon as config-poison.h moves from configure to
a meson custom_target (which is built at "ninja" time).

Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

configure: move non-command-line variables away from command-line parsing section

This makes it easier to identify candidates for moving to Meson.

Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

configure: parse --enable/--disable-strip automatically, flip default

Always include the STRIP variable in config-host.mak (it's only used
by the s390-ccw firmware build, and it adds a default if configure
omitted it), and use meson-buildoptions.sh to turn
--enable/--disable-strip into -Dstrip.

The default is now not to strip the binaries like for almost every other
package that has a configure script.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

configure, makefile: remove traces of really old files

These files have been removed for more than year in the best
case, or for more than ten years for some really old TCG files.
Remove any traces of it.

Acked-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

configure: do not set bsd_user/linux_user early

Similar to other optional features, leave the variables empty and compute
the actual value later.  Use the existence of include or source directories
to detect whether an OS or CPU supports respectively bsd-user and linux-user.

For now, BSD user-mode emulation is buildable even on TCI-only
architectures.  This probably will change once safe signals are
brought over from linux-user.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

configure: simplify creation of plugin symbol list

--dynamic-list is present on all supported ELF (not Windows or Darwin)
platforms, since it dates back to 2006; -exported_symbols_list is
likewise present on all supported versions of macOS.  Do not bother
doing a functional test in configure.

Remove the file creation from configure as well: for Darwin, move the
the creation of the Darwin-formatted symbols to meson; for ELF, use the
file in the source path directly and switch from -Wl, to -Xlinker to
not break weird paths that include a comma.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

block/file-posix: Simplify the XFS_IOC_DIOINFO handling

The handling for the XFS_IOC_DIOINFO ioctl is currently quite excessive:
This is not a "real" feature like the other features that we provide with
the "--enable-xxx" and "--disable-xxx" switches for the configure script,
since this does not influence lots of code (it's only about one call to
xfsctl() in file-posix.c), so people don't gain much with the ability to
disable this with "--disable-xfsctl".
It's also unfortunate that the ioctl will be disabled on Linux in case
the user did not install the right xfsprogs-devel package before running
configure. Thus let's simplify this by providing the ioctl definition
on our own, so we can completely get rid of the header dependency and
thus the related code in the configure script.

Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20211215125824.250091-1-thuth@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

meson: cleanup common-user/ build

It is not necessary to have a separate static_library just for common_user
files; using the one that already covers the rest of common_ss is enough
unless you need to reuse some source files between emulators and tests.
Just place common files for all user-mode emulators in common_ss,
similar to what is already done for softmmu_ss in full system emulators.

The only disadvantage is that the include_directories under bsd-user/include/
and linux-user/include/ are now enabled for all targets rather than only
user mode emulators.  This however is not different from how include/sysemu/
is available when building user mode emulators.

Tested-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

user: move common-user includes to a subdirectory of {bsd,linux}-user/

Avoid polluting the compilation of common-user/ with local include files;
making an include file available to common-user/ should be a deliberate
decision in order to keep a clear interface that can be used by both
bsd-user/ and linux-user/.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

meson: reuse common_user_inc when building files specific to user-mode emulators

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

linux-user: Fix clang warning for nios2-linux-user code

The clang in Ubuntu 18.04 (10.0.0-4ubuntu1) produces a warning
on the code added in commit f5ef0e518d03 where we use a
shifted expression in a boolean context:

../../linux-user/elfload.c:2423:16: error: converting the result of '<<' to a boolean always evaluates to true [-Werror,-Wtautological-constant-compare]
    } else if (LO_COMMPAGE) {
               ^
../../linux-user/elfload.c:1102:22: note: expanded from macro 'LO_COMMPAGE'
#define LO_COMMPAGE  TARGET_PAGE_SIZE
                     ^
/mnt/nvmedisk/linaro/qemu-from-laptop/qemu/include/exec/cpu-all.h:231:31: note: expanded from macro 'TARGET_PAGE_SIZE'
#define TARGET_PAGE_SIZE   (1 << TARGET_PAGE_BITS)
                              ^
1 error generated.

The warning is bogus because whether LO_COMMPAGE is zero or not
depends on compile-time ifdefs; shut the compiler up by adding
an explicit comparison to zero.

Fixes: f5ef0e518d0331 ("linux-user/nios2: Map a real kuser page")
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-id: 20220111082900.3341274-1-peter.maydell@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

Merge remote-tracking branch 'remotes/jsnow-gitlab/tags/python-pull-request' into staging

Python pull request

Fixes for the tests that broke during vacation, plus a simple syntax fix
for a python script.

# gpg: Signature made Mon 10 Jan 2022 23:24:47 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

* remotes/jsnow-gitlab/tags/python-pull-request:
  simplebench: Fix Python syntax error (reported by LGTM)
  python: update type hints for mypy 0.930
  Python/aqmp: fix type definitions for mypy 0.920
  python/aqmp: use absolute import statement

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

Merge remote-tracking branch 'remotes/philmd/tags/sdmmc-20220108' into staging

SD/MMC patches queue

- Add SDHC support for SD card SPI-mode (Frank Chang)

# gpg: Signature made Sat 08 Jan 2022 21:56:02 GMT
# gpg:                using RSA key FAABE75E12917221DCFD6BB2E3E32C2CDEADC0DE
# gpg: Good signature from "Philippe Mathieu-Daudé (F4BUG) <f4bug@amsat.org>" [unknown]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: FAAB E75E 1291 7221 DCFD  6BB2 E3E3 2C2C DEAD C0DE

* remotes/philmd/tags/sdmmc-20220108:
  hw/sd: Add SDHC support for SD card SPI-mode
  hw/sd/sdcard: Rename Write Protect Group variables

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

Merge remote-tracking branch 'remotes/mst/tags/for_upstream' into staging

virtio: revert config interrupt changes

Lots of fallout from config interrupt changes. Author wants to rework
the patches. Let's revert quickly so others don't suffer meanwhile.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
# gpg: Signature made Mon 10 Jan 2022 21:03:44 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

* remotes/mst/tags/for_upstream:
  Revert "virtio: introduce macro IRTIO_CONFIG_IRQ_IDX"
  Revert "virtio-pci: decouple notifier from interrupt process"
  Revert "virtio-pci: decouple the single vector from the interrupt process"
  Revert "vhost: introduce new VhostOps vhost_set_config_call"
  Revert "vhost-vdpa: add support for config interrupt"
  Revert "virtio: add support for configure interrupt"
  Revert "vhost: add support for configure interrupt"
  Revert "virtio-net: add support for configure interrupt"
  Revert "virtio-mmio: add support for configure interrupt"
  Revert "virtio-pci: add support for configure interrupt"

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

simplebench: Fix Python syntax error (reported by LGTM)

Fixes: b2fcb0c5754c2554b8406376e99a75e9e0a6b7bd
Signed-off-by: Stefan Weil <sw@weilnetz.de>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: John Snow <jsnow@redhat.com>
Message-id: 20220107153019.504124-1-sw@weilnetz.de
Signed-off-by: John Snow <jsnow@redhat.com>

python: update type hints for mypy 0.930

Mypy 0.930, released Dec 22, changes the way argparse objects are
considered. Crafting a definition that works under Python 3.6 and an
older mypy alongside newer versions simultaneously is ... difficult,
so... eh. Stub it out with an 'Any' definition to get the CI moving
again.

Oh well.

Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Beraldo Leal <bleal@redhat.com>
Message-id: 20220110191349.1841027-4-jsnow@redhat.com
Signed-off-by: John Snow <jsnow@redhat.com>

Python/aqmp: fix type definitions for mypy 0.920

0.920 (Released 2021-12-15) is not entirely happy with the
way that I was defining _FutureT:

qemu/aqmp/protocol.py:601: error: Item "object" of the upper bound
"Optional[Future[Any]]" of type variable "_FutureT" has no attribute
"done"

Update it with something a little mechanically simpler that works better
across a wider array of mypy versions.

Signed-off-by: John Snow <jsnow@redhat.com>
Message-id: 20220110191349.1841027-3-jsnow@redhat.com
Signed-off-by: John Snow <jsnow@redhat.com>

python/aqmp: use absolute import statement

pylint's dependency astroid appears to have bugs in 2.9.1 and 2.9.2 (Dec
31 and Jan 3) that appear to erroneously expect the qemu namespace to
have an __init__.py file. astroid 2.9.3 (Jan 9) avoids that problem, but
appears to not understand a relative import within a namespace package.

Update the relative import - it was worth changing anyway, because these
packages will eventually be packaged and distributed separately.

Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Beraldo Leal <bleal@redhat.com>
Message-id: 20220110191349.1841027-2-jsnow@redhat.com
Signed-off-by: John Snow <jsnow@redhat.com>

Revert "virtio: introduce macro IRTIO_CONFIG_IRQ_IDX"

This reverts commit bf1d85c166c19af95dbd27b1faba1d2909732323.

Fixes: bf1d85c166 ("virtio: introduce macro IRTIO_CONFIG_IRQ_IDX")
Cc: "Cindy Lu" <lulu@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

Revert "virtio-pci: decouple notifier from interrupt process"

This reverts commit e3480ef81f6fb61cc9c04e3b5be8b7e84484fc05.

Fixes: e3480ef81f ("virtio-pci: decouple notifier from interrupt process")
Cc: "Cindy Lu" <lulu@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

Revert "virtio-pci: decouple the single vector from the interrupt process"

This reverts commit 316011b8a74e777eb3ba03171cd701a291c28867.

Fixes: 316011b8a7 ("virtio-pci: decouple the single vector from the interrupt process")
Cc: "Cindy Lu" <lulu@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

Revert "vhost: introduce new VhostOps vhost_set_config_call"

This reverts commit 88062372340d33090b7f089ed3b1a00f530a8914.

Fixes: 8806237234 ("vhost: introduce new VhostOps vhost_set_config_call")
Cc: "Cindy Lu" <lulu@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

Revert "vhost-vdpa: add support for config interrupt"

This reverts commit 634f7c89fbd78f57d00d5d6b39c0ade9df1fe27f.

Fixes: 634f7c89fb ("vhost-vdpa: add support for config interrupt")
Cc: "Cindy Lu" <lulu@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

Revert "virtio: add support for configure interrupt"

This reverts commit 081f864f56307551f59c5e934e3f30a7290d0faa.

Fixes: 081f864f56 ("virtio: add support for configure interrupt")
Cc: "Cindy Lu" <lulu@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

Revert "vhost: add support for configure interrupt"

This reverts commit f7220a7ce21604a4bc6260ccca4dc9068c1f27f2.

Fixes: f7220a7ce2 ("vhost: add support for configure interrupt")
Cc: "Cindy Lu" <lulu@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

Revert "virtio-net: add support for configure interrupt"

This reverts commit 497679d51087090d5a22fd265d1b96cf92d49d9d.

Fixes: 497679d510 ("virtio-net: add support for configure interrupt")
Cc: "Cindy Lu" <lulu@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

Revert "virtio-mmio: add support for configure interrupt"

This reverts commit d48185f1a40d4e4ed2fa2873a42b2a5eb8748256.

Fixes: d48185f1a4 ("virtio-mmio: add support for configure interrupt")
Cc: "Cindy Lu" <lulu@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

Revert "virtio-pci: add support for configure interrupt"

This reverts commit d5d24d859c3957ea1674d0e102f96439cdbfe93a.

Fixes: d5d24d859c ("virtio-pci: add support for configure interrupt")
Cc: "Cindy Lu" <lulu@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

Merge remote-tracking branch 'remotes/vivier/tags/m68k-for-7.0-pull-request' into staging

M68k pull request 20220109

Add virt compat machine type for 7.0
fix q800 -bios parameter
fix VRAM refresh
fix M68K_FEATURE_UNALIGNED_DATA feature

# gpg: Signature made Sun 09 Jan 2022 11:41:19 GMT
# gpg:                using RSA key CD2F75DDC8E3A4DC2E4F5173F30C38BD3F2FBE3C
# gpg:                issuer "laurent@vivier.eu"
# gpg: Good signature from "Laurent Vivier <lvivier@redhat.com>" [full]
# gpg:                 aka "Laurent Vivier <laurent@vivier.eu>" [full]
# gpg:                 aka "Laurent Vivier (Red Hat) <lvivier@redhat.com>" [full]
# Primary key fingerprint: CD2F 75DD C8E3 A4DC 2E4F  5173 F30C 38BD 3F2F BE3C

* remotes/vivier/tags/m68k-for-7.0-pull-request:
  target/m68k: don't word align SP in stack frame if M68K_FEATURE_UNALIGNED_DATA feature enabled
  macfb: fix VRAM dirty memory region logging
  q800: fix segfault with invalid MacROM
  hw: m68k: Add virt compat machine type for 7.0

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/m68k: don't word align SP in stack frame if M68K_FEATURE_UNALIGNED_DATA feature enabled

Commit a9431a03f7 ("target/m68k: add M68K_FEATURE_UNALIGNED_DATA feature") added
a new feature for processors from the 68020 onwards which do not require data
accesses to be word aligned.

Unfortunately the original commit missed an additional case whereby the SP is
still word aligned when setting up an additional format 1 stack frame so add the
necessary M68K_FEATURE_UNALIGNED_DATA feature guard.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Fixes: a9431a03f7 ("target/m68k: add M68K_FEATURE_UNALIGNED_DATA feature")
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20220108180453.18680-1-mark.cave-ayland@ilande.co.uk>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>

macfb: fix VRAM dirty memory region logging

The macfb VRAM memory region was configured with coalescing rather than dirty
memory logging enabled, causing some areas of the screen not to redraw after
a full screen update.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Fixes: 8ac919a065 ("hw/m68k: add Nubus macfb video card")
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20220108164147.30813-1-mark.cave-ayland@ilande.co.uk>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>

q800: fix segfault with invalid MacROM

"qemu-system-m68k -M q800 -bios /dev/null" crashes with a segfault
in q800_init().
This happens because the code doesn't check that rom_ptr() returned
a non-NULL pointer .

To avoid NULL pointer, don't allow 0 sized file and use bios_size with
rom_ptr().

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/756
Reported-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Message-Id: <20220107105049.961489-1-laurent@vivier.eu>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>

hw: m68k: Add virt compat machine type for 7.0

Signed-off-by: Laurent Vivier <laurent@vivier.eu>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Message-Id: <20211218114340.1856757-1-laurent@vivier.eu>

Merge tag 'bsd-user-arm-pull-request' of gitlab.com:bsdimp/qemu into staging

bsd-user: arm (32-bit) support

This series of patches brings in 32-bit arm support for bsd-user.  It implements
all the bits needed to do image activation, signal handling, stack management
and threading. This allows us to get to the "Hello World" level. The arm and x86
code are now the same as in the bsd-user fork. For full context, the fork is at
https://github.com/qemu-bsd-user/qemu-bsd-user/tree/blitz (though the the recent
sig{bus,segv} needed updates are incomplete).

v5 changes:
   o Moved to using the CPUArchState typedef and move
     set_sigtramp_args, get_mcontext, set_mcontext, and
     get_ucontext_sigreturn prototypes to
     bsd-user/freebsd/target_os_ucontext.h
   o Fix issues with arm's set_mcontext related to masking
     and remove an unnecessary check.

We're down to only one hunk needing review:
    bsd-user/arm/target_arch_signal.c: arm set_mcontext

Warnings that should be ignored:
   o make checkpatch has a couple of complaints about the comments for the
     signal trampoline, since it's a false positive IMHO.
WARNING: Block comments use a leading /* on a separate line
+    /* 8 */ sys_sigreturn,
WARNING: Block comments use a leading /* on a separate line
+    /* 9 */ sys_exit

# gpg: Signature made Fri 07 Jan 2022 11:36:37 PM PST
# gpg:                using RSA key 2035F894B00AA3CF7CCDE1B76C1CD1287DB01100
# gpg: Good signature from "Warner Losh <wlosh@netflix.com>" [unknown]
# gpg:                 aka "Warner Losh <imp@bsdimp.com>" [unknown]
# gpg:                 aka "Warner Losh <imp@freebsd.org>" [unknown]
# gpg:                 aka "Warner Losh <imp@village.org>" [unknown]
# gpg:                 aka "Warner Losh <wlosh@bsdimp.com>" [unknown]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 2035 F894 B00A A3CF 7CCD  E1B7 6C1C D128 7DB0 1100

* tag 'bsd-user-arm-pull-request' of gitlab.com:bsdimp/qemu: (37 commits)
  bsd-user: add arm target build
  bsd-user/freebsd/target_os_ucontext.h: Require TARGET_*CONTEXT_SIZE
  bsd-user/arm/signal.c: arm get_ucontext_sigreturn
  bsd-user/arm/signal.c: arm set_mcontext
  bsd-user/arm/signal.c: arm get_mcontext
  bsd-user/arm/signal.c: arm set_sigtramp_args
  bsd-user/arm/target_arch_signal.h: Define size of *context_t
  bsd-user/arm/target_arch_signal.h: arm machine context and trapframe for signals
  bsd-user/arm/target_arch_signal.h: arm specific signal registers and stack
  bsd-user/arm/target_arch_elf.h: arm get_hwcap2 impl
  bsd-user/arm/target_arch_elf.h: arm get hwcap
  bsd-user/arm/target_arch_elf.h: arm defines for ELF
  bsd-user/arm/target_arch_thread.h: Routines to create and switch to a thread
  bsd-user/arm/target_arch_sigtramp.h: Signal Trampoline for arm
  bsd-user/arm/target_arch_vmparam.h: Parameters for arm address space
  bsd-user/arm/target_arch_reg.h: Implement core dump register copying
  bsd-user/arm/target_arch_cpu.h: Implement system call dispatch
  bsd-user/arm/target_arch_cpu.h: Implement data abort exceptions
  bsd-user/arm/target_arch_cpu.h: Implement trivial EXCP exceptions
  bsd-user/arm/target_arch_cpu.h: Dummy target_cpu_loop implementation
  ...

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

Merge tag 'pull-riscv-to-apply-20220108' of github.com:alistair23/qemu into staging

Second RISC-V PR for QEMU 7.0

 - Fix illegal instruction when PMP is disabled
 - SiFive PDMA 64-bit support
 - SiFive PLIC cleanups
 - Mark Hypervisor extension as non experimental
 - Enable Hypervisor extension by default
 - Support 32 cores on the virt machine
 - Corrections for the Vector extension
 - Experimental support for 128-bit CPUs
 - stval and mtval support for illegal instructions

# gpg: Signature made Fri 07 Jan 2022 09:50:11 PM PST
# gpg:                using RSA key F6C4AC46D4934868D3B8CE8F21E10D29DF977054
# gpg: Good signature from "Alistair Francis <alistair@alistair23.me>" [undefined]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: F6C4 AC46 D493 4868 D3B8  CE8F 21E1 0D29 DF97 7054

* tag 'pull-riscv-to-apply-20220108' of github.com:alistair23/qemu: (37 commits)
  target/riscv: Implement the stval/mtval illegal instruction
  target/riscv: Fixup setting GVA
  target/riscv: Set the opcode in DisasContext
  target/riscv: actual functions to realize crs 128-bit insns
  target/riscv: modification of the trans_csrxx for 128-bit support
  target/riscv: helper functions to wrap calls to 128-bit csr insns
  target/riscv: adding high part of some csrs
  target/riscv: support for 128-bit M extension
  target/riscv: support for 128-bit arithmetic instructions
  target/riscv: support for 128-bit shift instructions
  target/riscv: support for 128-bit U-type instructions
  target/riscv: support for 128-bit bitwise instructions
  target/riscv: accessors to registers upper part and 128-bit load/store
  target/riscv: moving some insns close to similar insns
  target/riscv: setup everything for rv64 to support rv128 execution
  target/riscv: array for the 64 upper bits of 128-bit registers
  target/riscv: separation of bitwise logic and arithmetic helpers
  target/riscv: additional macros to check instruction support
  qemu/int128: addition of div/rem 128-bit operations
  exec/memop: Adding signed quad and octo defines
  ...

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user: add arm target build

CC: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Acked-by: Kyle Evans <kevans@FreeBSD.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/freebsd/target_os_ucontext.h: Require TARGET_*CONTEXT_SIZE

Now that all architecutres define TARGET_[MU]CONTEXT_SIZE, enforce
requiring them and always check the sizeof target_{u,m}context_t
sizes.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/signal.c: arm get_ucontext_sigreturn

Update ucontext to implement sigreturn.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/signal.c: arm set_mcontext

Move the machine context to the CPU state.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@FreeBSD.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/signal.c: arm get_mcontext

Get the machine context from the CPU state.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@FreeBSD.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/signal.c: arm set_sigtramp_args

Implement set_sigtramp_args to setup the arguments to the sigtramp
calls.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/target_arch_signal.h: Define size of *context_t

Define the native sizes of mcontext_t and ucontext_t so that the tests
in target_os_ucontext.h ensure the size of arm's version of these
structures is correct.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/target_arch_signal.h: arm machine context and trapframe for signals

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@FreeBSD.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/target_arch_signal.h: arm specific signal registers and stack

Defines for registers and stack layout related to signals.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Kyle Evans <kevans@FreeBSD.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/target_arch_elf.h: arm get_hwcap2 impl

Implement the extended HW capabilities for HWCAP2.

Signed-off-by: Kyle Evans <kevans@FreeBSD.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Kyle Evans <kevans@FreeBSD.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/target_arch_elf.h: arm get hwcap

Implement get_elf_hwcap to get the first word of hardware capabilities.

Signed-off-by: Kyle Evans <kevans@FreeBSD.org>
Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Kyle Evans <kevans@FreeBSD.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/target_arch_elf.h: arm defines for ELF

Basic set of defines needed for arm ELF file activation.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Kyle Evans <kevans@FreeBSD.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/target_arch_thread.h: Routines to create and switch to a thread

Implement target_thread_init (to create a thread) and target_set_upcall
(to switch to a thread) for arm.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@FreeBSD.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Kyle Evans <kevans@FreeBSD.org>

bsd-user/arm/target_arch_sigtramp.h: Signal Trampoline for arm

Copy of the signal trampoline code for arm, as well as setup_sigtramp to
write it to the stack.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Kyle Evans <kevans@FreeBSD.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/target_arch_vmparam.h: Parameters for arm address space

Various parameters describing the layout of the ARM address space. In
addition, define routines to get the stack pointer and to set the second
return value.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@FreeBSD.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Kyle Evans <kevans@FreeBSD.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/target_arch_reg.h: Implement core dump register copying

Implement the register copying routines to extract registers from the
cpu for core dump generation.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Kyle Evans <kevans@FreeBSD.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/target_arch_cpu.h: Implement system call dispatch

Implement the system call dispatch. This implements all three kinds of
system call: direct and the two indirect variants. It handles all the
special cases for thumb as well.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@FreeBSD.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Kyle Evans <kevans@FreeBSD.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/target_arch_cpu.h: Implement data abort exceptions

Implement EXCP_PREFETCH_ABORT AND EXCP_DATA_ABORT. Both of these data
exceptions cause a SIGSEGV.

Signed-off-by: Kyle Evans <kevans@FreeBSD.org>
Signed-off-by: Olivier Houchard <cognet@ci0.org>
Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Kyle Evans <kevans@FreeBSD.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/target_arch_cpu.h: Implement trivial EXCP exceptions

Implement EXCP_UDEF, EXCP_DEBUG, EXCP_INTERRUPT, EXCP_ATOMIC and
EXCP_YIELD. The first two generate a signal to the emulated
binary. EXCP_ATOMIC handles atomic operations. The remainder are fancy
nops.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Mikaël Urankar <mikael.urankar@gmail.com>
Signed-off-by: Kyle Evans <kevans@FreeBSD.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/target_arch_cpu.h: Dummy target_cpu_loop implementation

Add a boiler plate CPU loop that does nothing except return an error for
all traps.

Signed-off-by: Sean Bruno <sbruno@FreeBSD.org>
Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Kyle Evans <kevans@FreeBSD.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/target_arch_cpu.h: Implement target_cpu_clone_regs

Implement target_cpu_clone_regs to clone the resister state on a fork.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Kyle Evans <kevans@FreeBSD.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/target_arch_cpu.h: CPU Loop definitions

target_arch_cpu.h is for CPU loop definitions. Create the file and
define target_cpu_init and target_cpu_reset for arm.

Signed-off-by: Olivier Houchard <cognet@ci0.org>
Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Kyle Evans <kevans@FreeBSD.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/target_arch_cpu.c: Target specific TLS routines

Target specific TLS routines to get and set the TLS values.

Signed-off-by: Kyle Evans <kevans@FreeBSD.org>
Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Kyle Evans <kevans@FreeBSD.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/target_syscall.h: Add copyright and update name

The preferred name for the 32-bit arm is now armv7. Update the name to
reflect that. In addition, add Stacey's copyright to this file and
update the include guards to the new convention.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Kyle Evans <kevans@FreeBSD.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/target_arch_sysarch.h: Use consistent include guards

As part of upstreaming, the include guards have been made more
consistent. Update this file to use the new guards.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Kyle Evans <kevans@FreeBSD.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/target_os_signal.h: Move signal prototypes to target_os_ucontext.h

Switch to the CPUArchState typedef and move target-provided prototypes
to target_os_ucontext.h.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/x86_64: Move functions into signal.c

Move the current inline functions into sigal.c. This will increate the
flexibility of implementation in the future.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/x86_64/target_arch_signal.h: Fill in mcontext_t

Fill in target_mcontext match the FreeBSD mcontext_t structure. Also
define the size correctly.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/x86_64/target_arch_signal.h: use new target_os_ucontext.h

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/x86_64/target_arch_signal.h: Remove target_sigcontext

In FreeBSD, sigcontext was retired in favor of ucontext/mcontext.
Remove vestigial target_sigcontext.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/i386: Move the inlines into signal.c

Move the (now stubbed out) inlines into bsd-user/i386/signal.c.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/i386/target_arch_signal.h: Update mcontext_t to match FreeBSD

Fill in target_mcontext_t to match the FreeBSD mcontex_t. Also tag the
current size of mcontext and ucontext to enable size checking for i386.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/i386/target_arch_signal.h: use new target_os_ucontext.h

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/i386/target_arch_signal.h: Remove target_sigcontext

In FreeBSD, sigcontext was retired in favor of ucontext/mcontext.
Remove vestigial target_sigcontext.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user: create a per-arch signal.c file

Create a place-holder signal.c file for each of the architectures that
are currently built. In the future, some code that's currently inlined
in target_arch_signal.h will live here.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/freebsd: Create common target_os_ucontext.h file

FreeBSD has a MI ucontext structure that contains the MD mcontext
machine state and other things that are machine independent. Create an
include file for all the ucontext stuff. It needs to be included in the
arch specific files after target_mcontext is defined. This is largely
copied from sys/_ucontext.h with the comments about layout removed
because we don't support ancient FreeBSD binaries.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/mips*: Remove mips support

FreeBSD is dropping support for mips starting with FreeBSD 14. mips
support has been removed from the bsd-user fork because updating it for
new signal requirements will take too much time. Remove it here since it
is a distraction.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Acked-by: Richard Henderson <richard.henderson@linaro.org>

target/riscv: Implement the stval/mtval illegal instruction

The stval and mtval registers can optionally contain the faulting
instruction on an illegal instruction exception. This patch adds support
for setting the stval and mtval registers.

The RISC-V spec states that "The stval register can optionally also be
used to return the faulting instruction bits on an illegal instruction
exception...". In this case we are always writing the value on an
illegal instruction.

This doesn't match all CPUs (some CPUs won't write the data), but in
QEMU let's just populate the value on illegal instructions. This won't
break any guest software, but will provide more information to guests.

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Message-id: 20211220064916.107241-4-alistair.francis@opensource.wdc.com

target/riscv: Fixup setting GVA

In preparation for adding support for the illegal instruction address
let's fixup the Hypervisor extension setting GVA logic and improve the
variable names.

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Message-id: 20211220064916.107241-3-alistair.francis@opensource.wdc.com

target/riscv: Set the opcode in DisasContext

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Message-id: 20211220064916.107241-2-alistair.francis@opensource.wdc.com

target/riscv: actual functions to realize crs 128-bit insns

The csrs are accessed through function pointers: we add 128-bit read
operations in the table for three csrs (writes fallback to the
64-bit version as the upper 64-bit information is handled elsewhere):
- misa, as mxl is needed for proper operation,
- mstatus and sstatus, to return sd
In addition, we also add read and write accesses to the machine and
supervisor scratch registers.

Signed-off-by: Frédéric Pétrot <frederic.petrot@univ-grenoble-alpes.fr>
Co-authored-by: Fabien Portas <fabien.portas@grenoble-inp.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220106210108.138226-19-frederic.petrot@univ-grenoble-alpes.fr
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: modification of the trans_csrxx for 128-bit support

As opposed to the gen_arith and gen_shift generation helpers, the csr insns
do not have a common prototype, so the choice to generate 32/64 or 128-bit
helper calls is done in the trans_csrxx functions.

Signed-off-by: Frédéric Pétrot <frederic.petrot@univ-grenoble-alpes.fr>
Co-authored-by: Fabien Portas <fabien.portas@grenoble-inp.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220106210108.138226-18-frederic.petrot@univ-grenoble-alpes.fr
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: helper functions to wrap calls to 128-bit csr insns

Given the side effects they have, the csr instructions are realized as
helpers. We extend this existing infrastructure for 128-bit sized csr.
We return 128-bit values using the same approach as for div/rem.
Theses helpers all call a unique function that is currently a fallback
on the 64-bit version.
The trans_csrxx functions supporting 128-bit are yet to be implemented.

Signed-off-by: Frédéric Pétrot <frederic.petrot@univ-grenoble-alpes.fr>
Co-authored-by: Fabien Portas <fabien.portas@grenoble-inp.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220106210108.138226-17-frederic.petrot@univ-grenoble-alpes.fr
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: adding high part of some csrs

Adding the high part of a very minimal set of csr.

Signed-off-by: Frédéric Pétrot <frederic.petrot@univ-grenoble-alpes.fr>
Co-authored-by: Fabien Portas <fabien.portas@grenoble-inp.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220106210108.138226-16-frederic.petrot@univ-grenoble-alpes.fr
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: support for 128-bit M extension

Mult are generated inline (using a cool trick pointed out by Richard), but
for div and rem, given the complexity of the implementation of these
instructions, we call helpers to produce their behavior. From an
implementation standpoint, the helpers return the low part of the results,
while the high part is temporarily stored in a dedicated field of cpu_env
that is used to update the architectural register in the generation wrapper.

Signed-off-by: Frédéric Pétrot <frederic.petrot@univ-grenoble-alpes.fr>
Co-authored-by: Fabien Portas <fabien.portas@grenoble-inp.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220106210108.138226-15-frederic.petrot@univ-grenoble-alpes.fr
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: support for 128-bit arithmetic instructions

Addition of 128-bit adds and subs in their various sizes,
"set if less than"s and branches.
Refactored the code to have a comparison function used for both stls and
branches.

Signed-off-by: Frédéric Pétrot <frederic.petrot@univ-grenoble-alpes.fr>
Co-authored-by: Fabien Portas <fabien.portas@grenoble-inp.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220106210108.138226-14-frederic.petrot@univ-grenoble-alpes.fr
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: support for 128-bit shift instructions

Handling shifts for 32, 64 and 128 operation length for RV128, following the
general framework for handling various olens proposed by Richard.

Signed-off-by: Frédéric Pétrot <frederic.petrot@univ-grenoble-alpes.fr>
Co-authored-by: Fabien Portas <fabien.portas@grenoble-inp.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220106210108.138226-13-frederic.petrot@univ-grenoble-alpes.fr
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: support for 128-bit U-type instructions

Adding the 128-bit version of lui and auipc, and introducing to that end
a "set register with immediat" function to handle extension on 128 bits.

Signed-off-by: Frédéric Pétrot <frederic.petrot@univ-grenoble-alpes.fr>
Co-authored-by: Fabien Portas <fabien.portas@grenoble-inp.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220106210108.138226-12-frederic.petrot@univ-grenoble-alpes.fr
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: support for 128-bit bitwise instructions

The 128-bit bitwise instructions do not need any function prototype change
as the functions can be applied independently on the lower and upper part of
the registers.

Signed-off-by: Frédéric Pétrot <frederic.petrot@univ-grenoble-alpes.fr>
Co-authored-by: Fabien Portas <fabien.portas@grenoble-inp.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220106210108.138226-11-frederic.petrot@univ-grenoble-alpes.fr
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: accessors to registers upper part and 128-bit load/store

Get function to retrieve the 64 top bits of a register, stored in the gprh
field of the cpu state. Set function that writes the 128-bit value at once.
The access to the gprh field can not be protected at compile time to make
sure it is accessed only in the 128-bit version of the processor because we
have no way to indicate that the misa_mxl_max field is const.

The 128-bit ISA adds ldu, lq and sq. We provide support for these
instructions. Note that (a) we compute only 64-bit addresses to actually
access memory, cowardly utilizing the existing address translation mechanism
of QEMU, and (b) we assume for now little-endian memory accesses.

Signed-off-by: Frédéric Pétrot <frederic.petrot@univ-grenoble-alpes.fr>
Co-authored-by: Fabien Portas <fabien.portas@grenoble-inp.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220106210108.138226-10-frederic.petrot@univ-grenoble-alpes.fr
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: moving some insns close to similar insns

lwu and ld are functionally close to the other loads, but were after the
stores in the source file.
Similarly, xor was away from or and and by two arithmetic functions, while
the immediate versions were nicely put together.
This patch moves the aforementioned loads after lhu, and xor above or,
where they more logically belong.

Signed-off-by: Frédéric Pétrot <frederic.petrot@univ-grenoble-alpes.fr>
Co-authored-by: Fabien Portas <fabien.portas@grenoble-inp.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220106210108.138226-9-frederic.petrot@univ-grenoble-alpes.fr
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: setup everything for rv64 to support rv128 execution

This patch adds the support of the '-cpu rv128' option to
qemu-system-riscv64 so that we can indicate that we want to run rv128
executables.
Still, there is no support for 128-bit insns at that stage so qemu fails
miserably (as expected) if launched with this option.

Signed-off-by: Frédéric Pétrot <frederic.petrot@univ-grenoble-alpes.fr>
Co-authored-by: Fabien Portas <fabien.portas@grenoble-inp.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220106210108.138226-8-frederic.petrot@univ-grenoble-alpes.fr
[ Changed by AF
 - Rename CPU to "x-rv128"
]
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: array for the 64 upper bits of 128-bit registers

The upper 64-bit of the 128-bit registers have now a place inside
the cpu state structure, and are created as globals for future use.

Signed-off-by: Frédéric Pétrot <frederic.petrot@univ-grenoble-alpes.fr>
Co-authored-by: Fabien Portas <fabien.portas@grenoble-inp.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220106210108.138226-7-frederic.petrot@univ-grenoble-alpes.fr
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: separation of bitwise logic and arithmetic helpers

Introduction of a gen_logic function for bitwise logic to implement
instructions in which no propagation of information occurs between bits and
use of this function on the bitwise instructions.

Signed-off-by: Frédéric Pétrot <frederic.petrot@univ-grenoble-alpes.fr>
Co-authored-by: Fabien Portas <fabien.portas@grenoble-inp.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220106210108.138226-6-frederic.petrot@univ-grenoble-alpes.fr
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: additional macros to check instruction support

Given that the 128-bit version of the riscv spec adds new instructions, and
that some instructions that were previously only available in 64-bit mode
are now available for both 64-bit and 128-bit, we added new macros to check
for the processor mode during translation.
Although RV128 is a superset of RV64, we keep for now the RV64 only tests
for extensions other than RVI and RVM.

Signed-off-by: Frédéric Pétrot <frederic.petrot@univ-grenoble-alpes.fr>
Co-authored-by: Fabien Portas <fabien.portas@grenoble-inp.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220106210108.138226-5-frederic.petrot@univ-grenoble-alpes.fr
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

qemu/int128: addition of div/rem 128-bit operations

Addition of div and rem on 128-bit integers, using the 128/64->128 divu and
64x64->128 mulu in host-utils.
These operations will be used within div/rem helpers in the 128-bit riscv
target.

Signed-off-by: Frédéric Pétrot <frederic.petrot@univ-grenoble-alpes.fr>
Co-authored-by: Fabien Portas <fabien.portas@grenoble-inp.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220106210108.138226-4-frederic.petrot@univ-grenoble-alpes.fr
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

exec/memop: Adding signed quad and octo defines

Adding defines to handle signed 64-bit and unsigned 128-bit quantities in
memory accesses.

Signed-off-by: Frédéric Pétrot <frederic.petrot@univ-grenoble-alpes.fr>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220106210108.138226-3-frederic.petrot@univ-grenoble-alpes.fr
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

exec/memop: Adding signedness to quad definitions

Renaming defines for quad in their various forms so that their signedness is
now explicit.
Done using git grep as suggested by Philippe, with a bit of hand edition to
keep assignments aligned.

Signed-off-by: Frédéric Pétrot <frederic.petrot@univ-grenoble-alpes.fr>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220106210108.138226-2-frederic.petrot@univ-grenoble-alpes.fr
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Fix position of 'experimental' comment

When commit 0643c12e4b dropped the 'x-' prefix for Zb[abcs] and set
them to be enabled by default, the comment about experimental
extensions was kept in place above them.  This moves it down a few
lines to only cover experimental extensions.

References: 0643c12e4b ("target/riscv: Enable bitmanip Zb[abcs] instructions")

Signed-off-by: Philipp Tomsich <philipp.tomsich@vrull.eu>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220106134020.1628889-1-philipp.tomsich@vrull.eu
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: rvv-1.0: Call the correct RVF/RVD check function for narrowing fp/int type-convert insns

vfncvt.f.xu.w, vfncvt.f.x.w convert double-width integer to single-width
floating-point. Therefore, should use require_rvf() to check whether
RVF/RVD is enabled.

vfncvt.f.f.w, vfncvt.rod.f.f.w convert double-width floating-point to
single-width integer. Therefore, should use require_scale_rvf() to check
whether RVF/RVD is enabled.

Signed-off-by: Frank Chang <frank.chang@sifive.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20220105022247.21131-4-frank.chang@sifive.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>