This series marks some more #defines as poisoned, which are
target-specific (declared in config-target.h) and thus must
not be used in common code.

v2:
 - First two patches are the same as in v1
 - Reworked the CONFIG_KVM patches according to Paolo's review feedback
 - Added two new patches to finally poison CONFIG_SOFTMMU, too
 - Added a final patch to move bootdevice.o to common-obj now
   (based on an earlier patch where I also tried to move numa.o and
   balloon.o, too - but these files are indirectly target-dependent as
   I now know, so they can't be moved)

Thomas Huth (7):
  include/exec/poison: Add missing TARGET defines
  include/exec/poison: Mark some CONFIG defines as poisoned, too
  Move CONFIG_KVM related definitions to kvm_i386.h
  include/exec/poison: Mark CONFIG_KVM as poisoned, too
  cpu: Introduce a wrapper for tlb_flush() that can be used in common
    code
  include/exec/poison: Mark CONFIG_SOFTMMU as poisoned
  Makefile: Move bootdevice.o to common-obj-y

 Makefile.objs             |  2 +-
 Makefile.target           |  2 +-
 bootdevice.c              |  2 +-
 hw/acpi/ich9.c            |  1 -
 hw/i386/pc_q35.c          |  1 +
 include/exec/cpu-common.h |  2 ++
 include/exec/poison.h     | 33 +++++++++++++++++++++++++++++++++
 include/hw/i386/pc.h      | 13 -------------
 include/qom/cpu.h         |  8 ++++++++
 include/sysemu/kvm.h      | 31 ++++++++++++-------------------
 qom/cpu.c                 |  5 ++---
 target/i386/kvm_i386.h    | 23 +++++++++++++++++++++++
 translate-all.c           |  8 ++++++++
 13 files changed, 92 insertions(+), 39 deletions(-)

-- 
1.8.3.1

On 16/06/2017 16:59, Thomas Huth wrote:
> This series marks some more #defines as poisoned, which are
> target-specific (declared in config-target.h) and thus must
> not be used in common code.
> 
> v2:
>  - First two patches are the same as in v1
>  - Reworked the CONFIG_KVM patches according to Paolo's review feedback
>  - Added two new patches to finally poison CONFIG_SOFTMMU, too
>  - Added a final patch to move bootdevice.o to common-obj now
>    (based on an earlier patch where I also tried to move numa.o and
>    balloon.o, too - but these files are indirectly target-dependent as
>    I now know, so they can't be moved)

Why do you say they can't be moved?  They can, and your series should
change nothing about it, thanks to what you're doing now with
CONFIG_KVM_IS_POSSIBLE (which was also done before with NEED_CPU_H).

If it compiles, it's perfect. :)  (Almost---your patches 5-6 show it's
not entirely true, but poisoning helps).

Paolo

> Thomas Huth (7):
>   include/exec/poison: Add missing TARGET defines
>   include/exec/poison: Mark some CONFIG defines as poisoned, too
>   Move CONFIG_KVM related definitions to kvm_i386.h
>   include/exec/poison: Mark CONFIG_KVM as poisoned, too
>   cpu: Introduce a wrapper for tlb_flush() that can be used in common
>     code
>   include/exec/poison: Mark CONFIG_SOFTMMU as poisoned
>   Makefile: Move bootdevice.o to common-obj-y
> 
>  Makefile.objs             |  2 +-
>  Makefile.target           |  2 +-
>  bootdevice.c              |  2 +-
>  hw/acpi/ich9.c            |  1 -
>  hw/i386/pc_q35.c          |  1 +
>  include/exec/cpu-common.h |  2 ++
>  include/exec/poison.h     | 33 +++++++++++++++++++++++++++++++++
>  include/hw/i386/pc.h      | 13 -------------
>  include/qom/cpu.h         |  8 ++++++++
>  include/sysemu/kvm.h      | 31 ++++++++++++-------------------
>  qom/cpu.c                 |  5 ++---
>  target/i386/kvm_i386.h    | 23 +++++++++++++++++++++++
>  translate-all.c           |  8 ++++++++
>  13 files changed, 92 insertions(+), 39 deletions(-)
> 

On 16.06.2017 17:05, Paolo Bonzini wrote:
> On 16/06/2017 16:59, Thomas Huth wrote:
>> This series marks some more #defines as poisoned, which are
>> target-specific (declared in config-target.h) and thus must
>> not be used in common code.
>>
>> v2:
>>  - First two patches are the same as in v1
>>  - Reworked the CONFIG_KVM patches according to Paolo's review feedback
>>  - Added two new patches to finally poison CONFIG_SOFTMMU, too
>>  - Added a final patch to move bootdevice.o to common-obj now
>>    (based on an earlier patch where I also tried to move numa.o and
>>    balloon.o, too - but these files are indirectly target-dependent as
>>    I now know, so they can't be moved)
> 
> Why do you say they can't be moved?  They can, and your series should
> change nothing about it, thanks to what you're doing now with
> CONFIG_KVM_IS_POSSIBLE (which was also done before with NEED_CPU_H).

numa.c and balloon.c both use ram_addr_t (numa.c uses it directly, and
balloon.c indirectly via the QEMUBalloonEvent typedef from the header).
And ram_addr_t is currently target-specific - it depends on
CONFIG_XEN_BACKEND. So this could break in certain subtle cases, e.g.
when compiling on a 32-bit host with XEN enabled.

 Thomas

On 16/06/2017 17:21, Thomas Huth wrote:
>> Why do you say they can't be moved?  They can, and your series should
>> change nothing about it, thanks to what you're doing now with
>> CONFIG_KVM_IS_POSSIBLE (which was also done before with NEED_CPU_H).
> numa.c and balloon.c both use ram_addr_t (numa.c uses it directly, and
> balloon.c indirectly via the QEMUBalloonEvent typedef from the header).
> And ram_addr_t is currently target-specific - it depends on
> CONFIG_XEN_BACKEND. So this could break in certain subtle cases, e.g.
> when compiling on a 32-bit host with XEN enabled.

Can ram_addr_t be poisoned on target-independent files?

Paolo

On 16.06.2017 17:55, Paolo Bonzini wrote:
> 
> 
> On 16/06/2017 17:21, Thomas Huth wrote:
>>> Why do you say they can't be moved?  They can, and your series should
>>> change nothing about it, thanks to what you're doing now with
>>> CONFIG_KVM_IS_POSSIBLE (which was also done before with NEED_CPU_H).
>> numa.c and balloon.c both use ram_addr_t (numa.c uses it directly, and
>> balloon.c indirectly via the QEMUBalloonEvent typedef from the header).
>> And ram_addr_t is currently target-specific - it depends on
>> CONFIG_XEN_BACKEND. So this could break in certain subtle cases, e.g.
>> when compiling on a 32-bit host with XEN enabled.
> 
> Can ram_addr_t be poisoned on target-independent files?

This is quite similar to trying to poison CONFIG_USER_ONLY ... every
common file that depends on include/exec/memory.h or
/include/exec/cpu-common.h then does not compile anymore - and as far as
I can see, this can't be fixed with some trivial "#ifdef NEED_CPU_H"s in
most cases anymore.

Since you've asked on IRC for a list of files which do not compile
anymore in this case, here we go (some of them are also quite easy to
fix, I think, but some need some bigger reworks, I guess):

stubs/xen-common.o
stubs/xen-hvm.o
stubs/qmp_pc_dimm_device_list.o
stubs/pc_madt_cpu_entry.o
device-hotplug.o
qdev-monitor.o
accel.o
dma-helpers.o
device_tree.o
qmp.o
cpus-common.o
hmp.o
vl.o
audio/audio.o
audio/wavcapture.o
backends/hostmem-ram.o
backends/hostmem.o
backends/hostmem-file.o
backends/cryptodev.o
backends/cryptodev-builtin.o
hw/9pfs/9p.o
hw/acpi/core.o
hw/acpi/pcihp.o
hw/acpi/piix4.o
hw/acpi/ich9.o
hw/acpi/cpu_hotplug.o
hw/acpi/memory_hotplug.o
hw/acpi/tco.o
hw/acpi/cpu.o
hw/acpi/nvdimm.o
hw/acpi/acpi_interface.o
hw/acpi/vmgenid.o
hw/acpi/aml-build.o
hw/acpi/ipmi.o
hw/audio/sb16.o
hw/acpi/acpi-stub.o
hw/audio/es1370.o
hw/audio/ac97.o
hw/audio/adlib.o
hw/audio/gus.o
hw/audio/cs4231a.o
hw/audio/pcspk.o
hw/audio/hda-codec.o
hw/audio/intel-hda.o
hw/audio/wm8750.o
hw/audio/lm4549.o
hw/audio/pl041.o
hw/audio/cs4231.o
hw/audio/marvell_88w8618.o
hw/audio/milkymist-ac97.o
hw/block/cdrom.o
hw/audio/soundhw.o
hw/block/fdc.o
hw/block/m25p80.o
hw/block/nand.o
hw/block/pflash_cfi01.o
hw/block/pflash_cfi02.o
hw/block/ecc.o
hw/block/onenand.o
hw/block/nvme.o
hw/char/ipoctal232.o
hw/bt/hci.o
hw/char/parallel.o
hw/char/pl011.o
hw/char/serial.o
hw/char/escc.o
hw/char/serial-isa.o
hw/char/serial-pci.o
hw/char/xilinx_uartlite.o
hw/char/virtio-console.o
hw/char/etraxfs_ser.o
hw/char/cadence_uart.o
hw/char/debugcon.o
hw/char/grlib_apbuart.o
hw/char/imx_serial.o
hw/char/lm32_juart.o
hw/char/lm32_uart.o
hw/char/milkymist-uart.o
hw/char/sclpconsole.o
hw/char/sclpconsole-lm.o
hw/core/qdev-properties.o
hw/core/qdev.o
hw/core/bus.o
hw/core/empty_slot.o
hw/core/ptimer.o
hw/core/sysbus.o
hw/core/machine.o
hw/core/loader.o
hw/core/loader-fit.o
hw/core/qdev-properties-system.o
hw/core/register.o
hw/core/or-irq.o
hw/core/platform-bus.o
hw/cpu/core.o
hw/display/ads7846.o
hw/display/g364fb.o
hw/display/jazz_led.o
hw/display/cirrus_vga.o
hw/display/ssd0303.o
hw/display/ssd0323.o
hw/display/pl110.o
hw/display/vga-pci.o
hw/display/vga-isa.o
hw/display/vga-isa-mm.o
hw/display/vmware_vga.o
hw/display/exynos4210_fimd.o
hw/display/framebuffer.o
hw/display/milkymist-vgafb.o
hw/dma/puv3_dma.o
hw/display/tc6393xb.o
hw/display/milkymist-tmu2.o
hw/dma/rc4030.o
hw/dma/pl080.o
hw/dma/i82374.o
hw/dma/i8257.o
hw/dma/pl330.o
hw/dma/xilinx_axidma.o
hw/dma/xlnx-zynq-devcfg.o
hw/dma/etraxfs_dma.o
hw/dma/sparc32_dma.o
hw/dma/sun4m_iommu.o
hw/gpio/max7310.o
hw/gpio/puv3_gpio.o
hw/gpio/pl061.o
hw/gpio/zaurus.o
hw/gpio/mpc8xxx.o
hw/gpio/gpio_key.o
hw/i2c/core.o
hw/i2c/smbus.o
hw/i2c/smbus_eeprom.o
hw/i2c/i2c-ddc.o
hw/i2c/versatile_i2c.o
hw/i2c/smbus_ich9.o
hw/i2c/pm_smbus.o
hw/i2c/bitbang_i2c.o
hw/i2c/exynos4210_i2c.o
hw/i2c/imx_i2c.o
hw/i2c/aspeed_i2c.o
hw/ide/core.o
hw/ide/atapi.o
hw/ide/qdev.o
hw/ide/pci.o
hw/ide/isa.o
hw/ide/piix.o
hw/ide/cmd646.o
hw/ide/macio.o
hw/ide/mmio.o
hw/ide/via.o
hw/ide/microdrive.o
hw/ide/ahci.o
hw/ide/ich.o
hw/input/adb.o
hw/input/pckbd.o
hw/input/hid.o
hw/input/lm832x.o
hw/input/pl050.o
hw/input/stellaris_input.o
hw/input/ps2.o
hw/input/tsc2005.o
hw/input/vmmouse.o
hw/input/virtio-input.o
hw/input/virtio-input-host.o
hw/intc/heathrow_pic.o
hw/input/virtio-input-hid.o
hw/intc/i8259_common.o
hw/intc/i8259.o
hw/intc/pl190.o
hw/intc/puv3_intc.o
hw/intc/xilinx_intc.o
hw/intc/etraxfs_pic.o
hw/intc/imx_avic.o
hw/intc/lm32_pic.o
hw/intc/realview_gic.o
hw/intc/ioapic_common.o
hw/intc/slavio_intctl.o
hw/intc/arm_gic_common.o
hw/intc/arm_gic.o
hw/intc/arm_gicv2m.o
hw/intc/arm_gicv3_common.o
hw/intc/arm_gicv3.o
hw/intc/arm_gicv3_dist.o
hw/intc/arm_gicv3_its_common.o
hw/intc/arm_gicv3_redist.o
hw/intc/openpic.o
hw/ipack/ipack.o
hw/ipack/tpci200.o
hw/ipmi/ipmi.o
hw/ipmi/ipmi_bmc_sim.o
hw/ipmi/ipmi_bmc_extern.o
hw/ipmi/isa_ipmi_kcs.o
hw/ipmi/isa_ipmi_bt.o
hw/isa/apm.o
hw/isa/isa-bus.o
hw/isa/i82378.o
hw/isa/pc87312.o
hw/isa/vt82c686.o
hw/isa/piix4.o
hw/mem/pc-dimm.o
hw/mem/nvdimm.o
hw/misc/max111x.o
hw/misc/tmp105.o
hw/misc/tmp421.o
hw/misc/applesmc.o
hw/misc/debugexit.o
hw/misc/sga.o
hw/misc/pc-testdev.o
hw/misc/pci-testdev.o
hw/misc/unimp.o
hw/misc/arm_l2x0.o
hw/misc/arm_integrator_debug.o
hw/misc/a9scu.o
hw/misc/puv3_pm.o
hw/misc/arm11scu.o
hw/misc/macio/macio.o
hw/misc/macio/cuda.o
hw/misc/macio/mac_dbdma.o
hw/net/dp8393x.o
hw/net/ne2000.o
hw/net/eepro100.o
hw/net/pcnet-pci.o
hw/net/pcnet.o
hw/net/e1000x_common.o
hw/net/e1000.o
hw/net/net_tx_pkt.o
hw/net/e1000e.o
hw/net/rtl8139.o
hw/net/e1000e_core.o
hw/net/smc91c111.o
hw/net/vmxnet3.o
hw/net/lan9118.o
hw/net/ne2000-isa.o
hw/net/opencores_eth.o
hw/net/xgmac.o
hw/net/mipsnet.o
hw/net/xilinx_axienet.o
hw/net/allwinner_emac.o
hw/net/imx_fec.o
hw/net/stellaris_enet.o
hw/net/cadence_gem.o
hw/net/lance.o
hw/net/ftgmac100.o
hw/net/rocker/rocker.o
hw/net/rocker/rocker_desc.o
hw/nvram/ds1225y.o
hw/nvram/eeprom93xx.o
hw/nvram/fw_cfg.o
hw/nvram/chrp_nvram.o
hw/nvram/mac_nvram.o
hw/pci-bridge/pci_bridge_dev.o
hw/pci-bridge/pcie_root_port.o
hw/pci-bridge/gen_pcie_root_port.o
hw/pci-bridge/xio3130_upstream.o
hw/pci-bridge/pci_expander_bridge.o
hw/pci-bridge/xio3130_downstream.o
hw/pci-bridge/ioh3420.o
hw/pci-bridge/dec.o
hw/pci-bridge/i82801b11.o
hw/pci-host/pam.o
hw/pci-host/prep.o
hw/pci-host/grackle.o
hw/pci-host/uninorth.o
hw/pci-host/ppce500.o
hw/pci-host/versatile.o
hw/pci-host/apb.o
hw/pci-host/bonito.o
hw/pci-host/piix.o
hw/pci-host/q35.o
hw/pci-host/gpex.o
hw/pci-host/xilinx-pcie.o
hw/pci/pci.o
hw/pci/pci_bridge.o
hw/pci/msix.o
hw/pci/msi.o
hw/pci/shpc.o
hw/pci/slotid_cap.o
hw/pci/pci_host.o
hw/pci/pcie_host.o
hw/pci/pcie.o
hw/pci/pcie_port.o
hw/pci/pcie_aer.o
hw/pci/pci-stub.o
hw/pcmcia/pcmcia.o
hw/scsi/scsi-generic.o
hw/scsi/scsi-disk.o
hw/scsi/scsi-bus.o
hw/scsi/lsi53c895a.o
hw/scsi/mptsas.o
hw/scsi/mptconfig.o
hw/scsi/mptendian.o
hw/scsi/megasas.o
hw/scsi/esp.o
hw/scsi/vmw_pvscsi.o
hw/scsi/esp-pci.o
hw/sd/pl181.o
hw/sd/ssi-sd.o
hw/sd/sd.o
hw/sd/core.o
hw/sd/sdhci.o
hw/smbios/smbios_type_38.o
hw/smbios/smbios.o
hw/ssi/pl022.o
hw/ssi/ssi.o
hw/ssi/xilinx_spi.o
hw/ssi/aspeed_smc.o
hw/ssi/xilinx_spips.o
hw/ssi/stm32f2xx_spi.o
hw/timer/arm_timer.o
hw/timer/arm_mptimer.o
hw/timer/armv7m_systick.o
hw/timer/a9gtimer.o
hw/timer/cadence_ttc.o
hw/timer/ds1338.o
hw/timer/i8254_common.o
hw/timer/i8254.o
hw/timer/hpet.o
hw/timer/m48t59.o
hw/timer/m48t59-isa.o
hw/timer/pl031.o
hw/timer/puv3_ost.o
hw/timer/xilinx_timer.o
hw/timer/slavio_timer.o
hw/timer/twl92230.o
hw/timer/etraxfs_timer.o
hw/timer/grlib_gptimer.o
hw/timer/imx_epit.o
hw/timer/imx_gpt.o
hw/timer/lm32_timer.o
hw/timer/milkymist-sysctl.o
hw/timer/stm32f2xx_timer.o
hw/timer/aspeed_timer.o
hw/timer/sun4v-rtc.o
hw/tpm/tpm_tis.o
hw/tpm/tpm_passthrough.o
hw/tpm/tpm_util.o
hw/usb/core.o
hw/usb/libhw.o
hw/usb/combined-packet.o
hw/usb/bus.o
hw/usb/desc.o
hw/usb/desc-msos.o
hw/usb/hcd-uhci.o
hw/usb/hcd-ohci.o
hw/usb/hcd-ehci-pci.o
hw/usb/hcd-ehci-sysbus.o
hw/usb/hcd-ehci.o
hw/usb/hcd-xhci-nec.o
hw/usb/hcd-xhci.o
hw/usb/hcd-musb.o
hw/usb/dev-hub.o
hw/usb/dev-hid.o
hw/usb/dev-wacom.o
hw/usb/dev-storage.o
hw/usb/dev-uas.o
hw/usb/dev-audio.o
hw/usb/dev-serial.o
hw/usb/dev-network.o
hw/usb/dev-bluetooth.o
hw/usb/dev-smartcard-reader.o
hw/usb/dev-mtp.o
hw/usb/host-legacy.o
hw/usb/host-libusb.o
hw/virtio/virtio-rng.o
hw/virtio/virtio-bus.o
hw/virtio/virtio-mmio.o
hw/virtio/vhost-stub.o
hw/virtio/virtio-pci.o
hw/watchdog/wdt_i6300esb.o
hw/watchdog/wdt_ib700.o
hw/watchdog/wdt_diag288.o
hw/watchdog/wdt_aspeed.o
migration/socket.o
migration/migration.o
migration/tls.o
migration/channel.o
migration/savevm.o
migration/colo-comm.o
migration/colo.o
migration/colo-failover.o
migration/vmstate.o
migration/vmstate-types.o
migration/qemu-file.o
migration/qemu-file-channel.o
migration/postcopy-ram.o
migration/block.o
net/net.o
net/vhost-user.o
net/filter.o
net/tap.o
qom/cpu.o
slirp/slirp.o
ui/console.o
ui/input.o
ui/vnc.o

 Thomas

On 16/06/2017 18:15, Thomas Huth wrote:
> This is quite similar to trying to poison CONFIG_USER_ONLY ... every
> common file that depends on include/exec/memory.h or
> /include/exec/cpu-common.h then does not compile anymore - and as far as
> I can see, this can't be fixed with some trivial "#ifdef NEED_CPU_H"s in
> most cases anymore.

Oh, indeed. :(  I'll put removing ram_addr_t from memory.h on my todo
list.  I think most of those can go under NEED_CPU_H or can be replaced
with size_t.

Those parts of cpu-common.h instead should move to
include/exec/ram_addr.h sooner rather than later.

Paolo

> Since you've asked on IRC for a list of files which do not compile
> anymore in this case, here we go (some of them are also quite easy to
> fix, I think, but some need some bigger reworks, I guess):