| Message ID | 1385140638-10444-1-git-send-email-peter.maydell@linaro.org |
|---|---|
| State | Accepted |
| Headers | show |
On Fri, Nov 22, 2013 at 05:17:07PM +0000, Peter Maydell wrote: > This patchset combines the 'virt' machine definition and > -cpu host support patchsets I've posted previous versions > of. I think these are now ready to go in once 1.8 opens > up; review appreciated. > > Changes since previous versions: > * added in -cpu host patchset, since it really only makes > sense with mach-virt > * rearranged virt address space a bit to allow space > for a reasonable sized boot flash device and possible > PCI window if a PCI controller model ever becomes possible > * fixed bug where we weren't starting secondary CPUs in > "PSCI power-down" state (by adding a CPU property which > lets the board specify that the CPU starts powered down) > [thanks to Giridhar Maruthy for tracking this down] > * includes a KVM header update against current mainline > * includes (previously RFC'd) patch which provides defines > of some kernel KVM constants we can use even if not CONFIG_KVM > * put the KVM CPU type into the CPU object rather than having > a silly 'QOM object type => constant' lookup table > * put dtb 'compatible' string into CPU object rather than > having the board model need to know about it > * abstracted out the "create a scratch vcpu for querying > for capabilities" code; this is currently only called in > one place, but the abstraction will be needed for 64 bit > KVM ARM support > > Mostly this is cleanup and streamlining which I noticed > was either possible or necessary in the course of putting > 64 bit KVM ARM control on top of this patchset. > > > The kernel patch to get the PL011 to work is still needed > (Christoffer tells me he's working on getting a proper fix > done for this): > I tell myself that too... Anyway, looked over the series and it looks good to me. So except for some of the QOM that I can't really wrap around my brain yet: Reviewed-by: Christoffer Dall <christoffer.dall@linaro.org>
On 22 November 2013 17:17, Peter Maydell <peter.maydell@linaro.org> wrote: > This patchset combines the 'virt' machine definition and > -cpu host support patchsets I've posted previous versions > of. I think these are now ready to go in once 1.8 opens > up; review appreciated. I'm applying this series to target-arm.next (with the typo of the kernel commit hash in the 'update headers' patch fixed). thanks! -- PMM
diff --git a/arch/arm/mach-virt/virt.c b/arch/arm/mach-virt/virt.c index b184e57..2b6aceb 100644 --- a/arch/arm/mach-virt/virt.c +++ b/arch/arm/mach-virt/virt.c @@ -21,11 +21,13 @@ #include <linux/of_irq.h> #include <linux/of_platform.h> #include <linux/smp.h> +#include <linux/clk-provider.h> #include <asm/mach/arch.h> static void __init virt_init(void) { + of_clk_init(NULL); of_platform_populate(NULL, of_default_bus_match_table, NULL, NULL); }
This patchset combines the 'virt' machine definition and -cpu host support patchsets I've posted previous versions of. I think these are now ready to go in once 1.8 opens up; review appreciated. Changes since previous versions: * added in -cpu host patchset, since it really only makes sense with mach-virt * rearranged virt address space a bit to allow space for a reasonable sized boot flash device and possible PCI window if a PCI controller model ever becomes possible * fixed bug where we weren't starting secondary CPUs in "PSCI power-down" state (by adding a CPU property which lets the board specify that the CPU starts powered down) [thanks to Giridhar Maruthy for tracking this down] * includes a KVM header update against current mainline * includes (previously RFC'd) patch which provides defines of some kernel KVM constants we can use even if not CONFIG_KVM * put the KVM CPU type into the CPU object rather than having a silly 'QOM object type => constant' lookup table * put dtb 'compatible' string into CPU object rather than having the board model need to know about it * abstracted out the "create a scratch vcpu for querying for capabilities" code; this is currently only called in one place, but the abstraction will be needed for 64 bit KVM ARM support Mostly this is cleanup and streamlining which I noticed was either possible or necessary in the course of putting 64 bit KVM ARM control on top of this patchset. The kernel patch to get the PL011 to work is still needed (Christoffer tells me he's working on getting a proper fix done for this): John Rigby (1): hw/arm/boot: Allow boards to provide an fdt blob Peter Maydell (10): target-arm: Provide mechanism for getting KVM constants even if not CONFIG_KVM device_tree.c: Terminate the empty reservemap in create_device_tree() target-arm: Provide PSCI constants to generic QEMU code target-arm: Add ARMCPU field for Linux device-tree 'compatible' string target-arm: Allow secondary KVM CPUs to be booted via PSCI hw/arm: Add 'virt' platform linux-headers: Update from mainline target-arm: Don't hardcode KVM target CPU to be A15 target-arm: Provide '-cpu host' when running KVM hw/arm/virt: Support -cpu host device_tree.c | 4 + hw/arm/Makefile.objs | 2 +- hw/arm/boot.c | 32 ++- hw/arm/virt.c | 452 ++++++++++++++++++++++++++++++ include/hw/arm/arm.h | 7 + linux-headers/asm-arm/kvm.h | 3 +- linux-headers/asm-powerpc/epapr_hcalls.h | 4 +- linux-headers/asm-powerpc/kvm.h | 86 +++++- linux-headers/asm-x86/hyperv.h | 19 ++ linux-headers/asm-x86/kvm.h | 6 +- linux-headers/linux/kvm.h | 11 + target-arm/cpu-qom.h | 11 + target-arm/cpu.c | 59 ++++ target-arm/cpu.h | 13 +- target-arm/helper.c | 6 + target-arm/kvm-consts.h | 64 +++++ target-arm/kvm.c | 243 +++++++++++++++- target-arm/kvm_arm.h | 55 ++++ 18 files changed, 1033 insertions(+), 44 deletions(-) create mode 100644 hw/arm/virt.c create mode 100644 target-arm/kvm-consts.h