From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([209.51.188.92]:39881) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1gw7lS-0003KA-Ra for qemu-devel@nongnu.org; Tue, 19 Feb 2019 10:56:59 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1gw7lL-0002Pb-59 for qemu-devel@nongnu.org; Tue, 19 Feb 2019 10:56:56 -0500 References: <20190205173306.20483-1-eric.auger@redhat.com> <20190205173306.20483-10-eric.auger@redhat.com> <20190218110701.7349c698@Igors-MacBook-Pro.local> From: Auger Eric Message-ID: Date: Tue, 19 Feb 2019 16:56:08 +0100 MIME-Version: 1.0 In-Reply-To: <20190218110701.7349c698@Igors-MacBook-Pro.local> Content-Type: text/plain; charset=utf-8 Content-Language: en-US Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH v6 09/18] hw/arm/virt: Implement kvm_type function for 4.0 machine List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Igor Mammedov Cc: eric.auger.pro@gmail.com, qemu-devel@nongnu.org, qemu-arm@nongnu.org, peter.maydell@linaro.org, shameerali.kolothum.thodi@huawei.com, david@redhat.com, drjones@redhat.com, dgilbert@redhat.com, david@gibson.dropbear.id.au Hi Igor, On 2/18/19 11:07 AM, Igor Mammedov wrote: > On Tue, 5 Feb 2019 18:32:57 +0100 > Eric Auger wrote: > >> This patch implements the machine class kvm_type() callback. >> It returns the max IPA shift needed to implement the whole GPA >> range including the RAM and IO regions located beyond. >> The returned value in passed though the KVM_CREATE_VM ioctl and >> this allows KVM to set the stage2 tables dynamically. >> >> At this stage the RAM limit still is limited to 255GB. >> >> Setting all the existing highmem IO regions beyond the RAM >> allows to have a single contiguous RAM region (initial RAM and >> possible hotpluggable device memory). That way we do not need >> to do invasive changes in the EDK2 FW to support a dynamic >> RAM base. >> >> Signed-off-by: Eric Auger >> >> --- >> >> v5 -> v6: >> - add some comments >> - high IO region cannot start before 256GiB >> --- >> hw/arm/virt.c | 52 +++++++++++++++++++++++++++++++++++++++++-- >> include/hw/arm/virt.h | 2 ++ >> 2 files changed, 52 insertions(+), 2 deletions(-) >> >> diff --git a/hw/arm/virt.c b/hw/arm/virt.c >> index 2b15839d0b..b90ffc2e5d 100644 >> --- a/hw/arm/virt.c >> +++ b/hw/arm/virt.c >> @@ -1366,6 +1366,7 @@ static uint64_t virt_cpu_mp_affinity(VirtMachineState *vms, int idx) >> >> static void virt_set_memmap(VirtMachineState *vms) >> { >> + MachineState *ms = MACHINE(vms); >> hwaddr base; >> int i; >> >> @@ -1375,7 +1376,17 @@ static void virt_set_memmap(VirtMachineState *vms) >> vms->memmap[i] = a15memmap[i]; >> } >> >> - vms->high_io_base = 256 * GiB; /* Top of the legacy initial RAM region */ >> + /* >> + * We now compute the base of the high IO region depending on the >> + * amount of initial and device memory. The device memory start/size >> + * is aligned on 1GiB. We never put the high IO region below 256GiB >> + * so that if maxram_size is < 255GiB we keep the legacy memory map >> + */ >> + vms->high_io_base = ROUND_UP(GiB + ms->ram_size, GiB) + >> + ROUND_UP(ms->maxram_size - ms->ram_size, GiB); >> + if (vms->high_io_base < 256 * GiB) { >> + vms->high_io_base = 256 * GiB; >> + } >> base = vms->high_io_base; >> >> for (i = VIRT_LOWMEMMAP_LAST; i < ARRAY_SIZE(extended_memmap); i++) { >> @@ -1386,6 +1397,7 @@ static void virt_set_memmap(VirtMachineState *vms) >> vms->memmap[i].size = size; >> base += size; >> } >> + vms->highest_gpa = base - 1; >> } >> >> static void machvirt_init(MachineState *machine) >> @@ -1402,7 +1414,9 @@ static void machvirt_init(MachineState *machine) >> bool firmware_loaded = bios_name || drive_get(IF_PFLASH, 0, 0); >> bool aarch64 = true; >> >> - virt_set_memmap(vms); >> + if (!vms->extended_memmap) { >> + virt_set_memmap(vms); >> + } >> >> /* We can probe only here because during property set >> * KVM is not available yet >> @@ -1784,6 +1798,36 @@ static HotplugHandler *virt_machine_get_hotplug_handler(MachineState *machine, >> return NULL; >> } >> >> +/* >> + * for arm64 kvm_type [7-0] encodes the IPA size shift >> + */ >> +static int virt_kvm_type(MachineState *ms, const char *type_str) >> +{ >> + VirtMachineState *vms = VIRT_MACHINE(ms); >> + int max_vm_phys_shift = kvm_arm_get_max_vm_phys_shift(ms); >> + int max_pa_shift; >> + >> + vms->extended_memmap = true; >> + >> + virt_set_memmap(vms); >> + >> + max_pa_shift = 64 - clz64(vms->highest_gpa); >> + >> + if (max_pa_shift > max_vm_phys_shift) { >> + error_report("-m and ,maxmem option values " >> + "require an IPA range (%d bits) larger than " >> + "the one supported by the host (%d bits)", >> + max_pa_shift, max_vm_phys_shift); >> + exit(1); >> + } >> + /* >> + * By default we return 0 which corresponds to an implicit legacy >> + * 40b IPA setting. Otherwise we return the actual requested IPA >> + * logsize >> + */ >> + return max_pa_shift > 40 ? max_pa_shift : 0; >> +} >> + >> static void virt_machine_class_init(ObjectClass *oc, void *data) >> { >> MachineClass *mc = MACHINE_CLASS(oc); >> @@ -1808,6 +1852,7 @@ static void virt_machine_class_init(ObjectClass *oc, void *data) >> mc->cpu_index_to_instance_props = virt_cpu_index_to_props; >> mc->default_cpu_type = ARM_CPU_TYPE_NAME("cortex-a15"); >> mc->get_default_cpu_node_id = virt_get_default_cpu_node_id; >> + mc->kvm_type = virt_kvm_type; >> assert(!mc->get_hotplug_handler); >> mc->get_hotplug_handler = virt_machine_get_hotplug_handler; >> hc->plug = virt_machine_device_plug_cb; >> @@ -1911,6 +1956,9 @@ static void virt_machine_3_1_options(MachineClass *mc) >> { >> virt_machine_4_0_options(mc); >> compat_props_add(mc->compat_props, hw_compat_3_1, hw_compat_3_1_len); >> + >> + /* extended memory map is enabled from 4.0 onwards */ >> + mc->kvm_type = NULL; > it's quite confusing, you have vms->extended_memmap and mc->kvm_type and > the later for some reason enables device memory. > > to me it seems that both are not related, device memory should work just fine > without kvm nor dynamic IPA (within TCG supported limits). > > I'd make extended_memmap virt machine class member the will enable pc-dimm support > and then it add checks for supported IPA range on top I agree I did not take into account the TCG use case and this series does not enable device memory with TCG which is a pitty. I will decorrelate things. Thanks! Eric > >> } >> DEFINE_VIRT_MACHINE(3, 1) >> >> diff --git a/include/hw/arm/virt.h b/include/hw/arm/virt.h >> index 3dc7a6c5d5..c88f67a492 100644 >> --- a/include/hw/arm/virt.h >> +++ b/include/hw/arm/virt.h >> @@ -132,6 +132,8 @@ typedef struct { >> uint32_t iommu_phandle; >> int psci_conduit; >> hwaddr high_io_base; >> + hwaddr highest_gpa; >> + bool extended_memmap; >> } VirtMachineState; >> >> #define VIRT_ECAM_ID(high) (high ? VIRT_HIGH_PCIE_ECAM : VIRT_PCIE_ECAM) >