[libvirt] [PATCH v7 0/4] ARM: add query-gic-capabilities QMP command
v7 changes: - patch 1 - add more to commit log, about how to use the results [Markus] - change interface from dict back to array [Markus, Eric] - patch 2 - kvm_arm_create_scratch_host_vcpu should raise error when init non-zero, but failed to find a good CPU model [Sergey] - patch 3 - rename function to kvm_device_supported [Sergey] - patch 4 - comment line from "FIXME" to "TODO" [Sergey] - make kvm-related part another function [Sergey] - change result to array to follow patch 1's change v6 changes: - patch 1 (squashed into patch 2) - explain more about the following in commit message: why we need this command, and what does the entries mean [Markus] - explain what GIC is [Eric] - squash this patch into patch 2 [Eric] - patch 2 (new patch 1) - fix "does not implement" english error [Eric] - remove useless headers in target-arm/monitor.c [Markus] - remove this command from whitelist [Markus, Eric] - return dict rather than array, with single key points to the original array results [Markus, Eric] - patch 5 (new patch 4) - tiny change in implementation to suite the new interface. v5 changes: - patch 2: moved to target-arm/monitor.c (from target-arm/machine.c) [Peter] - patch 3: splitted into three patches: [all from Peter's comments] - patch 3 (new): leverage kvm_arm_create_scratch_host_vcpu(), tiny enhancement of old one to suite our need - patch 4: introduce kvm_support_device() in kvm-all.c - patch 5: do the implementation. v4 changes: - all: rename query-gic-capability to query-gic-capabilities [Andrea] - patch 3: rename helper function to kvm_support_device, make it inline and lighter. [Drew] v3 changes: - patch 2: remove func declaration, add qmp header [Drew] - patch 3: being able to detect KVM GIC capabilities even without kvm enabled [Andrea]: this is a little bit hacky, need some more review on this. v2 changes: - result layout change: use array and dict for the capability bits rather than a single array of strings [Andrea/Markus] - spelling out what GIC is in doc [Eric] This patch is to add ARM-specific command "query-gic-capability". The new command can report which kind of GIC device the host/QEMU support. The returned result is in the form of array. Sample command and output: {"execute": "query-gic-capability"} {"return": [{"emulated": false, "version": 3, "kernel": false}, {"emulated": true, "version": 2, "kernel": true}]} Testing: Smoke tests on both x86 (emulated) and another moonshot ARM server. Peter Xu (4): arm: qmp: add query-gic-capabilities interface arm: enhance kvm_arm_create_scratch_host_vcpu kvm: add kvm_device_supported() helper function arm: implement query-gic-capabilities include/sysemu/kvm.h | 9 ++++++ kvm-all.c | 15 +++++++++ monitor.c | 8 +++++ qapi-schema.json | 36 +++++++++++++++++++++ qmp-commands.hx | 27 ++++++++++++++++ target-arm/Makefile.objs | 2 +- target-arm/kvm.c | 14 +++++++- target-arm/kvm_arm.h | 6 ++-- target-arm/monitor.c | 84 ++++++++++++++++++++++++++++++++++++++++++++++++ 9 files changed, 197 insertions(+), 4 deletions(-) create mode 100644 target-arm/monitor.c -- 2.4.3
This patch add "query-gic-capabilities" but does not implemnet it. The command is ARM-only. The command will return a list of GICCapability struct that describes all GIC versions that current QEMU and system support. Libvirt is possibly the first consumer of this new command. Before this patch, user will successfully configure all kinds of GIC devices for ARM guests, no matter whether current QEMU/kernel support it. If the specified GIC version/type is not supported, user will got an ambiguous "QEMU boot failure" when trying to start the VM. This is not user-friendly. With this patch, libvirt should be able to query which type (and which version) of GIC device that we support. Use this information, libvirt can warn the user during configuration of guests when specified GIC device type is not supported. Or better, we can just list those versions that we support, and filter out those not-supported ones. For example, if we got the query result: {"return": [{"emulated": false, "version": 3, "kernel": true}, {"emulated": true, "version": 2, "kernel": false}]} Then it means that we support emulated GIC version 2 using: qemu-system-aarch64 -M virt,accel=tcg,gic-version=2 ... or kvm-accelerated GIC version 3 using: qemu-system-aarch64 -M virt,accel=kvm,gic-version=3 ... If we specify other explicit GIC version rather than the above, QEMU will not be able to boot. Besides, the community is working on a more generic way to query these kind of information. However, due to the eagerness of this command, we decided to first implement this ad-hoc one, then when the generic method is ready, we can move on to that one smoothly. Signed-off-by: Peter Xu <peterx@redhat.com> --- monitor.c | 8 ++++++++ qapi-schema.json | 36 ++++++++++++++++++++++++++++++++++++ qmp-commands.hx | 27 +++++++++++++++++++++++++++ target-arm/Makefile.objs | 2 +- target-arm/monitor.c | 28 ++++++++++++++++++++++++++++ 5 files changed, 100 insertions(+), 1 deletion(-) create mode 100644 target-arm/monitor.c diff --git a/monitor.c b/monitor.c index 4c02f0f..5f2be8b 100644 --- a/monitor.c +++ b/monitor.c @@ -4258,3 +4258,11 @@ void qmp_dump_skeys(const char *filename, Error **errp) error_setg(errp, QERR_FEATURE_DISABLED, "dump-skeys"); } #endif + +#ifndef TARGET_ARM +GICCapabilityList *qmp_query_gic_capabilities(Error **errp) +{ + error_setg(errp, QERR_FEATURE_DISABLED, "query-gic-capabilities"); + return NULL; +} +#endif diff --git a/qapi-schema.json b/qapi-schema.json index 88f9b81..19921d5 100644 --- a/qapi-schema.json +++ b/qapi-schema.json @@ -4126,3 +4126,39 @@ ## { 'enum': 'ReplayMode', 'data': [ 'none', 'record', 'play' ] } + +## +# @GICCapability: +# +# The struct describes capability for a specific GIC (Generic +# Interrupt Controller) version. These bits are not only decided by +# QEMU/KVM software version, but also decided by the hardware that +# the program is running upon. +# +# @version: version of GIC to be described. Currently, only 2 and 3 +# are supported. +# +# @emulated: whether current QEMU/hardware supports emulated GIC +# device in user space. +# +# @kernel: whether current QEMU/hardware supports hardware +# accelerated GIC device in kernel. +# +# Since: 2.6 +## +{ 'struct': 'GICCapability', + 'data': { 'version': 'int', + 'emulated': 'bool', + 'kernel': 'bool' } } + +## +# @query-gic-capabilities: +# +# This command is ARM-only. It will return a list of GICCapability +# objects that describe its capability bits. +# +# Returns: a list of GICCapability objects. +# +# Since: 2.6 +## +{ 'command': 'query-gic-capabilities', 'returns': ['GICCapability'] } diff --git a/qmp-commands.hx b/qmp-commands.hx index 9e05365..de896a5 100644 --- a/qmp-commands.hx +++ b/qmp-commands.hx @@ -4853,3 +4853,30 @@ Example: {"type": 0, "out-pport": 0, "pport": 0, "vlan-id": 3840, "pop-vlan": 1, "id": 251658240} ]} + +EQMP + +#if defined TARGET_ARM + { + .name = "query-gic-capabilities", + .args_type = "", + .mhandler.cmd_new = qmp_marshal_query_gic_capabilities, + }, +#endif + +SQMP +query-gic-capabilities +--------------- + +Return a list of GICCapability objects, describing supported GIC +(Generic Interrupt Controller) versions. + +Arguments: None + +Example: + +-> { "execute": "query-gic-capabilities" } +<- { "return": [{ "version": 2, "emulated": true, "kernel": false }, + { "version": 3, "emulated": false, "kernel": true } ] } + +EQMP diff --git a/target-arm/Makefile.objs b/target-arm/Makefile.objs index a80eb39..334074c 100644 --- a/target-arm/Makefile.objs +++ b/target-arm/Makefile.objs @@ -8,4 +8,4 @@ obj-y += translate.o op_helper.o helper.o cpu.o obj-y += neon_helper.o iwmmxt_helper.o obj-y += gdbstub.o obj-$(TARGET_AARCH64) += cpu64.o translate-a64.o helper-a64.o gdbstub64.o -obj-y += crypto_helper.o +obj-y += crypto_helper.o monitor.o diff --git a/target-arm/monitor.c b/target-arm/monitor.c new file mode 100644 index 0000000..4c9bef3 --- /dev/null +++ b/target-arm/monitor.c @@ -0,0 +1,28 @@ +/* + * QEMU monitor.c for ARM. + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to deal + * in the Software without restriction, including without limitation the rights + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN + * THE SOFTWARE. + */ +#include "qemu/osdep.h" +#include "qmp-commands.h" + +GICCapabilityList *qmp_query_gic_capabilities(Error **errp) +{ + return NULL; +} -- 2.4.3
On 03/23/2016 08:55 PM, Peter Xu wrote:
This patch add "query-gic-capabilities" but does not implemnet it. The
s/implemnet/implement/
command is ARM-only. The command will return a list of GICCapability struct that describes all GIC versions that current QEMU and system support.
Libvirt is possibly the first consumer of this new command.
Before this patch, user will successfully configure all kinds of GIC devices for ARM guests, no matter whether current QEMU/kernel support it. If the specified GIC version/type is not supported, user will got an ambiguous "QEMU boot failure" when trying to start the VM. This is not user-friendly.
With this patch, libvirt should be able to query which type (and which version) of GIC device that we support. Use this information, libvirt can warn the user during configuration of guests when specified GIC device type is not supported. Or better, we can just list those versions that we support, and filter out those not-supported ones.
For example, if we got the query result:
{"return": [{"emulated": false, "version": 3, "kernel": true}, {"emulated": true, "version": 2, "kernel": false}]}
Then it means that we support emulated GIC version 2 using:
qemu-system-aarch64 -M virt,accel=tcg,gic-version=2 ...
or kvm-accelerated GIC version 3 using:
qemu-system-aarch64 -M virt,accel=kvm,gic-version=3 ...
This helps - it shows how libvirt will map the query into qemu command lines.
If we specify other explicit GIC version rather than the above, QEMU will not be able to boot.
Besides, the community is working on a more generic way to query these kind of information. However, due to the eagerness of this command, we decided to first implement this ad-hoc one, then when the generic method is ready, we can move on to that one smoothly.
Signed-off-by: Peter Xu <peterx@redhat.com> ---
Interface looks fine from QMP perspective: Reviewed-by: Eric Blake <eblake@redhat.com> Libvirt should be able to cope with the information here, and introspection will let us know if we ever expand the struct. -- Eric Blake eblake redhat com +1-919-301-3266 Libvirt virtualization library http://libvirt.org
Some more lines to make sure we allow NULL for 1st/3rd parameter. Signed-off-by: Peter Xu <peterx@redhat.com> --- target-arm/kvm.c | 14 +++++++++++++- target-arm/kvm_arm.h | 6 ++++-- 2 files changed, 17 insertions(+), 3 deletions(-) diff --git a/target-arm/kvm.c b/target-arm/kvm.c index 969ab0b..bb6935e 100644 --- a/target-arm/kvm.c +++ b/target-arm/kvm.c @@ -62,13 +62,17 @@ bool kvm_arm_create_scratch_host_vcpu(const uint32_t *cpus_to_try, goto err; } + if (!init) { + goto finish; + } + ret = ioctl(vmfd, KVM_ARM_PREFERRED_TARGET, init); if (ret >= 0) { ret = ioctl(cpufd, KVM_ARM_VCPU_INIT, init); if (ret < 0) { goto err; } - } else { + } else if (cpus_to_try) { /* Old kernel which doesn't know about the * PREFERRED_TARGET ioctl: we know it will only support * creating one kind of guest CPU which is its preferred @@ -85,8 +89,16 @@ bool kvm_arm_create_scratch_host_vcpu(const uint32_t *cpus_to_try, if (ret < 0) { goto err; } + } else { + /* + * Here, we got init != NULL but cpus_to_retry == NULL, + * meanwhile, we failed to probe one preferred target type + * (no matter what). So we fail. + */ + goto err; } +finish: fdarray[0] = kvmfd; fdarray[1] = vmfd; fdarray[2] = cpufd; diff --git a/target-arm/kvm_arm.h b/target-arm/kvm_arm.h index 07f0c72..4f01a99 100644 --- a/target-arm/kvm_arm.h +++ b/target-arm/kvm_arm.h @@ -124,9 +124,11 @@ void kvm_arm_reset_vcpu(ARMCPU *cpu); * kvm_arm_create_scratch_host_vcpu: * @cpus_to_try: array of QEMU_KVM_ARM_TARGET_* values (terminated with * QEMU_KVM_ARM_TARGET_NONE) to try as fallback if the kernel does not - * know the PREFERRED_TARGET ioctl + * know the PREFERRED_TARGET ioctl. If NULL is provided, will try + * to use perferred target, and fail if no preferred found. * @fdarray: filled in with kvmfd, vmfd, cpufd file descriptors in that order - * @init: filled in with the necessary values for creating a host vcpu + * @init: filled in with the necessary values for creating a host + * vcpu. If NULL is provided, will not init the vCPU. * * Create a scratch vcpu in its own VM of the type preferred by the host * kernel (as would be used for '-cpu host'), for purposes of probing it -- 2.4.3
On 24/03/16 05:55, Peter Xu wrote:
Some more lines to make sure we allow NULL for 1st/3rd parameter.
Signed-off-by: Peter Xu <peterx@redhat.com>
Acked-by: Sergey Fedorov <serge.fdrv@gmail.com> Kind regards, Sergey
--- target-arm/kvm.c | 14 +++++++++++++- target-arm/kvm_arm.h | 6 ++++-- 2 files changed, 17 insertions(+), 3 deletions(-)
diff --git a/target-arm/kvm.c b/target-arm/kvm.c index 969ab0b..bb6935e 100644 --- a/target-arm/kvm.c +++ b/target-arm/kvm.c @@ -62,13 +62,17 @@ bool kvm_arm_create_scratch_host_vcpu(const uint32_t *cpus_to_try, goto err; }
+ if (!init) { + goto finish; + } + ret = ioctl(vmfd, KVM_ARM_PREFERRED_TARGET, init); if (ret >= 0) { ret = ioctl(cpufd, KVM_ARM_VCPU_INIT, init); if (ret < 0) { goto err; } - } else { + } else if (cpus_to_try) { /* Old kernel which doesn't know about the * PREFERRED_TARGET ioctl: we know it will only support * creating one kind of guest CPU which is its preferred @@ -85,8 +89,16 @@ bool kvm_arm_create_scratch_host_vcpu(const uint32_t *cpus_to_try, if (ret < 0) { goto err; } + } else { + /* + * Here, we got init != NULL but cpus_to_retry == NULL, + * meanwhile, we failed to probe one preferred target type + * (no matter what). So we fail. + */ + goto err; }
+finish: fdarray[0] = kvmfd; fdarray[1] = vmfd; fdarray[2] = cpufd; diff --git a/target-arm/kvm_arm.h b/target-arm/kvm_arm.h index 07f0c72..4f01a99 100644 --- a/target-arm/kvm_arm.h +++ b/target-arm/kvm_arm.h @@ -124,9 +124,11 @@ void kvm_arm_reset_vcpu(ARMCPU *cpu); * kvm_arm_create_scratch_host_vcpu: * @cpus_to_try: array of QEMU_KVM_ARM_TARGET_* values (terminated with * QEMU_KVM_ARM_TARGET_NONE) to try as fallback if the kernel does not - * know the PREFERRED_TARGET ioctl + * know the PREFERRED_TARGET ioctl. If NULL is provided, will try + * to use perferred target, and fail if no preferred found. * @fdarray: filled in with kvmfd, vmfd, cpufd file descriptors in that order - * @init: filled in with the necessary values for creating a host vcpu + * @init: filled in with the necessary values for creating a host + * vcpu. If NULL is provided, will not init the vCPU. * * Create a scratch vcpu in its own VM of the type preferred by the host * kernel (as would be used for '-cpu host'), for purposes of probing it
On 24 March 2016 at 02:55, Peter Xu <peterx@redhat.com> wrote:
Some more lines to make sure we allow NULL for 1st/3rd parameter.
Signed-off-by: Peter Xu <peterx@redhat.com> --- target-arm/kvm.c | 14 +++++++++++++- target-arm/kvm_arm.h | 6 ++++-- 2 files changed, 17 insertions(+), 3 deletions(-)
diff --git a/target-arm/kvm.c b/target-arm/kvm.c index 969ab0b..bb6935e 100644 --- a/target-arm/kvm.c +++ b/target-arm/kvm.c @@ -62,13 +62,17 @@ bool kvm_arm_create_scratch_host_vcpu(const uint32_t *cpus_to_try, goto err; }
+ if (!init) { + goto finish; + } + ret = ioctl(vmfd, KVM_ARM_PREFERRED_TARGET, init); if (ret >= 0) { ret = ioctl(cpufd, KVM_ARM_VCPU_INIT, init); if (ret < 0) { goto err; } - } else { + } else if (cpus_to_try) { /* Old kernel which doesn't know about the * PREFERRED_TARGET ioctl: we know it will only support * creating one kind of guest CPU which is its preferred @@ -85,8 +89,16 @@ bool kvm_arm_create_scratch_host_vcpu(const uint32_t *cpus_to_try, if (ret < 0) { goto err; } + } else { + /* + * Here, we got init != NULL but cpus_to_retry == NULL, + * meanwhile, we failed to probe one preferred target type + * (no matter what). So we fail. + */ + goto err;
aka "treat a NULL cpus_to_retry like an empty list".
}
+finish: fdarray[0] = kvmfd; fdarray[1] = vmfd; fdarray[2] = cpufd; diff --git a/target-arm/kvm_arm.h b/target-arm/kvm_arm.h index 07f0c72..4f01a99 100644 --- a/target-arm/kvm_arm.h +++ b/target-arm/kvm_arm.h @@ -124,9 +124,11 @@ void kvm_arm_reset_vcpu(ARMCPU *cpu); * kvm_arm_create_scratch_host_vcpu: * @cpus_to_try: array of QEMU_KVM_ARM_TARGET_* values (terminated with * QEMU_KVM_ARM_TARGET_NONE) to try as fallback if the kernel does not - * know the PREFERRED_TARGET ioctl + * know the PREFERRED_TARGET ioctl. If NULL is provided, will try + * to use perferred target, and fail if no preferred found.
"preferred". Better phrased as "a NULL pointer is treated like an empty array", I think.
* @fdarray: filled in with kvmfd, vmfd, cpufd file descriptors in that order - * @init: filled in with the necessary values for creating a host vcpu + * @init: filled in with the necessary values for creating a host + * vcpu. If NULL is provided, will not init the vCPU. * * Create a scratch vcpu in its own VM of the type preferred by the host * kernel (as would be used for '-cpu host'), for purposes of probing it -- 2.4.3
(I'll fix this up in target-arm.next, it's a minor nit.) thanks -- PMM
This can be used when probing whether KVM support specific device. Here, a raw vmfd is used. Signed-off-by: Peter Xu <peterx@redhat.com> --- include/sysemu/kvm.h | 9 +++++++++ kvm-all.c | 15 +++++++++++++++ 2 files changed, 24 insertions(+) diff --git a/include/sysemu/kvm.h b/include/sysemu/kvm.h index 6695fa7..0e18f15 100644 --- a/include/sysemu/kvm.h +++ b/include/sysemu/kvm.h @@ -306,6 +306,15 @@ void kvm_device_access(int fd, int group, uint64_t attr, */ int kvm_create_device(KVMState *s, uint64_t type, bool test); +/** + * kvm_device_supported - probe whether KVM supports specific device + * + * @vmfd: The fd handler for VM + * @type: type of device + * + * @return: true if supported, otherwise false. + */ +bool kvm_device_supported(int vmfd, uint64_t type); /* Arch specific hooks */ diff --git a/kvm-all.c b/kvm-all.c index 44c0464..e7b66df 100644 --- a/kvm-all.c +++ b/kvm-all.c @@ -2339,6 +2339,21 @@ int kvm_create_device(KVMState *s, uint64_t type, bool test) return test ? 0 : create_dev.fd; } +bool kvm_device_supported(int vmfd, uint64_t type) +{ + struct kvm_create_device create_dev = { + .type = type, + .fd = -1, + .flags = KVM_CREATE_DEVICE_TEST, + }; + + if (ioctl(vmfd, KVM_CHECK_EXTENSION, KVM_CAP_DEVICE_CTRL) <= 0) { + return false; + } + + return (ioctl(vmfd, KVM_CREATE_DEVICE, &create_dev) >= 0); +} + int kvm_set_one_reg(CPUState *cs, uint64_t id, void *source) { struct kvm_one_reg reg; -- 2.4.3
On 24/03/16 05:55, Peter Xu wrote:
This can be used when probing whether KVM support specific device. Here, a raw vmfd is used.
Signed-off-by: Peter Xu <peterx@redhat.com>
Acked-by: Sergey Fedorov <serge.fdrv@gmail.com> Kind regards, Sergey
--- include/sysemu/kvm.h | 9 +++++++++ kvm-all.c | 15 +++++++++++++++ 2 files changed, 24 insertions(+)
diff --git a/include/sysemu/kvm.h b/include/sysemu/kvm.h index 6695fa7..0e18f15 100644 --- a/include/sysemu/kvm.h +++ b/include/sysemu/kvm.h @@ -306,6 +306,15 @@ void kvm_device_access(int fd, int group, uint64_t attr, */ int kvm_create_device(KVMState *s, uint64_t type, bool test);
+/** + * kvm_device_supported - probe whether KVM supports specific device + * + * @vmfd: The fd handler for VM + * @type: type of device + * + * @return: true if supported, otherwise false. + */ +bool kvm_device_supported(int vmfd, uint64_t type);
/* Arch specific hooks */
diff --git a/kvm-all.c b/kvm-all.c index 44c0464..e7b66df 100644 --- a/kvm-all.c +++ b/kvm-all.c @@ -2339,6 +2339,21 @@ int kvm_create_device(KVMState *s, uint64_t type, bool test) return test ? 0 : create_dev.fd; }
+bool kvm_device_supported(int vmfd, uint64_t type) +{ + struct kvm_create_device create_dev = { + .type = type, + .fd = -1, + .flags = KVM_CREATE_DEVICE_TEST, + }; + + if (ioctl(vmfd, KVM_CHECK_EXTENSION, KVM_CAP_DEVICE_CTRL) <= 0) { + return false; + } + + return (ioctl(vmfd, KVM_CREATE_DEVICE, &create_dev) >= 0); +} + int kvm_set_one_reg(CPUState *cs, uint64_t id, void *source) { struct kvm_one_reg reg;
For emulated GIC capabilities, currently only gicv2 is supported. We need to add gicv3 in when emulated gicv3 ready. For KVM accelerated ARM VM, we detect the capability bits by creating a scratch VM. Signed-off-by: Peter Xu <peterx@redhat.com> --- target-arm/monitor.c | 58 +++++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 57 insertions(+), 1 deletion(-) diff --git a/target-arm/monitor.c b/target-arm/monitor.c index 4c9bef3..1ee59a2 100644 --- a/target-arm/monitor.c +++ b/target-arm/monitor.c @@ -21,8 +21,64 @@ */ #include "qemu/osdep.h" #include "qmp-commands.h" +#include "hw/boards.h" +#include "kvm_arm.h" + +static GICCapability *gic_cap_new(int version) +{ + GICCapability *cap = g_new0(GICCapability, 1); + cap->version = version; + /* by default, support none */ + cap->emulated = false; + cap->kernel = false; + return cap; +} + +static GICCapabilityList *gic_cap_list_add(GICCapabilityList *head, + GICCapability *cap) +{ + GICCapabilityList *item = g_new0(GICCapabilityList, 1); + item->value = cap; + item->next = head; + return item; +} + +static inline void gic_cap_kvm_probe(GICCapability *v2, GICCapability *v3) +{ +#ifdef CONFIG_KVM + int fdarray[3]; + + if (!kvm_arm_create_scratch_host_vcpu(NULL, fdarray, NULL)) { + return; + } + + /* Test KVM GICv2 */ + if (kvm_device_supported(fdarray[1], KVM_DEV_TYPE_ARM_VGIC_V2)) { + v2->kernel = true; + } + + /* Test KVM GICv3 */ + if (kvm_device_supported(fdarray[1], KVM_DEV_TYPE_ARM_VGIC_V3)) { + v3->kernel = true; + } + + kvm_arm_destroy_scratch_host_vcpu(fdarray); +#endif +} GICCapabilityList *qmp_query_gic_capabilities(Error **errp) { - return NULL; + GICCapabilityList *head = NULL; + GICCapability *v2 = gic_cap_new(2), *v3 = gic_cap_new(3); + + v2->emulated = true; + /* TODO: we'd change to true after we get emulated GICv3. */ + v3->emulated = false; + + gic_cap_kvm_probe(v2, v3); + + head = gic_cap_list_add(head, v2); + head = gic_cap_list_add(head, v3); + + return head; } -- 2.4.3
On 24/03/16 05:55, Peter Xu wrote:
For emulated GIC capabilities, currently only gicv2 is supported. We need to add gicv3 in when emulated gicv3 ready. For KVM accelerated ARM VM, we detect the capability bits by creating a scratch VM.
Signed-off-by: Peter Xu <peterx@redhat.com>
Acked-by: Sergey Fedorov <serge.fdrv@gmail.com> Kind regards, Sergey
--- target-arm/monitor.c | 58 +++++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 57 insertions(+), 1 deletion(-)
diff --git a/target-arm/monitor.c b/target-arm/monitor.c index 4c9bef3..1ee59a2 100644 --- a/target-arm/monitor.c +++ b/target-arm/monitor.c @@ -21,8 +21,64 @@ */ #include "qemu/osdep.h" #include "qmp-commands.h" +#include "hw/boards.h" +#include "kvm_arm.h" + +static GICCapability *gic_cap_new(int version) +{ + GICCapability *cap = g_new0(GICCapability, 1); + cap->version = version; + /* by default, support none */ + cap->emulated = false; + cap->kernel = false; + return cap; +} + +static GICCapabilityList *gic_cap_list_add(GICCapabilityList *head, + GICCapability *cap) +{ + GICCapabilityList *item = g_new0(GICCapabilityList, 1); + item->value = cap; + item->next = head; + return item; +} + +static inline void gic_cap_kvm_probe(GICCapability *v2, GICCapability *v3) +{ +#ifdef CONFIG_KVM + int fdarray[3]; + + if (!kvm_arm_create_scratch_host_vcpu(NULL, fdarray, NULL)) { + return; + } + + /* Test KVM GICv2 */ + if (kvm_device_supported(fdarray[1], KVM_DEV_TYPE_ARM_VGIC_V2)) { + v2->kernel = true; + } + + /* Test KVM GICv3 */ + if (kvm_device_supported(fdarray[1], KVM_DEV_TYPE_ARM_VGIC_V3)) { + v3->kernel = true; + } + + kvm_arm_destroy_scratch_host_vcpu(fdarray); +#endif +}
GICCapabilityList *qmp_query_gic_capabilities(Error **errp) { - return NULL; + GICCapabilityList *head = NULL; + GICCapability *v2 = gic_cap_new(2), *v3 = gic_cap_new(3); + + v2->emulated = true; + /* TODO: we'd change to true after we get emulated GICv3. */ + v3->emulated = false; + + gic_cap_kvm_probe(v2, v3); + + head = gic_cap_list_add(head, v2); + head = gic_cap_list_add(head, v3); + + return head; }
On 24 March 2016 at 02:55, Peter Xu <peterx@redhat.com> wrote:
This patch is to add ARM-specific command "query-gic-capability".
The new command can report which kind of GIC device the host/QEMU support. The returned result is in the form of array.
Sample command and output:
{"execute": "query-gic-capability"} {"return": [{"emulated": false, "version": 3, "kernel": false}, {"emulated": true, "version": 2, "kernel": true}]}
I really strongly want this in 2.6 but we are very very nearly out of time. Eric or Markus, could I trouble one of you to provide an ack for this series for the QMP protocol aspects sometime before end of Wednesday, please? I also don't think I've seen an indication from the libvirt folks that this is definitely the API they want; did I miss it? thanks -- PMM
On Tue, 2016-03-29 at 21:10 +0100, Peter Maydell wrote:
On 24 March 2016 at 02:55, Peter Xu <peterx@redhat.com> wrote:
This patch is to add ARM-specific command "query-gic-capability".
The new command can report which kind of GIC device the host/QEMU support. The returned result is in the form of array.
Sample command and output:
{"execute": "query-gic-capability"} {"return": [{"emulated": false, "version": 3, "kernel": false}, {"emulated": true, "version": 2, "kernel": true}]}
I really strongly want this in 2.6 but we are very very nearly out of time. Eric or Markus, could I trouble one of you to provide an ack for this series for the QMP protocol aspects sometime before end of Wednesday, please?
I also don't think I've seen an indication from the libvirt folks that this is definitely the API they want; did I miss it?
Yes, this works nicely for libvirt; in fact, I've already posted an RFC implementation[1] to the list last week. Cheers. [1] https://www.redhat.com/archives/libvir-list/2016-March/msg00956.html -- Andrea Bolognani Software Engineer - Virtualization Team
On 24 March 2016 at 02:55, Peter Xu <peterx@redhat.com> wrote:
v7 changes: - patch 1 - add more to commit log, about how to use the results [Markus] - change interface from dict back to array [Markus, Eric] - patch 2 - kvm_arm_create_scratch_host_vcpu should raise error when init non-zero, but failed to find a good CPU model [Sergey] - patch 3 - rename function to kvm_device_supported [Sergey] - patch 4 - comment line from "FIXME" to "TODO" [Sergey] - make kvm-related part another function [Sergey] - change result to array to follow patch 1's change
Applied this version to target-arm.next, thanks. -- PMM
participants (5)
-
Andrea Bolognani -
Eric Blake -
Peter Maydell -
Peter Xu -
Sergey Fedorov