Yes, it's been a long time coming, but I really wasn't looking forward
to picking this up again. Anyway...
This (monster of a) series implements full support for GICv4, bringing
direct injection of MSIs to KVM on arm and arm64, assuming you have
the right hardware (which is quite unlikely).
To get an idea of the design, I'd recommend you start with patch #32,
which tries to shed some light on the approach that I've taken. And
before that, please digest some of the GICv3/GICv4 architecture
documentation[1] (less than 800 pages!). Once you feel reasonably
insane, you'll be in the right mood to read the code.
The structure of the series is fairly simple. The initial 34 patches
add some generic support for GICv4, while the rest of the code plugs
KVM into it. This series relies on Eric Auger's irq-bypass series[2],
which is a prerequisite for this work.
The stack has been *very lightly* tested on an arm64 model, with a PCI
virtio block device passed from the host to a guest (using kvmtool and
Jean-Philippe Brucker's excellent VFIO support patches[3]). As it has
never seen any HW, I expect things to be subtly broken, so go forward
and test if you can, though I'm mostly interested in people reviewing
the code at the moment.
I've pushed out a branch based on 4.12-rc6 containing the
dependencies (as well as a couple of debug patches):
git://git.kernel.org/pub/scm/linux/kernel/git/maz/arm-platforms.git kvm-arm64/gicv4-kvm
* From v1:
- The bulk of the 30-something initial patches have seen countless
bugs being fixed, and some key data structures have been subtly
tweaked (or killed altogether). They are still quite similar to
what I had in v1 though.
- The whole KVM code is brand new and as I said above, only lightly
tested.
- Collected a bunch a R-bs from Thomas and Eric (many thanks, guys).
[1] https://static.docs.arm.com/ihi0069/c/IHI0069C_gic_architecture_specification.pdf
[2] http://www.spinics.net/lists/kvm/msg151463.html
[3] http://www.spinics.net/lists/kvm/msg151823.html
Marc Zyngier (52):
genirq: Let irq_set_vcpu_affinity() iterate over hierarchy
irqchip/gic-v3: Add redistributor iterator
irqchip/gic-v3: Add VLPI/DirectLPI discovery
irqchip/gic-v3-its: Move LPI definitions around
irqchip/gic-v3-its: Add probing for VLPI properties
irqchip/gic-v3-its: Macro-ize its_send_single_command
irqchip/gic-v3-its: Implement irq_set_irqchip_state for pending state
irqchip/gic-v3-its: Split out property table allocation
irqchip/gic-v3-its: Allow use of indirect VCPU tables
irqchip/gic-v3-its: Split out pending table allocation
irqchip/gic-v3-its: Rework LPI freeing
irqchip/gic-v3-its: Generalize device table allocation
irqchip/gic-v3-its: Generalize LPI configuration
irqchip/gic-v4: Add management structure definitions
irqchip/gic-v3-its: Add GICv4 ITS command definitions
irqchip/gic-v3-its: Add VLPI configuration hook
irqchip/gic-v3-its: Add VLPI map/unmap operations
irqchip/gic-v3-its: Add VLPI configuration handling
irqchip/gic-v3-its: Add VPE domain infrastructure
irqchip/gic-v3-its: Add VPE irq domain allocation/teardown
irqchip/gic-v3-its: Add VPE irq domain [de]activation
irqchip/gic-v3-its: Add VPENDBASER/VPROPBASER accessors
irqchip/gic-v3-its: Add VPE scheduling
irqchip/gic-v3-its: Add VPE invalidation hook
irqchip/gic-v3-its: Add VPE affinity changes
irqchip/gic-v3-its: Add VPE interrupt masking
irqchip/gic-v3-its: Support VPE doorbell invalidation even when
!DirectLPI
irqchip/gic-v3-its: Set implementation defined bit to enable VLPIs
irqchip/gic-v4: Add per-VM VPE domain creation
irqchip/gic-v4: Add VPE command interface
irqchip/gic-v4: Add VLPI configuration interface
irqchip/gic-v4: Add some basic documentation
irqchip/gic-v4: Enable low-level GICv4 operations
irqchip/gic-v3: Advertise GICv4 support to KVM
KVM: arm/arm64: vgic: Move kvm_vgic_destroy call around
KVM: arm/arm64: vITS: Add MSI translation helpers
KVM: arm/arm64: GICv4: Add init and teardown of the vPE irq domain
KVM: arm/arm64: GICv4: Wire init/teardown of per-VM support
KVM: arm/arm64: GICv4: Wire mapping/unmapping of VLPIs in VFIO irq
bypass
KVM: arm/arm64: GICv4: Handle INT command applied to a VLPI
KVM: arm/arm64: GICv4: Unmap VLPI when freeing an LPI
KVM: arm/arm64: GICv4: Handle MOVI applied to a VLPI
KVM: arm/arm64: GICv4: Handle CLEAR applied to a VLPI
KVM: arm/arm64: GICv4: Handle MOVALL applied to a vPE
KVM: arm/arm64: GICv4: Propagate property updates to VLPIs
KVM: arm/arm64: GICv4: Handle INVALL applied to a vPE
KVM: arm/arm64: GICv4: Propagate VLPI properties at map time
KVM: arm/arm64: GICv4: Add doorbell interrupt handling
KVM: arm/arm64: GICv4: Hook vPE scheduling into vgic flush/sync
KVM: arm/arm64: GICv4: Enable virtual cpuif if VLPIs can be delivered
KVM: arm/arm64: GICv4: Use pending_last as a scheduling hint
KVM: arm/arm64: GICv4: Enable VLPI support
arch/arm/include/asm/arch_gicv3.h | 33 +
arch/arm/kvm/Makefile | 1 +
arch/arm64/include/asm/arch_gicv3.h | 6 +
arch/arm64/kvm/Makefile | 1 +
drivers/irqchip/Makefile | 2 +-
drivers/irqchip/irq-gic-v3-its.c | 1288 +++++++++++++++++++++++++++++---
drivers/irqchip/irq-gic-v3.c | 93 ++-
drivers/irqchip/irq-gic-v4.c | 209 ++++++
include/kvm/arm_vgic.h | 19 +
include/linux/irqchip/arm-gic-common.h | 2 +
include/linux/irqchip/arm-gic-v3.h | 84 +++
include/linux/irqchip/arm-gic-v4.h | 102 +++
kernel/irq/manage.c | 14 +-
virt/kvm/arm/arm.c | 31 +-
virt/kvm/arm/hyp/vgic-v3-sr.c | 9 +-
virt/kvm/arm/vgic/vgic-init.c | 11 +-
virt/kvm/arm/vgic/vgic-its.c | 165 ++--
virt/kvm/arm/vgic/vgic-v3.c | 6 +
virt/kvm/arm/vgic/vgic-v4.c | 210 ++++++
virt/kvm/arm/vgic/vgic.c | 7 +
virt/kvm/arm/vgic/vgic.h | 10 +
21 files changed, 2103 insertions(+), 200 deletions(-)
create mode 100644 drivers/irqchip/irq-gic-v4.c
create mode 100644 include/linux/irqchip/arm-gic-v4.h
create mode 100644 virt/kvm/arm/vgic/vgic-v4.c
--
2.11.0
Add helper functions that probe for VLPI and DirectLPI properties.
Reviewed-by: Eric Auger <[email protected]>
Reviewed-by: Thomas Gleixner <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3.c | 22 ++++++++++++++++++++++
include/linux/irqchip/arm-gic-v3.h | 3 +++
2 files changed, 25 insertions(+)
diff --git a/drivers/irqchip/irq-gic-v3.c b/drivers/irqchip/irq-gic-v3.c
index bc58a4339a97..efbd998baa9c 100644
--- a/drivers/irqchip/irq-gic-v3.c
+++ b/drivers/irqchip/irq-gic-v3.c
@@ -504,6 +504,24 @@ static int gic_populate_rdist(void)
return -ENODEV;
}
+static int __gic_update_vlpi_properties(struct redist_region *region,
+ void __iomem *ptr)
+{
+ u64 typer = gic_read_typer(ptr + GICR_TYPER);
+ gic_data.rdists.has_vlpis &= !!(typer & GICR_TYPER_VLPIS);
+ gic_data.rdists.has_direct_lpi &= !!(typer & GICR_TYPER_DirectLPIS);
+
+ return 1;
+}
+
+static void gic_update_vlpi_properties(void)
+{
+ gic_iterate_rdists(__gic_update_vlpi_properties);
+ pr_info("%sVLPI support, %sdirect LPI support\n",
+ !gic_data.rdists.has_vlpis ? "no " : "",
+ !gic_data.rdists.has_direct_lpi ? "no " : "");
+}
+
static void gic_cpu_sys_reg_init(void)
{
/*
@@ -965,6 +983,8 @@ static int __init gic_init_bases(void __iomem *dist_base,
gic_data.domain = irq_domain_create_tree(handle, &gic_irq_domain_ops,
&gic_data);
gic_data.rdists.rdist = alloc_percpu(typeof(*gic_data.rdists.rdist));
+ gic_data.rdists.has_vlpis = true;
+ gic_data.rdists.has_direct_lpi = true;
if (WARN_ON(!gic_data.domain) || WARN_ON(!gic_data.rdists.rdist)) {
err = -ENOMEM;
@@ -973,6 +993,8 @@ static int __init gic_init_bases(void __iomem *dist_base,
set_handle_irq(gic_handle_irq);
+ gic_update_vlpi_properties();
+
if (IS_ENABLED(CONFIG_ARM_GIC_V3_ITS) && gic_dist_supports_lpis())
its_init(handle, &gic_data.rdists, gic_data.domain);
diff --git a/include/linux/irqchip/arm-gic-v3.h b/include/linux/irqchip/arm-gic-v3.h
index 6a1f87ff94e2..20a553423ac7 100644
--- a/include/linux/irqchip/arm-gic-v3.h
+++ b/include/linux/irqchip/arm-gic-v3.h
@@ -204,6 +204,7 @@
#define GICR_TYPER_PLPIS (1U << 0)
#define GICR_TYPER_VLPIS (1U << 1)
+#define GICR_TYPER_DirectLPIS (1U << 3)
#define GICR_TYPER_LAST (1U << 4)
#define GIC_V3_REDIST_SIZE 0x20000
@@ -487,6 +488,8 @@ struct rdists {
struct page *prop_page;
int id_bits;
u64 flags;
+ bool has_vlpis;
+ bool has_direct_lpi;
};
struct irq_domain;
--
2.11.0
When assigning an interrupt to a vcpu, it is not unlikely that
the level of the hierarchy implementing irq_set_vcpu_affinity
is not the top level (think a generic MSI domain on top of a
virtualization aware interrupt controller).
In such a case, let's iterate over the hierarchy until we find
an irqchip implementing it.
Signed-off-by: Marc Zyngier <[email protected]>
---
kernel/irq/manage.c | 14 ++++++++++++--
1 file changed, 12 insertions(+), 2 deletions(-)
diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c
index 425170d4439b..804fb94e86e6 100644
--- a/kernel/irq/manage.c
+++ b/kernel/irq/manage.c
@@ -431,8 +431,18 @@ int irq_set_vcpu_affinity(unsigned int irq, void *vcpu_info)
return -EINVAL;
data = irq_desc_get_irq_data(desc);
- chip = irq_data_get_irq_chip(data);
- if (chip && chip->irq_set_vcpu_affinity)
+ do {
+ chip = irq_data_get_irq_chip(data);
+ if (chip && chip->irq_set_vcpu_affinity)
+ break;
+#ifdef CONFIG_IRQ_DOMAIN_HIERARCHY
+ data = data->parent_data;
+#else
+ data = NULL;
+#endif
+ } while (data);
+
+ if (data)
ret = chip->irq_set_vcpu_affinity(data, vcpu_info);
irq_put_desc_unlock(desc, flags);
--
2.11.0
Allow the pending state of an LPI to be set or cleared via
irq_set_irqchip_state.
Reviewed-by: Thomas Gleixner <[email protected]>
Reviewed-by: Eric Auger <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3-its.c | 78 ++++++++++++++++++++++++++++++++++++++++
1 file changed, 78 insertions(+)
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index bbf7bb89aa62..391b87572549 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -169,6 +169,11 @@ struct its_cmd_desc {
struct {
struct its_device *dev;
u32 event_id;
+ } its_clear_cmd;
+
+ struct {
+ struct its_device *dev;
+ u32 event_id;
} its_int_cmd;
struct {
@@ -382,6 +387,40 @@ static struct its_collection *its_build_inv_cmd(struct its_cmd_block *cmd,
return col;
}
+static struct its_collection *its_build_int_cmd(struct its_cmd_block *cmd,
+ struct its_cmd_desc *desc)
+{
+ struct its_collection *col;
+
+ col = dev_event_to_col(desc->its_int_cmd.dev,
+ desc->its_int_cmd.event_id);
+
+ its_encode_cmd(cmd, GITS_CMD_INT);
+ its_encode_devid(cmd, desc->its_int_cmd.dev->device_id);
+ its_encode_event_id(cmd, desc->its_int_cmd.event_id);
+
+ its_fixup_cmd(cmd);
+
+ return col;
+}
+
+static struct its_collection *its_build_clear_cmd(struct its_cmd_block *cmd,
+ struct its_cmd_desc *desc)
+{
+ struct its_collection *col;
+
+ col = dev_event_to_col(desc->its_clear_cmd.dev,
+ desc->its_clear_cmd.event_id);
+
+ its_encode_cmd(cmd, GITS_CMD_CLEAR);
+ its_encode_devid(cmd, desc->its_clear_cmd.dev->device_id);
+ its_encode_event_id(cmd, desc->its_clear_cmd.event_id);
+
+ its_fixup_cmd(cmd);
+
+ return col;
+}
+
static struct its_collection *its_build_invall_cmd(struct its_cmd_block *cmd,
struct its_cmd_desc *desc)
{
@@ -538,6 +577,26 @@ static void its_build_sync_cmd(struct its_cmd_block *sync_cmd,
static BUILD_SINGLE_CMD_FUNC(its_send_single_command, its_cmd_builder_t,
struct its_collection, its_build_sync_cmd)
+static void its_send_int(struct its_device *dev, u32 event_id)
+{
+ struct its_cmd_desc desc;
+
+ desc.its_int_cmd.dev = dev;
+ desc.its_int_cmd.event_id = event_id;
+
+ its_send_single_command(dev->its, its_build_int_cmd, &desc);
+}
+
+static void its_send_clear(struct its_device *dev, u32 event_id)
+{
+ struct its_cmd_desc desc;
+
+ desc.its_clear_cmd.dev = dev;
+ desc.its_clear_cmd.event_id = event_id;
+
+ its_send_single_command(dev->its, its_build_clear_cmd, &desc);
+}
+
static void its_send_inv(struct its_device *dev, u32 event_id)
{
struct its_cmd_desc desc;
@@ -701,6 +760,24 @@ static void its_irq_compose_msi_msg(struct irq_data *d, struct msi_msg *msg)
iommu_dma_map_msi_msg(d->irq, msg);
}
+static int its_irq_set_irqchip_state(struct irq_data *d,
+ enum irqchip_irq_state which,
+ bool state)
+{
+ struct its_device *its_dev = irq_data_get_irq_chip_data(d);
+ u32 event = its_get_event_id(d);
+
+ if (which != IRQCHIP_STATE_PENDING)
+ return -EINVAL;
+
+ if (state)
+ its_send_int(its_dev, event);
+ else
+ its_send_clear(its_dev, event);
+
+ return 0;
+}
+
static struct irq_chip its_irq_chip = {
.name = "ITS",
.irq_mask = its_mask_irq,
@@ -708,6 +785,7 @@ static struct irq_chip its_irq_chip = {
.irq_eoi = irq_chip_eoi_parent,
.irq_set_affinity = its_set_affinity,
.irq_compose_msi_msg = its_irq_compose_msi_msg,
+ .irq_set_irqchip_state = its_irq_set_irqchip_state,
};
/*
--
2.11.0
Move the LPI property table allocation into its own function, as
this is going to be required for those associated with VMs in
the future.
Reviewed-by: Eric Auger <[email protected]>
Reviewed-by: Thomas Gleixner <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3-its.c | 28 ++++++++++++++++++----------
1 file changed, 18 insertions(+), 10 deletions(-)
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index 391b87572549..2feecdf40216 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -897,15 +897,31 @@ static void its_lpi_free(struct event_lpi_map *map)
kfree(map->col_map);
}
+static struct page *its_allocate_prop_table(gfp_t gfp_flags)
+{
+ struct page *prop_page;
+
+ prop_page = alloc_pages(gfp_flags, get_order(LPI_PROPBASE_SZ));
+ if (!prop_page)
+ return NULL;
+
+ /* Priority 0xa0, Group-1, disabled */
+ memset(page_address(prop_page),
+ LPI_PROP_DEFAULT_PRIO | LPI_PROP_GROUP1,
+ LPI_PROPBASE_SZ);
+
+ /* Make sure the GIC will observe the written configuration */
+ gic_flush_dcache_to_poc(page_address(prop_page), LPI_PROPBASE_SZ);
+ return prop_page;
+}
static int __init its_alloc_lpi_tables(void)
{
phys_addr_t paddr;
- gic_rdists->prop_page = alloc_pages(GFP_NOWAIT,
- get_order(LPI_PROPBASE_SZ));
+ gic_rdists->prop_page = its_allocate_prop_table(GFP_NOWAIT);
if (!gic_rdists->prop_page) {
pr_err("Failed to allocate PROPBASE\n");
return -ENOMEM;
@@ -914,14 +930,6 @@ static int __init its_alloc_lpi_tables(void)
paddr = page_to_phys(gic_rdists->prop_page);
pr_info("GIC: using LPI property table @%pa\n", &paddr);
- /* Priority 0xa0, Group-1, disabled */
- memset(page_address(gic_rdists->prop_page),
- LPI_PROP_DEFAULT_PRIO | LPI_PROP_GROUP1,
- LPI_PROPBASE_SZ);
-
- /* Make sure the GIC will observe the written configuration */
- gic_flush_dcache_to_poc(page_address(gic_rdists->prop_page), LPI_PROPBASE_SZ);
-
return 0;
}
--
2.11.0
A long time ago, GITS_CTLR[1] used to be called GITC_CTLR.EnableVLPI.
It has been subsequently deprecated and is now an "Implementation
Defined" bit that may ot may not be set for GICv4. Brilliant.
And the current crop of the FastModel requires that bit for VLPIs
to be enabled. Oh well... Let's set it and find out what breaks.
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3-its.c | 7 +++++--
include/linux/irqchip/arm-gic-v3.h | 1 +
2 files changed, 6 insertions(+), 2 deletions(-)
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index 4979916cd3d7..3784bea2b994 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -2552,7 +2552,7 @@ static int its_force_quiescent(void __iomem *base)
return 0;
/* Disable the generation of all interrupts to this ITS */
- val &= ~GITS_CTLR_ENABLE;
+ val &= ~(GITS_CTLR_ENABLE | GITS_CTLR_ImDe);
writel_relaxed(val, base + GITS_CTLR);
/* Poll GITS_CTLR and wait until ITS becomes quiescent */
@@ -2818,7 +2818,10 @@ static int __init its_probe_one(struct resource *res,
gits_write_cwriter(0, its->base + GITS_CWRITER);
ctlr = readl_relaxed(its->base + GITS_CTLR);
- writel_relaxed(ctlr | GITS_CTLR_ENABLE, its->base + GITS_CTLR);
+ ctlr |= GITS_CTLR_ENABLE;
+ if (its->is_v4)
+ ctlr |= GITS_CTLR_ImDe;
+ writel_relaxed(ctlr, its->base + GITS_CTLR);
err = its_init_domain(handle, its);
if (err)
diff --git a/include/linux/irqchip/arm-gic-v3.h b/include/linux/irqchip/arm-gic-v3.h
index 6bc142cfa616..1ea576c8126f 100644
--- a/include/linux/irqchip/arm-gic-v3.h
+++ b/include/linux/irqchip/arm-gic-v3.h
@@ -298,6 +298,7 @@
#define GITS_TRANSLATER 0x10040
#define GITS_CTLR_ENABLE (1U << 0)
+#define GITS_CTLR_ImDe (1U << 1)
#define GITS_CTLR_ITS_NUMBER_SHIFT 4
#define GITS_CTLR_ITS_NUMBER (0xFU << GITS_CTLR_ITS_NUMBER_SHIFT)
#define GITS_CTLR_QUIESCENT (1U << 31)
--
2.11.0
When masking/unmasking a doorbell interrupt, it is necessary
to issue an invalidation to the corresponding redistributor.
We use the DirectLPI feature by writting directly to the corresponding
redistributor.
Reviewed-by: Thomas Gleixner <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
arch/arm/include/asm/arch_gicv3.h | 5 +++++
arch/arm64/include/asm/arch_gicv3.h | 1 +
drivers/irqchip/irq-gic-v3-its.c | 26 +++++++++++++++++++++++++-
3 files changed, 31 insertions(+), 1 deletion(-)
diff --git a/arch/arm/include/asm/arch_gicv3.h b/arch/arm/include/asm/arch_gicv3.h
index 8d45e88feac9..0735888fe42e 100644
--- a/arch/arm/include/asm/arch_gicv3.h
+++ b/arch/arm/include/asm/arch_gicv3.h
@@ -276,6 +276,11 @@ static inline u64 __gic_readq_nonatomic(const volatile void __iomem *addr)
#define gicr_write_pendbaser(v, c) __gic_writeq_nonatomic(v, c)
/*
+ * GICR_INVLPIR - only the lower bits are significant
+ */
+#define gic_write_invlpir(v, c) writel_relaxed(lower_32_bits(v), c)
+
+/*
* GITS_TYPER is an ID register and doesn't need atomicity.
*/
#define gits_read_typer(c) __gic_readq_nonatomic(c)
diff --git a/arch/arm64/include/asm/arch_gicv3.h b/arch/arm64/include/asm/arch_gicv3.h
index 0d2a53457c30..019128dd37e4 100644
--- a/arch/arm64/include/asm/arch_gicv3.h
+++ b/arch/arm64/include/asm/arch_gicv3.h
@@ -116,6 +116,7 @@ static inline void gic_write_bpr1(u32 val)
#define gic_read_typer(c) readq_relaxed(c)
#define gic_write_irouter(v, c) writeq_relaxed(v, c)
+#define gic_write_invlpir(v, c) writeq_relaxed(v, c)
#define gic_flush_dcache_to_poc(a,l) __flush_dcache_area((a), (l))
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index 281eed031e01..64893f0427f2 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -969,12 +969,12 @@ static inline u32 its_get_event_id(struct irq_data *d)
static void lpi_write_config(struct irq_data *d, u8 clr, u8 set)
{
- struct its_device *its_dev = irq_data_get_irq_chip_data(d);
irq_hw_number_t hwirq;
struct page *prop_page;
u8 *cfg;
if (irqd_is_forwarded_to_vcpu(d)) {
+ struct its_device *its_dev = irq_data_get_irq_chip_data(d);
u32 event = its_get_event_id(d);
prop_page = its_dev->event_map.vm->vprop_page;
@@ -2315,8 +2315,32 @@ static int its_vpe_set_vcpu_affinity(struct irq_data *d, void *vcpu_info)
}
}
+static void its_vpe_send_inv(struct irq_data *d)
+{
+ struct its_vpe *vpe = irq_data_get_irq_chip_data(d);
+ void __iomem *rdbase;
+
+ rdbase = per_cpu_ptr(gic_rdists->rdist, vpe->col_idx)->rd_base;
+ gic_write_invlpir(d->hwirq, rdbase + GICR_INVLPIR);
+}
+
+static void its_vpe_mask_irq(struct irq_data *d)
+{
+ lpi_write_config(d, LPI_PROP_ENABLED, 0);
+ its_vpe_send_inv(d);
+}
+
+static void its_vpe_unmask_irq(struct irq_data *d)
+{
+ lpi_write_config(d, 0, LPI_PROP_ENABLED);
+ its_vpe_send_inv(d);
+}
+
static struct irq_chip its_vpe_irq_chip = {
.name = "GICv4-vpe",
+ .irq_mask = its_vpe_mask_irq,
+ .irq_unmask = its_vpe_unmask_irq,
+ .irq_eoi = irq_chip_eoi_parent,
.irq_set_affinity = its_vpe_set_affinity,
.irq_set_vcpu_affinity = its_vpe_set_vcpu_affinity,
};
--
2.11.0
Add the required interfaces to schedule a VPE and perform a
VINVALL command.
Reviewed-by: Thomas Gleixner <[email protected]>
Reviewed-by: Eric Auger <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v4.c | 25 +++++++++++++++++++++++++
include/linux/irqchip/arm-gic-v4.h | 2 ++
2 files changed, 27 insertions(+)
diff --git a/drivers/irqchip/irq-gic-v4.c b/drivers/irqchip/irq-gic-v4.c
index 2ea12baf2e53..3c5ad70193ed 100644
--- a/drivers/irqchip/irq-gic-v4.c
+++ b/drivers/irqchip/irq-gic-v4.c
@@ -56,3 +56,28 @@ void its_free_vcpu_irqs(struct its_vm *vm)
irq_domain_free_irqs(vm->vpes[0]->irq, vm->nr_vpes);
irq_domain_remove(vm->domain);
}
+
+static int its_send_vpe_cmd(struct its_vpe *vpe, struct its_cmd_info *info)
+{
+ return irq_set_vcpu_affinity(vpe->irq, info);
+}
+
+int its_schedule_vpe(struct its_vpe *vpe, bool on)
+{
+ struct its_cmd_info info;
+
+ WARN_ON(preemptible());
+
+ info.cmd_type = on ? SCHEDULE_VPE : DESCHEDULE_VPE;
+
+ return its_send_vpe_cmd(vpe, &info);
+}
+
+int its_invall_vpe(struct its_vpe *vpe)
+{
+ struct its_cmd_info info = {
+ .cmd_type = INVALL_VPE,
+ };
+
+ return its_send_vpe_cmd(vpe, &info);
+}
diff --git a/include/linux/irqchip/arm-gic-v4.h b/include/linux/irqchip/arm-gic-v4.h
index 7474a981ed6d..97bb97a11fef 100644
--- a/include/linux/irqchip/arm-gic-v4.h
+++ b/include/linux/irqchip/arm-gic-v4.h
@@ -90,5 +90,7 @@ struct its_cmd_info {
int its_alloc_vcpu_irqs(struct its_vm *vm);
void its_free_vcpu_irqs(struct its_vm *vm);
+int its_schedule_vpe(struct its_vpe *vpe, bool on);
+int its_invall_vpe(struct its_vpe *vpe);
#endif
--
2.11.0
When we don't have the DirectLPI feature, we must work around the
architecture shortcomings to be able to perform the required
invalidation.
For this, we create a fake device whose sole purpose is to
provide a way to issue a map/inv/unmap sequence (and the corresponding
sync operations). That's 6 commands and a full serialization point
to be able to do this.
You just have to hope the hypervisor won't do that too often...
Reviewed-by: Thomas Gleixner <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3-its.c | 66 ++++++++++++++++++++++++++++++++++++++--
1 file changed, 63 insertions(+), 3 deletions(-)
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index 64893f0427f2..4979916cd3d7 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -138,6 +138,9 @@ struct its_device {
u32 device_id;
};
+static struct its_device *vpe_proxy_dev;
+static DEFINE_RAW_SPINLOCK(vpe_proxy_dev_lock);
+
static LIST_HEAD(its_nodes);
static DEFINE_SPINLOCK(its_lock);
static struct rdists *gic_rdists;
@@ -2074,6 +2077,14 @@ static int its_msi_prepare(struct irq_domain *domain, struct device *dev,
msi_info = msi_get_domain_info(domain);
its = msi_info->data;
+ if (its->is_v4 && !gic_rdists->has_direct_lpi &&
+ dev_id == vpe_proxy_dev->device_id) {
+ /* Bad luck. Get yourself a better implementation */
+ WARN_ONCE(1, "DevId %x clashes with GICv4 VPE proxy device\n",
+ dev_id);
+ return -EINVAL;
+ }
+
its_dev = its_find_device(its, dev_id);
if (its_dev) {
/*
@@ -2318,10 +2329,38 @@ static int its_vpe_set_vcpu_affinity(struct irq_data *d, void *vcpu_info)
static void its_vpe_send_inv(struct irq_data *d)
{
struct its_vpe *vpe = irq_data_get_irq_chip_data(d);
- void __iomem *rdbase;
- rdbase = per_cpu_ptr(gic_rdists->rdist, vpe->col_idx)->rd_base;
- gic_write_invlpir(d->hwirq, rdbase + GICR_INVLPIR);
+ if (gic_rdists->has_direct_lpi) {
+ void __iomem *rdbase;
+
+ rdbase = per_cpu_ptr(gic_rdists->rdist, vpe->col_idx)->rd_base;
+ gic_write_invlpir(d->hwirq, rdbase + GICR_INVLPIR);
+ } else {
+ /*
+ * This is insane.
+ *
+ * If a GICv4 doesn't implement Direct LPIs, the only
+ * way to perform an invalidate is to use a fake
+ * device to issue a MAP/INV/UNMAP sequence. Since
+ * each of these commands has a sync operation, this
+ * is really fast. Not.
+ *
+ * We always use event 0, and thus serialize all VPE
+ * invalidations in the system.
+ *
+ * Broken by design(tm).
+ */
+ unsigned long flags;
+
+ raw_spin_lock_irqsave(&vpe_proxy_dev_lock, flags);
+
+ vpe_proxy_dev->event_map.col_map[0] = vpe->col_idx;
+ its_send_mapti(vpe_proxy_dev, vpe->vpe_db_lpi, 0);
+ its_send_inv(vpe_proxy_dev, 0);
+ its_send_discard(vpe_proxy_dev, 0);
+
+ raw_spin_unlock_irqrestore(&vpe_proxy_dev_lock, flags);
+ }
}
static void its_vpe_mask_irq(struct irq_data *d)
@@ -2616,6 +2655,27 @@ static int its_init_domain(struct fwnode_handle *handle, struct its_node *its)
static int its_init_vpe_domain(void)
{
+ struct its_node *its;
+ u32 devid;
+
+ if (gic_rdists->has_direct_lpi) {
+ pr_info("ITS: Using DirectLPI for VPE invalidation\n");
+ return 0;
+ }
+
+ /* Any ITS will do, even if not v4 */
+ its = list_first_entry(&its_nodes, struct its_node, entry);
+
+ /* Use the last possible DevID */
+ devid = GENMASK(its->device_ids - 1, 0);
+ vpe_proxy_dev = its_create_device(its, devid, 1);
+ if (!vpe_proxy_dev) {
+ pr_err("ITS: Can't allocate GICv4 proxy device\n");
+ return -ENODEV;
+ }
+
+ pr_info("ITS: Allocated DevID %x as GICv4 proxy device\n", devid);
+
return 0;
}
--
2.11.0
The way we call kvm_vgic_destroy is a bit bizarre. We call it
*after* having freed the vcpus, which sort of defeats the point
of cleaning up things before that point.
Let's move kvm_vgic_destroy towards the beginning of kvm_arch_destroy_vm,
which seems more sensible.
Signed-off-by: Marc Zyngier <[email protected]>
---
virt/kvm/arm/arm.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/virt/kvm/arm/arm.c b/virt/kvm/arm/arm.c
index 7846a578db1a..ebab6c29e3be 100644
--- a/virt/kvm/arm/arm.c
+++ b/virt/kvm/arm/arm.c
@@ -177,6 +177,8 @@ void kvm_arch_destroy_vm(struct kvm *kvm)
{
int i;
+ kvm_vgic_destroy(kvm);
+
free_percpu(kvm->arch.last_vcpu_ran);
kvm->arch.last_vcpu_ran = NULL;
@@ -186,8 +188,6 @@ void kvm_arch_destroy_vm(struct kvm *kvm)
kvm->vcpus[i] = NULL;
}
}
-
- kvm_vgic_destroy(kvm);
}
int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
--
2.11.0
In order to control the GICv4 view of virtual CPUs, we rely
on an irqdomain allocated for that purpose. Let's add a couple
of helpers to that effect.
At the same time, the vgic data structures gain new fields to
track all this... erm... wonderful stuff.
Signed-off-by: Marc Zyngier <[email protected]>
---
arch/arm/kvm/Makefile | 1 +
arch/arm64/kvm/Makefile | 1 +
include/kvm/arm_vgic.h | 11 +++++++++
virt/kvm/arm/vgic/vgic-v4.c | 59 +++++++++++++++++++++++++++++++++++++++++++++
virt/kvm/arm/vgic/vgic.h | 3 +++
5 files changed, 75 insertions(+)
create mode 100644 virt/kvm/arm/vgic/vgic-v4.c
diff --git a/arch/arm/kvm/Makefile b/arch/arm/kvm/Makefile
index d9beee652d36..0a1dd2cdb928 100644
--- a/arch/arm/kvm/Makefile
+++ b/arch/arm/kvm/Makefile
@@ -31,6 +31,7 @@ obj-y += $(KVM)/arm/vgic/vgic-init.o
obj-y += $(KVM)/arm/vgic/vgic-irqfd.o
obj-y += $(KVM)/arm/vgic/vgic-v2.o
obj-y += $(KVM)/arm/vgic/vgic-v3.o
+obj-y += $(KVM)/arm/vgic/vgic-v4.o
obj-y += $(KVM)/arm/vgic/vgic-mmio.o
obj-y += $(KVM)/arm/vgic/vgic-mmio-v2.o
obj-y += $(KVM)/arm/vgic/vgic-mmio-v3.o
diff --git a/arch/arm64/kvm/Makefile b/arch/arm64/kvm/Makefile
index 5d9810086c25..c30fd388ef80 100644
--- a/arch/arm64/kvm/Makefile
+++ b/arch/arm64/kvm/Makefile
@@ -26,6 +26,7 @@ kvm-$(CONFIG_KVM_ARM_HOST) += $(KVM)/arm/vgic/vgic-init.o
kvm-$(CONFIG_KVM_ARM_HOST) += $(KVM)/arm/vgic/vgic-irqfd.o
kvm-$(CONFIG_KVM_ARM_HOST) += $(KVM)/arm/vgic/vgic-v2.o
kvm-$(CONFIG_KVM_ARM_HOST) += $(KVM)/arm/vgic/vgic-v3.o
+kvm-$(CONFIG_KVM_ARM_HOST) += $(KVM)/arm/vgic/vgic-v4.o
kvm-$(CONFIG_KVM_ARM_HOST) += $(KVM)/arm/vgic/vgic-mmio.o
kvm-$(CONFIG_KVM_ARM_HOST) += $(KVM)/arm/vgic/vgic-mmio-v2.o
kvm-$(CONFIG_KVM_ARM_HOST) += $(KVM)/arm/vgic/vgic-mmio-v3.o
diff --git a/include/kvm/arm_vgic.h b/include/kvm/arm_vgic.h
index ee8587c8c715..359eeffe9857 100644
--- a/include/kvm/arm_vgic.h
+++ b/include/kvm/arm_vgic.h
@@ -26,6 +26,8 @@
#include <linux/list.h>
#include <linux/jump_label.h>
+#include <linux/irqchip/arm-gic-v4.h>
+
#define VGIC_V3_MAX_CPUS 255
#define VGIC_V2_MAX_CPUS 8
#define VGIC_NR_IRQS_LEGACY 256
@@ -73,6 +75,9 @@ struct vgic_global {
/* Only needed for the legacy KVM_CREATE_IRQCHIP */
bool can_emulate_gicv2;
+ /* Does have GICv4? */
+ bool has_gicv4;
+
/* GIC system register CPU interface */
struct static_key_false gicv3_cpuif;
@@ -233,6 +238,9 @@ struct vgic_dist {
/* used by vgic-debug */
struct vgic_state_iter *iter;
+
+ /* GICv4 ITS per-VM stuff */
+ struct its_vm its_vm;
};
struct vgic_v2_cpu_if {
@@ -251,6 +259,9 @@ struct vgic_v3_cpu_if {
u32 vgic_ap0r[4];
u32 vgic_ap1r[4];
u64 vgic_lr[VGIC_V3_MAX_LRS];
+
+ /* GICv4 vPE info */
+ struct its_vpe its_vpe;
};
struct vgic_cpu {
diff --git a/virt/kvm/arm/vgic/vgic-v4.c b/virt/kvm/arm/vgic/vgic-v4.c
new file mode 100644
index 000000000000..81f607f2f83b
--- /dev/null
+++ b/virt/kvm/arm/vgic/vgic-v4.c
@@ -0,0 +1,59 @@
+/*
+ * Copyright (C) 2017 ARM Ltd.
+ * Author: Marc Zyngier <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/interrupt.h>
+#include <linux/irqdomain.h>
+#include <linux/kvm_host.h>
+
+#include "vgic.h"
+
+int vgic_v4_init(struct kvm *kvm)
+{
+ struct vgic_dist *dist = &kvm->arch.vgic;
+ struct kvm_vcpu *vcpu;
+ int i, nr_vcpus, ret;
+
+ nr_vcpus = atomic_read(&kvm->online_vcpus);
+
+ dist->its_vm.vpes = kzalloc(sizeof(*dist->its_vm.vpes) * nr_vcpus,
+ GFP_KERNEL);
+ if (!dist->its_vm.vpes)
+ return -ENOMEM;
+
+ dist->its_vm.nr_vpes = nr_vcpus;
+
+ kvm_for_each_vcpu(i, vcpu, kvm)
+ dist->its_vm.vpes[i] = &vcpu->arch.vgic_cpu.vgic_v3.its_vpe;
+
+ ret = its_alloc_vcpu_irqs(&dist->its_vm);
+ if (ret < 0) {
+ kvm_err("VPE IRQ allocation failure\n");
+ dist->its_vm.nr_vpes = 0;
+ kfree(dist->its_vm.vpes);
+ return ret;
+ }
+
+ return ret;
+}
+
+void vgic_v4_teardown(struct kvm *kvm)
+{
+ struct its_vm *its_vm = &kvm->arch.vgic.its_vm;
+
+ its_free_vcpu_irqs(its_vm);
+ kfree(its_vm->vpes);
+}
diff --git a/virt/kvm/arm/vgic/vgic.h b/virt/kvm/arm/vgic/vgic.h
index 3002b72d938b..4dc47edce672 100644
--- a/virt/kvm/arm/vgic/vgic.h
+++ b/virt/kvm/arm/vgic/vgic.h
@@ -229,4 +229,7 @@ int vgic_its_resolve_lpi(struct kvm *kvm, struct vgic_its *its,
u32 devid, u32 eventid, struct vgic_irq **irq);
struct vgic_its *vgic_msi_to_its(struct kvm *kvm, struct kvm_msi *msi);
+int vgic_v4_init(struct kvm *kvm);
+void vgic_v4_teardown(struct kvm *kvm);
+
#endif
--
2.11.0
As KVM needs to know about the availability of GICv4 to enable
direct injection of interrupts, let's advertise the feature in
the gic_kvm_info structure.
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3.c | 2 ++
include/linux/irqchip/arm-gic-common.h | 2 ++
2 files changed, 4 insertions(+)
diff --git a/drivers/irqchip/irq-gic-v3.c b/drivers/irqchip/irq-gic-v3.c
index efbd998baa9c..7ce88148da1f 100644
--- a/drivers/irqchip/irq-gic-v3.c
+++ b/drivers/irqchip/irq-gic-v3.c
@@ -1156,6 +1156,7 @@ static void __init gic_of_setup_kvm_info(struct device_node *node)
if (!ret)
gic_v3_kvm_info.vcpu = r;
+ gic_v3_kvm_info.has_v4 = gic_data.rdists.has_vlpis;
gic_set_kvm_info(&gic_v3_kvm_info);
}
@@ -1452,6 +1453,7 @@ static void __init gic_acpi_setup_kvm_info(void)
vcpu->end = vcpu->start + ACPI_GICV2_VCPU_MEM_SIZE - 1;
}
+ gic_v3_kvm_info.has_v4 = gic_data.rdists.has_vlpis;
gic_set_kvm_info(&gic_v3_kvm_info);
}
diff --git a/include/linux/irqchip/arm-gic-common.h b/include/linux/irqchip/arm-gic-common.h
index c647b0547bcd..0a83b4379f34 100644
--- a/include/linux/irqchip/arm-gic-common.h
+++ b/include/linux/irqchip/arm-gic-common.h
@@ -27,6 +27,8 @@ struct gic_kvm_info {
unsigned int maint_irq;
/* Virtual control interface */
struct resource vctrl;
+ /* vlpi support */
+ bool has_v4;
};
const struct gic_kvm_info *gic_get_kvm_info(void);
--
2.11.0
Let's use the irq bypass mechanism introduced for platform device
interrupts to intercept the virtual PCIe endpoint configuration
and establish our LPI->VLPI mapping.
Signed-off-by: Marc Zyngier <[email protected]>
---
include/kvm/arm_vgic.h | 8 ++++
virt/kvm/arm/arm.c | 27 +++++++++----
virt/kvm/arm/vgic/vgic-v4.c | 96 +++++++++++++++++++++++++++++++++++++++++++++
3 files changed, 123 insertions(+), 8 deletions(-)
diff --git a/include/kvm/arm_vgic.h b/include/kvm/arm_vgic.h
index 359eeffe9857..050f78d4fb42 100644
--- a/include/kvm/arm_vgic.h
+++ b/include/kvm/arm_vgic.h
@@ -367,4 +367,12 @@ int kvm_vgic_set_forwarding(struct kvm *kvm, unsigned int host_irq,
void kvm_vgic_unset_forwarding(struct kvm *kvm, unsigned int host_irq,
unsigned int vintid);
+struct kvm_kernel_irq_routing_entry;
+
+int kvm_vgic_v4_set_forwarding(struct kvm *kvm, int irq,
+ struct kvm_kernel_irq_routing_entry *irq_entry);
+
+int kvm_vgic_v4_unset_forwarding(struct kvm *kvm, int irq,
+ struct kvm_kernel_irq_routing_entry *irq_entry);
+
#endif /* __KVM_ARM_VGIC_H */
diff --git a/virt/kvm/arm/arm.c b/virt/kvm/arm/arm.c
index ebab6c29e3be..6803ea27c47d 100644
--- a/virt/kvm/arm/arm.c
+++ b/virt/kvm/arm/arm.c
@@ -1457,11 +1457,16 @@ int kvm_arch_irq_bypass_add_producer(struct irq_bypass_consumer *cons,
struct kvm_kernel_irqfd *irqfd =
container_of(cons, struct kvm_kernel_irqfd, consumer);
- if (prod->type != IRQ_BYPASS_VFIO_PLATFORM)
+ switch (prod->type) {
+ case IRQ_BYPASS_VFIO_PLATFORM:
+ return kvm_vgic_set_forwarding(irqfd->kvm, prod->irq,
+ irqfd->gsi + VGIC_NR_PRIVATE_IRQS);
+ case IRQ_BYPASS_VFIO_PCI_MSI:
+ return kvm_vgic_v4_set_forwarding(irqfd->kvm, prod->irq,
+ &irqfd->irq_entry);
+ default:
return 0;
-
- return kvm_vgic_set_forwarding(irqfd->kvm, prod->irq,
- irqfd->gsi + VGIC_NR_PRIVATE_IRQS);
+ }
}
void kvm_arch_irq_bypass_del_producer(struct irq_bypass_consumer *cons,
struct irq_bypass_producer *prod)
@@ -1469,11 +1474,17 @@ void kvm_arch_irq_bypass_del_producer(struct irq_bypass_consumer *cons,
struct kvm_kernel_irqfd *irqfd =
container_of(cons, struct kvm_kernel_irqfd, consumer);
- if (prod->type != IRQ_BYPASS_VFIO_PLATFORM)
- return;
+ switch (prod->type) {
+ case IRQ_BYPASS_VFIO_PLATFORM:
+ kvm_vgic_unset_forwarding(irqfd->kvm, prod->irq,
+ irqfd->gsi + VGIC_NR_PRIVATE_IRQS);
+ break;
- kvm_vgic_unset_forwarding(irqfd->kvm, prod->irq,
- irqfd->gsi + VGIC_NR_PRIVATE_IRQS);
+ case IRQ_BYPASS_VFIO_PCI_MSI:
+ kvm_vgic_v4_unset_forwarding(irqfd->kvm, prod->irq,
+ &irqfd->irq_entry);
+ break;
+ }
}
void kvm_arch_irq_bypass_stop(struct irq_bypass_consumer *cons)
diff --git a/virt/kvm/arm/vgic/vgic-v4.c b/virt/kvm/arm/vgic/vgic-v4.c
index 81f607f2f83b..22cc3871bf03 100644
--- a/virt/kvm/arm/vgic/vgic-v4.c
+++ b/virt/kvm/arm/vgic/vgic-v4.c
@@ -57,3 +57,99 @@ void vgic_v4_teardown(struct kvm *kvm)
its_free_vcpu_irqs(its_vm);
kfree(its_vm->vpes);
}
+
+static struct vgic_its *vgic_get_its(struct kvm *kvm,
+ struct kvm_kernel_irq_routing_entry *irq_entry)
+{
+ struct kvm_msi msi = (struct kvm_msi) {
+ .address_lo = irq_entry->msi.address_lo,
+ .address_hi = irq_entry->msi.address_hi,
+ .data = irq_entry->msi.data,
+ .flags = irq_entry->msi.flags,
+ .devid = irq_entry->msi.devid,
+ };
+
+ /*
+ * Get a reference on the LPI. If NULL, this is not a valid
+ * translation for any of our vITSs.
+ */
+ return vgic_msi_to_its(kvm, &msi);
+}
+
+int kvm_vgic_v4_set_forwarding(struct kvm *kvm, int virq,
+ struct kvm_kernel_irq_routing_entry *irq_entry)
+{
+ struct vgic_its *its;
+ struct vgic_irq *irq;
+ struct its_vlpi_map map;
+ int ret;
+
+ /*
+ * Get the LPI. If NULL, this is not a valid translation for
+ * any of our vITSs.
+ */
+ its = vgic_get_its(kvm, irq_entry);
+ if (!its)
+ return 0;
+
+ mutex_lock(&its->its_lock);
+
+ ret = vgic_its_resolve_lpi(kvm, its, irq_entry->msi.devid,
+ irq_entry->msi.data, &irq);
+ if (ret)
+ goto out;
+
+ /*
+ * Emit the mapping request. If it fails, the ITS probably
+ * isn't v4 compatible, so let's silently bail out. Holding
+ * the ITS lock should ensure that nothing can modify the
+ * target vcpu.
+ */
+ map = (struct its_vlpi_map) {
+ .vm = &kvm->arch.vgic.its_vm,
+ .vintid = irq->intid,
+ .db_enabled = true,
+ .vpe_idx = irq->target_vcpu->vcpu_id,
+ };
+
+ if (its_map_vlpi(virq, &map))
+ goto out;
+
+ irq->hw = true;
+ irq->host_irq = virq;
+
+out:
+ mutex_unlock(&its->its_lock);
+ return 0;
+}
+
+int kvm_vgic_v4_unset_forwarding(struct kvm *kvm, int virq,
+ struct kvm_kernel_irq_routing_entry *irq_entry)
+{
+ struct vgic_its *its;
+ struct vgic_irq *irq;
+ int ret;
+
+ /*
+ * Get the LPI. If NULL, this is not a valid translation for
+ * any of our vITSs.
+ */
+ its = vgic_get_its(kvm, irq_entry);
+ if (!its)
+ return 0;
+
+ mutex_lock(&its->its_lock);
+
+ ret = vgic_its_resolve_lpi(kvm, its, irq_entry->msi.devid,
+ irq_entry->msi.data, &irq);
+ if (ret)
+ goto out;
+
+ WARN_ON(!(irq->hw && irq->host_irq == virq));
+ irq->hw = false;
+ ret = its_unmap_vlpi(virq);
+
+out:
+ mutex_unlock(&its->its_lock);
+ return ret;
+}
--
2.11.0
The current implementation of MOVALL doesn't allow us to call
into the core ITS code as we hold a number of spinlocks.
Let's try a method used in other parts of the code, were we copy
the intids of the candicate interrupts, and then do whatever
we need to do with them outside of the critical section.
This allows us to move the interrupts one by one, at the expense
of a bit of CPU time. Who cares? MOVALL is such a stupid command
anyway...
Signed-off-by: Marc Zyngier <[email protected]>
---
virt/kvm/arm/vgic/vgic-its.c | 27 ++++++++++++++++++++-------
1 file changed, 20 insertions(+), 7 deletions(-)
diff --git a/virt/kvm/arm/vgic/vgic-its.c b/virt/kvm/arm/vgic/vgic-its.c
index 21a7b3f00107..b40f46e5a79d 100644
--- a/virt/kvm/arm/vgic/vgic-its.c
+++ b/virt/kvm/arm/vgic/vgic-its.c
@@ -1147,11 +1147,12 @@ static int vgic_its_cmd_handle_invall(struct kvm *kvm, struct vgic_its *its,
static int vgic_its_cmd_handle_movall(struct kvm *kvm, struct vgic_its *its,
u64 *its_cmd)
{
- struct vgic_dist *dist = &kvm->arch.vgic;
u32 target1_addr = its_cmd_get_target_addr(its_cmd);
u32 target2_addr = its_cmd_mask_field(its_cmd, 3, 16, 32);
struct kvm_vcpu *vcpu1, *vcpu2;
struct vgic_irq *irq;
+ u32 *intids;
+ int irq_count, i;
if (target1_addr >= atomic_read(&kvm->online_vcpus) ||
target2_addr >= atomic_read(&kvm->online_vcpus))
@@ -1163,19 +1164,31 @@ static int vgic_its_cmd_handle_movall(struct kvm *kvm, struct vgic_its *its,
vcpu1 = kvm_get_vcpu(kvm, target1_addr);
vcpu2 = kvm_get_vcpu(kvm, target2_addr);
- spin_lock(&dist->lpi_list_lock);
+ irq_count = vgic_copy_lpi_list(vcpu1, &intids);
+ if (irq_count < 0)
+ return irq_count;
- list_for_each_entry(irq, &dist->lpi_list_head, lpi_list) {
- spin_lock(&irq->irq_lock);
+ for (i = 0; i < irq_count; i++) {
+ irq = vgic_get_irq(kvm, NULL, intids[i]);
+ if (!irq)
+ continue;
if (irq->target_vcpu == vcpu1)
irq->target_vcpu = vcpu2;
- spin_unlock(&irq->irq_lock);
- }
+ if (irq->hw) {
+ struct its_vlpi_map map;
- spin_unlock(&dist->lpi_list_lock);
+ if (!its_get_vlpi(irq->host_irq, &map)) {
+ map.vpe_idx = vcpu2->vcpu_id;
+ its_map_vlpi(irq->host_irq, &map);
+ }
+ }
+ vgic_put_irq(kvm, irq);
+ }
+
+ kfree(intids);
return 0;
}
--
2.11.0
Upon updating a property, we propagate it all the way to the physical
ITS, and ask for an INV command to be executed there.
Signed-off-by: Marc Zyngier <[email protected]>
---
virt/kvm/arm/vgic/vgic-its.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/virt/kvm/arm/vgic/vgic-its.c b/virt/kvm/arm/vgic/vgic-its.c
index b40f46e5a79d..2015246ab9f9 100644
--- a/virt/kvm/arm/vgic/vgic-its.c
+++ b/virt/kvm/arm/vgic/vgic-its.c
@@ -297,6 +297,9 @@ static int update_lpi_config(struct kvm *kvm, struct vgic_irq *irq,
spin_unlock(&irq->irq_lock);
}
+ if (irq->hw)
+ return its_prop_update_vlpi(irq->host_irq, prop, true);
+
return 0;
}
--
2.11.0
When a vPE is not running, the delivery of a VLPI results in a
doorbell interrupt to be delivered. Let's handle this interrupt
and update the pending_last flag that indicates that VLPIs are
pending.
Signed-off-by: Marc Zyngier <[email protected]>
---
virt/kvm/arm/vgic/vgic-v4.c | 30 ++++++++++++++++++++++++++++++
1 file changed, 30 insertions(+)
diff --git a/virt/kvm/arm/vgic/vgic-v4.c b/virt/kvm/arm/vgic/vgic-v4.c
index a2d1a6db0490..fe0d0a8ed469 100644
--- a/virt/kvm/arm/vgic/vgic-v4.c
+++ b/virt/kvm/arm/vgic/vgic-v4.c
@@ -21,6 +21,14 @@
#include "vgic.h"
+static irqreturn_t vgic_v4_doorbell_handler(int irq, void *info)
+{
+ struct kvm_vcpu *vcpu = info;
+
+ vcpu->arch.vgic_cpu.vgic_v3.its_vpe.pending_last = true;
+ return IRQ_HANDLED;
+}
+
int vgic_v4_init(struct kvm *kvm)
{
struct vgic_dist *dist = &kvm->arch.vgic;
@@ -47,12 +55,34 @@ int vgic_v4_init(struct kvm *kvm)
return ret;
}
+ kvm_for_each_vcpu(i, vcpu, kvm) {
+ int irq = dist->its_vm.vpes[i]->irq;
+
+ ret = request_irq(irq, vgic_v4_doorbell_handler,
+ 0, "vcpu", vcpu);
+ if (ret) {
+ kvm_err("failed to allocate vcpu IRQ%d\n", irq);
+ dist->its_vm.nr_vpes = i;
+ break;
+ }
+ }
+
+ if (ret)
+ vgic_v4_teardown(kvm);
+
return ret;
}
void vgic_v4_teardown(struct kvm *kvm)
{
struct its_vm *its_vm = &kvm->arch.vgic.its_vm;
+ struct kvm_vcpu *vcpu;
+ int i;
+
+ for (i = 0; i < its_vm->nr_vpes; i++) {
+ vcpu = kvm_get_vcpu(kvm, i);
+ free_irq(its_vm->vpes[i]->irq, vcpu);
+ }
its_free_vcpu_irqs(its_vm);
kfree(its_vm->vpes);
--
2.11.0
The redistributor needs to be told which vPE is about to be run,
and tells us whether there is any pending VLPI on exit.
Let's add the scheduling calls to the vgic flush/sync functions,
allowing the VLPIs to be delivered to the guest.
Signed-off-by: Marc Zyngier <[email protected]>
---
virt/kvm/arm/vgic/vgic-v4.c | 23 +++++++++++++++++++++++
virt/kvm/arm/vgic/vgic.c | 4 ++++
virt/kvm/arm/vgic/vgic.h | 1 +
3 files changed, 28 insertions(+)
diff --git a/virt/kvm/arm/vgic/vgic-v4.c b/virt/kvm/arm/vgic/vgic-v4.c
index fe0d0a8ed469..6ede64be94f1 100644
--- a/virt/kvm/arm/vgic/vgic-v4.c
+++ b/virt/kvm/arm/vgic/vgic-v4.c
@@ -88,6 +88,29 @@ void vgic_v4_teardown(struct kvm *kvm)
kfree(its_vm->vpes);
}
+int vgic_v4_schedule(struct kvm_vcpu *vcpu, bool on)
+{
+ if (!(kvm_vgic_global_state.has_gicv4 && vgic_has_its(vcpu->kvm)))
+ return 0;
+
+ /*
+ * Before making the VPE resident, make sure the redistributor
+ * expects us here.
+ */
+ if (on) {
+ int irq = vcpu->arch.vgic_cpu.vgic_v3.its_vpe.irq;
+ int err;
+
+ err = irq_set_affinity(irq, cpumask_of(smp_processor_id()));
+ if (err) {
+ kvm_err("failed irq_set_affinity IRQ%d (%d)\n", irq, err);
+ return err;
+ }
+ }
+
+ return its_schedule_vpe(&vcpu->arch.vgic_cpu.vgic_v3.its_vpe, on);
+}
+
static struct vgic_its *vgic_get_its(struct kvm *kvm,
struct kvm_kernel_irq_routing_entry *irq_entry)
{
diff --git a/virt/kvm/arm/vgic/vgic.c b/virt/kvm/arm/vgic/vgic.c
index a96e566905d2..2106e4c06a12 100644
--- a/virt/kvm/arm/vgic/vgic.c
+++ b/virt/kvm/arm/vgic/vgic.c
@@ -721,6 +721,8 @@ void kvm_vgic_sync_hwstate(struct kvm_vcpu *vcpu)
{
struct vgic_cpu *vgic_cpu = &vcpu->arch.vgic_cpu;
+ WARN_ON(vgic_v4_schedule(vcpu, false));
+
/* An empty ap_list_head implies used_lrs == 0 */
if (list_empty(&vcpu->arch.vgic_cpu.ap_list_head))
return;
@@ -733,6 +735,8 @@ void kvm_vgic_sync_hwstate(struct kvm_vcpu *vcpu)
/* Flush our emulation state into the GIC hardware before entering the guest. */
void kvm_vgic_flush_hwstate(struct kvm_vcpu *vcpu)
{
+ WARN_ON(vgic_v4_schedule(vcpu, true));
+
/*
* If there are no virtual interrupts active or pending for this
* VCPU, then there is no work to do and we can bail out without
diff --git a/virt/kvm/arm/vgic/vgic.h b/virt/kvm/arm/vgic/vgic.h
index 974e382297f9..3ebb6c4443a7 100644
--- a/virt/kvm/arm/vgic/vgic.h
+++ b/virt/kvm/arm/vgic/vgic.h
@@ -233,5 +233,6 @@ int update_lpi_config(struct kvm *kvm, struct vgic_irq *irq,
int vgic_v4_init(struct kvm *kvm);
void vgic_v4_teardown(struct kvm *kvm);
+int vgic_v4_schedule(struct kvm_vcpu *vcpu, bool on);
#endif
--
2.11.0
Handling CLEAR is pretty easy. Just ask the ITS driver to clear
the corresponding pending bit (which will turn into a CLEAR
command on the physical side).
Signed-off-by: Marc Zyngier <[email protected]>
---
virt/kvm/arm/vgic/vgic-its.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/virt/kvm/arm/vgic/vgic-its.c b/virt/kvm/arm/vgic/vgic-its.c
index ee8355991bb1..21a7b3f00107 100644
--- a/virt/kvm/arm/vgic/vgic-its.c
+++ b/virt/kvm/arm/vgic/vgic-its.c
@@ -1069,6 +1069,10 @@ static int vgic_its_cmd_handle_clear(struct kvm *kvm, struct vgic_its *its,
ite->irq->pending_latch = false;
+ if (ite->irq->hw)
+ return irq_set_irqchip_state(ite->irq->host_irq,
+ IRQCHIP_STATE_PENDING, false);
+
return 0;
}
--
2.11.0
When freeing an LPI (on a DISCARD command, for example), we need
to unmap the VLPI down to the physical ITS level.
Signed-off-by: Marc Zyngier <[email protected]>
---
virt/kvm/arm/vgic/vgic-its.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/virt/kvm/arm/vgic/vgic-its.c b/virt/kvm/arm/vgic/vgic-its.c
index 255d29837514..ed52b94e8e6d 100644
--- a/virt/kvm/arm/vgic/vgic-its.c
+++ b/virt/kvm/arm/vgic/vgic-its.c
@@ -623,8 +623,12 @@ static void its_free_ite(struct kvm *kvm, struct its_ite *ite)
list_del(&ite->ite_list);
/* This put matches the get in vgic_add_lpi. */
- if (ite->irq)
+ if (ite->irq) {
+ if (ite->irq->hw)
+ its_unmap_vlpi(ite->irq->host_irq);
+
vgic_put_irq(kvm, ite->irq);
+ }
kfree(ite);
}
--
2.11.0
In order for VLPIs to be delivered to the guest, we must make
sure that the cpuif is always enabled, irrespective of the
presence of virtual interrupt in the LRs.
Signed-off-by: Marc Zyngier <[email protected]>
---
virt/kvm/arm/hyp/vgic-v3-sr.c | 9 ++++++---
1 file changed, 6 insertions(+), 3 deletions(-)
diff --git a/virt/kvm/arm/hyp/vgic-v3-sr.c b/virt/kvm/arm/hyp/vgic-v3-sr.c
index 91728faa13fd..f5c3d6d7019e 100644
--- a/virt/kvm/arm/hyp/vgic-v3-sr.c
+++ b/virt/kvm/arm/hyp/vgic-v3-sr.c
@@ -258,7 +258,8 @@ void __hyp_text __vgic_v3_save_state(struct kvm_vcpu *vcpu)
cpu_if->vgic_ap1r[0] = __vgic_v3_read_ap1rn(0);
}
} else {
- if (static_branch_unlikely(&vgic_v3_cpuif_trap))
+ if (static_branch_unlikely(&vgic_v3_cpuif_trap) ||
+ cpu_if->its_vpe.its_vm)
write_gicreg(0, ICH_HCR_EL2);
cpu_if->vgic_elrsr = 0xffff;
@@ -337,9 +338,11 @@ void __hyp_text __vgic_v3_restore_state(struct kvm_vcpu *vcpu)
/*
* If we need to trap system registers, we must write
* ICH_HCR_EL2 anyway, even if no interrupts are being
- * injected,
+ * injected. Same thing if GICv4 is used, as VLPI
+ * delivery is gated by ICH_HCR_EL2.En.
*/
- if (static_branch_unlikely(&vgic_v3_cpuif_trap))
+ if (static_branch_unlikely(&vgic_v3_cpuif_trap) ||
+ cpu_if->its_vpe.its_vm)
write_gicreg(cpu_if->vgic_hcr, ICH_HCR_EL2);
}
--
2.11.0
All it takes is for the has_v4 flag to be set in gic_kvm_info, and
we'll enable it...
Signed-off-by: Marc Zyngier <[email protected]>
---
virt/kvm/arm/vgic/vgic-v3.c | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/virt/kvm/arm/vgic/vgic-v3.c b/virt/kvm/arm/vgic/vgic-v3.c
index 96ea597db0e7..fb7bc3b93482 100644
--- a/virt/kvm/arm/vgic/vgic-v3.c
+++ b/virt/kvm/arm/vgic/vgic-v3.c
@@ -478,6 +478,12 @@ int vgic_v3_probe(const struct gic_kvm_info *info)
kvm_vgic_global_state.can_emulate_gicv2 = false;
kvm_vgic_global_state.ich_vtr_el2 = ich_vtr_el2;
+ /* GICv4 support? */
+ if (info->has_v4) {
+ kvm_vgic_global_state.has_gicv4 = true;
+ kvm_info("GICv4 support enabled\n");
+ }
+
if (!info->vcpu.start) {
kvm_info("GICv3: no GICV resource entry\n");
kvm_vgic_global_state.vcpu_base = 0;
--
2.11.0
When a vPE exits, the pending_last flag is set when there are
pending VLPIs stored in the pending table. Similarily, we set
this flag when a doorbell interrupt fires, as it indicates the
same condition.
Let's update kvm_vgic_vcpu_pending_irq() to account for that
flag as well, making a vcpu runnable when set.
Signed-off-by: Marc Zyngier <[email protected]>
---
virt/kvm/arm/vgic/vgic.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/virt/kvm/arm/vgic/vgic.c b/virt/kvm/arm/vgic/vgic.c
index 2106e4c06a12..9ab52108989d 100644
--- a/virt/kvm/arm/vgic/vgic.c
+++ b/virt/kvm/arm/vgic/vgic.c
@@ -785,6 +785,9 @@ int kvm_vgic_vcpu_pending_irq(struct kvm_vcpu *vcpu)
if (!vcpu->kvm->arch.vgic.enabled)
return false;
+ if (vcpu->arch.vgic_cpu.vgic_v3.its_vpe.pending_last)
+ return true;
+
spin_lock(&vgic_cpu->ap_list_lock);
list_for_each_entry(irq, &vgic_cpu->ap_list_head, ap_list) {
--
2.11.0
If the guest issues an INT command targetting a VLPI, let's
call into the irq_set_irqchip_state() helper to make it pending
on the physical side.
This works just as well if userspace decides to inject an interrupt
using the normal userspace API...
Signed-off-by: Marc Zyngier <[email protected]>
---
virt/kvm/arm/vgic/vgic-its.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/virt/kvm/arm/vgic/vgic-its.c b/virt/kvm/arm/vgic/vgic-its.c
index 5fbcf9ea8886..255d29837514 100644
--- a/virt/kvm/arm/vgic/vgic-its.c
+++ b/virt/kvm/arm/vgic/vgic-its.c
@@ -573,6 +573,10 @@ static int vgic_its_trigger_msi(struct kvm *kvm, struct vgic_its *its,
if (err)
return err;
+ if (irq->hw)
+ return irq_set_irqchip_state(irq->host_irq,
+ IRQCHIP_STATE_PENDING, true);
+
spin_lock(&irq->irq_lock);
irq->pending_latch = true;
vgic_queue_irq_unlock(kvm, irq);
--
2.11.0
Since when updating the properties one LPI at a time, there is no
need to perform an INV each time we read one. Instead, we rely
on the final VINVALL that gets sent to the ITS to do the work.
Signed-off-by: Marc Zyngier <[email protected]>
---
virt/kvm/arm/vgic/vgic-its.c | 15 +++++++++------
1 file changed, 9 insertions(+), 6 deletions(-)
diff --git a/virt/kvm/arm/vgic/vgic-its.c b/virt/kvm/arm/vgic/vgic-its.c
index 2015246ab9f9..b0790ca5e933 100644
--- a/virt/kvm/arm/vgic/vgic-its.c
+++ b/virt/kvm/arm/vgic/vgic-its.c
@@ -38,7 +38,7 @@ static int vgic_its_save_tables_v0(struct vgic_its *its);
static int vgic_its_restore_tables_v0(struct vgic_its *its);
static int vgic_its_commit_v0(struct vgic_its *its);
static int update_lpi_config(struct kvm *kvm, struct vgic_irq *irq,
- struct kvm_vcpu *filter_vcpu);
+ struct kvm_vcpu *filter_vcpu, bool needs_inv);
/*
* Creates a new (reference to a) struct vgic_irq for a given LPI.
@@ -106,7 +106,7 @@ static struct vgic_irq *vgic_add_lpi(struct kvm *kvm, u32 intid,
* However we only have those structs for mapped IRQs, so we read in
* the respective config data from memory here upon mapping the LPI.
*/
- ret = update_lpi_config(kvm, irq, NULL);
+ ret = update_lpi_config(kvm, irq, NULL, false);
if (ret)
return ERR_PTR(ret);
@@ -274,7 +274,7 @@ static struct its_collection *find_collection(struct vgic_its *its, int coll_id)
* VCPU. Unconditionally applies if filter_vcpu is NULL.
*/
static int update_lpi_config(struct kvm *kvm, struct vgic_irq *irq,
- struct kvm_vcpu *filter_vcpu)
+ struct kvm_vcpu *filter_vcpu, bool needs_inv)
{
u64 propbase = GICR_PROPBASER_ADDRESS(kvm->arch.vgic.propbaser);
u8 prop;
@@ -298,7 +298,7 @@ static int update_lpi_config(struct kvm *kvm, struct vgic_irq *irq,
}
if (irq->hw)
- return its_prop_update_vlpi(irq->host_irq, prop, true);
+ return its_prop_update_vlpi(irq->host_irq, prop, needs_inv);
return 0;
}
@@ -1095,7 +1095,7 @@ static int vgic_its_cmd_handle_inv(struct kvm *kvm, struct vgic_its *its,
if (!ite)
return E_ITS_INV_UNMAPPED_INTERRUPT;
- return update_lpi_config(kvm, ite->irq, NULL);
+ return update_lpi_config(kvm, ite->irq, NULL, true);
}
/*
@@ -1130,12 +1130,15 @@ static int vgic_its_cmd_handle_invall(struct kvm *kvm, struct vgic_its *its,
irq = vgic_get_irq(kvm, NULL, intids[i]);
if (!irq)
continue;
- update_lpi_config(kvm, irq, vcpu);
+ update_lpi_config(kvm, irq, vcpu, false);
vgic_put_irq(kvm, irq);
}
kfree(intids);
+ if (vcpu->arch.vgic_cpu.vgic_v3.its_vpe.its_vm)
+ its_invall_vpe(&vcpu->arch.vgic_cpu.vgic_v3.its_vpe);
+
return 0;
}
--
2.11.0
When the VLPI gets mapped, it must inherit the configuration of
LPI configured at the vITS level. FOr that purpose, let's make
update_lpi_config globally available and call it just after
having performed the VLPI map operation.
Signed-off-by: Marc Zyngier <[email protected]>
---
virt/kvm/arm/vgic/vgic-its.c | 6 ++----
virt/kvm/arm/vgic/vgic-v4.c | 2 ++
virt/kvm/arm/vgic/vgic.h | 2 ++
3 files changed, 6 insertions(+), 4 deletions(-)
diff --git a/virt/kvm/arm/vgic/vgic-its.c b/virt/kvm/arm/vgic/vgic-its.c
index b0790ca5e933..289c158aeee1 100644
--- a/virt/kvm/arm/vgic/vgic-its.c
+++ b/virt/kvm/arm/vgic/vgic-its.c
@@ -37,8 +37,6 @@
static int vgic_its_save_tables_v0(struct vgic_its *its);
static int vgic_its_restore_tables_v0(struct vgic_its *its);
static int vgic_its_commit_v0(struct vgic_its *its);
-static int update_lpi_config(struct kvm *kvm, struct vgic_irq *irq,
- struct kvm_vcpu *filter_vcpu, bool needs_inv);
/*
* Creates a new (reference to a) struct vgic_irq for a given LPI.
@@ -273,8 +271,8 @@ static struct its_collection *find_collection(struct vgic_its *its, int coll_id)
* If filter_vcpu is not NULL, applies only if the IRQ is targeting this
* VCPU. Unconditionally applies if filter_vcpu is NULL.
*/
-static int update_lpi_config(struct kvm *kvm, struct vgic_irq *irq,
- struct kvm_vcpu *filter_vcpu, bool needs_inv)
+int update_lpi_config(struct kvm *kvm, struct vgic_irq *irq,
+ struct kvm_vcpu *filter_vcpu, bool needs_inv)
{
u64 propbase = GICR_PROPBASER_ADDRESS(kvm->arch.vgic.propbaser);
u8 prop;
diff --git a/virt/kvm/arm/vgic/vgic-v4.c b/virt/kvm/arm/vgic/vgic-v4.c
index 22cc3871bf03..a2d1a6db0490 100644
--- a/virt/kvm/arm/vgic/vgic-v4.c
+++ b/virt/kvm/arm/vgic/vgic-v4.c
@@ -118,6 +118,8 @@ int kvm_vgic_v4_set_forwarding(struct kvm *kvm, int virq,
irq->hw = true;
irq->host_irq = virq;
+ /* Force the property update and invalidate */
+ update_lpi_config(kvm, irq, NULL, true);
out:
mutex_unlock(&its->its_lock);
return 0;
diff --git a/virt/kvm/arm/vgic/vgic.h b/virt/kvm/arm/vgic/vgic.h
index 4dc47edce672..974e382297f9 100644
--- a/virt/kvm/arm/vgic/vgic.h
+++ b/virt/kvm/arm/vgic/vgic.h
@@ -228,6 +228,8 @@ void unlock_all_vcpus(struct kvm *kvm);
int vgic_its_resolve_lpi(struct kvm *kvm, struct vgic_its *its,
u32 devid, u32 eventid, struct vgic_irq **irq);
struct vgic_its *vgic_msi_to_its(struct kvm *kvm, struct kvm_msi *msi);
+int update_lpi_config(struct kvm *kvm, struct vgic_irq *irq,
+ struct kvm_vcpu *filter_vcpu, bool needs_inv);
int vgic_v4_init(struct kvm *kvm);
void vgic_v4_teardown(struct kvm *kvm);
--
2.11.0
Should the HW support GICv4 and an ITS being associated with this
VM, let's init the its_vm and its_vpe structures.
Signed-off-by: Marc Zyngier <[email protected]>
---
virt/kvm/arm/vgic/vgic-init.c | 11 ++++++++++-
1 file changed, 10 insertions(+), 1 deletion(-)
diff --git a/virt/kvm/arm/vgic/vgic-init.c b/virt/kvm/arm/vgic/vgic-init.c
index 3a0b8999f011..0de1f0d986d4 100644
--- a/virt/kvm/arm/vgic/vgic-init.c
+++ b/virt/kvm/arm/vgic/vgic-init.c
@@ -285,8 +285,14 @@ int vgic_init(struct kvm *kvm)
if (ret)
goto out;
- if (vgic_has_its(kvm))
+ if (vgic_has_its(kvm)) {
dist->msis_require_devid = true;
+ if (kvm_vgic_global_state.has_gicv4) {
+ ret = vgic_v4_init(kvm);
+ if (ret)
+ goto out;
+ }
+ }
kvm_for_each_vcpu(i, vcpu, kvm)
kvm_vgic_vcpu_enable(vcpu);
@@ -323,6 +329,9 @@ static void kvm_vgic_dist_destroy(struct kvm *kvm)
kfree(dist->spis);
dist->nr_spis = 0;
+
+ if (kvm_vgic_global_state.has_gicv4 && vgic_has_its(kvm))
+ vgic_v4_teardown(kvm);
}
void kvm_vgic_vcpu_destroy(struct kvm_vcpu *vcpu)
--
2.11.0
When the guest issues a MOVI, we need to tell the physical ITS
that we're now targetting a new vcpu. This is done by extracting
the current mapping, updating the target, and reapplying the
mapping. The core ITS code should do the right thing.
Signed-off-by: Marc Zyngier <[email protected]>
---
virt/kvm/arm/vgic/vgic-its.c | 13 +++++++++++++
1 file changed, 13 insertions(+)
diff --git a/virt/kvm/arm/vgic/vgic-its.c b/virt/kvm/arm/vgic/vgic-its.c
index ed52b94e8e6d..ee8355991bb1 100644
--- a/virt/kvm/arm/vgic/vgic-its.c
+++ b/virt/kvm/arm/vgic/vgic-its.c
@@ -706,6 +706,19 @@ static int vgic_its_cmd_handle_movi(struct kvm *kvm, struct vgic_its *its,
ite->irq->target_vcpu = vcpu;
spin_unlock(&ite->irq->irq_lock);
+ if (ite->irq->hw) {
+ struct its_vlpi_map map;
+ int ret;
+
+ ret = its_get_vlpi(ite->irq->host_irq, &map);
+ if (ret)
+ return ret;
+
+ map.vpe_idx = vcpu->vcpu_id;
+
+ return its_map_vlpi(ite->irq->host_irq, &map);
+ }
+
return 0;
}
--
2.11.0
Get the show on the road...
Reviewed-by: Thomas Gleixner <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/Makefile | 2 +-
drivers/irqchip/irq-gic-v3-its.c | 3 ++-
drivers/irqchip/irq-gic-v4.c | 13 +++++++++++++
include/linux/irqchip/arm-gic-v4.h | 2 ++
4 files changed, 18 insertions(+), 2 deletions(-)
diff --git a/drivers/irqchip/Makefile b/drivers/irqchip/Makefile
index b64c59b838a0..4ca95310254b 100644
--- a/drivers/irqchip/Makefile
+++ b/drivers/irqchip/Makefile
@@ -28,7 +28,7 @@ obj-$(CONFIG_ARM_GIC_PM) += irq-gic-pm.o
obj-$(CONFIG_ARCH_REALVIEW) += irq-gic-realview.o
obj-$(CONFIG_ARM_GIC_V2M) += irq-gic-v2m.o
obj-$(CONFIG_ARM_GIC_V3) += irq-gic-v3.o irq-gic-common.o
-obj-$(CONFIG_ARM_GIC_V3_ITS) += irq-gic-v3-its.o irq-gic-v3-its-pci-msi.o irq-gic-v3-its-platform-msi.o
+obj-$(CONFIG_ARM_GIC_V3_ITS) += irq-gic-v3-its.o irq-gic-v3-its-pci-msi.o irq-gic-v3-its-platform-msi.o irq-gic-v4.o
obj-$(CONFIG_PARTITION_PERCPU) += irq-partition-percpu.o
obj-$(CONFIG_HISILICON_IRQ_MBIGEN) += irq-mbigen.o
obj-$(CONFIG_ARM_NVIC) += irq-nvic.o
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index 3784bea2b994..284e6e6785cf 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -2970,7 +2970,8 @@ int __init its_init(struct fwnode_handle *handle, struct rdists *rdists,
has_v4 |= its->is_v4;
if (has_v4 & rdists->has_vlpis) {
- if (its_init_vpe_domain()) {
+ if (its_init_vpe_domain() ||
+ its_init_v4(parent_domain, &its_vpe_domain_ops)) {
rdists->has_vlpis = false;
pr_err("ITS: Disabling GICv4 support\n");
}
diff --git a/drivers/irqchip/irq-gic-v4.c b/drivers/irqchip/irq-gic-v4.c
index 4ce5fdd1db15..7f573d65f440 100644
--- a/drivers/irqchip/irq-gic-v4.c
+++ b/drivers/irqchip/irq-gic-v4.c
@@ -194,3 +194,16 @@ int its_prop_update_vlpi(int irq, u8 config, bool inv)
return irq_set_vcpu_affinity(irq, &info);
}
+
+int its_init_v4(struct irq_domain *domain, const struct irq_domain_ops *ops)
+{
+ if (domain) {
+ pr_info("ITS: Enabling GICv4 support\n");
+ gic_domain = domain;
+ vpe_domain_ops = ops;
+ return 0;
+ }
+
+ pr_err("ITS: No GICv4 VPE domain allocated\n");
+ return -ENODEV;
+}
diff --git a/include/linux/irqchip/arm-gic-v4.h b/include/linux/irqchip/arm-gic-v4.h
index 843c9ff29269..862d20ab67d0 100644
--- a/include/linux/irqchip/arm-gic-v4.h
+++ b/include/linux/irqchip/arm-gic-v4.h
@@ -97,4 +97,6 @@ int its_get_vlpi(int irq, struct its_vlpi_map *map);
int its_unmap_vlpi(int irq);
int its_prop_update_vlpi(int irq, u8 config, bool inv);
+int its_init_v4(struct irq_domain *domain, const struct irq_domain_ops *ops);
+
#endif
--
2.11.0
Do a braindump of the way things are supposed to work.
Reviewed-by: Thomas Gleixner <[email protected]>
Reviewed-by: Eric Auger <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v4.c | 71 ++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 71 insertions(+)
diff --git a/drivers/irqchip/irq-gic-v4.c b/drivers/irqchip/irq-gic-v4.c
index 697d9dc1e005..4ce5fdd1db15 100644
--- a/drivers/irqchip/irq-gic-v4.c
+++ b/drivers/irqchip/irq-gic-v4.c
@@ -22,6 +22,77 @@
#include <linux/irqchip/arm-gic-v4.h>
+/*
+ * WARNING: The blurb below assumes that you understand the
+ * intricacies of GICv3, GICv4, and how a guest's view of a GICv3 gets
+ * translated into GICv4 commands. So it effectively targets at most
+ * two individuals. You know who you are.
+ *
+ * The core GICv4 code is designed to *avoid* exposing too much of the
+ * core GIC code (that would in turn leak into the hypervisor code),
+ * and instead provide a hypervisor agnostic interface to the HW (of
+ * course, the astute reader will quickly realize that hypervisor
+ * agnostic actually means KVM-specific - what were you thinking?).
+ *
+ * In order to achieve a modicum of isolation, we try to hide most of
+ * the GICv4 "stuff" behind normal irqchip operations:
+ *
+ * - Any guest-visible VLPI is backed by a Linux interrupt (and a
+ * physical LPI which gets unmapped when the guest maps the
+ * VLPI). This allows the same DevID/EventID pair to be either
+ * mapped to the LPI (host) or the VLPI (guest). Note that this is
+ * exclusive, and you cannot have both.
+ *
+ * - Enabling/disabling a VLPI is done by issuing mask/unmask calls.
+ *
+ * - Guest INT/CLEAR commands are implemented through
+ * irq_set_irqchip_state().
+ *
+ * - The *bizarre* stuff (mapping/unmapping an interrupt to a VLPI, or
+ * issuing an INV after changing a priority) gets shoved into the
+ * irq_set_vcpu_affinity() method. While this is quite horrible
+ * (let's face it, this is the irqchip version of an ioctl), it
+ * confines the crap to a single location. And map/unmap really is
+ * about setting the affinity of a VLPI to a vcpu, so only INV is
+ * majorly out of place. So there.
+ *
+ * A number of commands are simply not provided by this interface, as
+ * they do not make direct sense. For example, MAPD is purely local to
+ * the virtual ITS (because it references a virtual device, and the
+ * physical ITS is still very much in charge of the physical
+ * device). Same goes for things like MAPC (the physical ITS deals
+ * with the actual vPE affinity, and not the braindead concept of
+ * collection). SYNC is not provided either, as each and every command
+ * is followed by a VSYNC. This could be relaxed in the future, should
+ * this be seen as a bottleneck (yes, this means *never*).
+ *
+ * But handling VLPIs is only one side of the job of the GICv4
+ * code. The other (darker) side is to take care of the doorbell
+ * interrupts which are delivered when a VLPI targeting a non-running
+ * vcpu is being made pending.
+ *
+ * The choice made here is that each vcpu (VPE in old northern GICv4
+ * dialect) gets a single doorbell LPI, no matter how many interrupts
+ * are targeting it. This has a nice property, which is that the
+ * interrupt becomes a handle for the VPE, and that the hypervisor
+ * code can manipulate it through the normal interrupt API:
+ *
+ * - VMs (or rather the VM abstraction that matters to the GIC)
+ * contain an irq domain where each interrupt maps to a VPE. In
+ * turn, this domain sits on top of the normal LPI allocator, and a
+ * specially crafted irq_chip implementation.
+ *
+ * - mask/unmask do what is expected on the doorbell interrupt.
+ *
+ * - irq_set_affinity is used to move a VPE from one redistributor to
+ * another.
+ *
+ * - irq_set_vcpu_affinity once again gets hijacked for the purpose of
+ * creating a new sub-API, namely scheduling/descheduling a VPE
+ * (which involves programming GICR_V{PROP,PEND}BASER) and
+ * performing INVALL operations.
+ */
+
static struct irq_domain *gic_domain;
static const struct irq_domain_ops *vpe_domain_ops;
--
2.11.0
The whole MSI injection process is fairly monolithic. An MSI write
gets turned into an injected LPI in one swift go. But this is actually
a more fine-grained process:
- First, a virtual ITS gets selected using the doorbell address
- Then the DevID/EventID pair gets translated into an LPI
- Finally the LPI is injected
Since the GICv4 code needs the first two steps in order to match
an IRQ routing entry to an LPI, let's expose them as helpers,
and refactor the existing code to use them
Signed-off-by: Marc Zyngier <[email protected]>
---
virt/kvm/arm/vgic/vgic-its.c | 93 +++++++++++++++++++++++++-------------------
virt/kvm/arm/vgic/vgic.h | 4 ++
2 files changed, 57 insertions(+), 40 deletions(-)
diff --git a/virt/kvm/arm/vgic/vgic-its.c b/virt/kvm/arm/vgic/vgic-its.c
index 2dff288b3a66..5fbcf9ea8886 100644
--- a/virt/kvm/arm/vgic/vgic-its.c
+++ b/virt/kvm/arm/vgic/vgic-its.c
@@ -504,15 +504,8 @@ static unsigned long vgic_mmio_read_its_idregs(struct kvm *kvm,
return 0;
}
-/*
- * Find the target VCPU and the LPI number for a given devid/eventid pair
- * and make this IRQ pending, possibly injecting it.
- * Must be called with the its_lock mutex held.
- * Returns 0 on success, a positive error value for any ITS mapping
- * related errors and negative error values for generic errors.
- */
-static int vgic_its_trigger_msi(struct kvm *kvm, struct vgic_its *its,
- u32 devid, u32 eventid)
+int vgic_its_resolve_lpi(struct kvm *kvm, struct vgic_its *its,
+ u32 devid, u32 eventid, struct vgic_irq **irq)
{
struct kvm_vcpu *vcpu;
struct its_ite *ite;
@@ -531,26 +524,60 @@ static int vgic_its_trigger_msi(struct kvm *kvm, struct vgic_its *its,
if (!vcpu->arch.vgic_cpu.lpis_enabled)
return -EBUSY;
- spin_lock(&ite->irq->irq_lock);
- ite->irq->pending_latch = true;
- vgic_queue_irq_unlock(kvm, ite->irq);
-
+ *irq = ite->irq;
return 0;
}
-static struct vgic_io_device *vgic_get_its_iodev(struct kvm_io_device *dev)
+struct vgic_its *vgic_msi_to_its(struct kvm *kvm, struct kvm_msi *msi)
{
+ u64 address;
+ struct kvm_io_device *kvm_io_dev;
struct vgic_io_device *iodev;
- if (dev->ops != &kvm_io_gic_ops)
- return NULL;
+ if (!vgic_has_its(kvm))
+ return ERR_PTR(-ENODEV);
- iodev = container_of(dev, struct vgic_io_device, dev);
+ if (!(msi->flags & KVM_MSI_VALID_DEVID))
+ return ERR_PTR(-EINVAL);
+
+ address = (u64)msi->address_hi << 32 | msi->address_lo;
+
+ kvm_io_dev = kvm_io_bus_get_dev(kvm, KVM_MMIO_BUS, address);
+ if (!kvm_io_dev)
+ return ERR_PTR(-EINVAL);
+ if (kvm_io_dev->ops != &kvm_io_gic_ops)
+ return ERR_PTR(-EINVAL);
+
+ iodev = container_of(kvm_io_dev, struct vgic_io_device, dev);
if (iodev->iodev_type != IODEV_ITS)
- return NULL;
+ return ERR_PTR(-EINVAL);
- return iodev;
+ return iodev->its;
+}
+
+/*
+ * Find the target VCPU and the LPI number for a given devid/eventid pair
+ * and make this IRQ pending, possibly injecting it.
+ * Must be called with the its_lock mutex held.
+ * Returns 0 on success, a positive error value for any ITS mapping
+ * related errors and negative error values for generic errors.
+ */
+static int vgic_its_trigger_msi(struct kvm *kvm, struct vgic_its *its,
+ u32 devid, u32 eventid)
+{
+ struct vgic_irq *irq = NULL;
+ int err;
+
+ err = vgic_its_resolve_lpi(kvm, its, devid, eventid, &irq);
+ if (err)
+ return err;
+
+ spin_lock(&irq->irq_lock);
+ irq->pending_latch = true;
+ vgic_queue_irq_unlock(kvm, irq);
+
+ return 0;
}
/*
@@ -561,30 +588,16 @@ static struct vgic_io_device *vgic_get_its_iodev(struct kvm_io_device *dev)
*/
int vgic_its_inject_msi(struct kvm *kvm, struct kvm_msi *msi)
{
- u64 address;
- struct kvm_io_device *kvm_io_dev;
- struct vgic_io_device *iodev;
+ struct vgic_its *its;
int ret;
- if (!vgic_has_its(kvm))
- return -ENODEV;
-
- if (!(msi->flags & KVM_MSI_VALID_DEVID))
- return -EINVAL;
-
- address = (u64)msi->address_hi << 32 | msi->address_lo;
-
- kvm_io_dev = kvm_io_bus_get_dev(kvm, KVM_MMIO_BUS, address);
- if (!kvm_io_dev)
- return -EINVAL;
+ its = vgic_msi_to_its(kvm, msi);
+ if (IS_ERR(its))
+ return PTR_ERR(its);
- iodev = vgic_get_its_iodev(kvm_io_dev);
- if (!iodev)
- return -EINVAL;
-
- mutex_lock(&iodev->its->its_lock);
- ret = vgic_its_trigger_msi(kvm, iodev->its, msi->devid, msi->data);
- mutex_unlock(&iodev->its->its_lock);
+ mutex_lock(&its->its_lock);
+ ret = vgic_its_trigger_msi(kvm, its, msi->devid, msi->data);
+ mutex_unlock(&its->its_lock);
if (ret < 0)
return ret;
diff --git a/virt/kvm/arm/vgic/vgic.h b/virt/kvm/arm/vgic/vgic.h
index da254ae29aec..3002b72d938b 100644
--- a/virt/kvm/arm/vgic/vgic.h
+++ b/virt/kvm/arm/vgic/vgic.h
@@ -225,4 +225,8 @@ int vgic_debug_destroy(struct kvm *kvm);
bool lock_all_vcpus(struct kvm *kvm);
void unlock_all_vcpus(struct kvm *kvm);
+int vgic_its_resolve_lpi(struct kvm *kvm, struct vgic_its *its,
+ u32 devid, u32 eventid, struct vgic_irq **irq);
+struct vgic_its *vgic_msi_to_its(struct kvm *kvm, struct kvm_msi *msi);
+
#endif
--
2.11.0
When a VPE is scheduled to run, the corresponding redistributor must
be told so, by setting VPROPBASER to the VM's property table, and
VPENDBASER to the vcpu's pending table.
When scheduled out, we preserve the IDAI and PendingLast bits. The
latter is specially important, as it tells the hypervisor that
there are pending interrupts for this vcpu.
Reviewed-by: Eric Auger <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3-its.c | 73 ++++++++++++++++++++++++++++++++++++++
include/linux/irqchip/arm-gic-v3.h | 58 ++++++++++++++++++++++++++++++
2 files changed, 131 insertions(+)
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index 276602efdfa0..be8598354d5a 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -154,6 +154,7 @@ static DEFINE_IDA(its_vpeid_ida);
#define gic_data_rdist() (raw_cpu_ptr(gic_rdists->rdist))
#define gic_data_rdist_rd_base() (gic_data_rdist()->rd_base)
+#define gic_data_rdist_vlpi_base() (gic_data_rdist_rd_base() + SZ_128K)
static struct its_collection *dev_event_to_col(struct its_device *its_dev,
u32 event)
@@ -2144,8 +2145,80 @@ static const struct irq_domain_ops its_domain_ops = {
.deactivate = its_irq_domain_deactivate,
};
+static void its_vpe_schedule(struct its_vpe *vpe)
+{
+ void * __iomem vlpi_base = gic_data_rdist_vlpi_base();
+ u64 val;
+
+ /* Schedule the VPE */
+ val = virt_to_phys(page_address(vpe->its_vm->vprop_page)) &
+ GENMASK_ULL(51, 12);
+ val |= (LPI_NRBITS - 1) & GICR_VPROPBASER_IDBITS_MASK;
+ val |= GICR_VPROPBASER_RaWb;
+ val |= GICR_VPROPBASER_InnerShareable;
+ gits_write_vpropbaser(val, vlpi_base + GICR_VPROPBASER);
+
+ val = virt_to_phys(page_address(vpe->vpt_page)) &
+ GENMASK_ULL(51, 16);
+ val |= GICR_VPENDBASER_RaWaWb;
+ val |= GICR_VPENDBASER_NonShareable;
+ /*
+ * There is no good way of finding out if the pending table is
+ * empty as we can race against the doorbell interrupt very
+ * easily. So in the end, vpe->pending_last is only an
+ * indication that the vcpu has something pending, not one
+ * that the pending table is empty. A good implementation
+ * would be able to read its coarse map pretty quickly anyway,
+ * making this a tolerable issue.
+ */
+ val |= GICR_VPENDBASER_PendingLast;
+ val |= vpe->idai ? GICR_VPENDBASER_IDAI : 0;
+ val |= GICR_VPENDBASER_Valid;
+ gits_write_vpendbaser(val, vlpi_base + GICR_VPENDBASER);
+}
+
+static void its_vpe_deschedule(struct its_vpe *vpe)
+{
+ void * __iomem vlpi_base = gic_data_rdist_vlpi_base();
+ u64 val;
+
+ /* We're being scheduled out */
+ val = gits_read_vpendbaser(vlpi_base + GICR_VPENDBASER);
+ val &= ~GICR_VPENDBASER_Valid;
+ gits_write_vpendbaser(val, vlpi_base + GICR_VPENDBASER);
+
+ val = gits_read_vpendbaser(vlpi_base + GICR_VPENDBASER);
+ while (val & GICR_VPENDBASER_Dirty) {
+ cpu_relax();
+ val = gits_read_vpendbaser(vlpi_base + GICR_VPENDBASER);
+ }
+
+ vpe->idai = !!(val & GICR_VPENDBASER_IDAI);
+ vpe->pending_last = !!(val & GICR_VPENDBASER_PendingLast);
+}
+
+static int its_vpe_set_vcpu_affinity(struct irq_data *d, void *vcpu_info)
+{
+ struct its_vpe *vpe = irq_data_get_irq_chip_data(d);
+ struct its_cmd_info *info = vcpu_info;
+
+ switch (info->cmd_type) {
+ case SCHEDULE_VPE:
+ its_vpe_schedule(vpe);
+ return 0;
+
+ case DESCHEDULE_VPE:
+ its_vpe_deschedule(vpe);
+ return 0;
+
+ default:
+ return -EINVAL;
+ }
+}
+
static struct irq_chip its_vpe_irq_chip = {
.name = "GICv4-vpe",
+ .irq_set_vcpu_affinity = its_vpe_set_vcpu_affinity,
};
static int its_vpe_id_alloc(void)
diff --git a/include/linux/irqchip/arm-gic-v3.h b/include/linux/irqchip/arm-gic-v3.h
index 17ba0d732f12..6bc142cfa616 100644
--- a/include/linux/irqchip/arm-gic-v3.h
+++ b/include/linux/irqchip/arm-gic-v3.h
@@ -212,6 +212,64 @@
#define LPI_PROP_GROUP1 (1 << 1)
#define LPI_PROP_ENABLED (1 << 0)
+/*
+ * Re-Distributor registers, offsets from VLPI_base
+ */
+#define GICR_VPROPBASER 0x0070
+
+#define GICR_VPROPBASER_IDBITS_MASK 0x1f
+
+#define GICR_VPROPBASER_SHAREABILITY_SHIFT (10)
+#define GICR_VPROPBASER_INNER_CACHEABILITY_SHIFT (7)
+#define GICR_VPROPBASER_OUTER_CACHEABILITY_SHIFT (56)
+
+#define GICR_VPROPBASER_SHAREABILITY_MASK \
+ GIC_BASER_SHAREABILITY(GICR_VPROPBASER, SHAREABILITY_MASK)
+#define GICR_VPROPBASER_INNER_CACHEABILITY_MASK \
+ GIC_BASER_CACHEABILITY(GICR_VPROPBASER, INNER, MASK)
+#define GICR_VPROPBASER_OUTER_CACHEABILITY_MASK \
+ GIC_BASER_CACHEABILITY(GICR_VPROPBASER, OUTER, MASK)
+#define GICR_VPROPBASER_CACHEABILITY_MASK \
+ GICR_VPROPBASER_INNER_CACHEABILITY_MASK
+
+#define GICR_VPROPBASER_InnerShareable \
+ GIC_BASER_SHAREABILITY(GICR_VPROPBASER, InnerShareable)
+
+#define GICR_VPROPBASER_nCnB GIC_BASER_CACHEABILITY(GICR_VPROPBASER, INNER, nCnB)
+#define GICR_VPROPBASER_nC GIC_BASER_CACHEABILITY(GICR_VPROPBASER, INNER, nC)
+#define GICR_VPROPBASER_RaWt GIC_BASER_CACHEABILITY(GICR_VPROPBASER, INNER, RaWt)
+#define GICR_VPROPBASER_RaWb GIC_BASER_CACHEABILITY(GICR_VPROPBASER, INNER, RaWt)
+#define GICR_VPROPBASER_WaWt GIC_BASER_CACHEABILITY(GICR_VPROPBASER, INNER, WaWt)
+#define GICR_VPROPBASER_WaWb GIC_BASER_CACHEABILITY(GICR_VPROPBASER, INNER, WaWb)
+#define GICR_VPROPBASER_RaWaWt GIC_BASER_CACHEABILITY(GICR_VPROPBASER, INNER, RaWaWt)
+#define GICR_VPROPBASER_RaWaWb GIC_BASER_CACHEABILITY(GICR_VPROPBASER, INNER, RaWaWb)
+
+#define GICR_VPENDBASER 0x0078
+
+#define GICR_VPENDBASER_SHAREABILITY_SHIFT (10)
+#define GICR_VPENDBASER_INNER_CACHEABILITY_SHIFT (7)
+#define GICR_VPENDBASER_OUTER_CACHEABILITY_SHIFT (56)
+#define GICR_VPENDBASER_SHAREABILITY_MASK \
+ GIC_BASER_SHAREABILITY(GICR_VPENDBASER, SHAREABILITY_MASK)
+#define GICR_VPENDBASER_INNER_CACHEABILITY_MASK \
+ GIC_BASER_CACHEABILITY(GICR_VPENDBASER, INNER, MASK)
+#define GICR_VPENDBASER_OUTER_CACHEABILITY_MASK \
+ GIC_BASER_CACHEABILITY(GICR_VPENDBASER, OUTER, MASK)
+#define GICR_VPENDBASER_CACHEABILITY_MASK \
+ GICR_VPENDBASER_INNER_CACHEABILITY_MASK
+
+#define GICR_VPENDBASER_NonShareable \
+ GIC_BASER_SHAREABILITY(GICR_VPENDBASER, NonShareable)
+
+#define GICR_VPENDBASER_nCnB GIC_BASER_CACHEABILITY(GICR_VPENDBASER, INNER, nCnB)
+#define GICR_VPENDBASER_nC GIC_BASER_CACHEABILITY(GICR_VPENDBASER, INNER, nC)
+#define GICR_VPENDBASER_RaWt GIC_BASER_CACHEABILITY(GICR_VPENDBASER, INNER, RaWt)
+#define GICR_VPENDBASER_RaWb GIC_BASER_CACHEABILITY(GICR_VPENDBASER, INNER, RaWt)
+#define GICR_VPENDBASER_WaWt GIC_BASER_CACHEABILITY(GICR_VPENDBASER, INNER, WaWt)
+#define GICR_VPENDBASER_WaWb GIC_BASER_CACHEABILITY(GICR_VPENDBASER, INNER, WaWb)
+#define GICR_VPENDBASER_RaWaWt GIC_BASER_CACHEABILITY(GICR_VPENDBASER, INNER, RaWaWt)
+#define GICR_VPENDBASER_RaWaWb GIC_BASER_CACHEABILITY(GICR_VPENDBASER, INNER, RaWaWb)
+
#define GICR_VPENDBASER_Dirty (1ULL << 60)
#define GICR_VPENDBASER_PendingLast (1ULL << 61)
#define GICR_VPENDBASER_IDAI (1ULL << 62)
--
2.11.0
Add the required interfaces to map, unmap and update a VLPI.
Reviewed-by: Eric Auger <[email protected]>
Reviewed-by: Thomas Gleixner <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v4.c | 42 ++++++++++++++++++++++++++++++++++++++
include/linux/irqchip/arm-gic-v4.h | 4 ++++
2 files changed, 46 insertions(+)
diff --git a/drivers/irqchip/irq-gic-v4.c b/drivers/irqchip/irq-gic-v4.c
index 3c5ad70193ed..697d9dc1e005 100644
--- a/drivers/irqchip/irq-gic-v4.c
+++ b/drivers/irqchip/irq-gic-v4.c
@@ -81,3 +81,45 @@ int its_invall_vpe(struct its_vpe *vpe)
return its_send_vpe_cmd(vpe, &info);
}
+
+int its_map_vlpi(int irq, struct its_vlpi_map *map)
+{
+ struct its_cmd_info info = {
+ .cmd_type = MAP_VLPI,
+ .map = map,
+ };
+
+ /*
+ * The host will never see that interrupt firing again, so it
+ * is vital that we don't do any lazy masking.
+ */
+ irq_set_status_flags(irq, IRQ_DISABLE_UNLAZY);
+
+ return irq_set_vcpu_affinity(irq, &info);
+}
+
+int its_get_vlpi(int irq, struct its_vlpi_map *map)
+{
+ struct its_cmd_info info = {
+ .cmd_type = GET_VLPI,
+ .map = map,
+ };
+
+ return irq_set_vcpu_affinity(irq, &info);
+}
+
+int its_unmap_vlpi(int irq)
+{
+ irq_clear_status_flags(irq, IRQ_DISABLE_UNLAZY);
+ return irq_set_vcpu_affinity(irq, NULL);
+}
+
+int its_prop_update_vlpi(int irq, u8 config, bool inv)
+{
+ struct its_cmd_info info = {
+ .cmd_type = inv ? PROP_UPDATE_AND_INV_VLPI : PROP_UPDATE_VLPI,
+ .config = config,
+ };
+
+ return irq_set_vcpu_affinity(irq, &info);
+}
diff --git a/include/linux/irqchip/arm-gic-v4.h b/include/linux/irqchip/arm-gic-v4.h
index 97bb97a11fef..843c9ff29269 100644
--- a/include/linux/irqchip/arm-gic-v4.h
+++ b/include/linux/irqchip/arm-gic-v4.h
@@ -92,5 +92,9 @@ int its_alloc_vcpu_irqs(struct its_vm *vm);
void its_free_vcpu_irqs(struct its_vm *vm);
int its_schedule_vpe(struct its_vpe *vpe, bool on);
int its_invall_vpe(struct its_vpe *vpe);
+int its_map_vlpi(int irq, struct its_vlpi_map *map);
+int its_get_vlpi(int irq, struct its_vlpi_map *map);
+int its_unmap_vlpi(int irq);
+int its_prop_update_vlpi(int irq, u8 config, bool inv);
#endif
--
2.11.0
Add the basic GICv4 VPE (vcpu in GICv4 parlance) infrastructure
(irqchip, irq domain) that is going to be populated in the following
patches.
Reviewed-by: Eric Auger <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3-its.c | 24 ++++++++++++++++++++++++
1 file changed, 24 insertions(+)
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index 2e5e0499a0e6..0fd799c88d18 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -2023,6 +2023,13 @@ static const struct irq_domain_ops its_domain_ops = {
.deactivate = its_irq_domain_deactivate,
};
+static struct irq_chip its_vpe_irq_chip = {
+ .name = "GICv4-vpe",
+};
+
+static const struct irq_domain_ops its_vpe_domain_ops = {
+};
+
static int its_force_quiescent(void __iomem *base)
{
u32 count = 1000000; /* 1s */
@@ -2139,6 +2146,11 @@ static int its_init_domain(struct fwnode_handle *handle, struct its_node *its)
return 0;
}
+static int its_init_vpe_domain(void)
+{
+ return 0;
+}
+
static int __init its_compute_its_list_map(struct resource *res,
void __iomem *its_base)
{
@@ -2404,6 +2416,8 @@ int __init its_init(struct fwnode_handle *handle, struct rdists *rdists,
struct irq_domain *parent_domain)
{
struct device_node *of_node;
+ struct its_node *its;
+ bool has_v4 = false;
its_parent = parent_domain;
of_node = to_of_node(handle);
@@ -2421,5 +2435,15 @@ int __init its_init(struct fwnode_handle *handle, struct rdists *rdists,
its_alloc_lpi_tables();
its_lpi_init(rdists->id_bits);
+ list_for_each_entry(its, &its_nodes, entry)
+ has_v4 |= its->is_v4;
+
+ if (has_v4 & rdists->has_vlpis) {
+ if (its_init_vpe_domain()) {
+ rdists->has_vlpis = false;
+ pr_err("ITS: Disabling GICv4 support\n");
+ }
+ }
+
return 0;
}
--
2.11.0
When creating a VM, it is very convenient to have an irq domain
containing all the doorbell interrupts associated with that VM
(each interrupt representing a VPE).
Reviewed-by: Thomas Gleixner <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v4.c | 58 ++++++++++++++++++++++++++++++++++++++
include/linux/irqchip/arm-gic-v4.h | 3 ++
2 files changed, 61 insertions(+)
create mode 100644 drivers/irqchip/irq-gic-v4.c
diff --git a/drivers/irqchip/irq-gic-v4.c b/drivers/irqchip/irq-gic-v4.c
new file mode 100644
index 000000000000..2ea12baf2e53
--- /dev/null
+++ b/drivers/irqchip/irq-gic-v4.c
@@ -0,0 +1,58 @@
+/*
+ * Copyright (C) 2016 ARM Limited, All Rights Reserved.
+ * Author: Marc Zyngier <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/interrupt.h>
+#include <linux/irq.h>
+#include <linux/irqdomain.h>
+#include <linux/msi.h>
+
+#include <linux/irqchip/arm-gic-v4.h>
+
+static struct irq_domain *gic_domain;
+static const struct irq_domain_ops *vpe_domain_ops;
+
+int its_alloc_vcpu_irqs(struct its_vm *vm)
+{
+ int vpe_base_irq, i;
+
+ vm->domain = irq_domain_create_hierarchy(gic_domain, 0, vm->nr_vpes,
+ NULL, vpe_domain_ops, vm);
+ if (!vm->domain)
+ return -ENOMEM;
+
+ for (i = 0; i < vm->nr_vpes; i++) {
+ vm->vpes[i]->its_vm = vm;
+ vm->vpes[i]->idai = true;
+ }
+
+ vpe_base_irq = __irq_domain_alloc_irqs(vm->domain, -1, vm->nr_vpes,
+ NUMA_NO_NODE, vm,
+ false, NULL);
+ if (vpe_base_irq <= 0)
+ return -ENOMEM;
+
+ for (i = 0; i < vm->nr_vpes; i++)
+ vm->vpes[i]->irq = vpe_base_irq + i;
+
+ return 0;
+}
+
+void its_free_vcpu_irqs(struct its_vm *vm)
+{
+ irq_domain_free_irqs(vm->vpes[0]->irq, vm->nr_vpes);
+ irq_domain_remove(vm->domain);
+}
diff --git a/include/linux/irqchip/arm-gic-v4.h b/include/linux/irqchip/arm-gic-v4.h
index 4cac48e220da..7474a981ed6d 100644
--- a/include/linux/irqchip/arm-gic-v4.h
+++ b/include/linux/irqchip/arm-gic-v4.h
@@ -88,4 +88,7 @@ struct its_cmd_info {
};
};
+int its_alloc_vcpu_irqs(struct its_vm *vm);
+void its_free_vcpu_irqs(struct its_vm *vm);
+
#endif
--
2.11.0
When a guest issues a INVALL command targetting a collection, it must
be translated into a VINVALL for the VPE that has this collection.
This patch implements a hook that offers this functionallity to the
hypervisor.
Reviewed-by: Thomas Gleixner <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3-its.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index be8598354d5a..0eda36dc84b4 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -2211,6 +2211,10 @@ static int its_vpe_set_vcpu_affinity(struct irq_data *d, void *vcpu_info)
its_vpe_deschedule(vpe);
return 0;
+ case INVALL_VPE:
+ its_send_vinvall(vpe);
+ return 0;
+
default:
return -EINVAL;
}
--
2.11.0
When we're about to run a vcpu, it is crucial that the redistributor
associated with the physical CPU is being told about the new residency.
This is abstracted by hijacking the irq_set_affinity method for the
doorbell interrupt associated with the VPE. It is expected that the
hypervisor will call this method before scheduling the VPE.
Reviewed-by: Thomas Gleixner <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3-its.c | 96 ++++++++++++++++++++++++++++++++++++++++
1 file changed, 96 insertions(+)
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index 0eda36dc84b4..281eed031e01 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -150,6 +150,9 @@ static struct irq_domain *its_parent;
#define ITS_LIST_MAX 16
static unsigned long its_list_map;
+static u16 vmovp_seq_num;
+static DEFINE_RAW_SPINLOCK(vmovp_lock);
+
static DEFINE_IDA(its_vpeid_ida);
#define gic_data_rdist() (raw_cpu_ptr(gic_rdists->rdist))
@@ -240,6 +243,13 @@ struct its_cmd_desc {
u32 event_id;
bool db_enabled;
} its_vmovi_cmd;
+
+ struct {
+ struct its_vpe *vpe;
+ struct its_collection *col;
+ u16 seq_num;
+ u16 its_list;
+ } its_vmovp_cmd;
};
};
@@ -331,6 +341,16 @@ static void its_encode_db_valid(struct its_cmd_block *cmd, bool db_valid)
its_mask_encode(&cmd->raw_cmd[2], db_valid, 0, 0);
}
+static void its_encode_seq_num(struct its_cmd_block *cmd, u16 seq_num)
+{
+ its_mask_encode(&cmd->raw_cmd[0], seq_num, 47, 32);
+}
+
+static void its_encode_its_list(struct its_cmd_block *cmd, u16 its_list)
+{
+ its_mask_encode(&cmd->raw_cmd[1], its_list, 15, 0);
+}
+
static void its_encode_vpt_addr(struct its_cmd_block *cmd, u64 vpt_pa)
{
its_mask_encode(&cmd->raw_cmd[3], vpt_pa >> 16, 50, 16);
@@ -573,6 +593,20 @@ static struct its_vpe *its_build_vmovi_cmd(struct its_cmd_block *cmd,
return desc->its_vmovi_cmd.vpe;
}
+static struct its_vpe *its_build_vmovp_cmd(struct its_cmd_block *cmd,
+ struct its_cmd_desc *desc)
+{
+ its_encode_cmd(cmd, GITS_CMD_VMOVP);
+ its_encode_seq_num(cmd, desc->its_vmovp_cmd.seq_num);
+ its_encode_its_list(cmd, desc->its_vmovp_cmd.its_list);
+ its_encode_vpeid(cmd, desc->its_vmovp_cmd.vpe->vpe_id);
+ its_encode_target(cmd, desc->its_vmovp_cmd.col->target_address);
+
+ its_fixup_cmd(cmd);
+
+ return desc->its_vmovp_cmd.vpe;
+}
+
static u64 its_cmd_ptr_to_offset(struct its_node *its,
struct its_cmd_block *ptr)
{
@@ -867,6 +901,48 @@ static void its_send_vmapp(struct its_vpe *vpe, bool valid)
}
}
+static void its_send_vmovp(struct its_vpe *vpe)
+{
+ struct its_cmd_desc desc;
+ struct its_node *its;
+ unsigned long flags;
+ int col_id = vpe->col_idx;
+
+ desc.its_vmovp_cmd.vpe = vpe;
+ desc.its_vmovp_cmd.its_list = (u16)its_list_map;
+
+ if (!its_list_map) {
+ its = list_first_entry(&its_nodes, struct its_node, entry);
+ desc.its_vmovp_cmd.seq_num = 0;
+ desc.its_vmovp_cmd.col = &its->collections[col_id];
+ its_send_single_vcommand(its, its_build_vmovp_cmd, &desc);
+ return;
+ }
+
+ /*
+ * Yet another marvel of the architecture. If using the
+ * its_list "feature", we need to make sure that all ITSs
+ * receive all VMOVP commands in the same order. The only way
+ * to guarantee this is to make vmovp a serialization point.
+ *
+ * Wall <-- Head.
+ */
+ raw_spin_lock_irqsave(&vmovp_lock, flags);
+
+ desc.its_vmovp_cmd.seq_num = vmovp_seq_num++;
+
+ /* Emit VMOVPs */
+ list_for_each_entry(its, &its_nodes, entry) {
+ if (!its->is_v4)
+ continue;
+
+ desc.its_vmovp_cmd.col = &its->collections[col_id];
+ its_send_single_vcommand(its, its_build_vmovp_cmd, &desc);
+ }
+
+ raw_spin_unlock_irqrestore(&vmovp_lock, flags);
+}
+
static void its_send_vinvall(struct its_vpe *vpe)
{
struct its_cmd_desc desc;
@@ -2145,6 +2221,25 @@ static const struct irq_domain_ops its_domain_ops = {
.deactivate = its_irq_domain_deactivate,
};
+static int its_vpe_set_affinity(struct irq_data *d,
+ const struct cpumask *mask_val,
+ bool force)
+{
+ struct its_vpe *vpe = irq_data_get_irq_chip_data(d);
+ int cpu = cpumask_first(mask_val);
+
+ /*
+ * Changing affinity is mega expensive, so let's be as lazy as
+ * we can and only do it if we really have to.
+ */
+ if (vpe->col_idx != cpu) {
+ vpe->col_idx = cpu;
+ its_send_vmovp(vpe);
+ }
+
+ return IRQ_SET_MASK_OK_DONE;
+}
+
static void its_vpe_schedule(struct its_vpe *vpe)
{
void * __iomem vlpi_base = gic_data_rdist_vlpi_base();
@@ -2222,6 +2317,7 @@ static int its_vpe_set_vcpu_affinity(struct irq_data *d, void *vcpu_info)
static struct irq_chip its_vpe_irq_chip = {
.name = "GICv4-vpe",
+ .irq_set_affinity = its_vpe_set_affinity,
.irq_set_vcpu_affinity = its_vpe_set_vcpu_affinity,
};
--
2.11.0
When a VLPI is reconfigured (enabled, disabled, change in priority),
the full configuration byte must be written, and the caches invalidated.
Also, when using the irq_mask/irq_unmask methods, it is necessary
to disable the doorbell for that particular interrupt (by mapping it
to 1023) on top of clearing the Enable bit.
Reviewed-by: Thomas Gleixner <[email protected]>
Reviewed-by: Eric Auger <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3-its.c | 73 +++++++++++++++++++++++++++++++++++++---
1 file changed, 69 insertions(+), 4 deletions(-)
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index 527e4e95e10d..2e5e0499a0e6 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -808,18 +808,26 @@ static inline u32 its_get_event_id(struct irq_data *d)
return d->hwirq - its_dev->event_map.lpi_base;
}
-static void lpi_update_config(struct irq_data *d, u8 clr, u8 set)
+static void lpi_write_config(struct irq_data *d, u8 clr, u8 set)
{
struct its_device *its_dev = irq_data_get_irq_chip_data(d);
- irq_hw_number_t hwirq = d->hwirq;
+ irq_hw_number_t hwirq;
struct page *prop_page;
u8 *cfg;
- prop_page = gic_rdists->prop_page;
+ if (irqd_is_forwarded_to_vcpu(d)) {
+ u32 event = its_get_event_id(d);
+
+ prop_page = its_dev->event_map.vm->vprop_page;
+ hwirq = its_dev->event_map.vlpi_maps[event].vintid;
+ } else {
+ prop_page = gic_rdists->prop_page;
+ hwirq = d->hwirq;
+ }
cfg = page_address(prop_page) + hwirq - 8192;
*cfg &= ~clr;
- *cfg |= set;
+ *cfg |= set | LPI_PROP_GROUP1;
/*
* Make the above write visible to the redistributors.
@@ -830,16 +838,52 @@ static void lpi_update_config(struct irq_data *d, u8 clr, u8 set)
gic_flush_dcache_to_poc(cfg, sizeof(*cfg));
else
dsb(ishst);
+}
+
+static void lpi_update_config(struct irq_data *d, u8 clr, u8 set)
+{
+ struct its_device *its_dev = irq_data_get_irq_chip_data(d);
+
+ lpi_write_config(d, clr, set);
its_send_inv(its_dev, its_get_event_id(d));
}
+static void its_vlpi_set_doorbell(struct irq_data *d, bool enable)
+{
+ struct its_device *its_dev = irq_data_get_irq_chip_data(d);
+ u32 event = its_get_event_id(d);
+
+ if (its_dev->event_map.vlpi_maps[event].db_enabled == enable)
+ return;
+
+ its_dev->event_map.vlpi_maps[event].db_enabled = enable;
+
+ /*
+ * More fun with the architecture:
+ *
+ * Ideally, we'd issue a VMAPTI to set the doorbell to its LPI
+ * value or to 1023, depending on the enable bit. But that
+ * would be issueing a mapping for an /existing/ DevID+EventID
+ * pair, which is UNPREDICTABLE. Instead, let's issue a VMOVI
+ * to the /same/ vPE, using this opportunity to adjust the
+ * doorbell. Mouahahahaha. We loves it, Precious.
+ */
+ its_send_vmovi(its_dev, event);
+}
+
static void its_mask_irq(struct irq_data *d)
{
+ if (irqd_is_forwarded_to_vcpu(d))
+ its_vlpi_set_doorbell(d, false);
+
lpi_update_config(d, LPI_PROP_ENABLED, 0);
}
static void its_unmask_irq(struct irq_data *d)
{
+ if (irqd_is_forwarded_to_vcpu(d))
+ its_vlpi_set_doorbell(d, true);
+
lpi_update_config(d, 0, LPI_PROP_ENABLED);
}
@@ -852,6 +896,10 @@ static int its_set_affinity(struct irq_data *d, const struct cpumask *mask_val,
struct its_collection *target_col;
u32 id = its_get_event_id(d);
+ /* A forwarded interrupt should use irq_set_vcpu_affinity */
+ if (irqd_is_forwarded_to_vcpu(d))
+ return -EINVAL;
+
/* lpi cannot be routed to a redistributor that is on a foreign node */
if (its_dev->its->flags & ITS_FLAGS_WORKAROUND_CAVIUM_23144) {
if (its_dev->its->numa_node >= 0) {
@@ -1017,6 +1065,22 @@ static int its_vlpi_unmap(struct irq_data *d)
return ret;
}
+static int its_vlpi_prop_update(struct irq_data *d, struct its_cmd_info *info)
+{
+ struct its_device *its_dev = irq_data_get_irq_chip_data(d);
+
+ if (!its_dev->event_map.vm || !irqd_is_forwarded_to_vcpu(d))
+ return -EINVAL;
+
+ if (info->cmd_type == PROP_UPDATE_AND_INV_VLPI)
+ lpi_update_config(d, 0xff, info->config);
+ else
+ lpi_write_config(d, 0xff, info->config);
+ its_vlpi_set_doorbell(d, !!(info->config & LPI_PROP_ENABLED));
+
+ return 0;
+}
+
static int its_irq_set_vcpu_affinity(struct irq_data *d, void *vcpu_info)
{
struct its_device *its_dev = irq_data_get_irq_chip_data(d);
@@ -1039,6 +1103,7 @@ static int its_irq_set_vcpu_affinity(struct irq_data *d, void *vcpu_info)
case PROP_UPDATE_VLPI:
case PROP_UPDATE_AND_INV_VLPI:
+ return its_vlpi_prop_update(d, info);
default:
return -EINVAL;
--
2.11.0
V{PEND,PROP}BASER being 64bit registers, they need some ad-hoc
accessors on 32bit, specially given that VPENDBASER contains
a Valid bit, making the access a bit convoluted.
Reviewed-by: Thomas Gleixner <[email protected]>
Reviewed-by: Eric Auger <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
arch/arm/include/asm/arch_gicv3.h | 28 ++++++++++++++++++++++++++++
arch/arm64/include/asm/arch_gicv3.h | 5 +++++
include/linux/irqchip/arm-gic-v3.h | 5 +++++
3 files changed, 38 insertions(+)
diff --git a/arch/arm/include/asm/arch_gicv3.h b/arch/arm/include/asm/arch_gicv3.h
index 27475904e096..8d45e88feac9 100644
--- a/arch/arm/include/asm/arch_gicv3.h
+++ b/arch/arm/include/asm/arch_gicv3.h
@@ -291,5 +291,33 @@ static inline u64 __gic_readq_nonatomic(const volatile void __iomem *addr)
*/
#define gits_write_cwriter(v, c) __gic_writeq_nonatomic(v, c)
+/*
+ * GITS_VPROPBASER - hi and lo bits may be accessed independently.
+ */
+#define gits_write_vpropbaser(v, c) __gic_writeq_nonatomic(v, c)
+
+/*
+ * GITS_VPENDBASER - the Valid bit must be cleared before changing
+ * anything else.
+ */
+static inline void gits_write_vpendbaser(u64 val, void * __iomem addr)
+{
+ u32 tmp;
+
+ tmp = readl_relaxed(addr + 4);
+ if (tmp & (GICR_VPENDBASER_Valid >> 32)) {
+ tmp &= ~(GICR_VPENDBASER_Valid >> 32);
+ writel_relaxed(tmp, addr + 4);
+ }
+
+ /*
+ * Use the fact that __gic_writeq_nonatomic writes the second
+ * half of the 64bit quantity after the first.
+ */
+ __gic_writeq_nonatomic(val, addr);
+}
+
+#define gits_read_vpendbaser(c) __gic_readq_nonatomic(c)
+
#endif /* !__ASSEMBLY__ */
#endif /* !__ASM_ARCH_GICV3_H */
diff --git a/arch/arm64/include/asm/arch_gicv3.h b/arch/arm64/include/asm/arch_gicv3.h
index 8cef47fa2218..0d2a53457c30 100644
--- a/arch/arm64/include/asm/arch_gicv3.h
+++ b/arch/arm64/include/asm/arch_gicv3.h
@@ -133,5 +133,10 @@ static inline void gic_write_bpr1(u32 val)
#define gicr_write_pendbaser(v, c) writeq_relaxed(v, c)
#define gicr_read_pendbaser(c) readq_relaxed(c)
+#define gits_write_vpropbaser(v, c) writeq_relaxed(v, c)
+
+#define gits_write_vpendbaser(v, c) writeq_relaxed(v, c)
+#define gits_read_vpendbaser(c) readq_relaxed(c)
+
#endif /* __ASSEMBLY__ */
#endif /* __ASM_ARCH_GICV3_H */
diff --git a/include/linux/irqchip/arm-gic-v3.h b/include/linux/irqchip/arm-gic-v3.h
index 7c6fd8f3e36c..17ba0d732f12 100644
--- a/include/linux/irqchip/arm-gic-v3.h
+++ b/include/linux/irqchip/arm-gic-v3.h
@@ -212,6 +212,11 @@
#define LPI_PROP_GROUP1 (1 << 1)
#define LPI_PROP_ENABLED (1 << 0)
+#define GICR_VPENDBASER_Dirty (1ULL << 60)
+#define GICR_VPENDBASER_PendingLast (1ULL << 61)
+#define GICR_VPENDBASER_IDAI (1ULL << 62)
+#define GICR_VPENDBASER_Valid (1ULL << 63)
+
/*
* ITS registers, offsets from ITS_base
*/
--
2.11.0
When creating a VM, the low level GICv4 code is responsible for:
- allocating each VPE a unique VPEID
- allocating a doorbell interrupt for each VPE
- allocating the pending tables for each VPE
- allocating the property table for the VM
This of course has to be reversed when the VM is brought down.
All of this is wired into the irq domain alloc/free methods.
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3-its.c | 169 +++++++++++++++++++++++++++++++++++++++
1 file changed, 169 insertions(+)
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index 0fd799c88d18..58f3b19ec462 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -150,6 +150,7 @@ static struct irq_domain *its_parent;
#define ITS_LIST_MAX 16
static unsigned long its_list_map;
+static DEFINE_IDA(its_vpeid_ida);
#define gic_data_rdist() (raw_cpu_ptr(gic_rdists->rdist))
#define gic_data_rdist_rd_base() (gic_data_rdist()->rd_base)
@@ -1247,6 +1248,11 @@ static struct page *its_allocate_prop_table(gfp_t gfp_flags)
return prop_page;
}
+static void its_free_prop_table(struct page *prop_page)
+{
+ free_pages((unsigned long)page_address(prop_page),
+ get_order(LPI_PROPBASE_SZ));
+}
static int __init its_alloc_lpi_tables(void)
{
@@ -1548,6 +1554,12 @@ static struct page *its_allocate_pending_table(gfp_t gfp_flags)
return pend_page;
}
+static void its_free_pending_table(struct page *pt)
+{
+ free_pages((unsigned long)page_address(pt),
+ get_order(max(LPI_PENDBASE_SZ, SZ_64K)));
+}
+
static void its_cpu_init_lpis(void)
{
void __iomem *rbase = gic_data_rdist_rd_base();
@@ -1770,6 +1782,34 @@ static bool its_alloc_device_table(struct its_node *its, u32 dev_id)
return its_alloc_table_entry(baser, dev_id);
}
+static bool its_alloc_vpe_table(u32 vpe_id)
+{
+ struct its_node *its;
+
+ /*
+ * Make sure the L2 tables are allocated on *all* v4 ITSs. We
+ * could try and only do it on ITSs corresponding to devices
+ * that have interrupts targeted at this VPE, but the
+ * complexity becomes crazy (and you have tons of memory
+ * anyway, right?).
+ */
+ list_for_each_entry(its, &its_nodes, entry) {
+ struct its_baser *baser;
+
+ if (!its->is_v4)
+ continue;
+
+ baser = its_get_baser(its, GITS_BASER_TYPE_VCPU);
+ if (!baser)
+ return false;
+
+ if (!its_alloc_table_entry(baser, vpe_id))
+ return false;
+ }
+
+ return true;
+}
+
static struct its_device *its_create_device(struct its_node *its, u32 dev_id,
int nvecs)
{
@@ -2027,7 +2067,136 @@ static struct irq_chip its_vpe_irq_chip = {
.name = "GICv4-vpe",
};
+static int its_vpe_id_alloc(void)
+{
+ return ida_simple_get(&its_vpeid_ida, 0, 1 << 16, GFP_KERNEL);
+}
+
+static void its_vpe_id_free(u16 id)
+{
+ ida_simple_remove(&its_vpeid_ida, id);
+}
+
+static int its_vpe_init(struct its_vpe *vpe)
+{
+ struct page *vpt_page;
+ int vpe_id;
+
+ /* Allocate vpe_id */
+ vpe_id = its_vpe_id_alloc();
+ if (vpe_id < 0)
+ return vpe_id;
+
+ /* Allocate VPT */
+ vpt_page = its_allocate_pending_table(GFP_KERNEL);
+ if (!vpt_page) {
+ its_vpe_id_free(vpe_id);
+ return -ENOMEM;
+ }
+
+ if (!its_alloc_vpe_table(vpe_id)) {
+ its_vpe_id_free(vpe_id);
+ its_free_pending_table(vpe->vpt_page);
+ return -ENOMEM;
+ }
+
+ vpe->vpe_id = vpe_id;
+ vpe->vpt_page = vpt_page;
+
+ return 0;
+}
+
+static void its_vpe_teardown(struct its_vpe *vpe)
+{
+ its_vpe_id_free(vpe->vpe_id);
+ its_free_pending_table(vpe->vpt_page);
+}
+
+static void its_vpe_irq_domain_free(struct irq_domain *domain,
+ unsigned int virq,
+ unsigned int nr_irqs)
+{
+ struct its_vm *vm = domain->host_data;
+ int i;
+
+ irq_domain_free_irqs_parent(domain, virq, nr_irqs);
+
+ for (i = 0; i < nr_irqs; i++) {
+ struct irq_data *data = irq_domain_get_irq_data(domain,
+ virq + i);
+ struct its_vpe *vpe = irq_data_get_irq_chip_data(data);
+
+ BUG_ON(vm != vpe->its_vm);
+
+ clear_bit(data->hwirq, vm->db_bitmap);
+ its_vpe_teardown(vpe);
+ irq_domain_reset_irq_data(data);
+ }
+
+ if (bitmap_empty(vm->db_bitmap, vm->nr_db_lpis)) {
+ its_lpi_free_chunks(vm->db_bitmap, vm->db_lpi_base, vm->nr_db_lpis);
+ its_free_prop_table(vm->vprop_page);
+ }
+}
+
+static int its_vpe_irq_domain_alloc(struct irq_domain *domain, unsigned int virq,
+ unsigned int nr_irqs, void *args)
+{
+ struct its_vm *vm = args;
+ unsigned long *bitmap;
+ struct page *vprop_page;
+ int base, nr_ids, i, err = 0;
+
+ BUG_ON(!vm);
+
+ bitmap = its_lpi_alloc_chunks(nr_irqs, &base, &nr_ids);
+ if (!bitmap)
+ return -ENOMEM;
+
+ if (nr_ids < nr_irqs) {
+ its_lpi_free_chunks(bitmap, base, nr_ids);
+ return -ENOMEM;
+ }
+
+ vprop_page = its_allocate_prop_table(GFP_KERNEL);
+ if (!vprop_page) {
+ its_lpi_free_chunks(bitmap, base, nr_ids);
+ return -ENOMEM;
+ }
+
+ vm->db_bitmap = bitmap;
+ vm->db_lpi_base = base;
+ vm->nr_db_lpis = nr_ids;
+ vm->vprop_page = vprop_page;
+
+ for (i = 0; i < nr_irqs; i++) {
+ vm->vpes[i]->vpe_db_lpi = base + i;
+ err = its_vpe_init(vm->vpes[i]);
+ if (err)
+ break;
+ err = its_irq_gic_domain_alloc(domain, virq + i,
+ vm->vpes[i]->vpe_db_lpi);
+ if (err)
+ break;
+ irq_domain_set_hwirq_and_chip(domain, virq + i, i,
+ &its_vpe_irq_chip, vm->vpes[i]);
+ set_bit(i, bitmap);
+ }
+
+ if (err) {
+ if (i > 0)
+ its_vpe_irq_domain_free(domain, virq, i - 1);
+
+ its_lpi_free_chunks(bitmap, base, nr_ids);
+ its_free_prop_table(vprop_page);
+ }
+
+ return err;
+}
+
static const struct irq_domain_ops its_vpe_domain_ops = {
+ .alloc = its_vpe_irq_domain_alloc,
+ .free = its_vpe_irq_domain_free,
};
static int its_force_quiescent(void __iomem *base)
--
2.11.0
On activation, a VPE is mapped using the VMAPP command, followed
by a VINVALL for a good measure. On deactivation, the VPE is
simply unmapped.
Reviewed-by: Thomas Gleixner <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3-its.c | 102 +++++++++++++++++++++++++++++++++++++++
1 file changed, 102 insertions(+)
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index 58f3b19ec462..276602efdfa0 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -217,6 +217,16 @@ struct its_cmd_desc {
struct {
struct its_vpe *vpe;
+ } its_vinvall_cmd;
+
+ struct {
+ struct its_vpe *vpe;
+ struct its_collection *col;
+ bool valid;
+ } its_vmapp_cmd;
+
+ struct {
+ struct its_vpe *vpe;
struct its_device *dev;
u32 virt_id;
u32 event_id;
@@ -320,6 +330,16 @@ static void its_encode_db_valid(struct its_cmd_block *cmd, bool db_valid)
its_mask_encode(&cmd->raw_cmd[2], db_valid, 0, 0);
}
+static void its_encode_vpt_addr(struct its_cmd_block *cmd, u64 vpt_pa)
+{
+ its_mask_encode(&cmd->raw_cmd[3], vpt_pa >> 16, 50, 16);
+}
+
+static void its_encode_vpt_size(struct its_cmd_block *cmd, u8 vpt_size)
+{
+ its_mask_encode(&cmd->raw_cmd[3], vpt_size, 4, 0);
+}
+
static inline void its_fixup_cmd(struct its_cmd_block *cmd)
{
/* Let's fixup BE commands */
@@ -478,6 +498,36 @@ static struct its_collection *its_build_invall_cmd(struct its_cmd_block *cmd,
return NULL;
}
+static struct its_vpe *its_build_vinvall_cmd(struct its_cmd_block *cmd,
+ struct its_cmd_desc *desc)
+{
+ its_encode_cmd(cmd, GITS_CMD_VINVALL);
+ its_encode_vpeid(cmd, desc->its_vinvall_cmd.vpe->vpe_id);
+
+ its_fixup_cmd(cmd);
+
+ return desc->its_vinvall_cmd.vpe;
+}
+
+static struct its_vpe *its_build_vmapp_cmd(struct its_cmd_block *cmd,
+ struct its_cmd_desc *desc)
+{
+ unsigned long vpt_addr;
+
+ vpt_addr = virt_to_phys(page_address(desc->its_vmapp_cmd.vpe->vpt_page));
+
+ its_encode_cmd(cmd, GITS_CMD_VMAPP);
+ its_encode_vpeid(cmd, desc->its_vmapp_cmd.vpe->vpe_id);
+ its_encode_valid(cmd, desc->its_vmapp_cmd.valid);
+ its_encode_target(cmd, desc->its_vmapp_cmd.col->target_address);
+ its_encode_vpt_addr(cmd, vpt_addr);
+ its_encode_vpt_size(cmd, LPI_NRBITS - 1);
+
+ its_fixup_cmd(cmd);
+
+ return desc->its_vmapp_cmd.vpe;
+}
+
static struct its_vpe *its_build_vmapti_cmd(struct its_cmd_block *cmd,
struct its_cmd_desc *desc)
{
@@ -799,6 +849,37 @@ static void its_send_vmovi(struct its_device *dev, u32 id)
its_send_single_vcommand(dev->its, its_build_vmovi_cmd, &desc);
}
+static void its_send_vmapp(struct its_vpe *vpe, bool valid)
+{
+ struct its_cmd_desc desc;
+ struct its_node *its;
+
+ desc.its_vmapp_cmd.vpe = vpe;
+ desc.its_vmapp_cmd.valid = valid;
+
+ list_for_each_entry(its, &its_nodes, entry) {
+ if (!its->is_v4)
+ continue;
+
+ desc.its_vmapp_cmd.col = &its->collections[vpe->col_idx];
+ its_send_single_vcommand(its, its_build_vmapp_cmd, &desc);
+ }
+}
+
+static void its_send_vinvall(struct its_vpe *vpe)
+{
+ struct its_cmd_desc desc;
+ struct its_node *its;
+
+ desc.its_vinvall_cmd.vpe = vpe;
+
+ list_for_each_entry(its, &its_nodes, entry) {
+ if (!its->is_v4)
+ continue;
+ its_send_single_vcommand(its, its_build_vinvall_cmd, &desc);
+ }
+}
+
/*
* irqchip functions - assumes MSI, mostly.
*/
@@ -2194,9 +2275,30 @@ static int its_vpe_irq_domain_alloc(struct irq_domain *domain, unsigned int virq
return err;
}
+static void its_vpe_irq_domain_activate(struct irq_domain *domain,
+ struct irq_data *d)
+{
+ struct its_vpe *vpe = irq_data_get_irq_chip_data(d);
+
+ /* Map the VPE to the first possible CPU */
+ vpe->col_idx = cpumask_first(cpu_online_mask);
+ its_send_vmapp(vpe, true);
+ its_send_vinvall(vpe);
+}
+
+static void its_vpe_irq_domain_deactivate(struct irq_domain *domain,
+ struct irq_data *d)
+{
+ struct its_vpe *vpe = irq_data_get_irq_chip_data(d);
+
+ its_send_vmapp(vpe, false);
+}
+
static const struct irq_domain_ops its_vpe_domain_ops = {
.alloc = its_vpe_irq_domain_alloc,
.free = its_vpe_irq_domain_free,
+ .activate = its_vpe_irq_domain_activate,
+ .deactivate = its_vpe_irq_domain_deactivate,
};
static int its_force_quiescent(void __iomem *base)
--
2.11.0
In order to let a VLPI being injected into a guest, the VLPI must
be mapped using the VMAPTI command. When moved to a different vcpu,
it must be moved with the VMOVI command.
These commands are issued via the irq_set_vcpu_affinity method,
making sure we unmap the corresponding host LPI first.
The reverse is also done when the VLPI is unmapped from the guest.
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3-its.c | 250 ++++++++++++++++++++++++++++++++++++++-
1 file changed, 247 insertions(+), 3 deletions(-)
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index c02e1e6cd169..527e4e95e10d 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -117,11 +117,17 @@ struct event_lpi_map {
u16 *col_map;
irq_hw_number_t lpi_base;
int nr_lpis;
+ struct mutex vlpi_lock;
+ struct its_vm *vm;
+ struct its_vlpi_map *vlpi_maps;
+ int nr_vlpis;
};
/*
- * The ITS view of a device - belongs to an ITS, a collection, owns an
- * interrupt translation table, and a list of interrupts.
+ * The ITS view of a device - belongs to an ITS, owns an interrupt
+ * translation table, and a list of interrupts. If it some of its
+ * LPIs are injected into a guest (GICv4), the event_map.vm field
+ * indicates which one.
*/
struct its_device {
struct list_head entry;
@@ -207,6 +213,21 @@ struct its_cmd_desc {
struct {
struct its_collection *col;
} its_invall_cmd;
+
+ struct {
+ struct its_vpe *vpe;
+ struct its_device *dev;
+ u32 virt_id;
+ u32 event_id;
+ bool db_enabled;
+ } its_vmapti_cmd;
+
+ struct {
+ struct its_vpe *vpe;
+ struct its_device *dev;
+ u32 event_id;
+ bool db_enabled;
+ } its_vmovi_cmd;
};
};
@@ -223,6 +244,9 @@ struct its_cmd_block {
typedef struct its_collection *(*its_cmd_builder_t)(struct its_cmd_block *,
struct its_cmd_desc *);
+typedef struct its_vpe *(*its_cmd_vbuilder_t)(struct its_cmd_block *,
+ struct its_cmd_desc *);
+
static void its_mask_encode(u64 *raw_cmd, u64 val, int h, int l)
{
u64 mask = GENMASK_ULL(h, l);
@@ -275,6 +299,26 @@ static void its_encode_collection(struct its_cmd_block *cmd, u16 col)
its_mask_encode(&cmd->raw_cmd[2], col, 15, 0);
}
+static void its_encode_vpeid(struct its_cmd_block *cmd, u16 vpeid)
+{
+ its_mask_encode(&cmd->raw_cmd[1], vpeid, 47, 32);
+}
+
+static void its_encode_virt_id(struct its_cmd_block *cmd, u32 virt_id)
+{
+ its_mask_encode(&cmd->raw_cmd[2], virt_id, 31, 0);
+}
+
+static void its_encode_db_phys_id(struct its_cmd_block *cmd, u32 db_phys_id)
+{
+ its_mask_encode(&cmd->raw_cmd[2], db_phys_id, 63, 32);
+}
+
+static void its_encode_db_valid(struct its_cmd_block *cmd, bool db_valid)
+{
+ its_mask_encode(&cmd->raw_cmd[2], db_valid, 0, 0);
+}
+
static inline void its_fixup_cmd(struct its_cmd_block *cmd)
{
/* Let's fixup BE commands */
@@ -433,6 +477,50 @@ static struct its_collection *its_build_invall_cmd(struct its_cmd_block *cmd,
return NULL;
}
+static struct its_vpe *its_build_vmapti_cmd(struct its_cmd_block *cmd,
+ struct its_cmd_desc *desc)
+{
+ u32 db;
+
+ if (desc->its_vmapti_cmd.db_enabled)
+ db = desc->its_vmapti_cmd.vpe->vpe_db_lpi;
+ else
+ db = 1023;
+
+ its_encode_cmd(cmd, GITS_CMD_VMAPTI);
+ its_encode_devid(cmd, desc->its_vmapti_cmd.dev->device_id);
+ its_encode_vpeid(cmd, desc->its_vmapti_cmd.vpe->vpe_id);
+ its_encode_event_id(cmd, desc->its_vmapti_cmd.event_id);
+ its_encode_db_phys_id(cmd, db);
+ its_encode_virt_id(cmd, desc->its_vmapti_cmd.virt_id);
+
+ its_fixup_cmd(cmd);
+
+ return desc->its_vmapti_cmd.vpe;
+}
+
+static struct its_vpe *its_build_vmovi_cmd(struct its_cmd_block *cmd,
+ struct its_cmd_desc *desc)
+{
+ u32 db;
+
+ if (desc->its_vmovi_cmd.db_enabled)
+ db = desc->its_vmovi_cmd.vpe->vpe_db_lpi;
+ else
+ db = 1023;
+
+ its_encode_cmd(cmd, GITS_CMD_VMOVI);
+ its_encode_devid(cmd, desc->its_vmovi_cmd.dev->device_id);
+ its_encode_vpeid(cmd, desc->its_vmovi_cmd.vpe->vpe_id);
+ its_encode_event_id(cmd, desc->its_vmovi_cmd.event_id);
+ its_encode_db_phys_id(cmd, db);
+ its_encode_db_valid(cmd, true);
+
+ its_fixup_cmd(cmd);
+
+ return desc->its_vmovi_cmd.vpe;
+}
+
static u64 its_cmd_ptr_to_offset(struct its_node *its,
struct its_cmd_block *ptr)
{
@@ -578,6 +666,18 @@ static void its_build_sync_cmd(struct its_cmd_block *sync_cmd,
static BUILD_SINGLE_CMD_FUNC(its_send_single_command, its_cmd_builder_t,
struct its_collection, its_build_sync_cmd)
+static void its_build_vsync_cmd(struct its_cmd_block *sync_cmd,
+ struct its_vpe *sync_vpe)
+{
+ its_encode_cmd(sync_cmd, GITS_CMD_VSYNC);
+ its_encode_vpeid(sync_cmd, sync_vpe->vpe_id);
+
+ its_fixup_cmd(sync_cmd);
+}
+
+static BUILD_SINGLE_CMD_FUNC(its_send_single_vcommand, its_cmd_vbuilder_t,
+ struct its_vpe, its_build_vsync_cmd)
+
static void its_send_int(struct its_device *dev, u32 event_id)
{
struct its_cmd_desc desc;
@@ -671,6 +771,33 @@ static void its_send_invall(struct its_node *its, struct its_collection *col)
its_send_single_command(its, its_build_invall_cmd, &desc);
}
+static void its_send_vmapti(struct its_device *dev, u32 id)
+{
+ struct its_vlpi_map *map = &dev->event_map.vlpi_maps[id];
+ struct its_cmd_desc desc;
+
+ desc.its_vmapti_cmd.vpe = map->vm->vpes[map->vpe_idx];
+ desc.its_vmapti_cmd.dev = dev;
+ desc.its_vmapti_cmd.virt_id = map->vintid;
+ desc.its_vmapti_cmd.event_id = id;
+ desc.its_vmapti_cmd.db_enabled = map->db_enabled;
+
+ its_send_single_vcommand(dev->its, its_build_vmapti_cmd, &desc);
+}
+
+static void its_send_vmovi(struct its_device *dev, u32 id)
+{
+ struct its_vlpi_map *map = &dev->event_map.vlpi_maps[id];
+ struct its_cmd_desc desc;
+
+ desc.its_vmovi_cmd.vpe = map->vm->vpes[map->vpe_idx];
+ desc.its_vmovi_cmd.dev = dev;
+ desc.its_vmovi_cmd.event_id = id;
+ desc.its_vmovi_cmd.db_enabled = map->db_enabled;
+
+ its_send_single_vcommand(dev->its, its_build_vmovi_cmd, &desc);
+}
+
/*
* irqchip functions - assumes MSI, mostly.
*/
@@ -780,19 +907,135 @@ static int its_irq_set_irqchip_state(struct irq_data *d,
return 0;
}
+static int its_vlpi_map(struct irq_data *d, struct its_cmd_info *info)
+{
+ struct its_device *its_dev = irq_data_get_irq_chip_data(d);
+ u32 event = its_get_event_id(d);
+ int ret = 0;
+
+ if (!info->map)
+ return -EINVAL;
+
+ mutex_lock(&its_dev->event_map.vlpi_lock);
+
+ if (!its_dev->event_map.vm) {
+ struct its_vlpi_map *maps;
+
+ maps = kzalloc(sizeof(*maps) * its_dev->event_map.nr_lpis,
+ GFP_KERNEL);
+ if (!maps) {
+ ret = -ENOMEM;
+ goto out;
+ }
+
+ its_dev->event_map.vm = info->map->vm;
+ its_dev->event_map.vlpi_maps = maps;
+ } else if (its_dev->event_map.vm != info->map->vm) {
+ ret = -EINVAL;
+ goto out;
+ }
+
+ /* Get our private copy of the mapping information */
+ its_dev->event_map.vlpi_maps[event] = *info->map;
+
+ if (irqd_is_forwarded_to_vcpu(d)) {
+ /* Already mapped, move it around */
+ its_send_vmovi(its_dev, event);
+ } else {
+ /* Drop the physical mapping */
+ its_send_discard(its_dev, event);
+
+ /* and install the virtual one */
+ its_send_vmapti(its_dev, event);
+ irqd_set_forwarded_to_vcpu(d);
+
+ /* Increment the number of VLPIs */
+ its_dev->event_map.nr_vlpis++;
+ }
+
+out:
+ mutex_unlock(&its_dev->event_map.vlpi_lock);
+ return ret;
+}
+
+static int its_vlpi_get(struct irq_data *d, struct its_cmd_info *info)
+{
+ struct its_device *its_dev = irq_data_get_irq_chip_data(d);
+ u32 event = its_get_event_id(d);
+ int ret = 0;
+
+ mutex_lock(&its_dev->event_map.vlpi_lock);
+
+ if (!its_dev->event_map.vm ||
+ !its_dev->event_map.vlpi_maps[event].vm) {
+ ret = -EINVAL;
+ goto out;
+ }
+
+ /* Copy our mapping information to the incoming request */
+ *info->map = its_dev->event_map.vlpi_maps[event];
+
+out:
+ mutex_unlock(&its_dev->event_map.vlpi_lock);
+ return ret;
+}
+
+static int its_vlpi_unmap(struct irq_data *d)
+{
+ struct its_device *its_dev = irq_data_get_irq_chip_data(d);
+ u32 event = its_get_event_id(d);
+ int ret = 0;
+
+ mutex_lock(&its_dev->event_map.vlpi_lock);
+
+ if (!its_dev->event_map.vm || !irqd_is_forwarded_to_vcpu(d)) {
+ ret = -EINVAL;
+ goto out;
+ }
+
+ /* Drop the virtual mapping */
+ its_send_discard(its_dev, event);
+
+ /* and restore the physical one */
+ irqd_clr_forwarded_to_vcpu(d);
+ its_send_mapti(its_dev, d->hwirq, event);
+ lpi_update_config(d, 0xff, (LPI_PROP_DEFAULT_PRIO |
+ LPI_PROP_ENABLED |
+ LPI_PROP_GROUP1));
+
+ /*
+ * Drop the refcount and make the device available again if
+ * this was the last VLPI.
+ */
+ if (!--its_dev->event_map.nr_vlpis) {
+ its_dev->event_map.vm = NULL;
+ kfree(its_dev->event_map.vlpi_maps);
+ }
+
+out:
+ mutex_unlock(&its_dev->event_map.vlpi_lock);
+ return ret;
+}
+
static int its_irq_set_vcpu_affinity(struct irq_data *d, void *vcpu_info)
{
struct its_device *its_dev = irq_data_get_irq_chip_data(d);
struct its_cmd_info *info = vcpu_info;
/* Need a v4 ITS */
- if (!its_dev->its->is_v4 || !info)
+ if (!its_dev->its->is_v4)
return -EINVAL;
+ /* Unmap request? */
+ if (!info)
+ return its_vlpi_unmap(d);
+
switch (info->cmd_type) {
case MAP_VLPI:
+ return its_vlpi_map(d, info);
case GET_VLPI:
+ return its_vlpi_get(d, info);
case PROP_UPDATE_VLPI:
case PROP_UPDATE_AND_INV_VLPI:
@@ -1509,6 +1752,7 @@ static struct its_device *its_create_device(struct its_node *its, u32 dev_id,
dev->event_map.col_map = col_map;
dev->event_map.lpi_base = lpi_base;
dev->event_map.nr_lpis = nr_lpis;
+ mutex_init(&dev->event_map.vlpi_lock);
dev->device_id = dev_id;
INIT_LIST_HEAD(&dev->entry);
--
2.11.0
Add the skeleton irq_set_vcpu_affinity method that will be used
to configure VLPIs.
Reviewed-by: Eric Auger <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3-its.c | 24 ++++++++++++++++++++++++
1 file changed, 24 insertions(+)
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index 25e53958a618..c02e1e6cd169 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -36,6 +36,7 @@
#include <linux/irqchip.h>
#include <linux/irqchip/arm-gic-v3.h>
+#include <linux/irqchip/arm-gic-v4.h>
#include <asm/cputype.h>
#include <asm/exception.h>
@@ -779,6 +780,28 @@ static int its_irq_set_irqchip_state(struct irq_data *d,
return 0;
}
+static int its_irq_set_vcpu_affinity(struct irq_data *d, void *vcpu_info)
+{
+ struct its_device *its_dev = irq_data_get_irq_chip_data(d);
+ struct its_cmd_info *info = vcpu_info;
+
+ /* Need a v4 ITS */
+ if (!its_dev->its->is_v4 || !info)
+ return -EINVAL;
+
+ switch (info->cmd_type) {
+ case MAP_VLPI:
+
+ case GET_VLPI:
+
+ case PROP_UPDATE_VLPI:
+ case PROP_UPDATE_AND_INV_VLPI:
+
+ default:
+ return -EINVAL;
+ }
+}
+
static struct irq_chip its_irq_chip = {
.name = "ITS",
.irq_mask = its_mask_irq,
@@ -787,6 +810,7 @@ static struct irq_chip its_irq_chip = {
.irq_set_affinity = its_set_affinity,
.irq_compose_msi_msg = its_irq_compose_msi_msg,
.irq_set_irqchip_state = its_irq_set_irqchip_state,
+ .irq_set_vcpu_affinity = its_irq_set_vcpu_affinity,
};
/*
--
2.11.0
Add the new GICv4 ITS command definitions, most of them, being
defined in terms of their physical counterparts.
Reviewed-by: Eric Auger <[email protected]>
Reviewed-by: Thomas Gleixner <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
include/linux/irqchip/arm-gic-v3.h | 12 ++++++++++++
1 file changed, 12 insertions(+)
diff --git a/include/linux/irqchip/arm-gic-v3.h b/include/linux/irqchip/arm-gic-v3.h
index af8c55105fc2..7c6fd8f3e36c 100644
--- a/include/linux/irqchip/arm-gic-v3.h
+++ b/include/linux/irqchip/arm-gic-v3.h
@@ -348,6 +348,18 @@
#define GITS_CMD_SYNC 0x05
/*
+ * GICv4 ITS specific commands
+ */
+#define GITS_CMD_GICv4(x) ((x) | 0x20)
+#define GITS_CMD_VINVALL GITS_CMD_GICv4(GITS_CMD_INVALL)
+#define GITS_CMD_VMAPP GITS_CMD_GICv4(GITS_CMD_MAPC)
+#define GITS_CMD_VMAPTI GITS_CMD_GICv4(GITS_CMD_MAPTI)
+#define GITS_CMD_VMOVI GITS_CMD_GICv4(GITS_CMD_MOVI)
+#define GITS_CMD_VSYNC GITS_CMD_GICv4(GITS_CMD_SYNC)
+/* VMOVP is the odd one, as it doesn't have a physical counterpart */
+#define GITS_CMD_VMOVP GITS_CMD_GICv4(2)
+
+/*
* ITS error numbers
*/
#define E_ITS_MOVI_UNMAPPED_INTERRUPT 0x010107
--
2.11.0
We're are going to need to change a bit more than just the enable
bit in the LPI property table in the future. So let's change the
LPI configuration funtion to take a set of bits to be cleared,
and a set of bits to be set.
This way, we'll be able to use it when a guest updates an LPI
property (priority, for example).
Reviewed-by: Eric Auger <[email protected]>
Reviewed-by: Thomas Gleixner <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3-its.c | 21 +++++++++++----------
1 file changed, 11 insertions(+), 10 deletions(-)
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index 40682f2a5e1f..25e53958a618 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -680,17 +680,18 @@ static inline u32 its_get_event_id(struct irq_data *d)
return d->hwirq - its_dev->event_map.lpi_base;
}
-static void lpi_set_config(struct irq_data *d, bool enable)
+static void lpi_update_config(struct irq_data *d, u8 clr, u8 set)
{
struct its_device *its_dev = irq_data_get_irq_chip_data(d);
irq_hw_number_t hwirq = d->hwirq;
- u32 id = its_get_event_id(d);
- u8 *cfg = page_address(gic_rdists->prop_page) + hwirq - 8192;
+ struct page *prop_page;
+ u8 *cfg;
- if (enable)
- *cfg |= LPI_PROP_ENABLED;
- else
- *cfg &= ~LPI_PROP_ENABLED;
+ prop_page = gic_rdists->prop_page;
+
+ cfg = page_address(prop_page) + hwirq - 8192;
+ *cfg &= ~clr;
+ *cfg |= set;
/*
* Make the above write visible to the redistributors.
@@ -701,17 +702,17 @@ static void lpi_set_config(struct irq_data *d, bool enable)
gic_flush_dcache_to_poc(cfg, sizeof(*cfg));
else
dsb(ishst);
- its_send_inv(its_dev, id);
+ its_send_inv(its_dev, its_get_event_id(d));
}
static void its_mask_irq(struct irq_data *d)
{
- lpi_set_config(d, false);
+ lpi_update_config(d, LPI_PROP_ENABLED, 0);
}
static void its_unmask_irq(struct irq_data *d)
{
- lpi_set_config(d, true);
+ lpi_update_config(d, 0, LPI_PROP_ENABLED);
}
static int its_set_affinity(struct irq_data *d, const struct cpumask *mask_val,
--
2.11.0
The VCPU tables can be quite sparse as well, and it makes sense
to use indirect tables as well if possible.
Reviewed-by: Thomas Gleixner <[email protected]>
Reviewed-by: Eric Auger <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3-its.c | 25 +++++++++++++++++--------
1 file changed, 17 insertions(+), 8 deletions(-)
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index 2feecdf40216..27f57c528026 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -1068,10 +1068,13 @@ static int its_setup_baser(struct its_node *its, struct its_baser *baser,
return 0;
}
-static bool its_parse_baser_device(struct its_node *its, struct its_baser *baser,
- u32 psz, u32 *order)
+static bool its_parse_indirect_baser(struct its_node *its,
+ struct its_baser *baser,
+ u32 psz, u32 *order)
{
- u64 esz = GITS_BASER_ENTRY_SIZE(its_read_baser(its, baser));
+ u64 tmp = its_read_baser(its, baser);
+ u64 type = GITS_BASER_TYPE(tmp);
+ u64 esz = GITS_BASER_ENTRY_SIZE(tmp);
u64 val = GITS_BASER_InnerShareable | GITS_BASER_RaWaWb;
u32 ids = its->device_ids;
u32 new_order = *order;
@@ -1110,8 +1113,9 @@ static bool its_parse_baser_device(struct its_node *its, struct its_baser *baser
if (new_order >= MAX_ORDER) {
new_order = MAX_ORDER - 1;
ids = ilog2(PAGE_ORDER_TO_SIZE(new_order) / (int)esz);
- pr_warn("ITS@%pa: Device Table too large, reduce ids %u->%u\n",
- &its->phys_base, its->device_ids, ids);
+ pr_warn("ITS@%pa: %s Table too large, reduce ids %u->%u\n",
+ &its->phys_base, its_base_type_string[type],
+ its->device_ids, ids);
}
*order = new_order;
@@ -1159,11 +1163,16 @@ static int its_alloc_tables(struct its_node *its)
u32 order = get_order(psz);
bool indirect = false;
- if (type == GITS_BASER_TYPE_NONE)
+ switch (type) {
+ case GITS_BASER_TYPE_NONE:
continue;
- if (type == GITS_BASER_TYPE_DEVICE)
- indirect = its_parse_baser_device(its, baser, psz, &order);
+ case GITS_BASER_TYPE_DEVICE:
+ case GITS_BASER_TYPE_VCPU:
+ indirect = its_parse_indirect_baser(its, baser,
+ psz, &order);
+ break;
+ }
err = its_setup_baser(its, baser, cache, shr, psz, order, indirect);
if (err < 0) {
--
2.11.0
Add a bunch of GICv4-specific data structures that will get used in
subsequent patches.
Reviewed-by: Thomas Gleixner <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
include/linux/irqchip/arm-gic-v4.h | 91 ++++++++++++++++++++++++++++++++++++++
1 file changed, 91 insertions(+)
create mode 100644 include/linux/irqchip/arm-gic-v4.h
diff --git a/include/linux/irqchip/arm-gic-v4.h b/include/linux/irqchip/arm-gic-v4.h
new file mode 100644
index 000000000000..4cac48e220da
--- /dev/null
+++ b/include/linux/irqchip/arm-gic-v4.h
@@ -0,0 +1,91 @@
+/*
+ * Copyright (C) 2016 ARM Limited, All Rights Reserved.
+ * Author: Marc Zyngier <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef __LINUX_IRQCHIP_ARM_GIC_V4_H
+#define __LINUX_IRQCHIP_ARM_GIC_V4_H
+
+struct its_vpe;
+
+/* Embedded in kvm.arch */
+struct its_vm {
+ struct irq_domain *domain;
+ struct page *vprop_page;
+ struct its_vpe **vpes;
+ int nr_vpes;
+ irq_hw_number_t db_lpi_base;
+ unsigned long *db_bitmap;
+ int nr_db_lpis;
+};
+
+/* Embedded in kvm_vcpu.arch */
+struct its_vpe {
+ struct page *vpt_page;
+ struct its_vm *its_vm;
+ /* Doorbell interrupt */
+ int irq;
+ irq_hw_number_t vpe_db_lpi;
+ /*
+ * This collection ID is used to indirect the target
+ * redistributor for this VPE. The ID itself isn't involved in
+ * programming of the ITS.
+ */
+ u16 col_idx;
+ /* Unique (system-wide) VPE identifier */
+ u16 vpe_id;
+ /* Implementation Defined Area Invalid */
+ bool idai;
+ /* Pending VLPIs on schedule out? */
+ bool pending_last;
+};
+
+/*
+ * struct its_vlpi_map: structure describing the mapping of a
+ * VLPI. Only to be interpreted in the context of a physical interrupt
+ * it complements. To be used as the vcpu_info passed to
+ * irq_set_vcpu_affinity().
+ *
+ * @vm: Pointer to the GICv4 notion of a VM
+ * @vintid: Virtual LPI number
+ * @db_enabled: Is the VPE doorbell to be generated?
+ * @vpe_idx: Index (0-based) of the VPE in this VM. Not the vpe_id!
+ */
+struct its_vlpi_map {
+ struct its_vm *vm;
+ u32 vintid;
+ bool db_enabled;
+ u16 vpe_idx;
+};
+
+enum its_vcpu_info_cmd_type {
+ MAP_VLPI,
+ GET_VLPI,
+ PROP_UPDATE_VLPI,
+ PROP_UPDATE_AND_INV_VLPI,
+ SCHEDULE_VPE,
+ DESCHEDULE_VPE,
+ INVALL_VPE,
+};
+
+struct its_cmd_info {
+ enum its_vcpu_info_cmd_type cmd_type;
+ union {
+ struct its_vlpi_map *map;
+ u8 config;
+ };
+};
+
+#endif
--
2.11.0
As we want to use 2-level tables for VCPUs, let's hack the device
table allocator in order to make it slightly more generic. It
will get reused in subsequent patches.
Reviewed-by: Thomas Gleixner <[email protected]>
Reviewed-by: Eric Auger <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3-its.c | 26 ++++++++++++++++----------
1 file changed, 16 insertions(+), 10 deletions(-)
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index 3011403438c3..40682f2a5e1f 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -1383,26 +1383,19 @@ static struct its_baser *its_get_baser(struct its_node *its, u32 type)
return NULL;
}
-static bool its_alloc_device_table(struct its_node *its, u32 dev_id)
+static bool its_alloc_table_entry(struct its_baser *baser, u32 id)
{
- struct its_baser *baser;
struct page *page;
u32 esz, idx;
__le64 *table;
- baser = its_get_baser(its, GITS_BASER_TYPE_DEVICE);
-
- /* Don't allow device id that exceeds ITS hardware limit */
- if (!baser)
- return (ilog2(dev_id) < its->device_ids);
-
/* Don't allow device id that exceeds single, flat table limit */
esz = GITS_BASER_ENTRY_SIZE(baser->val);
if (!(baser->val & GITS_BASER_INDIRECT))
- return (dev_id < (PAGE_ORDER_TO_SIZE(baser->order) / esz));
+ return (id < (PAGE_ORDER_TO_SIZE(baser->order) / esz));
/* Compute 1st level table index & check if that exceeds table limit */
- idx = dev_id >> ilog2(baser->psz / esz);
+ idx = id >> ilog2(baser->psz / esz);
if (idx >= (PAGE_ORDER_TO_SIZE(baser->order) / GITS_LVL1_ENTRY_SIZE))
return false;
@@ -1431,6 +1424,19 @@ static bool its_alloc_device_table(struct its_node *its, u32 dev_id)
return true;
}
+static bool its_alloc_device_table(struct its_node *its, u32 dev_id)
+{
+ struct its_baser *baser;
+
+ baser = its_get_baser(its, GITS_BASER_TYPE_DEVICE);
+
+ /* Don't allow device id that exceeds ITS hardware limit */
+ if (!baser)
+ return (ilog2(dev_id) < its->device_ids);
+
+ return its_alloc_table_entry(baser, dev_id);
+}
+
static struct its_device *its_create_device(struct its_node *its, u32 dev_id,
int nvecs)
{
--
2.11.0
Rework LPI deallocation so that it can be reused by the v4 support
code.
Reviewed-by: Eric Auger <[email protected]>
Reviewed-by: Thomas Gleixner <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3-its.c | 13 +++++++------
1 file changed, 7 insertions(+), 6 deletions(-)
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index d6f9ccb36f4d..3011403438c3 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -873,16 +873,15 @@ static unsigned long *its_lpi_alloc_chunks(int nr_irqs, int *base, int *nr_ids)
return bitmap;
}
-static void its_lpi_free(struct event_lpi_map *map)
+static void its_lpi_free_chunks(unsigned long *bitmap, int base, int nr_ids)
{
- int base = map->lpi_base;
- int nr_ids = map->nr_lpis;
int lpi;
spin_lock(&lpi_lock);
for (lpi = base; lpi < (base + nr_ids); lpi += IRQS_PER_CHUNK) {
int chunk = its_lpi_to_chunk(lpi);
+
BUG_ON(chunk > lpi_chunks);
if (test_bit(chunk, lpi_bitmap)) {
clear_bit(chunk, lpi_bitmap);
@@ -893,8 +892,7 @@ static void its_lpi_free(struct event_lpi_map *map)
spin_unlock(&lpi_lock);
- kfree(map->lpi_map);
- kfree(map->col_map);
+ kfree(bitmap);
}
static struct page *its_allocate_prop_table(gfp_t gfp_flags)
@@ -1665,7 +1663,10 @@ static void its_irq_domain_free(struct irq_domain *domain, unsigned int virq,
/* If all interrupts have been freed, start mopping the floor */
if (bitmap_empty(its_dev->event_map.lpi_map,
its_dev->event_map.nr_lpis)) {
- its_lpi_free(&its_dev->event_map);
+ its_lpi_free_chunks(its_dev->event_map.lpi_map,
+ its_dev->event_map.lpi_base,
+ its_dev->event_map.nr_lpis);
+ kfree(its_dev->event_map.col_map);
/* Unmap device/itt */
its_send_mapd(its_dev, 0);
--
2.11.0
Just as for the property table, let's move the pending table
allocation to a separate function.
Reviewed-by: Thomas Gleixner <[email protected]>
Reviewed-by: Eric Auger <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3-its.c | 29 ++++++++++++++++++++---------
1 file changed, 20 insertions(+), 9 deletions(-)
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index 27f57c528026..d6f9ccb36f4d 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -1199,6 +1199,24 @@ static int its_alloc_collections(struct its_node *its)
return 0;
}
+static struct page *its_allocate_pending_table(gfp_t gfp_flags)
+{
+ struct page *pend_page;
+ /*
+ * The pending pages have to be at least 64kB aligned,
+ * hence the 'max(LPI_PENDBASE_SZ, SZ_64K)' below.
+ */
+ pend_page = alloc_pages(gfp_flags | __GFP_ZERO,
+ get_order(max(LPI_PENDBASE_SZ, SZ_64K)));
+ if (!pend_page)
+ return NULL;
+
+ /* Make sure the GIC will observe the zero-ed page */
+ gic_flush_dcache_to_poc(page_address(pend_page), LPI_PENDBASE_SZ);
+
+ return pend_page;
+}
+
static void its_cpu_init_lpis(void)
{
void __iomem *rbase = gic_data_rdist_rd_base();
@@ -1209,21 +1227,14 @@ static void its_cpu_init_lpis(void)
pend_page = gic_data_rdist()->pend_page;
if (!pend_page) {
phys_addr_t paddr;
- /*
- * The pending pages have to be at least 64kB aligned,
- * hence the 'max(LPI_PENDBASE_SZ, SZ_64K)' below.
- */
- pend_page = alloc_pages(GFP_NOWAIT | __GFP_ZERO,
- get_order(max(LPI_PENDBASE_SZ, SZ_64K)));
+
+ pend_page = its_allocate_pending_table(GFP_NOWAIT);
if (!pend_page) {
pr_err("Failed to allocate PENDBASE for CPU%d\n",
smp_processor_id());
return;
}
- /* Make sure the GIC will observe the zero-ed page */
- gic_flush_dcache_to_poc(page_address(pend_page), LPI_PENDBASE_SZ);
-
paddr = page_to_phys(pend_page);
pr_info("CPU%d: using LPI pending table @%pa\n",
smp_processor_id(), &paddr);
--
2.11.0
In order to discover the VLPI properties, we need to iterate over
the redistributor regions. As we already have code that does this,
let's factor it out and make it slightly more generic.
Reviewed-by: Thomas Gleixner <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3.c | 69 ++++++++++++++++++++++++++++++--------------
1 file changed, 47 insertions(+), 22 deletions(-)
diff --git a/drivers/irqchip/irq-gic-v3.c b/drivers/irqchip/irq-gic-v3.c
index c132f29322cc..bc58a4339a97 100644
--- a/drivers/irqchip/irq-gic-v3.c
+++ b/drivers/irqchip/irq-gic-v3.c
@@ -421,24 +421,14 @@ static void __init gic_dist_init(void)
gic_write_irouter(affinity, base + GICD_IROUTER + i * 8);
}
-static int gic_populate_rdist(void)
+static int gic_iterate_rdists(int (*fn)(struct redist_region *, void __iomem *))
{
- unsigned long mpidr = cpu_logical_map(smp_processor_id());
- u64 typer;
- u32 aff;
+ int ret = -ENODEV;
int i;
- /*
- * Convert affinity to a 32bit value that can be matched to
- * GICR_TYPER bits [63:32].
- */
- aff = (MPIDR_AFFINITY_LEVEL(mpidr, 3) << 24 |
- MPIDR_AFFINITY_LEVEL(mpidr, 2) << 16 |
- MPIDR_AFFINITY_LEVEL(mpidr, 1) << 8 |
- MPIDR_AFFINITY_LEVEL(mpidr, 0));
-
for (i = 0; i < gic_data.nr_redist_regions; i++) {
void __iomem *ptr = gic_data.redist_regions[i].redist_base;
+ u64 typer;
u32 reg;
reg = readl_relaxed(ptr + GICR_PIDR2) & GIC_PIDR2_ARCH_MASK;
@@ -450,15 +440,9 @@ static int gic_populate_rdist(void)
do {
typer = gic_read_typer(ptr + GICR_TYPER);
- if ((typer >> 32) == aff) {
- u64 offset = ptr - gic_data.redist_regions[i].redist_base;
- gic_data_rdist_rd_base() = ptr;
- gic_data_rdist()->phys_base = gic_data.redist_regions[i].phys_base + offset;
- pr_info("CPU%d: found redistributor %lx region %d:%pa\n",
- smp_processor_id(), mpidr, i,
- &gic_data_rdist()->phys_base);
+ ret = fn(gic_data.redist_regions + i, ptr);
+ if (!ret)
return 0;
- }
if (gic_data.redist_regions[i].single_redist)
break;
@@ -473,9 +457,50 @@ static int gic_populate_rdist(void)
} while (!(typer & GICR_TYPER_LAST));
}
+ return ret ? -ENODEV : 0;
+}
+
+static int __gic_populate_rdist(struct redist_region *region, void __iomem *ptr)
+{
+ unsigned long mpidr = cpu_logical_map(smp_processor_id());
+ u64 typer;
+ u32 aff;
+
+ /*
+ * Convert affinity to a 32bit value that can be matched to
+ * GICR_TYPER bits [63:32].
+ */
+ aff = (MPIDR_AFFINITY_LEVEL(mpidr, 3) << 24 |
+ MPIDR_AFFINITY_LEVEL(mpidr, 2) << 16 |
+ MPIDR_AFFINITY_LEVEL(mpidr, 1) << 8 |
+ MPIDR_AFFINITY_LEVEL(mpidr, 0));
+
+ typer = gic_read_typer(ptr + GICR_TYPER);
+ if ((typer >> 32) == aff) {
+ u64 offset = ptr - region->redist_base;
+ gic_data_rdist_rd_base() = ptr;
+ gic_data_rdist()->phys_base = region->phys_base + offset;
+
+ pr_info("CPU%d: found redistributor %lx region %d:%pa\n",
+ smp_processor_id(), mpidr,
+ (int)(region - gic_data.redist_regions),
+ &gic_data_rdist()->phys_base);
+ return 0;
+ }
+
+ /* Try next one */
+ return 1;
+}
+
+static int gic_populate_rdist(void)
+{
+ if (gic_iterate_rdists(__gic_populate_rdist) == 0)
+ return 0;
+
/* We couldn't even deal with ourselves... */
WARN(true, "CPU%d: mpidr %lx has no re-distributor!\n",
- smp_processor_id(), mpidr);
+ smp_processor_id(),
+ (unsigned long)cpu_logical_map(smp_processor_id()));
return -ENODEV;
}
--
2.11.0
The various LPI definitions are in the middle of the code, and
would be better placed at the beginning, given that we're going
to use some of them much earlier.
Reviewed-by: Thomas Gleixner <[email protected]>
Reviewed-by: Eric Auger <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3-its.c | 27 +++++++++++++++------------
1 file changed, 15 insertions(+), 12 deletions(-)
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index 45ea193325d2..335bd21c6f41 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -49,6 +49,21 @@
#define RDIST_FLAGS_PROPBASE_NEEDS_FLUSHING (1 << 0)
/*
+ * We allocate 64kB for PROPBASE. That gives us at most 64K LPIs to
+ * deal with (one configuration byte per interrupt). PENDBASE has to
+ * be 64kB aligned (one bit per LPI, plus 8192 bits for SPI/PPI/SGI).
+ */
+#define LPI_PROPBASE_SZ SZ_64K
+#define LPI_PENDBASE_SZ (LPI_PROPBASE_SZ / 8 + SZ_1K)
+
+/*
+ * This is how many bits of ID we need, including the useless ones.
+ */
+#define LPI_NRBITS ilog2(LPI_PROPBASE_SZ + SZ_8K)
+
+#define LPI_PROP_DEFAULT_PRIO 0xa0
+
+/*
* Collection structure - just an ID, and a redistributor address to
* ping. We use one per CPU as a bag of interrupts assigned to this
* CPU.
@@ -785,20 +800,8 @@ static void its_lpi_free(struct event_lpi_map *map)
kfree(map->col_map);
}
-/*
- * We allocate 64kB for PROPBASE. That gives us at most 64K LPIs to
- * deal with (one configuration byte per interrupt). PENDBASE has to
- * be 64kB aligned (one bit per LPI, plus 8192 bits for SPI/PPI/SGI).
- */
-#define LPI_PROPBASE_SZ SZ_64K
-#define LPI_PENDBASE_SZ (LPI_PROPBASE_SZ / 8 + SZ_1K)
-/*
- * This is how many bits of ID we need, including the useless ones.
- */
-#define LPI_NRBITS ilog2(LPI_PROPBASE_SZ + SZ_8K)
-#define LPI_PROP_DEFAULT_PRIO 0xa0
static int __init its_alloc_lpi_tables(void)
{
--
2.11.0
Add the probing code for the ITS VLPI support. This includes
configuring the ITS number if not supporting the single VMOVP
command feature.
Reviewed-by: Eric Auger <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3-its.c | 68 +++++++++++++++++++++++++++++++++++---
include/linux/irqchip/arm-gic-v3.h | 5 +++
2 files changed, 69 insertions(+), 4 deletions(-)
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index 335bd21c6f41..006a5259609d 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -103,6 +103,7 @@ struct its_node {
u32 ite_size;
u32 device_ids;
int numa_node;
+ bool is_v4;
};
#define ITS_ITT_ALIGN SZ_256
@@ -135,6 +136,14 @@ static DEFINE_SPINLOCK(its_lock);
static struct rdists *gic_rdists;
static struct irq_domain *its_parent;
+/*
+ * We have a maximum number of 16 ITSs in the whole system if we're
+ * using the ITSList mechanism
+ */
+#define ITS_LIST_MAX 16
+
+static unsigned long its_list_map;
+
#define gic_data_rdist() (raw_cpu_ptr(gic_rdists->rdist))
#define gic_data_rdist_rd_base() (gic_data_rdist()->rd_base)
@@ -1673,13 +1682,51 @@ static int its_init_domain(struct fwnode_handle *handle, struct its_node *its)
return 0;
}
+static int __init its_compute_its_list_map(struct resource *res,
+ void __iomem *its_base)
+{
+ int its_number;
+ u32 ctlr;
+
+ /*
+ * This is assumed to be done early enough that we're
+ * guaranteed to be single-threaded, hence no
+ * locking. Should this change, we should address
+ * this.
+ */
+ its_number = find_first_zero_bit(&its_list_map, ITS_LIST_MAX);
+ if (its_number >= ITS_LIST_MAX) {
+ pr_err("ITS@%pa: No ITSList entry available!\n",
+ &res->start);
+ return -EINVAL;
+ }
+
+ ctlr = readl_relaxed(its_base + GITS_CTLR);
+ ctlr &= ~GITS_CTLR_ITS_NUMBER;
+ ctlr |= its_number << GITS_CTLR_ITS_NUMBER_SHIFT;
+ writel_relaxed(ctlr, its_base + GITS_CTLR);
+ ctlr = readl_relaxed(its_base + GITS_CTLR);
+ if ((ctlr & GITS_CTLR_ITS_NUMBER) != (its_number << GITS_CTLR_ITS_NUMBER_SHIFT)) {
+ its_number = ctlr & GITS_CTLR_ITS_NUMBER;
+ its_number >>= GITS_CTLR_ITS_NUMBER_SHIFT;
+ }
+
+ if (test_and_set_bit(its_number, &its_list_map)) {
+ pr_err("ITS@%pa: Duplicate ITSList entry %d\n",
+ &res->start, its_number);
+ return -EINVAL;
+ }
+
+ return its_number;
+}
+
static int __init its_probe_one(struct resource *res,
struct fwnode_handle *handle, int numa_node)
{
struct its_node *its;
void __iomem *its_base;
- u32 val;
- u64 baser, tmp;
+ u32 val, ctlr;
+ u64 baser, tmp, typer;
int err;
its_base = ioremap(res->start, resource_size(res));
@@ -1712,9 +1759,21 @@ static int __init its_probe_one(struct resource *res,
raw_spin_lock_init(&its->lock);
INIT_LIST_HEAD(&its->entry);
INIT_LIST_HEAD(&its->its_device_list);
+ typer = gic_read_typer(its_base + GITS_TYPER);
its->base = its_base;
its->phys_base = res->start;
- its->ite_size = ((gic_read_typer(its_base + GITS_TYPER) >> 4) & 0xf) + 1;
+ its->ite_size = GITS_TYPER_ITT_ENTRY_SIZE(typer);
+ its->is_v4 = !!(typer & GITS_TYPER_VLPIS);
+ if (its->is_v4 && !(typer & GITS_TYPER_VMOVP)) {
+ err = its_compute_its_list_map(res, its_base);
+ if (err < 0)
+ goto out_free_its;
+
+ pr_info("ITS@%pa: Using ITS number %d\n", &res->start, err);
+ } else {
+ pr_info("ITS@%pa: Single VMOVP capable\n", &res->start);
+ }
+
its->numa_node = numa_node;
its->cmd_base = (void *)__get_free_pages(GFP_KERNEL | __GFP_ZERO,
@@ -1761,7 +1820,8 @@ static int __init its_probe_one(struct resource *res,
}
gits_write_cwriter(0, its->base + GITS_CWRITER);
- writel_relaxed(GITS_CTLR_ENABLE, its->base + GITS_CTLR);
+ ctlr = readl_relaxed(its->base + GITS_CTLR);
+ writel_relaxed(ctlr | GITS_CTLR_ENABLE, its->base + GITS_CTLR);
err = its_init_domain(handle, its);
if (err)
diff --git a/include/linux/irqchip/arm-gic-v3.h b/include/linux/irqchip/arm-gic-v3.h
index 20a553423ac7..af8c55105fc2 100644
--- a/include/linux/irqchip/arm-gic-v3.h
+++ b/include/linux/irqchip/arm-gic-v3.h
@@ -235,15 +235,20 @@
#define GITS_TRANSLATER 0x10040
#define GITS_CTLR_ENABLE (1U << 0)
+#define GITS_CTLR_ITS_NUMBER_SHIFT 4
+#define GITS_CTLR_ITS_NUMBER (0xFU << GITS_CTLR_ITS_NUMBER_SHIFT)
#define GITS_CTLR_QUIESCENT (1U << 31)
#define GITS_TYPER_PLPIS (1UL << 0)
+#define GITS_TYPER_VLPIS (1UL << 1)
#define GITS_TYPER_ITT_ENTRY_SIZE_SHIFT 4
+#define GITS_TYPER_ITT_ENTRY_SIZE(r) ((((r) >> GITS_TYPER_ITT_ENTRY_SIZE_SHIFT) & 0x1f) + 1)
#define GITS_TYPER_IDBITS_SHIFT 8
#define GITS_TYPER_DEVBITS_SHIFT 13
#define GITS_TYPER_DEVBITS(r) ((((r) >> GITS_TYPER_DEVBITS_SHIFT) & 0x1f) + 1)
#define GITS_TYPER_PTA (1UL << 19)
#define GITS_TYPER_HWCOLLCNT_SHIFT 24
+#define GITS_TYPER_VMOVP (1ULL << 37)
#define GITS_IIDR_REV_SHIFT 12
#define GITS_IIDR_REV_MASK (0xf << GITS_IIDR_REV_SHIFT)
--
2.11.0
Most ITS commands do operate on a collection object, and require
a SYNC command to be performed on that collection in order to
guarantee the execution of the first command.
With GICv4 ITS, another set of commands perform similar operations
on a VPE object, and a VSYNC operations must be executed to guarantee
their execution.
Given the similarities (post a command, perform a synchronization
operation on a sync object), it makes sense to reuse the same
mechanism for both class of commands.
Let's start with turning its_send_single_command into a huge macro
that performs the bulk of the work, and a set of helpers that
make this macro usable for the GICv3 ITS commands.
Signed-off-by: Marc Zyngier <[email protected]>
---
drivers/irqchip/irq-gic-v3-its.c | 84 ++++++++++++++++++++++------------------
1 file changed, 47 insertions(+), 37 deletions(-)
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index 006a5259609d..bbf7bb89aa62 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -490,43 +490,53 @@ static void its_wait_for_range_completion(struct its_node *its,
}
}
-static void its_send_single_command(struct its_node *its,
- its_cmd_builder_t builder,
- struct its_cmd_desc *desc)
-{
- struct its_cmd_block *cmd, *sync_cmd, *next_cmd;
- struct its_collection *sync_col;
- unsigned long flags;
-
- raw_spin_lock_irqsave(&its->lock, flags);
-
- cmd = its_allocate_entry(its);
- if (!cmd) { /* We're soooooo screewed... */
- pr_err_ratelimited("ITS can't allocate, dropping command\n");
- raw_spin_unlock_irqrestore(&its->lock, flags);
- return;
- }
- sync_col = builder(cmd, desc);
- its_flush_cmd(its, cmd);
-
- if (sync_col) {
- sync_cmd = its_allocate_entry(its);
- if (!sync_cmd) {
- pr_err_ratelimited("ITS can't SYNC, skipping\n");
- goto post;
- }
- its_encode_cmd(sync_cmd, GITS_CMD_SYNC);
- its_encode_target(sync_cmd, sync_col->target_address);
- its_fixup_cmd(sync_cmd);
- its_flush_cmd(its, sync_cmd);
- }
-
-post:
- next_cmd = its_post_commands(its);
- raw_spin_unlock_irqrestore(&its->lock, flags);
-
- its_wait_for_range_completion(its, cmd, next_cmd);
-}
+/* Warning, macro hell follows */
+#define BUILD_SINGLE_CMD_FUNC(name, buildtype, synctype, buildfn) \
+void name(struct its_node *its, \
+ buildtype builder, \
+ struct its_cmd_desc *desc) \
+{ \
+ struct its_cmd_block *cmd, *sync_cmd, *next_cmd; \
+ synctype *sync_obj; \
+ unsigned long flags; \
+ \
+ raw_spin_lock_irqsave(&its->lock, flags); \
+ \
+ cmd = its_allocate_entry(its); \
+ if (!cmd) { /* We're soooooo screewed... */ \
+ raw_spin_unlock_irqrestore(&its->lock, flags); \
+ return; \
+ } \
+ sync_obj = builder(cmd, desc); \
+ its_flush_cmd(its, cmd); \
+ \
+ if (sync_obj) { \
+ sync_cmd = its_allocate_entry(its); \
+ if (!sync_cmd) \
+ goto post; \
+ \
+ buildfn(sync_cmd, sync_obj); \
+ its_flush_cmd(its, sync_cmd); \
+ } \
+ \
+post: \
+ next_cmd = its_post_commands(its); \
+ raw_spin_unlock_irqrestore(&its->lock, flags); \
+ \
+ its_wait_for_range_completion(its, cmd, next_cmd); \
+}
+
+static void its_build_sync_cmd(struct its_cmd_block *sync_cmd,
+ struct its_collection *sync_col)
+{
+ its_encode_cmd(sync_cmd, GITS_CMD_SYNC);
+ its_encode_target(sync_cmd, sync_col->target_address);
+
+ its_fixup_cmd(sync_cmd);
+}
+
+static BUILD_SINGLE_CMD_FUNC(its_send_single_command, its_cmd_builder_t,
+ struct its_collection, its_build_sync_cmd)
static void its_send_inv(struct its_device *dev, u32 event_id)
{
--
2.11.0
Hi Marc,
I've verified the basic GICv4 functionality with v2 series + Eric's IRQ
bypass patches on QDF2400 platform with a minor change in vgic-init.c
successfully. Nice, I don't see any deadlock or catastrophic issues
running on QCOM hardware. You can add my tested-by, I'll provide comments
after reviewing giant v2 series.
Tested-by: Shanker Donthineni <[email protected]>
-----Original Message-----
From: linux-arm-kernel [mailto:[email protected]]
On Behalf Of Marc Zyngier
Sent: Wednesday, June 28, 2017 10:03 AM
To: [email protected]; [email protected];
[email protected]
Cc: Mark Rutland <[email protected]>; Jason Cooper
<[email protected]>; Eric Auger <[email protected]>; Christoffer Dall
<[email protected]>; Thomas Gleixner <[email protected]>; Shanker
Donthineni <[email protected]>
Subject: [PATCH v2 00/52] irqchip: KVM: Add support for GICv4
Yes, it's been a long time coming, but I really wasn't looking forward to
picking this up again. Anyway...
This (monster of a) series implements full support for GICv4, bringing
direct injection of MSIs to KVM on arm and arm64, assuming you have the
right hardware (which is quite unlikely).
To get an idea of the design, I'd recommend you start with patch #32, which
tries to shed some light on the approach that I've taken. And before that,
please digest some of the GICv3/GICv4 architecture documentation[1] (less
than 800 pages!). Once you feel reasonably insane, you'll be in the right
mood to read the code.
The structure of the series is fairly simple. The initial 34 patches add
some generic support for GICv4, while the rest of the code plugs KVM into
it. This series relies on Eric Auger's irq-bypass series[2], which is a
prerequisite for this work.
The stack has been *very lightly* tested on an arm64 model, with a PCI
virtio block device passed from the host to a guest (using kvmtool and
Jean-Philippe Brucker's excellent VFIO support patches[3]). As it has never
seen any HW, I expect things to be subtly broken, so go forward and test if
you can, though I'm mostly interested in people reviewing the code at the
moment.
I've pushed out a branch based on 4.12-rc6 containing the dependencies (as
well as a couple of debug patches):
git://git.kernel.org/pub/scm/linux/kernel/git/maz/arm-platforms.git
kvm-arm64/gicv4-kvm
* From v1:
- The bulk of the 30-something initial patches have seen countless
bugs being fixed, and some key data structures have been subtly
tweaked (or killed altogether). They are still quite similar to
what I had in v1 though.
- The whole KVM code is brand new and as I said above, only lightly
tested.
- Collected a bunch a R-bs from Thomas and Eric (many thanks, guys).
[1]
https://static.docs.arm.com/ihi0069/c/IHI0069C_gic_architecture_specificatio
n.pdf
[2] http://www.spinics.net/lists/kvm/msg151463.html
[3] http://www.spinics.net/lists/kvm/msg151823.html
Marc Zyngier (52):
genirq: Let irq_set_vcpu_affinity() iterate over hierarchy
irqchip/gic-v3: Add redistributor iterator
irqchip/gic-v3: Add VLPI/DirectLPI discovery
irqchip/gic-v3-its: Move LPI definitions around
irqchip/gic-v3-its: Add probing for VLPI properties
irqchip/gic-v3-its: Macro-ize its_send_single_command
irqchip/gic-v3-its: Implement irq_set_irqchip_state for pending state
irqchip/gic-v3-its: Split out property table allocation
irqchip/gic-v3-its: Allow use of indirect VCPU tables
irqchip/gic-v3-its: Split out pending table allocation
irqchip/gic-v3-its: Rework LPI freeing
irqchip/gic-v3-its: Generalize device table allocation
irqchip/gic-v3-its: Generalize LPI configuration
irqchip/gic-v4: Add management structure definitions
irqchip/gic-v3-its: Add GICv4 ITS command definitions
irqchip/gic-v3-its: Add VLPI configuration hook
irqchip/gic-v3-its: Add VLPI map/unmap operations
irqchip/gic-v3-its: Add VLPI configuration handling
irqchip/gic-v3-its: Add VPE domain infrastructure
irqchip/gic-v3-its: Add VPE irq domain allocation/teardown
irqchip/gic-v3-its: Add VPE irq domain [de]activation
irqchip/gic-v3-its: Add VPENDBASER/VPROPBASER accessors
irqchip/gic-v3-its: Add VPE scheduling
irqchip/gic-v3-its: Add VPE invalidation hook
irqchip/gic-v3-its: Add VPE affinity changes
irqchip/gic-v3-its: Add VPE interrupt masking
irqchip/gic-v3-its: Support VPE doorbell invalidation even when
!DirectLPI
irqchip/gic-v3-its: Set implementation defined bit to enable VLPIs
irqchip/gic-v4: Add per-VM VPE domain creation
irqchip/gic-v4: Add VPE command interface
irqchip/gic-v4: Add VLPI configuration interface
irqchip/gic-v4: Add some basic documentation
irqchip/gic-v4: Enable low-level GICv4 operations
irqchip/gic-v3: Advertise GICv4 support to KVM
KVM: arm/arm64: vgic: Move kvm_vgic_destroy call around
KVM: arm/arm64: vITS: Add MSI translation helpers
KVM: arm/arm64: GICv4: Add init and teardown of the vPE irq domain
KVM: arm/arm64: GICv4: Wire init/teardown of per-VM support
KVM: arm/arm64: GICv4: Wire mapping/unmapping of VLPIs in VFIO irq
bypass
KVM: arm/arm64: GICv4: Handle INT command applied to a VLPI
KVM: arm/arm64: GICv4: Unmap VLPI when freeing an LPI
KVM: arm/arm64: GICv4: Handle MOVI applied to a VLPI
KVM: arm/arm64: GICv4: Handle CLEAR applied to a VLPI
KVM: arm/arm64: GICv4: Handle MOVALL applied to a vPE
KVM: arm/arm64: GICv4: Propagate property updates to VLPIs
KVM: arm/arm64: GICv4: Handle INVALL applied to a vPE
KVM: arm/arm64: GICv4: Propagate VLPI properties at map time
KVM: arm/arm64: GICv4: Add doorbell interrupt handling
KVM: arm/arm64: GICv4: Hook vPE scheduling into vgic flush/sync
KVM: arm/arm64: GICv4: Enable virtual cpuif if VLPIs can be delivered
KVM: arm/arm64: GICv4: Use pending_last as a scheduling hint
KVM: arm/arm64: GICv4: Enable VLPI support
arch/arm/include/asm/arch_gicv3.h | 33 +
arch/arm/kvm/Makefile | 1 +
arch/arm64/include/asm/arch_gicv3.h | 6 +
arch/arm64/kvm/Makefile | 1 +
drivers/irqchip/Makefile | 2 +-
drivers/irqchip/irq-gic-v3-its.c | 1288
+++++++++++++++++++++++++++++---
drivers/irqchip/irq-gic-v3.c | 93 ++-
drivers/irqchip/irq-gic-v4.c | 209 ++++++
include/kvm/arm_vgic.h | 19 +
include/linux/irqchip/arm-gic-common.h | 2 +
include/linux/irqchip/arm-gic-v3.h | 84 +++
include/linux/irqchip/arm-gic-v4.h | 102 +++
kernel/irq/manage.c | 14 +-
virt/kvm/arm/arm.c | 31 +-
virt/kvm/arm/hyp/vgic-v3-sr.c | 9 +-
virt/kvm/arm/vgic/vgic-init.c | 11 +-
virt/kvm/arm/vgic/vgic-its.c | 165 ++--
virt/kvm/arm/vgic/vgic-v3.c | 6 +
virt/kvm/arm/vgic/vgic-v4.c | 210 ++++++
virt/kvm/arm/vgic/vgic.c | 7 +
virt/kvm/arm/vgic/vgic.h | 10 +
21 files changed, 2103 insertions(+), 200 deletions(-) create mode 100644
drivers/irqchip/irq-gic-v4.c create mode 100644
include/linux/irqchip/arm-gic-v4.h
create mode 100644 virt/kvm/arm/vgic/vgic-v4.c
--
2.11.0
_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
On Wed, 28 Jun 2017, Marc Zyngier wrote:
> When assigning an interrupt to a vcpu, it is not unlikely that
> the level of the hierarchy implementing irq_set_vcpu_affinity
> is not the top level (think a generic MSI domain on top of a
> virtualization aware interrupt controller).
>
> In such a case, let's iterate over the hierarchy until we find
> an irqchip implementing it.
>
> Signed-off-by: Marc Zyngier <[email protected]>
Reviewed-by: Thomas Gleixner <[email protected]>
Hi Marc,
On 06/28/2017 10:03 AM, Marc Zyngier wrote:
> Should the HW support GICv4 and an ITS being associated with this
> VM, let's init the its_vm and its_vpe structures.
>
> Signed-off-by: Marc Zyngier <[email protected]>
> ---
> virt/kvm/arm/vgic/vgic-init.c | 11 ++++++++++-
> 1 file changed, 10 insertions(+), 1 deletion(-)
>
> diff --git a/virt/kvm/arm/vgic/vgic-init.c b/virt/kvm/arm/vgic/vgic-init.c
> index 3a0b8999f011..0de1f0d986d4 100644
> --- a/virt/kvm/arm/vgic/vgic-init.c
> +++ b/virt/kvm/arm/vgic/vgic-init.c
> @@ -285,8 +285,14 @@ int vgic_init(struct kvm *kvm)
> if (ret)
> goto out;
>
> - if (vgic_has_its(kvm))
> + if (vgic_has_its(kvm)) {
> dist->msis_require_devid = true;
> + if (kvm_vgic_global_state.has_gicv4) {
> + ret = vgic_v4_init(kvm);
> + if (ret)
> + goto out;
> + }
This is not quite right, ITS virtual device may not be initialized at the time of
calling vgic-init(). This change breaks the existing KVM functionality with QEMU
hypervisor tool. In later patches, code assumes vgic_v4_init(kvm) was called when
vgic_has_its(kvm) returns a true value.
The right change would be move this logic to inside vgic_its_create() something like this.
--- a/virt/kvm/arm/vgic/vgic-init.c
+++ b/virt/kvm/arm/vgic/vgic-init.c
@@ -285,14 +285,8 @@ int vgic_init(struct kvm *kvm)
if (ret)
goto out;
- if (vgic_has_its(kvm)) {
+ if (vgic_has_its(kvm))
dist->msis_require_devid = true;
- if (kvm_vgic_global_state.has_gicv4) {
- ret = vgic_v4_init(kvm);
- if (ret)
- goto out;
- }
- }
kvm_for_each_vcpu(i, vcpu, kvm)
kvm_vgic_vcpu_enable(vcpu);
--- a/virt/kvm/arm/vgic/vgic-its.c
+++ b/virt/kvm/arm/vgic/vgic-its.c
@@ -1637,6 +1637,7 @@ static int vgic_register_its_iodev(struct kvm *kvm, struct
static int vgic_its_create(struct kvm_device *dev, u32 type)
{
struct vgic_its *its;
+ int ret;
if (type != KVM_DEV_TYPE_ARM_VGIC_ITS)
return -ENODEV;
@@ -1657,6 +1658,12 @@ static int vgic_its_create(struct kvm_device *dev, u32 ty
its->enabled = false;
its->dev = dev;
+ if (kvm_vgic_global_state.has_gicv4) {
+ ret = vgic_v4_init(dev->kvm);
+ if (ret)
+ return -ENOMEM;
+ }
+
> + }
>
> kvm_for_each_vcpu(i, vcpu, kvm)
> kvm_vgic_vcpu_enable(vcpu);
> @@ -323,6 +329,9 @@ static void kvm_vgic_dist_destroy(struct kvm *kvm)
>
> kfree(dist->spis);
> dist->nr_spis = 0;
> +
> + if (kvm_vgic_global_state.has_gicv4 && vgic_has_its(kvm))
> + vgic_v4_teardown(kvm);
> }
>
> void kvm_vgic_vcpu_destroy(struct kvm_vcpu *vcpu)
>
--
Shanker Donthineni
Qualcomm Datacenter Technologies, Inc. as an affiliate of Qualcomm Technologies, Inc.
Qualcomm Technologies, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project.
On 08/07/17 12:26, Shanker Donthineni wrote:
> Hi Marc,
>
> On 06/28/2017 10:03 AM, Marc Zyngier wrote:
>> Should the HW support GICv4 and an ITS being associated with this
>> VM, let's init the its_vm and its_vpe structures.
>>
>> Signed-off-by: Marc Zyngier <[email protected]>
>> ---
>> virt/kvm/arm/vgic/vgic-init.c | 11 ++++++++++-
>> 1 file changed, 10 insertions(+), 1 deletion(-)
>>
>> diff --git a/virt/kvm/arm/vgic/vgic-init.c b/virt/kvm/arm/vgic/vgic-init.c
>> index 3a0b8999f011..0de1f0d986d4 100644
>> --- a/virt/kvm/arm/vgic/vgic-init.c
>> +++ b/virt/kvm/arm/vgic/vgic-init.c
>> @@ -285,8 +285,14 @@ int vgic_init(struct kvm *kvm)
>> if (ret)
>> goto out;
>>
>> - if (vgic_has_its(kvm))
>> + if (vgic_has_its(kvm)) {
>> dist->msis_require_devid = true;
>> + if (kvm_vgic_global_state.has_gicv4) {
>> + ret = vgic_v4_init(kvm);
>> + if (ret)
>> + goto out;
>> + }
>
> This is not quite right, ITS virtual device may not be initialized at the time of
> calling vgic-init(). This change breaks the existing KVM functionality with QEMU
> hypervisor tool. In later patches, code assumes vgic_v4_init(kvm) was called when
> vgic_has_its(kvm) returns a true value.
>
> The right change would be move this logic to inside vgic_its_create() something like this.
>
> --- a/virt/kvm/arm/vgic/vgic-init.c
> +++ b/virt/kvm/arm/vgic/vgic-init.c
> @@ -285,14 +285,8 @@ int vgic_init(struct kvm *kvm)
> if (ret)
> goto out;
>
> - if (vgic_has_its(kvm)) {
> + if (vgic_has_its(kvm))
> dist->msis_require_devid = true;
> - if (kvm_vgic_global_state.has_gicv4) {
> - ret = vgic_v4_init(kvm);
> - if (ret)
> - goto out;
> - }
> - }
>
> kvm_for_each_vcpu(i, vcpu, kvm)
> kvm_vgic_vcpu_enable(vcpu);
>
> --- a/virt/kvm/arm/vgic/vgic-its.c
> +++ b/virt/kvm/arm/vgic/vgic-its.c
> @@ -1637,6 +1637,7 @@ static int vgic_register_its_iodev(struct kvm *kvm, struct
> static int vgic_its_create(struct kvm_device *dev, u32 type)
> {
> struct vgic_its *its;
> + int ret;
>
> if (type != KVM_DEV_TYPE_ARM_VGIC_ITS)
> return -ENODEV;
> @@ -1657,6 +1658,12 @@ static int vgic_its_create(struct kvm_device *dev, u32 ty
> its->enabled = false;
> its->dev = dev;
>
> + if (kvm_vgic_global_state.has_gicv4) {
> + ret = vgic_v4_init(dev->kvm);
> + if (ret)
> + return -ENOMEM;
> + }
> +
This approach is pretty busted if you have more than a single virtual
ITS, as you now leak memory from the VPE array and IRQ domain allocations.
Also, there is no guarantee that we've created all vcpus by the time we
create an ITS, so this approach fails as well ("create vcpu 1, create
ITS, create vcpu 2, init vgic" is a valid sequence).
I guess we need to call vgic_v4_init() on both vgic_init() and
its_create(), and only perform the GICv4 init if the vgic has already
been initialized.
Or we simply restrict the order to be vgic_init() last, and fix
userspace to respect that order. Not enabling GICv4 won't break SW. It
may even be faster.
M.
--
Jazz is not dead. It just smells funny...