File s390x-pci-Honor-DMA-limits-set-by-vfio.patch of Package qemu.20153
From: Matthew Rosato <mjrosato@linux.ibm.com>
Date: Mon, 26 Oct 2020 11:34:35 -0400
Subject: s390x/pci: Honor DMA limits set by vfio
Git-commit: 37fa32de707340f3a93959ad5a1ebc41ba1520ee
References: bsc#1179719
When an s390 guest is using lazy unmapping, it can result in a very
large number of oustanding DMA requests, far beyond the default
limit configured for vfio. Let's track DMA usage similar to vfio
in the host, and trigger the guest to flush their DMA mappings
before vfio runs out.
Signed-off-by: Matthew Rosato <mjrosato@linux.ibm.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
[aw: non-Linux build fixes]
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Liang Yan <lyan@suse.com>
---
hw/s390x/s390-pci-bus.c | 20 +++++++++-----
hw/s390x/s390-pci-bus.h | 10 +++++++
hw/s390x/s390-pci-inst.c | 45 +++++++++++++++++++++++++++-----
hw/s390x/s390-pci-inst.h | 3 +++
hw/s390x/s390-pci-vfio.c | 42 +++++++++++++++++++++++++++++
include/hw/s390x/s390-pci-vfio.h | 12 +++++++++
6 files changed, 119 insertions(+), 13 deletions(-)
diff --git a/hw/s390x/s390-pci-bus.c b/hw/s390x/s390-pci-bus.c
index 2d2f4a7c419c63e0e1fa7d26e399..a9f6f55047273e235d3ba89ba5f0 100644
--- a/hw/s390x/s390-pci-bus.c
+++ b/hw/s390x/s390-pci-bus.c
@@ -15,8 +15,9 @@
#include "qapi/error.h"
#include "qapi/visitor.h"
#include "cpu.h"
-#include "s390-pci-bus.h"
-#include "s390-pci-inst.h"
+#include "hw/s390x/s390-pci-bus.h"
+#include "hw/s390x/s390-pci-inst.h"
+#include "hw/s390x/s390-pci-vfio.h"
#include "hw/pci/pci_bus.h"
#include "hw/qdev-properties.h"
#include "hw/pci/pci_bridge.h"
@@ -771,6 +772,7 @@ static void s390_pcihost_realize(DeviceState *dev, Error **errp)
s->bus_no = 0;
QTAILQ_INIT(&s->pending_sei);
QTAILQ_INIT(&s->zpci_devs);
+ QTAILQ_INIT(&s->zpci_dma_limit);
css_register_io_adapters(CSS_IO_ADAPTER_PCI, true, false,
S390_ADAPTER_SUPPRESSIBLE, &local_err);
@@ -951,17 +953,18 @@ static void s390_pcihost_plug(HotplugHandler *hotplug_dev, DeviceState *dev,
}
}
+ pbdev->pdev = pdev;
+ pbdev->iommu = s390_pci_get_iommu(s, pci_get_bus(pdev), pdev->devfn);
+ pbdev->iommu->pbdev = pbdev;
+ pbdev->state = ZPCI_FS_DISABLED;
+
if (object_dynamic_cast(OBJECT(dev), "vfio-pci")) {
pbdev->fh |= FH_SHM_VFIO;
+ pbdev->iommu->dma_limit = s390_pci_start_dma_count(s, pbdev);
} else {
pbdev->fh |= FH_SHM_EMUL;
}
- pbdev->pdev = pdev;
- pbdev->iommu = s390_pci_get_iommu(s, pci_get_bus(pdev), pdev->devfn);
- pbdev->iommu->pbdev = pbdev;
- pbdev->state = ZPCI_FS_DISABLED;
-
if (s390_pci_msix_init(pbdev)) {
error_setg(errp, "MSI-X support is mandatory "
"in the S390 architecture");
@@ -1014,6 +1017,9 @@ static void s390_pcihost_unplug(HotplugHandler *hotplug_dev, DeviceState *dev,
pbdev->fid = 0;
QTAILQ_REMOVE(&s->zpci_devs, pbdev, link);
g_hash_table_remove(s->zpci_table, &pbdev->idx);
+ if (pbdev->iommu->dma_limit) {
+ s390_pci_end_dma_count(s, pbdev->iommu->dma_limit);
+ }
object_property_set_bool(OBJECT(dev), false, "realized", NULL);
}
}
diff --git a/hw/s390x/s390-pci-bus.h b/hw/s390x/s390-pci-bus.h
index 550f3cc5e92076cdb8a28b932265..c554aa951ace4e293854e716b05a 100644
--- a/hw/s390x/s390-pci-bus.h
+++ b/hw/s390x/s390-pci-bus.h
@@ -266,6 +266,14 @@ typedef struct S390IOTLBEntry {
} S390IOTLBEntry;
typedef struct S390PCIBusDevice S390PCIBusDevice;
+
+typedef struct S390PCIDMACount {
+ int id;
+ int users;
+ uint32_t avail;
+ QTAILQ_ENTRY(S390PCIDMACount) link;
+} S390PCIDMACount;
+
typedef struct S390PCIIOMMU {
Object parent_obj;
S390PCIBusDevice *pbdev;
@@ -277,6 +285,7 @@ typedef struct S390PCIIOMMU {
uint64_t pba;
uint64_t pal;
GHashTable *iotlb;
+ S390PCIDMACount *dma_limit;
} S390PCIIOMMU;
typedef struct S390PCIIOMMUTable {
@@ -352,6 +361,7 @@ typedef struct S390pciState {
GHashTable *zpci_table;
QTAILQ_HEAD(, SeiContainer) pending_sei;
QTAILQ_HEAD(, S390PCIBusDevice) zpci_devs;
+ QTAILQ_HEAD(, S390PCIDMACount) zpci_dma_limit;
} S390pciState;
S390pciState *s390_get_phb(void);
diff --git a/hw/s390x/s390-pci-inst.c b/hw/s390x/s390-pci-inst.c
index 92c7e45df5f5c421dfb043ef7dbb..a683749df713892a068986ca01be 100644
--- a/hw/s390x/s390-pci-inst.c
+++ b/hw/s390x/s390-pci-inst.c
@@ -32,6 +32,20 @@
} \
} while (0)
+static inline void inc_dma_avail(S390PCIIOMMU *iommu)
+{
+ if (iommu->dma_limit) {
+ iommu->dma_limit->avail++;
+ }
+}
+
+static inline void dec_dma_avail(S390PCIIOMMU *iommu)
+{
+ if (iommu->dma_limit) {
+ iommu->dma_limit->avail--;
+ }
+}
+
static void s390_set_status_code(CPUS390XState *env,
uint8_t r, uint64_t status_code)
{
@@ -572,7 +586,8 @@ int pcistg_service_call(S390CPU *cpu, uint8_t r1, uint8_t r2, uintptr_t ra)
return 0;
}
-static void s390_pci_update_iotlb(S390PCIIOMMU *iommu, S390IOTLBEntry *entry)
+static uint32_t s390_pci_update_iotlb(S390PCIIOMMU *iommu,
+ S390IOTLBEntry *entry)
{
S390IOTLBEntry *cache = g_hash_table_lookup(iommu->iotlb, &entry->iova);
IOMMUTLBEntry notify = {
@@ -585,14 +600,15 @@ static void s390_pci_update_iotlb(S390PCIIOMMU *iommu, S390IOTLBEntry *entry)
if (entry->perm == IOMMU_NONE) {
if (!cache) {
- return;
+ goto out;
}
g_hash_table_remove(iommu->iotlb, &entry->iova);
+ inc_dma_avail(iommu);
} else {
if (cache) {
if (cache->perm == entry->perm &&
cache->translated_addr == entry->translated_addr) {
- return;
+ goto out;
}
notify.perm = IOMMU_NONE;
@@ -606,9 +622,13 @@ static void s390_pci_update_iotlb(S390PCIIOMMU *iommu, S390IOTLBEntry *entry)
cache->len = PAGE_SIZE;
cache->perm = entry->perm;
g_hash_table_replace(iommu->iotlb, &cache->iova, cache);
+ dec_dma_avail(iommu);
}
memory_region_notify_iommu(&iommu->iommu_mr, 0, notify);
+
+out:
+ return iommu->dma_limit ? iommu->dma_limit->avail : 1;
}
int rpcit_service_call(S390CPU *cpu, uint8_t r1, uint8_t r2, uintptr_t ra)
@@ -620,6 +640,7 @@ int rpcit_service_call(S390CPU *cpu, uint8_t r1, uint8_t r2, uintptr_t ra)
S390PCIIOMMU *iommu;
S390IOTLBEntry entry;
hwaddr start, end;
+ uint32_t dma_avail;
if (env->psw.mask & PSW_MASK_PSTATE) {
s390_program_interrupt(env, PGM_PRIVILEGED, ra);
@@ -658,6 +679,11 @@ int rpcit_service_call(S390CPU *cpu, uint8_t r1, uint8_t r2, uintptr_t ra)
}
iommu = pbdev->iommu;
+ if (iommu->dma_limit) {
+ dma_avail = iommu->dma_limit->avail;
+ } else {
+ dma_avail = 1;
+ }
if (!iommu->g_iota) {
error = ERR_EVENT_INVALAS;
goto err;
@@ -675,8 +701,9 @@ int rpcit_service_call(S390CPU *cpu, uint8_t r1, uint8_t r2, uintptr_t ra)
}
start += entry.len;
- while (entry.iova < start && entry.iova < end) {
- s390_pci_update_iotlb(iommu, &entry);
+ while (entry.iova < start && entry.iova < end &&
+ (dma_avail > 0 || entry.perm == IOMMU_NONE)) {
+ dma_avail = s390_pci_update_iotlb(iommu, &entry);
entry.iova += PAGE_SIZE;
entry.translated_addr += PAGE_SIZE;
}
@@ -689,7 +716,13 @@ err:
s390_pci_generate_error_event(error, pbdev->fh, pbdev->fid, start, 0);
} else {
pbdev->fmb.counter[ZPCI_FMB_CNT_RPCIT]++;
- setcc(cpu, ZPCI_PCI_LS_OK);
+ if (dma_avail > 0) {
+ setcc(cpu, ZPCI_PCI_LS_OK);
+ } else {
+ /* vfio DMA mappings are exhausted, trigger a RPCIT */
+ setcc(cpu, ZPCI_PCI_LS_ERR);
+ s390_set_status_code(env, r1, ZPCI_RPCIT_ST_INSUFF_RES);
+ }
}
return 0;
}
diff --git a/hw/s390x/s390-pci-inst.h b/hw/s390x/s390-pci-inst.h
index fa3bf8b5aad11e03376774f8fa41..8ee3a3c237576757f99dc1adef14 100644
--- a/hw/s390x/s390-pci-inst.h
+++ b/hw/s390x/s390-pci-inst.h
@@ -254,6 +254,9 @@ typedef struct ClpReqRspQueryPciGrp {
#define ZPCI_STPCIFC_ST_INVAL_DMAAS 28
#define ZPCI_STPCIFC_ST_ERROR_RECOVER 40
+/* Refresh PCI Translations status codes */
+#define ZPCI_RPCIT_ST_INSUFF_RES 16
+
/* FIB function controls */
#define ZPCI_FIB_FC_ENABLED 0x80
#define ZPCI_FIB_FC_ERROR 0x40
diff --git a/hw/s390x/s390-pci-vfio.c b/hw/s390x/s390-pci-vfio.c
index cb3f4d98adf8e7f1b104ce1e775e..0621fa386cedf3feb14f448aae91 100644
--- a/hw/s390x/s390-pci-vfio.c
+++ b/hw/s390x/s390-pci-vfio.c
@@ -12,7 +12,9 @@
#include <sys/ioctl.h>
#include "qemu/osdep.h"
+#include "hw/s390x/s390-pci-bus.h"
#include "hw/s390x/s390-pci-vfio.h"
+#include "hw/vfio/pci.h"
#include "hw/vfio/vfio-common.h"
/*
@@ -52,3 +54,43 @@ retry:
return vfio_get_info_dma_avail(info, avail);
}
+S390PCIDMACount *s390_pci_start_dma_count(S390pciState *s,
+ S390PCIBusDevice *pbdev)
+{
+ S390PCIDMACount *cnt;
+ uint32_t avail;
+ VFIOPCIDevice *vpdev = container_of(pbdev->pdev, VFIOPCIDevice, pdev);
+ int id;
+
+ assert(vpdev);
+
+ id = vpdev->vbasedev.group->container->fd;
+
+ if (!s390_pci_update_dma_avail(id, &avail)) {
+ return NULL;
+ }
+
+ QTAILQ_FOREACH(cnt, &s->zpci_dma_limit, link) {
+ if (cnt->id == id) {
+ cnt->users++;
+ return cnt;
+ }
+ }
+
+ cnt = g_new0(S390PCIDMACount, 1);
+ cnt->id = id;
+ cnt->users = 1;
+ cnt->avail = avail;
+ QTAILQ_INSERT_TAIL(&s->zpci_dma_limit, cnt, link);
+ return cnt;
+}
+
+void s390_pci_end_dma_count(S390pciState *s, S390PCIDMACount *cnt)
+{
+ assert(cnt);
+
+ cnt->users--;
+ if (cnt->users == 0) {
+ QTAILQ_REMOVE(&s->zpci_dma_limit, cnt, link);
+ }
+}
diff --git a/include/hw/s390x/s390-pci-vfio.h b/include/hw/s390x/s390-pci-vfio.h
index 1727292e9b5d019ac2218a54eac2..539bcf04eb5bcc29f0f54ef0cda2 100644
--- a/include/hw/s390x/s390-pci-vfio.h
+++ b/include/hw/s390x/s390-pci-vfio.h
@@ -12,13 +12,25 @@
#ifndef HW_S390_PCI_VFIO_H
#define HW_S390_PCI_VFIO_H
+#include "hw/s390x/s390-pci-bus.h"
+
#ifdef CONFIG_LINUX
bool s390_pci_update_dma_avail(int fd, unsigned int *avail);
+S390PCIDMACount *s390_pci_start_dma_count(S390pciState *s,
+ S390PCIBusDevice *pbdev);
+void s390_pci_end_dma_count(S390pciState *s, S390PCIDMACount *cnt);
#else
static inline bool s390_pci_update_dma_avail(int fd, unsigned int *avail)
{
return false;
}
+static inline S390PCIDMACount *s390_pci_start_dma_count(S390pciState *s,
+ S390PCIBusDevice *pbdev)
+{
+ return NULL;
+}
+static inline void s390_pci_end_dma_count(S390pciState *s,
+ S390PCIDMACount *cnt) { }
#endif
#endif
